PAGE 1
The Costs and Effectiveness of Neonatal Intensive Care August 1981 NTIS order #PB82-101411
PAGE 2
CASE STUDY #10 THE IMPLICATIONS OF COST-EFFECTIVENESS ANALYSIS OF MEDICAL TECHNOLOGY AUGUST 1981 BACKGROUND PAPER #2: CASE STUDIES OF MEDICAL TECHNOLOGIES CASE STUDY #10: THE COSTS AND EFFECTIVENESS OF NEONATAL INTENSIVE CARE Peter Budetti, M. D., J.D. Nancy Barrand, B.A. Peggy McManus, M.H.S. Lu Ann Heinen, B.A. Health Policy Program, University of California, San Francisco OTA Background Papers are documents that contain information believed to be useful to various parties. The information undergirds formal OTA assessments or is an outcome of internal exploratory planning and evaluation. The material is usually not of immediate policy interest such as is contained in an OTA Report or Technical Memorandum, nor does it present options for Congress to consider. CONGRESS OF THE UNITED STATES Office of Technology Assessment Washington D C 20510
PAGE 3
Library of Congress Catalog Card Number 80-600161 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402
PAGE 4
Foreword This case study is one of 17 studies comprising Background Paper #2 for OTAs assessment, The Implications of Cost-Effectiveness Analysis of Medical Technology. That assessment analyzes the feasibility, implications, and value of using cost-effectiveness and cost-benefit analysis (CEA/CBA) in health care decisionmaking. The major, policy-oriented report of the assessment was published in August 1980. In addition to Background Paper #2, there are four other background papers being published in conjunction with the assessment: 1) a document which addresses methodological issues and reviews the CEA/CBA literature, published in September 1980; 2) a case study of the efficacy and cost-effectiveness of psychotherapy, published in October 1980; 3) a case study of four common diagnostic X-ray procedures, to be published in summer 1981; and 4) a review of international experience in managing medical technology, published in October 1980. Another related report was published in September of 1979: A Review of Selected Federal Vaccine and Immunization Policies. The case studies in Background Paper #2: Case Studies of Medical Technologies are being published individually. They were commissioned by OTA both to provide information on the specific technologies and to gain lessons that could be applied to the broader policy aspects of the use of CEA/CBA. Several of the studies were specifically requested by the Senate Committee on Finance. Drafts of each case study were reviewed by OTA staff; by members of the advisory panel to the overall assessment, chaired by Dr. John Hogness; by members of the Health Program Advisory Committee, chaired by Dr. Frederick Robbins; and by numerous other experts in clinical medicine, health policy, Government, and economics. We are grateful for their assistance. However, responsibility for the case studies remains with the authors. JOHN H. GIBBONS Director 111
PAGE 5
Advisory Panel on The Implications of Cost-Effectiveness Analysis of Medical Technology John R. Hogness, Panel Chairman President, Association of Academic Health Centers Stuart H. Altman Dean Florence Heller School Brandeis University James L. Bennington Chairman Department of Anatomic Pathology and Clinic a/ Laboratories Children Hospital of San Francisco John D. Chase Associate Dean for Clinical Affairs University of Washington School of Medicine Joseph Fletcher Visiting Scholar Medical Ethics School of Medicine University of Virginia Clark C. Havighurst Professor of Law School of Law Duke University Sheldon Leonard Manager Regulatory Affairs General Electric Co. Barbara J. McNeil Department of Radiolog y Peter Bent Brigham Hospital Robert H. Moser Executive Vice President American College of Physicians Frederick Mosteller Chairman Department of Biostatistics Harvard University Robert M. Sigmond Advisor on Hospital Affairs Blue Cross and Blue Shield Associations Jane Sisk Willems VA Scholar Veterans Administration
PAGE 6
OTA Staff for Background Paper #2 Joyce C. Lashof, Assistant Director, OTA Health and Life Sciences Division H. David Banta, Health Program Manager Clyde J. Behney, Project Director Kerry Britten Kemp, Editor Virginia Cwalina, Research Assistant Shirley Ann Gayheart, Secretary Nancy L. Kenney, Secretary Martha Finney, Assistant Editor Other Contributing Staff Bryan R. Luce Lawrence Miike Michael A. Riddiough Leonard Saxe Chester Strobe] OTA Publishing Staff John C. Holmes, Publishing Officer John Bergling Kathie S. Boss Debra M. Datcher Joe Henson OTA contract personnel
PAGE 7
Preface This case study is one of 17 that comprise Background Paper #2 to the OTA project on the Implicatiom of Cost-Effectiveness Analysis of Medical Technology. The overall project was requested by the Senate Committee on Labor and Human Resources. In all, 19 case studies of technological applications were commissioned as part of that project. Three of the 19 were specifically requested by the Senate Committee on Finance: psychotherapy, which was issued separately as Background Paper #3; diagnostic Xray, which will be issued as Background Paper #5; and respiratory therapies, which will be included as part of this series. The other 16 case studies were selected by OTA staff. In order to select those 16 case studies, OTA, in consultation with the advisory panel to the overall project, developed a set of selection criteria. Those criteria were designed to ensure that l l l l l l l as a group the case studies would provide: examples of types of technologies by function (preventive, diagnostic, therapeutic, and rehabilitative); examples of types of technologies by physical nature (drugs, devices, and procedures); examples of technologies in different stages of development and diffusion (new, emerging, and established); examples from different areas of medicine (such as general medical practice, pediatrics, radiology, and surgery); examples addressing medical problems that are important because of their high frequency or significant impacts (such as cost ) ; examples of technologies with associated high costs either because of high volume (for low-cost technologies) or high individual costs; examples that could provide informative material relating to the broader policy and methodological issues of cost-effectiveness or cost-benefit analysis (CEA/CBA); and l examples with sufficient evaluable literature. On the basis of these criteria and recommendations by panel members and other experts, OTA staff selected the other case studies. These 16 plus the respiratory therapy case study requested by the Finance Committee make up the 17 studies in this background paper. All case studies were commissioned by OTA and performed under contract by experts in academia. They are authored studies. OTA subjected each case study to an extensive review process. Initial drafts of cases were reviewed by OTA staff and by members of the advisory panel to the project. Comments were provided to authors, along with OTAs suggestions for revisions. Subsequent drafts were sent by OTA to numerous experts for review and comment. Each case was seen by at least 20, and some by 40 or more, outside reviewers. These reviewers were from relevant Government agencies, professional societies, consumer and public interest groups, medical practice, and academic medicine. Academicians such as economists and decision analysts also reviewed the cases. In all, over 400 separate individuals or organizations reviewed one or more case studies. Although all these reviewers cannot be acknowledged individually, OTA is very grateful for their comments and advice. In addition, the authors of the case studies themselves often sent drafts to reviewers and incorporated their comments. These case studies are authored works commissioned by OTA. The authors are responsible for the conclusions of their specific case study. These cases are not statements of official OTA position. OTA does not make recommendations or endorse particular technologies. During the various stages of the review and revision process, therefore, OTA encouraged the authors to present balanced information and to recognize divergent points of view. In two cases, OTA decided that in order to more fully present divergent views on particular technologies a commentary should be added to the case study. Thus, following the case :,,
PAGE 8
studies on gastrointestinal endoscopy and on the Keyes technique for periodontal disease, commentaries from experts in the appropriate health care specialty have been included, followed by responses from the authors. The case studies were selected and designed to fulfill two functions. The first, and primary, purpose was to provide OTA with specific information that could be used in formulating general conclusions regarding the feasibility and implications of applying CEA/CBA in health care. By examining the 19 cases as a group and looking for common problems or strengths in the techniques of CEA/CBA, OTA was able to better analyze the potential contribution that these techniques might make to the management of medical technologies and health care costs and quality. The second function of the cases was to provide useful information on the specific technologies covered. However, this was not the major intent of the cases, and they should not be regarded as complete and definitive studies of the individual technologies. In many instances, the case studies do represent excellent reviews of the literature pertaining to the specific technologies and as such can stand on their own as a useful contribution to the field. In general, though, the design and the funding levels of these case studies were such that they should be read primarily in the context of the overall OTA project on CEA/CBA in health care. Some of the case studies are formal CEAs or CBAs; most are not. Some are primarily concerned with analysis of costs; others are more concerned with analysis of efficacy or effectiveness. Some, such as the study on end-stage renal disease, examine the role that formal analysis of costs and benefits can play in policy formulation. Others, such as the one on breast cancer surgery, illustrate how influences other than costs can determine the patterns of use of a technology. In other words, each looks at evaluation of the costs and the benefits of medical technologies from a slightly different perspective. The reader is encouraged to read this study in the context of the overall assessments objectives in order to gain a feeling for the potential role that CEA/CBA can or cannot play in health care and to better understand the difficulties and complexities involved in applying CEA/CBA to specific medical technologies. The 17 case studies comprising Background Paper #2 (short titles) and their authors are: Artificial Heart: Deborah P. Lubeck and John P. Bunker Automated Multichannel Chemistry Analyzers: Milton C. Weinstein and Laurie A. Pearlman Bone Marrow Transplants: Stuart O. Schweitzer and C. C. Scalzi Breast Cancer Surgery: Karen Schachter and Duncan Neuhauser Cardiac Radionuclide Imaging: William B. Stason and Eric Fortess Cervical Cancer Screening: Bryan R. Luce Cimetidine and Peptic Ulcer Disease: Harvey V. Fineberg and Laurie A. Pearlman Colon Cancer Screening: David M. Eddy CT Scanning: Judith L. Wagner Elective Hysterectomy: Carol Korenbrot, Ann B. Flood, Michael Higgins, Noralou Roos, and John P. Bunker End-Stage Renal Disease: Richard A. Rettig Gastrointestinal Endoscopy: Jonathan A. Showstack and Steven A. Schroeder Neonatal Intensive Care: Peter Budetti, Peggy McManus, Nancy Barrand, and Lu An n Heinen Nurse Practitioners: Lauren LeRoy and Sharon Solkowitz Orthopedic Joint Prosthetic Implants: Judith D. Bentkover and Philip G. Drew Periodontal Disease Interventions: Richard M. Scheffler and Sheldon Rovin Selected Respiratory Therapies: Richard M. Scheffler and Morgan Delaney These studies will be available for sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402. Call OTAs Publishing Office (224-8996) for availability and ordering information.
PAGE 9
Case Study #10 The Costs and Effectiveness of Neonatal Intensive Care Peter Budetti, M. D., J.D. Peggy McManus, M.H.S. Nancy Barrand, B.A. Lu Ann Heinen, B.A. Health Policy Program University of California, San Francisco AUTHORS ACKNOWLEDGMENTS We are grateful to many individuals for their assistance in preparing this report. Special thanks are due to our coworkers at the Health Policy Program, in particular, to Philip R. Lee, Albert R. Jensen, Paul Newacheck, Barbara Johnson, and Suzanne Stenmark. We also thank Ronald L. Williams, Roderic Phibbs, Ciaran Phibbs, and the anonymous reviewers of the first and second drafts. In addition, we appreciate the many investigators and organizations who provided us with unpublished data, papers, and other information. We take full responsibility for the interpretation of all data, both published and unpublished, other than any analysis credited to specific authors.
PAGE 10
. Contents Page Summary . . . . . . . . 4 Definitions . . . . . . . . 4 Need and Demand . . . . . . 4 Supply and Utilization. . . . . . 4 Cost and Reimbursement . . . . . 4 Effectiveness. ...,,.., . . . . . 5 Economic Analysis . . . . . . 6 Definition of Neonatal Intensive Care . . . 7 History of Neonatal Intensive Care. . . . 7 Levels of Neonatal Intensive Care. . . . 8 Definitional Problems . . . . . 9 Data Collection Problems Based on Definitional Confusions . . . . . . . 10 Summary . . . . . . . . 11 Infants Receiving Neonatal Intensive Care: Factors Influencing Present and FutureDemand, 11 Birthweight . . . . . . . 11 Prematurity . . . . . . . 11 Race. . . . . . . . . . 12 Socioeconomic Status. . . . . . 13 Birthrate. . . ...:..... . . . 13 Congenital Anomalies . . . . . 13 Maternal Age . . . . . . . 14 Prenatal Care . . . . . . . 14 Medical Practices . . . . . . 15 Utilization of Neonatal Intensive Care, . . . 15 NICU Admission Rates.. . . . . . 15 Length of Stay . . . . . . . 16 Total Days of Care. . . . . . . 17 Supply of NICU Beds . . . . . . 18 Costs of Neonatal Intensive Care . . . . 19 Findings and Limitations of Cost Data . . 19 Cost sby Birthweight, Diagnosis, and Outcome . 21 Total Costs and Charges, . . . . . 23 Reimbursement for Neonatal Intensive Care. . 23 Effectiveness of Neonatal Intensive Care . . 27 Mortality .,..,..., . . . . . 28 Morbidity. . . . . . . . . 34 Conclusion. ...,. . . . . . . 38 Economic Analysis of Neonatal Intensive Care. . 39 Application of One Method of Economic Analysis to the Results of This Case Study . . . 40 Federal Policies Related to Neonatal Intensive Care 43 References . . . . . . . . 47 LIST OF TABLES Table No. Page l. Birth Rate, Fertility Rate, and Total Number of Births, 1975-79. .~. . . . . . . 12 2. Percentage of Low and Very Low Birthweight infants Among All Births, by Race, 1950-77. ...., 12 Table No. Page 3. Low and VeryLowBirthweightRates, by Race, 1950-77 . . . . . . . . 13 4. NICU Admission Rates, 1975-78 . . . 16 5. 6. 7. 8. 9. Average Length of Stay in Childrens Hospitals, by Diagnosis, 1976-77 . . . . . 17 Average Length of Stay, by Hospital . . 17 Average Length of Stay, by Region . . . 17 Supply of NICUs and Beds in the United States, 1976-78 . . . . . . . . 18 Number of NICU Beds in the United States . 19 10. Summary of Cost Studies on NICUs . . 20 11. Cost by Birthweight GroupAll Patients . 21 12. Cost by Birthweight Group-Inborn . . 22 13. Average Cost by Diagnostic Group and Outcome . . . . . 0 . . 22 14. Alternative Estimates of Total Annual Costs for Neonatal Intensive Care, 1978 . . . . 24 15. Examples of Existing Systems for Reimbursement of Neonatal Intensive Care. . . . . 25 16. Who Pays the Bill?. . . . . . . 25 17. Inborn Neonatal Mortality Rates, Birthweight l,001 to l,500 Grams. . . . . . 31 18. Inborn Neonatal Mortality Rates, Birthweight 1,000 Grams . . . . . . . 31 19. Serious Handicaps Birthweightght >=51,500 Grams . . . . . . . . 34 20. Serious Handicaps,Birthweight >=1,000 Grams . . . . . . . . 35 21. Total Number and Distribution of Survivors, Birthweight >=l,500 Grams,1978 . . . 39 22. Information Required for a CEA or CBA of Neonatal Intensive Care. . . . . . 40 23. Kramers Cost-Effectiveness Formula . . 41 24. National Guidelines for Health Planning: Obstetrical and Neonatal Resource Standards . 44 25. Budgeted Use of Federal Maternal and Child Health Formula Grant Funds for Infant Intensive Care, by State, Fiscal Year 1978 . . . . . 46 LIST OF FIGURES Figure No. Page 1. 2. 3. 4. 5. 6. 7. Comparison of Average Daily Costand Reimbursement From Various Sourcesforlo Infants in Neonatal Intensive Care . . 26 Infant, Neonatal, and Postneonatal Mortality Rates: United States, 1915-77 . 28 Pooled Neonatal Mortality Data, 1961.76. 32 Serious Handicaps:" Birthweight <=1,500 Grams . . . . . . . 35 Serious Handicaps,Birthweight <=1,000 Grams . . . . . . . 36 Survival of Newborns, Birthweight <=1,500 Grams . . . . . . . 38 Survival of Newborns, Birthweight <=1,000 Grams. ..,..... . . . . . 38
PAGE 11
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care Peter Budetti, M. D., J.D. Peggy McManus, M.H.S. Nancy Barrand, B.A. Lu Ann Heinen, B.A. Health Policy Program University of California, San Francisco PREFACE It is frequently stated that neonatal intensive care gives rise to serious ethical problems. These problems have been discussed in an extensive literature (39,62,108, 120). In general, the ethical discussions have focused on the problem of deciding whether to initiate or to terminate life-preserving treatment in the case of a particular infant (39). These discussions do not make clear how, and to what extent, the ethicaI problems affect policy determinations about the efficiency of neonatal intensive care (56,70). Presumably, the endemic presence of such tragic decisionsto live or let die leads commentators to suggest that ethics foreclose the possibility of cost-benefit analysis in neonatal intensive care. Is it ethical to allow cost considerations to dictate life or death? Is it ethical to subordinate the care of an individual to criteria for maximizing social benefit? Certain authors wholeheartedly repudiate any thought of a positive answer (104); others are quite ready to embrace an affirmative answer as the only ethical one (46). The majority of authors in the ethical literature raise the questions but refrain from extensive analysis. At present, several problems impede further analysis of the ethical issues as they relate to public policy. The first problem is the lack of factual information or clear implications of information about neonatal intensive care. Without knowing whether infants are being helped or harmed, in what numbers, to what extent, and at what costs, debates over ethics are entirely speculative. Second, the lack of conceptual clarity about the nature of benefit and costs, about risks, about prognoses, etc., often makes ethical discussions unrealistic (63). Third, the ethical, legal, and political significance of tragic choices that expose some individuals to the risk of death in order to benefit others are only now beginning to be discussed (22). Thus, although there are certainly ethical problems within neonatal intensive care, it is unclear how they influence policy decisions. For example, it is certainly not obvious that cost-benefit studies and. policies based on conclusions of these studies are necessarily excluded by ethical considerations (15). Careful analysis of the ethical issues requires more complete factual information, clearer concepts, and a firmer grasp of the values which public policy does and should promote. Prolonged discussion of the clinical dilemmas of neonatal intensive care has led to more definitive statements of moral policy in the nursery (31). Intense and critical discussion of the ethical issues in public policy concerning neonatal intensive care might be similarly productive. Albert R. Jonsen, Ph. D. 3
PAGE 12
4 l Background Paper #2: Case Studies of Medical Technologies SUMMARY The widespread application of neonatal intensive care appears to have played a major part in producing the improved survival, as well as the improved physical condition, of very low birthweight infants in recent years. Unfortunately, the number of such very small infants is increasing every year, and some survivors continue to suffer from severe handicaps such as mental retardation and cerebral palsy. Rising birth rates and the high level of risk factors, associated with low birthweight, such as teenage pregnancy, smoking, and low socioeconomic status, are likely to create an increasing need for intensive medical care of the high-risk newborn over the next decade. These findings are based on the present study conducted by the Health Policy Program of the University of California, San Francisco, for OTA. Included in the study and summarized below are sections on the definitions of neonatal intensive care, need and demand, supply and utilization, costs and reimbursement, effectiveness, and economic analysis. Definitions Neonatal services reflect a complex mix of people and technologies. In many hospitals, the organization of these services does not reflect the three levels of care defined by the Committee on Perinatal Health. As a consequence, the services provided in different facilities classified at the same level can vary considerably, making a standard level of care difficult to identify in practice. The absence of uniform definitions of levels of neonatal intensive care has complicated data collection, making statistical analysis difficult, especially when comparing cost or utilization data for different hospitals, Need and Demand The incidence of low birthweight is the most important predictor of illness or death in early infancy and of the need for neonatal intensive care. Since 1966, there has been a 15-percent decline in the overall incidence of low birthweight (2,500 g or less, about 5 lbs) infants as a proportion of all births, associated with improvements in many of the risk factors (e.g., age of mother, socioeconomic levels, maternal nutrition, and personal health practices). Still, some 230,000 low birthweight infants are born annually. Moreover, the birth rate for the United States has increased by nearly 7 percent since 1975, mainly owing to larger numbers of women entering the childbearing ages, and this has resulted in a new increase in the absolute number of very low birthweight (l,500 g or less, about 3 1 / 4 lbs) infants born each year since 1974. The main determinants of future demand for newborn intensive care will most likely be the duration of the current baby boom and the rates of prematurity and low birthweight. Continued increases in the number of very low birthweight infants will expand the need for neonatal intensive care. Supply and Utilization No national data exist that describe the amount of neonatal intensive care currently being delivered in the United States. Only rough estimates based on studies with small sample sizes and variations in definitions of levels of care can be computed. The following estimates of neonatal intensive care supply and use in the United States were extrapolated from data available in the literature and from individual neonatal intensive care units (NICUs): l l l l l NICU admission rates: 6 percent of all live births go to intensive care (about 200,000 admissions annually; range 3.8 to 8.9 percent of all births). Estimated average length of stay (ALOS): 8 to 18 days per patient (mean 13). Estimated total patient days: 2.6 million. Number of hospitals with NICUS: approximately 600. Number of intensive care beds (Levels II and III): 7,500 (approximately 2.3 beds per 1,000 live births). Cost and Reimbursement The total costs of neonatal intensive care are
PAGE 13
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care l 5 similar to the costs of end-stage renal disease and coronary artery bypass surgery. Various sources report neonatal intensive care costs that range from $1,800 to over $40,000 per patient. We estimated average expenditures per patient in 1978 to be about $8,000. For the United States as a whole, this amounts to approximately $1.5 billion. To estimate the total annual costs for neonatal intensive care, we used the following two alternative calculations. Effectiveness Numerous recent reports claim to demonstrate the effectiveness of intensive care of the newborn. In fact, much of the literature consists of studies with sample sizes too small and populations too diverse for generalization. For the present study, we combined all available data in 5-year blocks and were able to conclude that neonatal intensive care has played a major role 1. Number of Percentage of all births Mean cost / births x admitted to NICUS x patient = Total cost (3,300,000) (0.06) ($8,000) $1.58 billion (1978 dollars) 2. Number of Level III beds Estimated reported by Ross occupancy Laboratories x rate x Days/ year x Mean cost/day = Total cost (7,387) (0.90) (365) ($545) $1.3 billion (1978 dollars) Cost data are plagued with even greater problems than utilization and supply data (e.g., NICUs are seldom separate cost centers in hospitals, and reported costs often exclude ancillary services and physician fees). Nevertheless, existing studies do show: 1) total costs for survivors are higher than for nonsurvivors; 2) as birthweight decreases, cost increases; and 3) total costs increase with complications such as hyaline membrane disease or anomalies that require surgery. The present system of reimbursing neonatal intensive care according to a uniform per diem rate encourages cross-subsidies, so that costs properly attributable to one patient may be borne by other patients. Hospital charges for neonatal intensive care are often not fully reimbursed by medicaid or by insurance plans that pay only for allowable costs, increasing the incentives for cross-subsidization among payers. Moreover, because it is difficult to adjust charges continuously with varying levels of care, expected revenues often are below costs at the beginning of a stay and exceed costs at the end, allowing for cross-subsidies based on variations in the length of stay. in improving the chances of survival of many newborns, particularly those of very low birthweight. Mortality rates within birthweight groups have declined over time, strongly supporting the conclusion that neonatal intensive care has helped improve survival. Given the scarcity of randomized clinical trials and the possible contribution of additional medical and nonmedical factors, however, it is impossible to quantify precisely how much of the improvement in survival is due to intensive medical care of the newborn. On the basis of limited morbidity data, it appears that the incidence of serious problems in survivors of neonatal intensive care is probably declining. In any case, the rate of serious handicaps has not increased as had been feared when neonatal intensive care first began to produce survivors of extremely low birthweight. It should be noted, however, that at the same time the incidence of serious problems is decreasing, the absolute number of severely handicapped individuals may be increasing. This seemingly contradictory situation may be occurring, especially with regard to infants weighing 1,000 g or less. Even though the number of normal sur-
PAGE 14
6 l Background Paper #2: Case Studies of Medical Technologies vivors has increased eightfold to twentyfold since 1960, the small number of severely handicapped individuals has risen. These results illustrate the dilemma of trying to determine whether intensive care of the newborn is effective. Every year several thousand babies who without neonatal intensive care would have died are now surviving to lead normal lives. Part of the price for this success, however, is a persistently high number of abnormal survivors. Economic Analysis Cost-benefit and cost-effectiveness analyses (CBAs and CEAs) conducted to date are of limited value. For example, many reports contain statements that the costs of hospitalization in an intensive care nursery are far less than the costs of life-long institutionalization for a severely defective survivor. Such analyses presume fully beneficial outcomes with treatment and totally unavoidable, severe handicaps without treatment. A method of economic analysis developed by economist Marcia Kramer is neither CBA nor CEA, but uses elements of both to estimate and compare the actual dollar costs and benefits of different levels of intensity of medical care, each with different outcomes. For purposes of illustration, we applied this methodology to the aforementioned effectiveness data. The tentative findings yielded were that neonatal intensive care of infants weighing 1,500 g or less is marginally cost effective, but that treatment of the subgroup of infants weighing 1,000 g or less is not yet cost effective unless only the most recent reports are used to estimate present outcomes. Data and methodological limitations common to all such analyses preclude developing an estimate of the cost effectiveness of neonatal intensive care about which one could be confident. Unresolved questions include: what percentage of, how long, and at what price abnormal infants are institutionalized, and what discount rate is most appropriate. Without such information, an accurate CEA of neonatal intensive care is impossible. In spite of these limitations, the economic analysis presented in this case study does highlight certain important aspects of the present return on the investment in neonatal intensive care. For example, care of the birthweight group 1,000 g or less does not turn out to be cost effective. The primary reason is that the small increase in the chance that a severely abnormal individual in this birthweight group would survivean increase that occurred between 1960 and 1970-75outweighed, in economic terms, the fact that the odds of a normal survivor in this group increased from 17 per 1,000 live births to 135 per 1,000 live births during that same period. Withholding care from all newborns weighing 1,000 g or less to avert the exceptional costs of the severely abnormal survivors would take the lives of many potentially normal babies. Clearly, a decision to withhold care from such infants would not be made on cost-effectiveness grounds alone. The considerations in this situation contrast with those in a hypothetical outcome, often discussed in the past: Neonatal intensive care was not cost effective because it resulted primarily in increased survival of defective individuals. In that situation, the hypothetical tradeoff was not between normal and abnormal survivors, but between fewer or greater numbers of defective survivors. Neither situation would be financially cost effective for society, but the factors to be weighed are quite different in each. The economic analysis in this case study speaks only to the question of whether neonatal intensive care is cost effective when compared with less intensive care of small or ill newborns. It does not address the larger question of whether such care is cost effective when compared with alternative programs to reduce the levels of prematurit y and other risk factors in the population. The larger question would require a separate analysis of the costs and effectiveness of socioeconomic initiatives and prenatal medical care. The question concerning the results that could be expected from trading off some intensive postnatal care in favor of prevention-oriented programs is one which has important racial implications. Our analysis of present utilization and outcomes by race concludes that marked reductions in the availability of intensive care
PAGE 15
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care 7 would have a greater adverse impact on blacks than on whites unless all correctable factors that predispose to low birthweight had previously been dealt with. This result would be predicted, because the disproportionate number of black births in the high-risk very low birthweight groups accounts for virtually all of the blackwhite differences in neonatal mortality rates. Thus, it appears that medical care of the newborn may be partially compensating for the socioeconomic, nutritional, and other inequities that play a large role in determining interracial differences in prematurity rates. For this reason, it is critical to be able to predict the result of programs aimed at reducing those inequities before considering reductions in the availability of neonatal intensive care. DEFINITION OF NEONATAL INTENSIVE CARE Neonatal intensive care is defined by the American Academy of Pediatrics as the constant and continuous care of the critically ill newborn (4). This type of care involves many individual medical technologies, highly specialized physicians and nurses, and proximity or linkage to obstetric services. Neonatal intensive care is typically delivered in organized hospital units. The facilities delivering neonatal care are classified into three groups or levels, depending on the sophistication and scope of the services the facilities are equipped and staffed to provide. The distinctions between the three levels blur considerably in the field, however, and the definitional confusions that result substantially complicate collecting and analyzing data on costs and utilization. History of Neonatal Intensive Care In the early part of this century, most sick newborns died within the first few hours of life. Premature newborns were not expected to live more than a few days. In 1878, Dr. E. S. Tarnier instituted the use of a warming chamberthe first incubatorand was able to increase the chance of survival for a large number of premature newborns (117). Similar machinery was widely used in international expositions and later became common at amusement park sideshows featuring incubator babies. Eventually, incubators gained acceptance in the medical community and became standard equipment in most hospitals with maternity services. As the survival period for premature infants lengthened, problems of nutrition and disease to which such infants are prone became evident, and research on these problems began. Facilities and medical techniques for care of the newborn have progressed most rapidly since 1965 with the evolution of perinatal 1 medicine and the development of associated medical technology. Care provided the newborn has passed through many phases and now involves the use of highly technological diagnostic and therapeutic techniques and sophisticated life-support systems. Most newborn medical care problems arise in severely premature 2 infants and for that reason, a large part of neonatal intensive care consists of using machines and other therapies to compensate for the lack of full development of the infant. The most common technologies are respirators and positive pressure breathing devices for treatment of respiratory distress syndrome 1 Perinatal: The period around the time of birth, now generally defined as from 20 weeks of gestation up to 28 days of life. Neonatal: The period from the moment of live birth up to but not including the moment at which the infant completes the 28th day of life. Premature infants comprise two groups: I ) infants who were born too soon (before the 38th week of gestation), but whose weight and development are Appropriate for their Gestational Age (preterm AGA); 2) infants who were born early and who are underdeveloped, or Small for their Gestational Age (preterm SGA), due to intrauterine growth retardation. In addition, some infants born at term (after 37 weeks gestation) are Small for their Gestational Age (term SGA) and have birthweights in the same range as truly premature infants but are relatively more mature. Most of the literature relevant to this paper focuses on birthweight alone, and does not mention gestational age. Consequently, we wi]] most often use the terms low birthweight (2,500 g or less) or very low birthweight ( I s00 g or less); when we use the term premature, it is generally in reference to very low birthweight intants who may be SGA or AGA, but are most often preterm. For ease of reference, 2,500 g is approximately 5 lb 8 O Z while 1,500 g is about 3 lb 4 oz.
PAGE 16
8 l Background Paper #2: Case Studies of Medical Technologies (RDS), also called hyaline membrane disease (HMD). 3 Once it was discovered and accepted that oxygen therapy can cause blindness in premature infants, many new technologies were developed to monitor more carefully the levels of inspired and circulating oxygen. The oxygen content of a premature infants arterial blood is often measured several hundred times during the serious stages of illness. Newer techniques that monitor oxygen saturation continuously are now being used. Other therapeutic and diagnostic innovations are also highly technological. Intravenous hyperalimentation, 4 advanced thermoregulatory apparatuses, cardiac catheterization, and microchemistry laboratory tests using only minute samples of infant blood are all associated with expensive, sophisticated machinery. Computerized ventilator systems may further refine mechanical ventilation in the future (101). Technology and research made it possible to treat severe jaundice with exchange blood transfusions, representing a major breakthrough in neonatal care. More recently, the need for this complicated and dangerous procedure has been reduced by two other innovations: phototherapy (exposure of the jaundiced infant to artificial light), and passive immunization of mothers at risk for Rh disease. Progress in the diagnosis of intracranial hemorrhage has led at least one pediatric radiologist to recommend computer-assisted tomographic scans in the first week of life for all babies weighing less than 1,500 g at birth (34). Along with changes in neonatal intensive care have come changes in the training of the physicians and nurses who deliver such care. Before 1960, responsibility for the care of the newborn was shared among general pediatricians, obstetricians and general practitioners. In the early to mid-1960s, pediatricians with special interest in caring for sick newborns began to specialize in RDS is a clinical diagnosis of pulmonary immaturity; HMD was originally a pathological diagnosis made at postmortem exam ination. The terms are now often confused and used interchangeably. In this text, the}~ should be taken to reter to the clinical entity requiring neonatal medical care, unless otherwise specified. The intravenous administra ion CJ} nutrients. early infant care, leading to the development of neonatology as a pediatric subspecialty (111). By 1979, 835 neonatologists had been certified by the American Board of Pediatrics since the first exam was held in 1975 (6). Although 200 neonatology training programs exist in the country today, none is subject to formal accreditation (6). Levels of Neonatal Intensive Care Hospitals delivering neonatal intensive care are generally separated into three levels based on the intensity of care each is equipped and staffed to provide. Level I hospitals provide minimal or normal newborn care, Level II hospitals are those considered to provide intermediate care, and Level III hospitals are those considered to provide the most intensive care of the three levels. Definitions of the three levels of neonatal care are found in the recommendations of the 1976 report of the Committee on Perinatal Health (32) and in the 1977 report of the American Academy of Pediatrics (4). 5 These recommendations were developed as guidelines for the regional development of perinatal health services. The definitions of each level are briefly summarized below. Level I hospitals are those whose function is to provide services primarily for uncomplicated maternity and newborn patients, and those with minor complications (32). Level I hospitals must also be able to provide for detection and identification of existing and potential problems. Their emergency services may include some forms of temporary intensive care techniques to manage unexpected complications before the patient can be transferred to a higher level facility. Level I hospitals also include nurseries able to provide supportive or recovery care for infants transferred back from Level II or III units after the infants acute problems have been resolved. 5 The American Academy of Pediatrics was a participant on the Committee on Perinatal Health, but had published a separate set c~f recommendations and standards prior to the Committee report, Although the recommendations are virtually identical, the presentations differ,
PAGE 17
Level II hospitals are those that can provide a full range of maternal and neonatal services for uncomplicated patients, for the majority of complicated obstetrical problems, and certain types of neonatal illnesses (32). Infants treated in a Level II intermediate care unit may include those with mild RDS, unstabilized respiratory function (periodic apnea), hyperbilirubinemia (jaundice), hypoglycemia, and superficial and localized infections. In addition to all the services of a Level I hospital, Level 11 hospitals provide more 24-hour services, more sophisticated equipment, and more medical personnel trained in specialized care of the newborn. Level III hospitals have intensive care units which must be able to provide the full range of resources and expertise required for the management of any complication of pregnancy or of the newborn (32). Level III units in hospitals with obstetric services have facilities for extremely ill infants born in the hospital or transported from surrounding regions and facilities for moderately ill and normal infants. In addition to serving as referral centers, these hospitals provide consultation services, conduct continuing education programs, and coordinate and direct transport of referred patients. Transport can also be classified as a special care service by itself. Transport of a sick newborn requires specialized, portable equipment to provide care, and trained medical and nursing staff to manage the move. The major difference between the services of a Level II unit and those of a Level III unit is the Level III units capacity for continuous and constant long-term intensive care and immediate availability of subspecialty consultants in fields such as cardiology and surgery. Level III services include continuous cardiopulmonary support and capability to treat those infants requiring long-term intravenous therapy, hyperalimentation, major surgery, and treatment of sepsis (widespread infection). Definitional Problems The terminology for defining levels of care, unfortunately, is not precise. A Level III hospital, for example, is generally expected to provide care at all levels of intensity. Such a hospital might have as many as four nursery units: normal newborn, continuing care, intermediate care, and maximal care. In that case, only the maximal care site might be called the neonatal intensive care unit (NICU). More commonly, however, levels would be combined into one or two units, so that the NICU would include all but the normal newborn area. Moreover, Level III hospitals generally have Level II and Level I beds, as well as Level 111 beds. Level II hospitals often refer to their newborn special care center as a neonatal intensive care unit, and may claim to have some Level III beds. Neonatal services in many hospitals do not reflect the three defined levels of care. The rapid advancement of medical technology, escalating costs of medical equipment, training requirements for medical personnel, and a rush to apply new knowledge with minimal planning to guide development has led to a diversity of services provided at various hospitals offering neonatal care in a region. As a consequence, the services provided in different facilities classified at the same level can vary considerably, making a standard level of care difficult to define in practice. The diversity of services, personnel, and equipment has several causes. In some areas, one or two pediatric subspecialists might be available at a Level II hospital enabling certain infants to be cared for at that facility. Similar infants would require transport from a Level II unit without the same subspecialists to a Level III unit. Also, a hospital may maintain specialized equipment (e. g., a newborn mechanical respirator) which is not required for the facilitys designated level, but which allows that hospital to provide a service normally performed at a higher level facility, In addition, a hospital may attempt to treat most serious neonatal problems even though the facility may not include one or more of the ancillary services or subspecialty consultants recommended for Level III hospitals by the Committee on Perinatal Health (32). Regulatory programs and reimbursement policies further complicate the definitional problems by creating incentives for hospitals to
PAGE 18
10 Background Paper #2: Case Studies of Medical Technologies classify their units inappropriately. For example, ratesetting programs may be more likely to approve higher rates for a Level III unit than for a Level 11 unit. Certificate-of-need programs, however, may apply very restrictive criteria to Level III facilities while not even recognizing Level II nurseries as special units (23). Additional confusion results from the lack of agreement among hospitals, professional medical groups, State planning, regulatory and funding agencies, and the Department of Health and Human Services (DHHS) 6 on a uniform definition of each level of care. This confusion is evident throughout the country. A 1979 report prepared for the Health Resources Administration of the Department of Health, Education, and Welfare (HEW) (113) showed that only 12 of 33 States with certificate-of-need programs responding to a survey had published standards for obstetrical and neonatal care; 9 States provided no definitions for Level I neonatal care; 5 States provided no definitions for Level II neonatal care; and 5 States provided no definitions for Level III neonatal care. Many of the definitions were the same as those used for obstetrical standards. Moreover, State standards for resource requirements (e.g., neonatal bed supply, occupancy rate, volume of deliveries, and travel time), service requirements (e.g., special laboratory tests and anesthesiology), and personnel, consultation, and facility requirements ranged from general policy statements to detailed specifications. Not only are the definitions sparse and inadequate to characterize most existing facilities, but there is no uniform application of the standards that do exist. A 1978 HEW study of the National Guidelines for Health Planning (82) noted that responsibilities for setting and monitoring standards related to Levels I, II, and III were unknown or not yet formalized in 30 States, and only 4 States used licensure and certificateof-need authorities to ensure adherence to specific standards. The Department of Health and Human Services (DHHS) is the new name of what was former y the Department of Health, Education, and Welfare (HEW). The new name, which reflects the transfer of most of HEWs education programs to a separate Department of Education, became 01 ficial on May 7, 1980. Both names are used in this paper. Data Collection Problems Based on Definitional Confusions The absence of uniform standards has complicated data collection and has made comparative statistical analysis difficult, especially when comparing cost or utilization data for different hospitals. For example, a certain facility maybe classified as a Level III hospital because it maintains long-term mechanical respirators and is staffed and equipped to provide treatment for the most seriously ill infants, even though it does not accept transports from the region and may not provide consultation, continuing education, or transport facilities for lower level hospitals within its region. The cost data for services rendered by this Level 111 hospital could not be directly compared to those of a Level 111 hospital meeting the Committee on Perinatal Healths definition for a Level 111 facility, because the two facilities would not have the same overhead for equipment and personnel. Also, utilization data would only reflect infants born at that particular hospital. On the other hand, a facility may be classified as a Level 11 hospital even though it can provide treatment for most serious neonatal illnesses and contains some equipment designated as necessary for a Level III hospital. If any comparable data are to be obtained for neonatal intensive care, the classification system must accurately reflect the treatment modes, the types of personnel, and the patient population served. These definitional problems are evident in the major surveys undertaken to obtain data for neonatal care. Even the most basic data, such as the number of neonatal intensive care beds in the country, cannot be corroborated by the two independent surveys done to date (6,7,8,14 ).7 Several State surveys of NICUs have also been conducted, but these are generally limited to the data collected, and they suffer from numerous definitional ambiguities. 8 One State has made an attempt to avoid confusion and to describe accurately services being These surveys of the supply of NICU beds are discussed below in the part of this case study on utilization of neonatal intensive care. The available State data are presented below in the part of this case study on util izat ion of neonatal intensive care.
PAGE 19
Case Study #10: The Costs and Effectiveness of Neonatal lntensive Care l 11 provided at existing facilities by developing a unique approach to classifying NICUs. The State Advisory Committee on Perinatal Care of the Maryland Department of Health and Mental Hygiene (78) has rejected the idea of hospital classification by level of care, and instead has drafted standards for patient care situations to be dealt with at each nursery. Thus, if a hospital has the equipment and staff deemed necessary to handle a particular situation, then it can provide the services regardless of whether it qualifies as a Level I, II, or III facility. Maryland then defines an NICU as being able to provide for any neonatal patient care situation, but does not specify that the hospital maintain all the facilities recommended by the Committee on Perinatal Health. The State approved these standards for neonatal intensive care in June 1978, but the data collected have not yet been analyzed. Unfortunately, although this may represent a future model for other States, the initial An example of a specific patient situation might be a newborn of 37 or more weeks gestation, requiring less than 40 percent oxygen for less than 24 hours. reports from Maryland will be difficult to compare with reports from other States because of the lack of consistent definitions used elsewhere. Summary Because of ambiguities in commonly used definitions, we adopt in the present study a broad concept of what constitutes neonatal intensive care, i.e., we consider neonatal intensive care to be the care provided infants in Level III or Level II nurseries. Ideally, one would analyze only Level 111 nurseries. As we have seen, however, that approach would leave out a significant amount of high-technology, truly intensive care. Furthermore, it is necessary to include both Levels II and III, because data on costs, utilization, and effectiveness are often not separated by the intensity of care provided. Although neonatal intensive care generally is, and preferably should be, practiced as part of a comprehensive perinatal care system, we consider in this study only the neonatal aspects of intensive care. INFANTS RECEIVING NEONATAL INTENSIVE CARE: FACTORS INFLUENCING PRESENT AND FUTURE DEMAND Birthweight Birthweight is the most important predictor of illness or death in early infancy (73). The neonatal mortality rate is, in general, directly related to the incidence and severity of prematurity. Mortality among low birthweight infants, those weighing 2,500 g or less, ranges from nearly 100 percent for newborns of birthweights less than 750 g (about 1 lb 10 O Z ) to approximately 10 percent for newborns of birthweights between 2,000 to 2,500 g (4 lbs 7 oz to 5 lbs 8 oz)(28). Newborns weighing 1,500 g (3 lbs 4 O Z ) or less, the very low birthweight infants, are an important subgroup, because although they represent only 1 percent of the newborn population, they account for nearly half of all infant deaths. Some 230,000 low birthweight infants are born annual] y (see table 1). Prematurity Conditions associated with prematurity are the most common reasons for the provision to infants of neonatal intensive care. The number one problem for premature infants, by a considerable margin, is RDS. Nearly 20 percent of all neonatal deaths are caused by this disorder, which is primarily due to the infants being born before the lungs are ready for breathing air. Highly technological methods for keeping newborns alive while their lungs mature are the major components of neonatal intensive care. RDS increases in incidence with the degree of prematurity. In a large study done in Norway (106), RDS was diagnosed in 5.5 percent of neonates of less than 32 weeks gestation; 2.7 percent of those of 32 to 35 weeks; and 0.3 percent of those of 36 to 38 weeks gestation. Be-
PAGE 20
12 l Background Paper #2: Case Studies of Medical Technologies Table 1. Birth Rate, Fertility Rate, and Total Number of Births, 1975.79 1975 1976 1977 1978 1979a Birth rate b . . . . 14.8 14.8 15.4 15.3 15.8 Fertility rate c . . . 66.7 65.8 67.8 66.6 68.0 Total births. . . . . 3,144,198 3,167,788 3,326,632 3,333,279 3,473,000 2,500 g or less . . . 231,627 229,375 234,884 236,342 N A d 1,500 g or less . . . 36,297 36,449 37,602 38,752 NAd a Provisional bBirths/1,000 population. cBirths/1,000 women 15 to 44 years of age dNot available as of December 1980. SOURCES 1975 dataNational Center for Health Statistics, Monthly Vital Statistics Reporft Final Natality Statistics, 1975, HEW publication No (HRA) 77-t 120, VO l. 25, No 10, suppl., Dec. 30, 1976 1976 data National Center for Health Statistics, Monthly Vital Statistics Report, Final Natality Statistics, 1976, HEW publication No (HRA) 78-1120, Vol. 26, No 12 (suppl.), Mar 29, 1978 1977 dataNational Center for Health Statistics, Monthly Vital Statistics Report, Advance Report, Final Natality Statistics, 1977, HEW publication No (PHS) 79.1120, vol. 27, No. 11, suppl. Feb. 5, 1979. 1978 data National Center for Health Statistics, Monthly Vital Statistics Report, Advance Report, Final Natality Statistics, 7978, HEW publication No (PHS) 80-1120, vol. 29. N O 1, suppl., Apr. 28, 1980 1979 dataNational Center for Health Statistlcs. Monthly Vital Statistics Report, Provisional Statistics, Annual Summary for the United States, 1979, DHHS publication No (PHS) 81.1120, vol. 28, No 13, Nov. 13, 1980 tween 1968 and 1972, RDS-related infant mortality increased somewhat. Since 1972, however, the mortality from RDS has decreased by nearly 30 percent (90). In addition to those with diagnosed respiratory disease, a number of premature will die without a specific disorder other than immaturity. This group accounts for over 10 percent of all neonatal deaths and often receives intensive care. Over the past 10 years, the mortality rate for the International Classification of Diseases category immaturity, unqualified has experienced a steady decline, from 2.69 deaths per 1,000 live births in 1968 to 1.11 deaths per 1,000 live births in 1977 (91). This change could be due in part to improved specification of diagnoses, but no study of such a trend has been identified. Race Since 1966, there has been a 15-percent decline in the overall incidence of low birthweight as a proportion of all births (see table 2). This decline is associated with improvements in some of the risk factors predisposing to prematurity. Major risk factors for prematurity include race and age of mother, socioeconomic level, and maternal nutrition and health practices. Race is a serious risk factor, because blacks are far more likely to have low birthweight infants than whites (see table 3). Although blacks account for only about 16 percent of all births, they account for more than one-third of the very low birthweight infants. Moreover, the rate of very low birthweight infants has not declined in recent years (see table 3). Because of the vast differences in very low birthweight rates, blacks Table 2.Percentage of Low and Very Low Birthweight Infants Among All Births, by Race, 1950-77 ... 1950 1956 1960 1966 1970 1971 1972 1973 1974 1975 1976 1977 Percentage of low birth weight infants ( <=2,500 g) Total. . . . 7.0%1 7.23% 7.69% 8.32% 7.89% 7.64% 7.65% 7.54% 7.39% 7.37% 7.24% 7.06% 1.26% White . . . 5.68 5.49 5.72 5.98 5.63 5.38 5.28 5.20 5.12 5.07 4.95 4.79 .19 Black . . . 1.32 1.74 1.97 2.35 2.11 2.11 2.21 2.15 2.10 2.12 2.10 2.09 .26 Percentage of very low birth weight infants (<=1,500 g) Total. . . . 0.95% 1.08% 1.17% 1.25% 1.17% 1.14% 1.18% 1.18% 1.16% 1.13%1.15% 1.150/0 1.13/o.12% White . . . 0.77 0.80 0.85 0.84 0.78 0.76 0.77 0.76 0.74 0.75 0.74 0.72 .12 Black . . . 0.19 0.27 0.32 0.40 0.36 0.36 0.39 0.37 0.36 0.38 0.39 0.39 .01 NOTE 1950$6 black birthweights reported as nonwhite SOURCE National Center for Health Statistics, Vital Statistics of the United States, selected years
PAGE 21
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care l 13 Table 3. Low and Very Low Birthweight Rates, by Race, 1950-77 1970-77 net 1950 1956 1960 1966 1970 1971 1972 1973 1974 1975 1976 1977 change Low birth weight rate a ( <=2,500 g) White. . . . 65.9 64.4 Black. . . . 100.8 123.7 Very low birthweight rateb (<= 1,500 g) White. . . . 8.9 9.4 Black. . . . 14.4 19.8 67.7 72.1 68.0 65.5 64.7 63.9 62.8 62.4 61.1 59.2 8.8 39.0 151.6 137.6 133.1 135.3 131.9 130.9 130.5 129.3 127.7 9.9 10.0 10.1 9.4 9.2 9.4 9.4 9.1 9.2 9.1 8.8 0.6 22.9 25.9 23.8 22.8 23.7 22.7 22.6 23.7 24.0 23.7 0.1 NOTE: 1950-66 black birthweights reported as nonwhite aLow birthweight rate (number of infants <= 2,500 g per 1,000 live births in specified racial group.) bVery low birthweight rate (number of Infants <= 1,500 g per 1,000 live births in specified racial groups) SOURCE: National Center for Health Statistics, Vital Statistics of the United States, selected years have a much higher neonatal mortality rate than whites. Black males have the highest infant mortality rate, with 28.3 deaths per 1,000 live births in 1975, a figure 2.3 times higher than the mortality rate for white females (90). The black-white mortality difference is almost entirely due to the higher incidence of prematurity and low birthweight among blacks. The excess of low birthweight infants among blacks is impressive. In 1977, blacks had two to three times the rate for whites in every birthweight category under 2,500 g. On the other hand, gram for gram black low birthweight newborns do better than white newborns of the same birthweight, although the reasons for this are unknown. Much of the black-white difference in prematurity and low birthweight disappears when one corrects for socioeconomic factors such as education and income. In all reported studies, however, black prematurity rates are higher within every subgroup. Racial differences in mortality rates have narrowed significantly in the past few years, but have by no means disappeared. Socioeconomic Status Socioeconomic status is also closely related to the incidence of low birthweight. Infant mortality is higher among families in which the mother and father have an education of 8 years or less, and in which the family income averages less than $3,000 annually (86,123). Birth Rate The recent decline in the incidence of low birthweight has been offset by a baby boom, principally due to increased numbers of women in the childbearing ages, together with a small increase in fertility rates (see table 1). Since 1975, the birth rate has increased 6.7 percent. This has resulted in a net increase in the absolute number of low and very low birthweight infants born each year (see table 1), Clearly, any increases in the number of very low birthweight infants will expand the number of newborns receiving neonatal intensive care. The Census Bureau projects an 8-percent increase in the number of women in the childbearing ages between 1979 and 1985 (93). Thus, increased demand for intensive newborn care can be expected. Congenital Anomalies Not all neonatal problems requiring intensive care are due to prematurity or low birthweight. Other risk factors can increase the number of problems that infants will have, whether or not they are premature. Often, infants suffering severe congenital anomalies are born at term. These anomalies are physical or metabolic abnormalities that include anencephalus (no brain), spina bifida, and other serious problems affecting the major body systems. Together, they account for nearly 20 percent of infant deaths (93).
PAGE 22
14 l Background Paper #2: Case Studies of Medical Technologies Congenital heart diseases present the largest group of congenital anomalies, accounting for over 5 percent of all neonatal deaths in 1977 (87). The development of surgical techniques for curing or ameliorating many previously fatal congenital heart defects has revolutionized treatment of these conditions over the past 10 to 15 years. The infants on whom surgery is performed are generally cared for in NICUS both preoperatively and postoperatively. The neonatal mortality rate for congenital anomalies has also shown a steady decline over the past decade, from 2.14 deaths per 1,000 live births in 1968 (85) to 1.79 deaths per 1,000 live births in 1977 (87). Other serious problems treated in neonatal intensive care units include diarrhea and malabsorption diseases, meningitis, hemorrhagic disease of the newborn, idiopathic jaundice of the newborn, and septicemia (blood poisoning). Certain risk factors associated with many of these neonatal problems have been improving in recent years, although a few have been getting worse. The identification of risk factors associated with neonatal mortality and morbidity has led to interest in programs aimed at preventing newborn disease. Some risk factors, such as maternal socioeconomic status, require complex and costly interventions with uncertain benefits, and are addressed through welfare, education, and job programs. Other risk factors can be traced to specific maternal illnesses or behavior, or to events occurring during the prenatal period. For example, smoking and alcohol and drug abuse can cause intrauterine growth retardation and a number of necrologic and metabolic disorders. These factors have direct causal links with certain newborn problems and might allow for targeted interventions with reasonably predictable benefits. Because so many of the problems of the newborn are associated with maternal factors that can be detected during pregnancy by routine prenatal care and the use of specialized diagnostic procedures, about two-thirds of infants who will require special care can be identified before birth (4). Unfortunately, such screening is costly and has high false-positive and falsenegative rates. In other words, some high-risk women will deliver normal infants, while some low-risk women will produce sick ones. Maternal Age The incidence of abnormal birth can be directly related to maternal age. Mothers over the age of 35 have an increased risk of having fetuses with genetic malformations and of experiencing fetal death in utero (84). A strong correlation exists between pregnancies in older women and the incidence of Downs syndrome (mongolism) (12). Older mothers are also more likely to produce a low birthweight baby, but they contribute an insignificant proportion of this risk group. Between 1950 and 1977, birth rates for mothers aged 35 to 39 have declined by two-thirds, from a rate of 60 per 1,000 women in that age group to less than 20 per 1,000. Birth rates for mothers aged 40 to 45 have also declined, from 15 per 1,000 to less than 3 per 1,000 (92). Teenage pregnancy also produces infants at high risk. Very young mothers commonly have hypertension of pregnancy, premature labor, or a small pelvis contributing to birth injury (4). Mothers under 15 years old have the highest risk of any group for producing a low birthweight baby (84). This group of mothers remains small, with very low birth rates, and accounts for only one-third of 1 percent of births (86). Among slightly older teenagers 15 to 17 years old, the number of births rose by over 20 percent between 1966 and 1975 (86), but has since fallen by nearly 11 percent, to pre-1970 levels. Prenatal Care Several studies have tried to assess the role of prenatal medical care in determining the outcome of pregnancy (42,50). unfortunately, it proves difficult to separate prenatal care from the other factors that influence outcomes. A recent analysis based on relatively old (1968) data for New York City suggested that if prenatal care exerts any effect on infant mortality, it is likely that it does so by reducing the incidence of low birthweight newborns (50), It has also been
PAGE 23
Case Study #10: The Costs and Effectiveneness of Neonatal lntensive Care l 15 reported that the greatest risk of having a low birthweight baby comes from pregnancies with no prenatal medical care (42). Only a small number of pregnancies in the United States are now at risk due to inadequate prenatal care, and the proportion of pregnancies with inadequate prenatal care has been declining in recent years. In States reporting to the National Center for Health Statistics, the percentage of pregnancies with late prenatal care 10 or no prenatal care fell from 8.1 to 6.0 percent between 1969 and 1975 (89). Much of the decline represented major increases in access to prenatal care by blacks, among whom inadequate prenatal care fell from 18.2 percent in 1969 to 10.5 percent in 1975 (89). Medical Practices Finally, risks have also been associated with medical practices such as electronic fetal monitoring (11), amniocentesis, cesarean section, Care initiated in the third trimester of pregnancy and induced labor (77), all of which are becoming more common. Between 1971 and 1975, the cesarean section rate in the United States nearly tripled, from 5.5 percent of births to 15.2 percent (100). In California alone, 15.4 percent of births in 1977 were delivered by cesarean section, a greater than threefold increase from the 5.1 percent rate in 1965 (131). The aforementioned medical practices could result in more premature births, but it is not yet clear whether they will increase or decrease the need for intensive care in the newborn period. If early delivery after signs of fetal distress does indeed reduce the incidence of complications such as asphyxia, less care after birth might be necessary. This could partially offset the increased demand likely to result from the current baby boom. Studies on the impact of fetal monitoring, however, do not appear to justify any hope for a dramatic reduction in the demand for newborn intensive care. The main determinant of the demand for intensive newborn care will probably be the duration of the present increase in the birth rate and the total number of births. UTILIZATION OF NEONATAL INTENSIVE CARE No national data that describe the amount of newborn intensive care now being provided in the United States are available. Rough estimates can be computed, however, by extrapolating from the data that are available. The estimates are necessarily rough because of definitional inconsistencies in the data that are available and small sample sizes on which the data are based. As noted earlier, the definition of what constitutes an NICU varies from study to study, and reports often fail to separate units by level of care. More importantly, however, the existing data are based on limited experience, often in small geographic areas with small or restricted population samples, and case-mix severity is not controlled. We calculated the following estimates of neonatal intensive care availability and use in the United States based on utilization data available in the literature and submitted by individual nurseries: l NICU admission rates: 6 percent of all live births go to intensive care (about 200,000 admissions annually, range 3.8 to 8.9 percent of all births). l Estimated ALOS: 8 to 18 days per patient (mean 13). l Estimated total patient days: 2.6 million. l Number of hospitals with NICUs: approximately 600. l Number of intensive care beds (Levels II and III): 7,500 (approximately 2.3 beds per 1,000 live births). The data we used and the manner in which we arrived at these estimates are discussed in the sections below. NICU Admission Rates Admission rates for NICUs vary according to availability of facilities, staffing capabilities, physician referral patterns, and risk factors
PAGE 24
16 l Background Paper #2: Case Studies of Medical Technologies present in different areas. Data from three large population-based studies and reports from five individual neonatal centers are summarized in table 4. The range of NICU admission rates across the three population-based studies shown in table 4 is substantial and is probably part real and part an artifact of reporting procedures. The New York City area has a number of factors associated with high-risk pregnancies: a large black population, high prematurity rates, and a high incidence of inadequate prenatal care. Because of reporting vagaries, the figure reported for New York City could be as low as 7.5 percent, still considerably above the figures for Ohio and California. Less than 4 percent of Ohios births and just under 6 percent of Californias births received intensive care in the newborn period. These figures reflect Ohio and Californias lower rates of prematurity and other risk factors (13). When data for all three regions are combined, the weighted average admission rate is about 5.9 percent (see table 4). The sources of this figure are quite disparate. However, the fact that it is based on over 20 percent of all births in the United States makes 6 percent seem a reasonable estimate for the country as a whole. Level 111 referral hospitals consistently reported higher NICU admission rates than the large population surveys, with rates ranging from less than 10 percent to more than 20 percent of all births (see table 4). These hospitals have relatively high-risk inborn populations because they serve as regional per]. natal referral centers. These figures only reflect current practice, and they are not necessaril y indicative of ideal patterns of care. It is not possible to conclude whether or not the present levels of utilization are appropriate, because the limitations of currently available data make it impossible to analyze the reasons for the wide discrepanc y in utilization. The American Academy of Pediatrics (5) has estimated that 9 to 11 percent of all live births would require special care to 4 percent at the intensive level and 7 percent at the intermediate and continuing care levels. This figure has been widely cited even though it was simpl y a consensus approximation by members of the Manpower Subcommittee of the Academys Committee on the Fetus and Newborn. Clearly, admission rates will vary with the incidence of very low birthweight infants in a population, but it is not yet reasonable to attempt to define precise, ideal utilization rates for different frequencies of prematurity y. Length of Stay Estimating ALOS is even more difficult than estimating NICU admissions. In addition to the problems associated with differential risk factors and a lack of comparable definitions, extensive transfers of infants in and out of hospitals and among beds and units of different intensity levels causes double counting of the same infants. The literature and our estimate may therefore underestimate ALOS and, correspondingly, overestimate admission rates. Table 4.NICU Admission Rates, 1975-78 Region or hospital and year a Births NICU admissions Percentage of births Large surveys Southern District New York State, 1977 (76) . . . 131,83 4 11,12 8 8.4 % California, 1977 (24) . . . . . . . . 347,426 20,551 5.9 Ohio, 1977 (103) . . . . . . . . . 160,850 6,058 3.8 Total . . . . . . . . . . 640,110 37,737 5.9 % Individual centers b UCSF Medical Center, 1976-77 (99). . . . . . 1,921 276 14.4 /0 University of Washington, 1978 (60). . . . . . 1,500 317 21.0 Bellevue Hospital, 1975-76 (43). . . . . . . 612 88 14.4 Brooklyn Hospital, 1975 (43) . . . . . . 2,485 263 10.6 Milton S. Hershey Medical Center, 1977-78 (55) . . 1,350 300 22.2 aNumbers in parentheses refer to references in the list that appears at the end of this case study bNICU admissions include in born babies only (no transfers)
PAGE 25
Case Study #10: The Costs and Effectiveness of Neonatal intensive Care l 17 On the basis of the available data collected from individual NICUS (25,54,55,60,76,94,97, 98,103,122), and the Childrens Hospitals Automated Medical Programs (29), we estimated an ALOS of 8 to 18 days (mean, 13 days) per patient in Level II and 111 nurseries combined. Because of the problem of double counting of transferred infants, however, utilization of NICUS can better be expressed in terms of total patient days, number of beds, and occupancy rates (as shown below). To summarize, ALOS varies widely from one State to another, among hospitals, by birthweight, and by diagnosis. Data from the abstracting service for childrens hospitals shown in table 5 demonstrate the variation in ALOS by diagnosis. Data for different NICUS are summarized in table 6; those for different regions are summarized in table 7. Table 5.Average Length of Stay (ALOS) in Childrens Hospitals, by Diagnosis, 1976-77 ALOS (in days) Diagnosis 1976 1977 HMD/RDS . . . . . . 20.3 18.5 Immaturity . . . . . . 29.8 29.6 Asphyxia. . . . . . . 14.2 15.9 SOURCE: Data in this table were supplied by the Childrens Hospitals Auto. mated Medical Programs (CHAMP), Columbus, Ohio (29) Any analYSIS interpretation, or conclusion based on these data or any other data cited elsewhere in this study as having been supplied by CHAMP is solely that of the Health Policy Program CHAMP specifically disclaims responsibility for any such analysis, Interpretation, or conclusion. Total Days of Care The total number of days of intensive care in NICUs for the United States can be approximated by multiplying estimated NICU admissions by estimated ALOS. Applying an estimated NICU admission rate of 6 percent to the number of live births registered in the United Table 6.Average Length of Stay (ALOS), by Hospital Hospital and year a Admissions Patient days ALOS b (in days) University of Washington Hospital, Seattle, 1978 (60) . . . 461 9,128 19.8 c Childrens Orthopedic Hospital, Seattle, 1978 (60). . . . . 407 7,326 18.0 Hospital for Sick Children, Toronto, Canada, 1978 (122) d . . . 17.oe UCSF Medical Center, 1978 (98) . . . . . . . . 1,185 14.8 Milton S. Hershey Medical Center, Pennsylvania State University, 1977 (55) . . . . . . . . . . 527 8,485 16.1 Milton S. Hershey Medical Center, 1978 (55). . . . . . 475 9,358 19.7 aNumbers in parentheses refer 10 references in the list that appears at the end of this case study bTotal stay for Levels II and Ill combined cSurvivors, 22 days (83.7%), nonsurvivors, 85 days (16.3%) dSwyer reports ALOS was similar at McMaster University Medical Center and Emory University in Canada for the same Year eALOS for infants admitted directly to Level II is 7 days Table 7.Average Length of Stay (ALOS), by Region Region and year a Admissions Discharges Patient days ALOS (in days) Southern District of New York State, 1977 (76). . . . 9,867 170,233 17.3 b 15.3 C Ohio, 1978 (103) . . . . . . . . . . 6,058 129,013 21.3 California, 1977 (26). . . . . . . . . . 20,551 178,744 8.7 California, 1972-73 (54) . . . . . . . . 6,863 10.8 d 11 .3 e Massachusetts, 1976 (94) . . . . . . . . 12.9 d Pennsylvania, 1977-78 (97) . . . . . . . 11.8 aNumbers in parentheses refer to references in the Iist that appears at the end of this case study bBased on reported discharges c Based on the larger of admissions of discharges where discrepancies exist n hospital reports to the United Hospital Fund of Greater New York Level Ill Level II
PAGE 26
18 l Background Paper #2: Case Studies of Medical Technologies States for 1978, 3,329,000 (79) yields as the total number of newborns requiring intensive care in 1978 approximately 200,000. If one uses an approximate ALOS of 13 days, total patient days (admissions x ALOS) would be 200,000 X 13 = 2,600,000. This total, 2,600,000 days, would represent approximately 0.7 percent of the U.S. total of hospital days for 1978 (380,152,083 total hospital days) (32). Supply of NICU Beds A limited number of surveys of NICU facilities are currently available (see table 8). Th e results are contradictor y and are difficult to reconcile. Table 8.Supply of NICUs and Beds in the United States, 1976-78 AHA MFIS R OS S 1976: 6,500 hospitals reporting Number of hospitals with premature nursery facilities . . . 1,923 2,017 Number of hospitals with NICUS. 529 540 Number of hospital beds used for neonatal intensive care,. . . 6,602 6,668 1977: 6,495 hospitals reporting Number of hospitals with premature nursery facilities . . . 1,821 2,014 Number of hospitals with NICUS. 591 655 Number of hospital beds used for neonatal intensive care. . . 7,553 7,792 1978: 6,321 hospitals reporting Number of hospitals with premature nursery facilities . . . 1,726 NA Number of hospitals with NICUS. 448 NA 259 Number of hospital beds used for neonatal intensive care. . . 6,252 NA 7,387 NA = not available SOURCES 1976 data: AHA (7), MFIS (79) 1977 data AHA (8), MFIS (79). 1978 data: AHA (9), Ross (14). The American Hospital Associations (AHA) Annual Survey of Hospitals counted NICUs and beds for the first time in 1976, and reported the results in 1977 (7). The 1977 (8) and 1978 (9) surveys have now been published, as well. The apparent increase and decline in NICU supply over those 3 years cannot be explained by AHA staff with confidence, although the pattern may simply reflect the vagaries of the first years of reporting a new survey item. The AHA figures reflect all hospitals in the United States except the 200-plus institutions that are not registered with AHA. In its Master Facility Inventory Survey (MFIS), the National Center for Health Statistics uses AHA survey results and adds an estimate for the hospitals that are not registered (79). AHA has not yet separated NICUs into different levels, although it had intended to do so in the most recent survey, Thus, AHA and MFIS figures should include all NICUs and all NICU beds at every level of intensity. Ross Laboratories, which surveyed only Level 111 regional referral centers, counted virtually the same number of NICU beds as AHA in only half the number of hospitals (14). Although counting beds is complicated by differences among beds in use, licensed beds, and specially designated NICU beds, Ross tried to report the standard number of infant positions normally accommodated in the neonatal units surveyed. A 1980 update of this survey should be available soon. Since the discrepancies among the reports show no consistent pattern (see table 8), the relationship between the Ross data and AHA and MFIS reports cannot be determined. Moreover, none of the published national surveys has publicly identified individual hospitals, making it impossible to cross-check the results directly. Our estimate of 7,500 beds was reached by combining the results of the bed surveys (see table 9) with the estimated number of days of care being provided. That is, our estimate of 2.6 million patient days would require 7,500 beds at a 95-percent occupancy rate ([2,600,000/(365 x 7,500)] = 0.95). Our estimate of 7,500 beds is, in fact, a conservative figure. If the Ross data accurately portray the number of beds in Level 111 hospitals alone, the national total for both Levels II and III could be closer to 14,000 beds.
PAGE 27
Case Study #10 The Costs and Effectiveness of Neonatal Intensive Care l 19 Table 9.Number of NICU Beds in the United States Number of Number of beds Survey and year a hospitals with NICU Maximum care Intermediate care Total Beds/hospital California RoSS, 1978 (14) . . . 28 330 394 724 25.9 MCH, 1976 (53). . . . 23 (of above) 192 221 413 18.0 OSHP, 1977 (25). ., . . 23 (of above) 467 20.3 MCH, 1976 (53). . . 43 236 308 544 12.7 OSHP, 1977 (25). . . . 54 697 12.9 Ohio R OSS 1978 (14) . . 9 142 202 344 38.2 SDH, 1978(103)). . . . 17 472 27.8 Massachusetts Ross, 1978 (14) . . . 6 65 47 112 18.7 SHP, 1979(94) . . . 8 95 69 164 20.5 New York Ross, 1978 (14) . . 21 (15-NYC area) 292 468 760 36.2 UHF/NY, 1978 (75) . . 27 (NYC area) 505 18.7 UHF/NY, 1978 (75) . . 14 (of 15 above) 392 28.0 Washington R OSS 1978 (14) . . . 4 52 38 90 22.5 UW, 1978 (60) . . . 6 71 39 110 18.3 aNumbers in parentheses refer to references in the Iist that appears at the end of thIs case study COSTS OF NEONATAL INTENSIVE The costs of neonatal intensive care are great. Among patients with high cost hospitalization ($4,000 or more in 1 year) studied by Schroeder, et al. (112), neonatal cases were by far the most expensive, averaging over $20,000 each. In fact, neonatal costs were higher than those for neoplastic and circulatory diseases, two of the most expensive adult services. They were similar to the total cost of end-stage renal disease and coronary bypass surgery, both of which require specific costly medical technology. Various reports of neonatal intensive care costs range from $1,800 to $40,000 per patient. We estimate average expenditures in 1978 to be about $8,000 per case. In the United States as a whole, this amounts to approximately $1.5 billion (19,98). Findings and Limitations of Cost Data Cost data are plagued with even greater problems than are utilization and supply data. NICUS are often not separate cost centers in hospital reports, but are mingled with other intensive care or pediatric services. Even when the units are identifiable cost centers, the costs CARE that generally are reported exclude so-called ancillary services such as laboratory tests, Xrays, and physician fees. Since diagnostic and therapeutic services are a major part of the total costs of caring for neonatal patients, such figures greatly underestimate per capita costs. Total costs per patient are not readily accessible except from the limited number of special studies that have been undertaken (see table 10). In order to compare nurseries studied at different times, we have updated many of the figures to 1978 dollars, as specified. Many of the data report hospital charges that may not in any way reflect actual costs. Hospital charges allocate overhead and other costs not directly attributable to an individual patient according to reimbursement practices and overall institutional revenue targets. 11 Moreover most States do not require uniform accounting, and comparability among hospitals is limited as a result. Additional problems arise because costs for See section below on reimbursement for neonatal intensive care.
PAGE 28
20 l Background Paper #2 Case Studies of Medical Technologies Table 10.Summary of Cost Studies on NICUs Average cost Average cost Total cost Study a /year/population Characteristics of data per day updated to 1978 per patient Cotton, Vanderbilt University, 1, 2 $700 NS $832 NS $5,691 S Term. (33) 4No information if the charges $2,966 NS o 1976, 3 months cover both S & NS. Also all but l S & NS 8.3% of total was spent on S. 4 Nothing on cost/charges per patient care or ALOS for S. Kaufman & Shepard, Tufts New 1, 2 $345 Ill $345 Ill England Medical Center, Mass. (66) 1978 l Level I I I services Kaufman & Shepard, 2 $188 I $188 I Johns Hopkins, Md. (66) $200 II $200 II l 1978 $340 Ill $340 Ill Kaufman & Shepard, Women & 1 $74 I $74 I Infants Hospital, R.I. (66) $130 II $130 II 1978 $175 Ill $175 Ill Hawes, 19 tertiary NICUs & 1 1, 3a, 3b community (Ii) hospital, 4Total costs/charges will be $123 II Cal if. (54) underestimated since over 50% $198 Ill l 1973 of infants were transferred and l 54io of 6,863 admissions were some 12 of the 20 hospitals are transfers returning babies to community hospitals for convalescent care. 4Reported from 11 hospitals. N O information on what it covers. $24511, a b $1,390 II $394 Ill, a b (ALOS = 11.3 days) $2,138 Ill (A LOS = 10.8 days) $5,1 78= average total cost, b Kaufman & Shepard, Model Budgets (64) l 1977 McCarthy, Childrens Hospital, Denver, Colo. (80) l 1976, 4-month period l 84% S & 16% N S l All transport infants 3a 4Based on Swyers (121) estimate of total patient days (45/1,000 live births need Level II; 45/1,000 need Level Ill; 6 day ALOS). 4 Includes obstetric and neonatal services. 1, 3b 4Charges include transport fees. $125 II (80% occ.) $213 Ill (83%occ.) $15511, a $266 Ill, a $338 S (mean) $442 S, b $607 NS (mean) $793 NS, b Kaufman & Shepard, Boston 4Costs allocated based on ac$579 S & NS $628 $14.654-$40.752 S Hospital for Women, Mass. (65) counting costs and services (62-110 days S) l 1977 rendered. Costs cover MD fees, $0-$7,594 NS l 10 infants: 750-999 g ancillary services and overhead, l 40% S & 60% NS Meier, University of Washington 1 $308 S & NS $308 $14,190 (46 days) NICU (83) l 1978 l Infants with HMD l 12 of 31 infants were transfers Childrens Hospital Medical 3a, 3b, 4 No information on $243 $361 a & b Center of Northern California number of infants or ALOS. [$430 charge] (30) 1976
PAGE 29
Case Study #10 The Costs and Effectivess of Neonatal Intensive Care l 21 Table 10.Summary of Cost Studies on NICUsContinued Study a /year/populat ion Characteristics of data Pomerance, NICU at Cedars-Sinai 1, 3b Medical Center (102) 4 Used 940/0 of collected Adjusted to 1976 hospital charges as actual cost Infants 1,000 g of care. l 40% S & 60% NS Shannon, Massachusetts General 1,2 NICU (1 14) l 1974 Infants without RDS l 70/0 S & 300/ 0 NS c 330/ 0 transfers Phibbs, UCSF, NICU (98) 1 l 1978 (over 30-month period) by S & NS, birth weight, diagnosis & treatment, inborn v. outborn s sample = 1,185 infants Childrens Hospital Automated 1 Medical Programs (29) 1975,770 Infants 1976,645 infants 1977 7,714 infants Average cost Average cost Total cost per day updated to 1978 per patient $825 NS, b $1,078 NS, b $14,236 (17 days) $450 S, b $588 S, b $40,287 (89 days) $299 S $436 $720 NS $1,050 $545 1975$229 $298 1976$291 $346 1977$340 $369 aNumbers in parentheses refer to references in the list that appears at the end of this case study $8,842 S (29.6 days) $2,448 NS (3.4 days) . $8,069 (ALOS = 16 days) S = $7,620 NS = $11,624 <= 1,500g = $16,684 1975$3,1 77 (ALOS = 14.8 days) 1976$4,203 (ALOS = 15 days) 1977$5,283 (ALOS = 15 days) KEY 1 = study reports charges rather than costs a = Ancillaries = 15% S = survivors 2 = no Information on what cost/charges cover b = M D fees = 100,. NS = nonsurvivors 3a = costs/charges do not include ancillaries I = Level I bed ALOS = average length of stay 3b = costs/charges do not include physician fees II = Level II bed HMD = hyaline membrane disease 4 = other characteristics Ill = Level Ill bed RDS = respiratory distress syndrome different levels of care and diagnoses are typically aggregated as a single composite figure. Together with the frequent need to transfer patients among hospitals with different levels of care, this nearly precludes collecting total costs per patient. Patterns of practice and case mix are not reflected in existing systems for collecting or reporting financial data. Costs by Birthweight, Diagnosis, and Outcome Their analysis emphasizes the great variation in costs of care among patients. All admissions were classified by birthweight, diagnosis and treatment, outcome (survival v. nonsurvival), and inborn v. outborn. Their results show that average costs increased as assisted ventilation was employed, as surgical procedures were used, and as birthweight fell (see tables 11, 12, and 13). Table 11 .Cost by Birthweight GroupAll Patients Phibbs, et al. (98) analyzed total charges billed to all NICU patients at the University of California, San Francisco (UCSF), over a 30month period through the end of 1978. They found the average charge for the whole sample (N = 1,185) to be $8,069, the average charge to produce a survivor to be $9,089, the average daily charge to be $545, and the ALOS to be 16 Cost to produce Birth weight group Average cost a survivor a <= 1,000 g . . $19,213 $31,621 1,001 to 1,500 g 15,204 18,659 1,501 to 2,000 g 9,516 10,140 2,001 to 2,500 g 5,908 6,499 2,501 to 3,000 g 5,445 5,874 >3,001 g 5,649 6,157 aCost to produce a survivor = total cost/number of survivors SOURCES C S Phibbs, et al Analysis of Factors Associated With Costs of Neonatal Intensive Care 1980 (98), and R H Phibbs, UCSF unpublished data. 1979 (99) days.
PAGE 30
22 Background Paper #2: Case Studies of Medical Technologies Table 12.Cost by Birthweight Grouplnborn a Cost to produce Birthweight group Average cost a survivor b <= 1,000 g . . $22,508 $46,340 1,001 to 1,500 g . 15,457 17,630 1,501 to 2,000 g . 7,645 8,038 2,001 to 2,500 g . 4,307 4,397 2,501 to 3,000 g . 3,378 3,626 >=3,001 g . . 2,123 2,138 aUniversity of California, San Francisco, 1976-78. bCost to produce a survivor = total cost/number of survivors SOURCES: C S. Phibbs, et al Anlalysis of Factors Associated With Costs of Neonatal Intensive Cale, 1980 (98); and R H Phibbs, UCSF, unpublished data, 1979 (99) The Phibbs, et al., stud y (98) also demonstrated that neonatal intensive care is often used for relatively brief periods to observe and stabilize high-risk patients. Nearly one-half of the infants in their study were in the category normal birthweight, never requiring assisted ventilation or surgery. That group of infants was relatively low cost, averaging $2,190 per case and collectively accounting for only 13 percent of the total charges. Another group, the 24 percent of infants with the greatest medical and surgical complications, represented nearly 60 percent of the total charges. Additional cost studies have reported similar relationships: 1) total costs for survivors are higher than for nonsurvivors; 2) as birthweights decrease, costs increase; and 3) total costs increase with complications such as HMD (see table 10). Meier (83) analyzed the in-hospital costs of treating HMD in the NICU at the University of Washington Hospital between 1977 and 1978. For survivors, the average cost per patient day was $435, compared to $1,050 a da y for nonsurvivors. The cost per day among survivors did not differ greatly within the disease categories, but total costs for treating surviving infants were strongly correlated with the length of hospitalization. Thus, the mean cost per surviving infant ranged from $8,560 for those in the mild disease category to $35,210 for those with severe disease. Shannon, et al. (114) also studied the costs of treating RDS at the Massachusetts General Hospital in 1974. Shannon found that for survivors, with an average of 29.6 days of intensive care, the total cost was $8,842 in 1974 dollars. For nonsurvivors, averaging 3.4 days of intensive care, the total cost was $2,448. Expressed as 1978 dollars, these figures would be $12,890 and $3,569, respectively. Support services, personnel, and other resources are heavily consumed in the earliest stages of neonatal intensive care; the later days for observation and recovery require less intensive care. For example, one study noted that the first few days of neonatal intensive care require a nurse-to-infant care ratio of 1:1 or 1:2, plus many ancillary procedures (80). During recovery days, the nurse-to-infant care ratio was about 1:4. Fewer daily services and tests were performed. The high mean charge per day for nonsurvivors reflects the early intensive care days without the averaging effect of the survivors recovery days (80). This finding that cost per day for nonsurvivors is almost twice as high as for survivors has been demonstrated in Table 13.Average Cost by Diagnostic Group and Outcome Total sample Survivors Non survivors Diagnostic group All patients Inborn Outborn Inborn Outborn Inborn Outborn Primary medical with major surgical complications. . $23,952 $38,844 $19,874 $36,540 $14,096 $12,639 $37,522 Primary medical . . . . . . . . 5,822 5,038 7,936 5,017 7,994 7,114 4,985 Noncardiac anomaly, treated medically. . . . 4,804 3,051 5,839 4,053 5,912 2,186 7,541 Cardiac, treated medically. . . . . . . 3,151 5,259 2,934 6,946 3,109 1,297 1,761 Primary surgery not cardiac or anomaly . . . 10,207 16,786 9,195 16,786 9,195 Noncardiac anomaly, treated surgically . . . 21,077 26,020 20,273 29,871 20,117 2,912 18,594 Cardiac treated surgically . . . . . . 17,227 15,848 17,320 15,501 16,897 10,547 Average cost. . . . . . . . . $ 8,069 $ 5,952 $10,872 $ 5,824 $10,332 $ 7,870 $13,357 aExcept patent ductus arteriosus SOURCES : C. S Phibbs, et al Analysis of Factors Associated With Costs of Neonatal Intensive Care, 1980 (98), and R H Phibbs, UCSF, unpublished data, 1979 (99)
PAGE 31
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care 23 additional studies (65,102). Other studies comparing neonatal intensive care costs to birthweight demonstrate a clear relationship: As birthweight decreases, cost increases. Total Costs and Charges A model NICU budget in the 1976 report of the Committee on Perinatal Health (32) used 1974 financial data from hospitals in the midAtlantic region. In the absence of alternatives, this budget has been the main planning tool for hospitals and policy makers alike. The budget has been criticized, though, for its low overhead and unrealistic physician salaries (64). An updated model NICU budget prepared by Kaufman and Shepard (64) showed that the average costs per patient day would be $213.40 at 83-percent occupancy for a Level III NICU. A Level II unit at 80-percent occupancy would have average daily per patient day costs of $124.78. These figures do not include ancillary costs and are at least $50 less per patient day than reports in the literature from actual nurseries. Table 10 displays existing studies on costs for NICUs by level of care and outcome, birthweight, and diagnosis. The following discussion summarizes the findings from these cost studies. Data collected in 1976 by the Massachusetts Rate Setting Commission (64) show total costs ranging between $80 and $488 per patient day for eight Level III NICUs in Massachusetts. The wide range was due to a number of factors, including case mix, occupancy levels, accounting methods, size, specialization, and the mix of intensive and less intensive beds. For example, the mean cost of NICUs with intensive beds only was $287 per patient day, while NICUs with multiple levels of care averaged $133 per patient day, a figure much closer to the updated model budget estimate. Hawes (54) collected charge data for patients in California NICUs in 1973. In that study, estimated daily charges averaged $198 for Level III units and $123 for Level 11 units, or an average of $160 per day for a newborn staying an equal number of days at each level. Updated to 1978 dollars, this would be equal to $255 per day on average, or $315 for Level III beds and $196 for Level II beds (54). (Adjusting for physician fees and ancillary services adds approximately 25 percent to these figures. ) As is evident from these existing studies, costs and charges vary by accounting methods, case mix, and level of care provided. Given the wide range of costs and charges evident in the data, it is difficult to estimate total costs with confidence. Each of the studies described above has its limitations, as summarized in table 10. The most comprehensive and recent data are those from the study by Phibbs, et al. (98). Based on their work, we expect that total costs will average about $8,000 (in 1978 dollars) for the wide range of patients in a Level 111 full-service nursery that has both inborn and transported-in infants. As those authors point out, newborns with different diagnoses and birthweights will have widely varying costs, and the particular case mix must be taken into account in estimating total costs for a specific nursery. Using the figures reported by Phibbs, et al. (98) for purposes of illustration, we can make a crude estimate that about $1.5 billion was spent on neonatal intensive care in 1978 (see table 14). Their estimate of mean charge per case was chosen because it represents a large, diverse, and virtually complete sample of patients. The authors took great care to track down all charges, including physician fees. The UCSF estimate (98) may or may not be representative of other areas of the country. However, since neonatal intensive care is available primarily in urban centers, the figures for San Francisco should not be unreasonably high. For example, the UCSF estimate of mean charges for outborn patients transported in was $10,872, remarkably close to the $10,513 reported for patients transported to The Childrens Hospital in Denver, Colo. (80). Reimbursement for Neonatal Intensive Care The diversity and complexity of hospital billing and third-party reimbursement practices further complicate the attempt to measure the costs of neonatal intensive care. One particular problem is the degree to which the present sys-
PAGE 32
. 24 l Background Paper #2: Case Studies of Medical Technologies Table 14.Alternative Estimates of Total Annual Costs for Neonatal Intensive Care, 1978 Percentage of all births Mean cost/ Total cost 1. Number of births x admitted to NICUS x Patient = $1.58 billion (3,300,000) (0.06) 2. Number of Level Ill beds reported by Estimated Ross Laboratories x occupancy rate x (7,387) (0.90) SOURCE Text and previous tables tern encourages cross-subsidies, so that costs properly attributable to one patient may be borne by other patients. Hospital charges for neonatal intensive care are often not fully reimbursed by medicaid or by insurance plans that pay only for allowable costs. This situation creates incentives for hospitals to adjust charges to cover their expenditures by cross-subsidizin g among payers. Moreover f because it is difficult to adjust charges continuously with varying levels of care, expected revenues often are below costs at the beginning of a stay and exceed costs at the end, allowing for cross-subsidies based on variations in the length of stay. Problems created by cross-subsidies include encouraging NICU admission of infants with less serious problems and extending stays after treatment when it would be adequate to continue care in a lower level and less costly setting. Common reimbursement methods used by individual hospitals in five Eastern States, all of which have State ratesetting programs, illustrate the followin g variations: 1) units may not be the same as cost centers recorded in the hospitals accounting scheme; 2) levels of care are often not differentiated according to cost; 3) intensive care for adults and newborns are sometimes grouped together for cost purposes; and 4) services included in the per diem rate vary (66). Table 15 summarizes the examples of neonatal intensive care reimbursement systems. These reimbursement methods often result in a less expensive level of neonatal cares crosssubsidizing other more expensive care. The major villain in cross-subsidization, as far as hospitals are concerned, is medicaid, becau s e medicaid pays a fixed amount per day for any type of hospital care in some States, and reimburses costs rather than charges in other States, even when the State formula pays separately for ($8,000) (1978 dollars) Total cost Days/year x Mean cost/day = $1.3 billion (365) ($545) (1978 dollars) intensive care. It is difficult to identify which is the chicken and which is the egg in the cycle. Hospitals adjust charges to cover total expenditures, medicaid and other cost-payers then pay at the level of allowable costs, and hospitals raise charges even more to cover the difference. Neonatal intensive care, because of its high-cost nature, presents a telling picture of hospital practices in maximizing reimbursement. In the limited number of studies in which costs and charges could be compared, costs represented about 68 percent of the charges (27,55). A limited number of studies have been conducted on the sources of payment. In a 1978 study, McCarthy, et al. (80), examined hospital charges (excluding physician fees) by source of payment for 174 admissions to Denver s Childrens Hospital during 4 months in 1976. As described earlier, the amount paid was dependent on the type of insurance coverage. Of the accounts closed within 2 years of discharge from the NICU, third-party payers paid 85 percent of the bills, direct or individual payers paid 4 percent, and 11 percent was uncollectible. Infants covered by medicaid (15 percent) were responsible for 51 percent of the uncollectible or writeoffs (see table 16). Kaufman and Shepard (65) compared their own estimates of the cost per day for 10 low birthweight infants to hospital charges and to expected revenues (see fig. 1). Blue Cross paid about 80 percent of charges. Medicaid, accounting for 20 percent of the hospital billings, paid a flat per diem rate that amounted to about onehalf of the hospitals charges. Uncollectible (including discounts, bad debts, and regulated fee schedules) were primarily among chargepayers and amounted to 14 percent of the total charges, a figure very close to that reported by McCarthy, et al. Blue Cross reimbursement and
PAGE 33
Case Study #10: The Costs and Effectiveness of Neonatal intensive Care l 25 Table 15.Examples of Existing Systems of Reimbursement of Neonatal Intensive Care Hospitals Massachusetts Boston Hospital for Women; Tufts New England Medical Center; Memorial Hospital, Worcester New York All hospitals Connecticut Yale-New Haven Hospita l Maryland Johns Hopkins Hospital, others in passing Rhode Island Women and Infants Hospital of Rhode Island Cost centers NIC as distinct from other units No distinction of units NIC as distinct from other units, Yet, this is not required in State reporting forms NIC can be a separate cost center, but not required. Hopkins considers NICU separately All newborn care IS a single cost center Commercial Hosapital or private Self-pay charges Insurance Blue Cross Medicaid patients Hospitals have flexibility to set charges Same daily charge for all types of inpatient care, computed yearly for each hospital Pay directly for Contracts rate Uniform per Pay directly for care with individual diem rate for care hospitals dean services pending on ratio provided to all of costs to patients charges. Can set specific unit of care All inclusive per All inclusive per All inclusive per Pay directly for diem. Exclusive diem. Exclusion diem. Exclusion care of care of well for care of well for care of well newborns reimnewbornsreimnewborns reimbursed at 1/3 th e bursed at 1/3 the bursed at 1/3 the hospital per diem hospital per diem hospital per diem Two charges No information generally: sick newborns and well newborns. Yale-New Haven sets one level for all infant special care Cost Review No information Commission sets average rate for each of three levels of neonatal Intensive care. Hospitals can adjust rates to cover costs Each hospital No information negotiates with Blue Cross and medicaid and establishes charges to be uniformly applied to all patients 95%. of charges Retrospective Pay directly for and based on care costs Set by Cost Set by Cost Pay directly for Review CommisReview Commiscare sion sion Contract with inContract with inPay direct for dividual hospidividual hoscare tals to reimburse pitals to reimservices based burse services on prospective based on proCosts spective costs. Medicare reimburses hospital based on actual costs SOURCE S L Kaufman and D S Shepard, Reimbursement of Neonatal Intensive Care A Descriptive Overview, in A Review of Planning Methods for Neonatal Care Units. vol. II, HRA contract No 231-770108 study prepared by the Boston University Center for Health Planning for the Health Resources Administration, Hyattsville. Md 1979 Table 16.Who Pays the Bill? (accounts paid in full) Number of Total hospital Amount paid by Amount paid by Amount written Third-part y payer accounts charges third-party family off by hospital Blue Cross-. . . . . 54 (360/. ) $491,119 $457,757 $24,994 $8,368 Commercial insurance . 53 (36/0) 490,982 439,355 27,703 23,924 Medicaid. . . . . 23 (150/o) 256,750 159,726 1,000 96,024 Kaiser . . . 6 (4/0) 148,748 148,748 0 0 Other third-party payer a . 11 (70/o) 109,692 80,135 1,002 28,555 No coverage. . . . 3 (2/0) 10,210 0 3,502 6,708 a lncludes Handicapped Children Program SOURCE :J. T. McCarthy, et al., Who Pays the Bill for Neonatal Intensive Care? J. Pediatrics 95 :757, 1979
PAGE 34
26 l Background Paper #2: Case Studies of Medical Technologies hospital charges both exceeded the average cost per patient day overall, while medicaid reimbursement was considerably less than costs. Using a weighted average of all three reimbursement methods based on the hospitals patient mix, the authors calculated on a preliminary basis the expected revenue from all sources and found it to be $2.00 per day below costs. The same authors also demonstrated how length of stay produces cross-subsidies. Blue Cross and charge-payer reimbursements were highest during the first week, and decreased significantly through the fourth week. Medicaid per diem, on the other hand, steadily reimbursed below costs through the eighth week of hospitalization. Short or long stays determine whether total reimbursement calculated on a fixed per diem basis will be above or below costs, dependin g on the payer. Kaufman and Shepard (66) recommended two options for improving the present reimbursement system to create incentives for cost Figure 1. Comparison of Average Daily Cost and Reimbursement From Various Sources for 10 Infants (birthweight 750 to 999 g) in Neonatal Intensive Care $428 $312 cost Blue Cross -------$217 Medicaid Total charges Collected charges (adjusted for bad debt) $310 Expected revenue from all sourcesa awelghted average of the three reimbursement sources based on hospitals patient mix Blue Cross, 41 7 percent, medlcald, 208 Percent, charge PaYerS, 375 Percent SOURCE S L Kaufman and D S Shepard, Cost by Day of Neonatal Intenswe Care, speech dellvered at the 108th Annual Meettng of the American Public Health Assoclatlon, Detroit, Mlch Oct 19.23, 1980
PAGE 35
Case Study #10, The Costs and Effectiveness of Neonatal Intensive Care l 2 7 savings and for quality. The first would be to reimburse at a flat rate for each case and eliminate per diem payments completely. A New Jersey experiment with such a payment system is about to begin, but this will have only limited application to neonatal intensive care because very broad diagnostic groupings are used (e.g., with and without distress). A second reimbursement option is a per diem rate that declines over the course of an infants hospital stay. This proposed option might create incentives for return transports to community hospitals for growth and recovery (66). Whatever alternatives are selected, it is evident from experience that the present approach, a uniform per diem rate, provides incentives for inequitable cross-subsidization in the reimbursement of neonatal intensive care. EFFECTIVENESS OF NEONATAL INTENSIVE CARE Neonatal intensive care brings together complex medical technologies and highly specialized personnel to improve the survival of premature and ill newborns. The first indicator of the effectiveness of this care must be whether such infants do in fact have a better chance of survival with intensive care than without it. Randomized clinical trials are very limited, but more extensive mortality reports are available. Taken together, the data strongly suggest that neonatal intensive care is effective in improving survival. All reports from this country show remarkable declines in mortality during the years in which neonatal intensive care proliferated. Survival statistics alone, however, leave unanswered the central question of whether the recent improvements in infant survival actually are due to intensive medical care of the newborn, or whether they are the result of other factors such as lower prematurity rates or improved maternal health and nutrition. This complex question is dealt with in some detail in the discussion presented below, but is not easily resolved. On balance, the available evidence and some recently applied analytic methods allow us to conclude that neonatal intensive care has played a major role in improving the chances of survival of many newborns, particularly those of very low birthweight, even though it is impossible to quantify precisely how much of the improvement is due to medical care. A second and equally important question left unanswered by survival figures is whether the increased number of infants being saved will be able to lead normal lives, or whether they will suffer from serious handicaps such as cerebral palsy and mental retardation. This question is perhaps even more perplexing than the issue of survival, for two reasons. First, the data on morbidity of survivors of intensive newborn care are limited and do not allow definitive conclusions. It appears reasonable, however, to conclude that the incidence of serious problems has not increased and is probably decreasing. Second, the absolute number of severely handicapped individuals can increase at the same time the incidence of serious problems is decreasing. This unpredicted and seemingly contradictory situation would result if mortality rates fell faster than the incidence of serious complications. Our analysis shows that this may be happening, especially in the subgroup of the smallest infants, those weighing 1,000 g or less. Thus, while the number of normal survivors has increased eightfold to twentyfold since 1960, the small but worrisome number of severely handicapped individuals may have doubled. Morbidity studies are made even more difficult by the fact that some handicaps are not discovered until the children are of school age. As discussed in the next part of this case study, these results illustrate the dilemma of trying to determine whether intensive care of the newborn is cost effective. Every year, several thousand babies who would have died are now surviving to lead normal lives. Part of the price for this success, however, is a persistently significant number of abnormal survivors.
PAGE 36
Mortality Remarkable improvements in infant mortality have occurred in this country since 1915, as demonstrated in figure 2. Infant mortality includes all deaths during the first year of life. Deaths in the neonatal period (the first 28 days) are referred to as neonatal mortality, and the rate of neonatal mortality should be influenced by improvements due to the intensive care of newborns. Most of the decline in infant mortality had taken place by 1950 and cannot be attributed to neonatal intensive care. Between 1950 and 1965, the infant mortality rate continued to decline, but at a slower rate. During that period, nearly 60 percent of the decline in infant mortality was in deaths during the neonatal period. The end of that era coincides with the introduction of NICUs and of perinatology as a medical subspecialty. Since 1965, infant mortality has fallen by over 40 percent, declining at 3.6 percent annually, as compared with the 1 percent per year fall during the 1950 to 1965 period. Nearly all (90 percent) of the decline in infant mortality since 1970 has occurred in the neonatal period. Because the most recent improvements in the neonatal mortality rate were concurrent with the proliferation of intensive care of the newborn, a number of studies have been undertaken to see whether the changes are, in fact, due to modern medical care. The evidence collected falls into four categories: 1) clinical trials, 2 ) epidemiological surveys, 3) reports from indiFigure 2. Infant, Neonatal, and Postneonatal Mortality Rates: United States, 1915-77 Infant mortality rate (IMR) deaths per 1,000 live births (under 1 year) Neonatal mortality rate (NMR) (under 28 days) Post neonatal mortality rate (PNMR) (28 days to year) I 00 90 80 70 60 1 50 40 30 20 10 b Year IMR NMR PNMR 1915 1920 1930 1940 1950 1955 1960 1965 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 99.9 85.8 64.6 47.0 29.2 26.4 26.0 24.7 20.9 20.0 19.1 18.5 17.7 16.7 16.1 15.2 14.1 13.8 13.0 44.4 41.5 35.7 28.8 20.5 19.1 18.7 17.7 15.6 15.1 14.2 13.6 13.0 12.3 11.6 10.9 9.9 9.5 8.7 55.5 44.3 28.9 18.2 8.7 7.3 7.3 7.0 5.3 4.9 4.9 4.8 4.8 4.4 4.5 4.3 4.2 4.3 4.2 1915 20 30 40 50 55 60 65 69 70 71 72 73 74 75 76 1977 Year SOURCE National Center for Health Statistics, Vital .Sfat/.st/cs of the Un/ted States, selected years
PAGE 37
vidual intensive care units, and 4) analyses of birthweight-specific mortality rates. Clinical Trials There has been only trolled clinical trial of routine care of neonates in Australia in 1966-70, vival as intensive care one randomized conintensive care versus (68). That study, done showed improved surexperience developed, particularly for infants with respiratory distress. At least one leading neonatologist attributes the paucity of such studies to the notion that controlled clinical trials of perinatal intensive care and regionalization have been neither feasible nor morally acceptable (127). There also have been clinical trials of some individual intensive care therapeutic techniques. A recent clinical trial demonstrated improved mortality and shorter periods of hospitalization for infants who were transported under the care of a specialized neonatal team compared with those transported by the local ambulance authority (26). The recent debate over the efficacy of continuous positive airway pressure (CPAP) devices in the treatment of RDS illustrates the status of most of the clinical evidence, however. This therapy has been hailed as a breakthrough responsible for marked improvement in the survival of infants with RDS. However, the first controlled clinical trial was not conducted until several years after widespread adoption of CPAP, and that study was not able to document statistically significant improvement in mortality (16). Publication of that report in the pediatric literature was accompanied by a long editorial reiterating the abundant evidence for the effectiveness of CPAP and detailing possible deficiencies of the controlled study, while expressing general support for clinical trials (133). In short, many newborn therapies are unproven and even controversial. Thus, no overall conclusions can be reached from the clinical literature alone. Epidemiological Surveys Epidemiological evidence suggests that medical care has helped lower mortality rates, but for the reasons discussed below, the evidence is not convincing. The Canadian experience showed a .50-percent decline in neonatal mortality in Toronto in the decade following organization of special neonatal care facilities (121). An analysis of data for the United States between 1955 and 1973 demonstrated that, in general, mortality rates fell with increasing urbanization (41), but urban areas were not stratified in terms of levels of intensive care available for newborns. Thus, no conclusions could be drawn about the relationship between urban medical care and mortal it y. A major review article summarized reports from various parts of the United States, all of which showed declines of 23 to 42 percent after the development of regional intensive care facilities (125). Our own analysis confirms that these declines were generally greater than were experienced for the United States as a whole during the same years. Although these data are suggestive that development of regional intensive care centers improved the outlook for infants in surrounding areas, the studies do not consider whether prematurity rates fell during the same period. Since declines in prematurity can markedly reduce neonatal mortality rates independent of changes in medical care, this omission greatly reduces the conclusions that can be drawn from these reports. We conducted an analysis of other crude epidemiological evidence and failed to find any support for the hypothesis that increasing access to neonatal intensive care has had a large-scale effect on neonatal mortality rates. Comparing the reduction in neonatal mortality rates for each State between 1971 and 1977 with the number of tertiary care neonatal intensive care beds per 1,000 live births that were reported at the end of that period revealed no consistent relationship between the two variables. Our analysis is inconclusive since the development of NICUs in many areas might have been disproportionate to actual need. Moreover, the data on the number of beds are not fully reliable.
PAGE 38
Sequential Reports From Individual Intensive Care Nurseries Most of the literature on the effectiveness of neonatal intensive care consists of reports detailing the experience of individual nurseries. Twelve of these reports were recently reviewed; all demonstrated reductions in inborn neonatal mortality, ranging from 17 to 56 percent (125). The impact that changes in prematurity rates can have on mortality experience, however, limits the value of these reports. If the distribution of births among birthweight groups changes from year to year, so that very small infants represent a lower proportion of total births, mortality experience will fall substantially. The proportion of infants in the very low birthweight category (1,500 g or less) has the greatest impact on mortality rates (69). Because very low birthweight infants have a very high mortality rate, any decrease in the proportion of such infants will have a significant impact on neonatal mortality. The experience of the Bronx Municipal Hospital Center illustrates this phenomenon clearly (71). In that hospital, between 1966-71 and 1972-73, the in-hospital neonatal mortality rate declined from 16.9 per 1,000 live births to 11.2 per 1,000 live births. Over 75 percent of the decline could be attributed to changes in the proportion of very low birthweight infants (1,500 g or less). This institutions experience demonstrates that even if no improvements due to medical care after birth are taking place, there can be impressive declines in neonatal mortality if a smaller proportion of high-risk infants is being born. Clearly, reports on overall mortality must specify birthweight distribution if they are to help determine the effectiveness of newborn medical care. If mortality rates within specific birthweight groups improve over time, this would be a strong indication that neonatal intensive care is having a beneficial impact. For example, if infants of 1,500 g birthweight or less have a better chance of surviving after introduction of neonatal units, such a change must be due to either improved neonatal medical care or to improved health status of the infants at birth, or both (71). A large number of nurseries have reported mortality rates by birthweight before and after the introduction of an intensive care facility (2,57,99,107,110,125,126,127). The results for every nursery but one (61) show impressive declines in birthweight-specific mortality over time. Attributing most of these improvements to newborn medical care, Thompson and Reynolds stated: NICUs have more than justified their reason for existence, reducing by approximately one-half the risk of neonatal death (125). Even birthweight-specific reports from individual nurseries are of limited usefulness, however, because of their small population samples. Often, only 20 to 30 infants in the very low birthweight groups are born during the study period, making it difficult to determine whether improvements over time are statistically significant. For this analysis, we grouped all available reports of mortality rates for inborn infants into 5-year periods beginning in 1961. These pooled data demonstrate significant declines over time for very low birthweight infants (see tables 17 and 18 and figure 3). Mortality for infants with birthweights of 1,001 to 1,500 g has fallen from more than 50 percent to less than 20 percent, while that for the extremely small newborns (1,000 g or less) has fallen from nearly 94 percent to approximately 50 percent. Birthweight-Specific Mortality Rate Analyses It is not possible to compare the experience in individual nurseries to the survival rates of infants of similar birthweights who were not cared for in intensive newborn units. As pointed out above, it is difficult to identify with confidence the levels of care available at different hospitals. Moreover, not all States link birth certificates with death certificates, so it is impossible to know on a routine basis the birthweight of individual infants who die. Data for all births in California (where birth and death certificate data are compiled) have been collected, however, and show major improvements in neonatal mortality between 1960 and the present (128),
PAGE 39
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care l 31 Table 17.lnborn Neonatal Mortality Rates, Birthweight 1,001 to 1,500 Grams Years reported 1961-65 1966-70 1971-75 1976Hospital a UCSF (99). . . . . . . . Cambridge Maternity (107) . . . . Royal Victoria (127). . . . . . Harvard (135), . . . . . . . University of Washington, Seattle (60). . Medical Center, Columbus, Ga. (126). . . Bronx Municipal (71). . . . . . Simpson Memorial Edinburgh (37) . . E. Hospital, Goteberg, Sweden (1 10) . . Total. . . . . . . . . if other than headings 1976-78 1975-78 1978 1959-68 1969-70 1971-72 1966-71 1972-73 1966-70 1969-70 Deaths/ Deaths/ Deaths/ Deaths/ births Rat e b births Rat e b births Rat e b births Rat e b 26/55 473 23/63 365 15/80 188 9/68 132 23/96 240 25189 281 14/90 156 37/213 174 14/75 187 116/219 530 14/44 318 14/41 341 69/193 358 11/42 262 71/04 461 10/24 417 142/274 518 212/567 374 C 54/253 213 c 831452 184 d aNumbers in parentheses refer t O references in the Iist that appears at the end of this case study bRate = deaths/1,000 Iive births cSignificantly different from preceding 5-year rate (p<.01) dNot significantly different from 213 (1971-75). Table 18.lnborn Neonatal Mortality Rates, Birthweight <= 1,000 Grams Hospital a UCSF (99). . . . . . . . Childrens, San Francisco (57) . . . Cambridge Maternity Hospital (107). . . Royal Victoria (127). . . . . . University of Washington, Seattle (60). . Medical Center, Columbus, Ga. (126) . . Bronx Municipal (71). . . . . . University of lllinois (17) . . . . Simpson Memorial Edinburgh (37) . . University College Hospital, London (1 19). . E. Hospital, Goteberg, Sweden (1 10) . . Total. . . . . . . . . Years reported 1961-65 1966-70 1971-75 1976if other than headings Deaths/ Deaths/ Deaths/ Deaths/ births Rat e b births Rat e b births Rat e b births Rat e b 28/29 966 26/34 765 1972-75 1976-77 1976-78 65/73 890 1978 1959-68 1969-70 1971-72 1966-71 1972-73 1974-76 1969-70 22/33 667 15/30 500 30/35 857 8124 333 9/25 360 58175 773 25/41 610 157/168 935 27134 794 26135 743 119/138 862 18/25 720 29/38 763 84/92 913 29/36 805 417 571 185/197 939 381/443 860 C 26/33 788 209/274 763 c 57/120 475 c aNumbers in parentheses refer to references in the list that appears at the end of this case study bRate = deaths/1,000 Iive births cSignificantly different from preceding 5-year rate (P<.01) Analysis of these data by Ronald L. Williams of the University of California, Santa Barbara, supports the hypothesis that medical care was a major factor in the lowering of mortality rates (129,130). For infants in all hospitals in California, Williams found that once the risks faced at each hospital are taken into account, Level III hospitals have far better survival rates than Level I hospitals, and hospitals with large delivery services and high specialist-to-generalist ratios have the best performance. The fact that mortality rates within birthweight groups have declined over time strongly supports the conclusion that neonatal intensive care has helped improve survival. One could question, though, whether there were also underlying changes in the health status of infants within each weight group that improved
PAGE 40
32 l Background paper #2: Case Studies of Medical Technologies Figure 3.Pooled Neonatal Mortality Data, 1961.76 Birth weight <= 1,000 grams 1961-65 1966-70 1971-75 1976SOURCE. See tables 17 and 18; neonatal mortality independent of intensive newborn medical care. Such changes could be the result of medical or nonmedical factors. Medical factors, such as intensive intrapartum obstetric care or steroid therapy during premature labor to prevent HMD, could produce infants more likely to survive in the extremely fragile low birthweight groups. Obstetric care during pregnancy prior to labor would not be expected to affect birthweight-specific mortality rates, but to reduce the chances of low birthweight. Nonmedical factors such as improved maternal nutrition and socioeconomic status also have their primary beneficial effect through higher birthweights (86,123). However, it is possible to speculate that better maternal nutrition and general health results in subtle biochemical or other changes that further increase the chance of survival for infants in each weight group. Unfortunately, none of these hypotheses can be established or rejected with confidence from the available data. As noted above, Williams has analyzed changes in mortality rates among newborns in California over the past two decades (128,130). Between 1960 and 1977, the neonatal mortality rate in California fell by 57 percent. During the same period, prematurity rates also declined. It was necessary, therefore, to partition the improvement in neonatal mortality into the part due to changes in birthweight distribution and the part due to other factors, including medical care. In contrast with the result noted above for Bronx Municipal Hospital Center (7I), Williams analysis showed nearly 85 percent of the remarkable decline in neonatal mortality in California was due to birthweight-specific declines. That is, the reduction in prematurity accounted for only about 15 percent of the decline in neonatal mortality, while medical care and/or better health status of individual infants accounted for the remainder. More than half of the overall decline could be attributed to improvements in the mortality rates of low birthweight infants. Blacks experienced lower mortality rates than whites, even though their birthweight distribution actually became less favorable over the 1960 to 1977 period. Kleinman and others at the National Center for Health Statistics analyzed data from six States and reached similar conclusions (69). They found that weight-specific changes accounted for nearly 80 percent of the decline in early neonatal mortality rates. More than half of the overall decline was due to improvements in mortality among the low birthweight groups. These investigators estimated that for the nation as a whole, about 83 percent of the decline in early neonatal mortality for whites and 97 percent of the decline for nonwhites could be attributed to factors other than changes in birthweight. The same investigators discussed factors that could be causing the decline in birthweightspecific mortality rates. They rejected a suggestion that lower death rates were merely an artifact due to improved registration. They also concluded it was unlikely that any major shift in
PAGE 41
Case Study #10 The Costs and Effectiveness of Neonatal Intensive Care l 3 3 distinctions between live births and fetal deaths (stillbirths) had taken place, since fetal deaths were also declining, Finally, they attempted to deal with the question of whether infants within given birthweight groups might have different risk characteristics over time. Kleinman and his colleagues found that comparing gestational age distributions made no difference for the calculation, and they concluded (69): The most likely explanation for at least part of the recent decline in mortality levels among low birthweight infants is the development of the medical technology for the successful management of premature infants and the consequent proliferation of this technology. While there are many components of the technology which have not been critically evaluated, there are indications of effectiveness. Lee, et al. (72) examined the effect of birthweight changes on the improvements in neonatal mortality in the United States between 1950 and 1975. They determined that no improvement had taken place in the weight distribution or, with the exception of perinatal medical care, other factors known to affect survival. Their conclusion, similar to that of Williams (129) and Kleinman, et. al. (69), was that (72): The most plausible explanation is the steady improvement in perinatal medical care which has made for greater infant survival at a given birthweight. Most recently, these same investigators used birthweight-specific analysis to examine variations in the neonatal mortality among the 50 States and the District of Columbia, and among 13 industrialized nations. They confirmed the role of the very low birthweight rate in predicting neonatal mortalitythat rate accounted for about three-quarters of the variance among the populations studied (73). They suggest that variance of neonatal mortality after holding constant the very low birthweight rate might be a useful preliminary indicator of the quality of newborn care. Birthweight-specific analyses are helpful in answering part of the question regarding the role of medical care in bringing about improvements in infant mortality, but they do not settle all the issues. By estimating that about 15 percent of the decline in overall mortality rates is simply the result of changes in birthweight distribution, these analyses define the maximum impact on survival that could be attributed to medical care of the newborn. However, this is not to say that all of the remaining 85 percent is necessarily due to neonatal medical care. As noted above, there might be medical or other factors producing infants more likely to survive. Some limited evidence that mortality rates can improve even in the absence of high grade neonatal medical care precludes our assuming that medical care is responsible for all the improvement in birthweight-specific rates. One recent and unfortunately rather sketchy report describes a British hospitals experience in managing ill and premature babies with little or no intensive care (59). No respirators, phototherapy, CPAP device, or extensive mask and glove isolation were used in the nursery. Some modern techniques, including initial resuscitation, catheterization of umbilical vessels, exchange transfusions, and incubators were applied. The absolute levels of mortality achieved in each weight category would clearly be unacceptable at a major nursery in the United States. However, there was very significant annual improvement within birthweight groups. On the basis of this limited report, one is not able to conclude whether this nursery is, in fact, able to improve mortality without changing neonatal care and without delivering the most advanced methods. No other evidence is available to help determine the precise contribution of neonatal care in improving birthweight-specific mortality rates. In sum, while it may be impossible to determine the precise role that neonatal intensive care has played in lowering mortality rates among low birthweight infants, the evidence supports the conclusion that neonatal care has been a significant factor. The degree, if any, to which infants are somehow improving in their inherent survivability from year to year and the role of intensive intrapartum obstetric care could be tested only through controlled clinicaI trials, and such trials are unlikely to be conducted,
PAGE 42
34 l Background Paper #2: Case Studies of Medical Technologies Morbidity When it first appeared that neonatal intensive care would save ill and premature infants who otherwise would have died, many concerns were voiced that the net result would be a large number of highly defective survivors. Some early studies, particularly those by Lubchenco (74,75) and Drillien (38), seemed to corroborate this fear. They reported that many of the survivors of their premature nurseries in the 1940s and 1950s were afflicted with serious problems. These included necrologic defects, particularly cerebral palsy, mental retardation of varying degrees, blindness and other visual defects, growth failure, hearing loss, and chronic respiratory disease. All of these problems are still found to a significant degree in the survivors of present day intensive care nurseries. The incidence of problems appears to have fallen in recent years, but the available data for the most part permit only tentative estimates, not definitive conclusions. Incidence of Serious Handicaps Many articles have claimed that the incidence of severely handicapping conditions has declined remarkably in recent years (see tables 19 and 20). However, these reports all suffer from serious limitations. The most critical shortcoming is that the studies are based on survivors of individual nurseries. Morbidity data are not collected on a routine basis in the United States. Thus, the incidence of problems and trends over time in the general population cannot be determined. Second, the reports cover a variety of different aspects of morbidity, some emphasizing neurological diseases and others focusing on problems specific to the cardiovascular system or respiratory tract. It is often impossible to determine, for example, whether survivors with neurological disease also had chronic respiratory disease, or whether normal survivors in a study emphasizing neurological disease were afflicted with cardiovascular problems. In addition, the reports have used different measures of Table 19. Serious Handicaps, Birthweight <= 1,500 Grams a Serious Major Serious Retrolental Survivors with 1 or more Number of retardation neurological hearing fibroplasia c serious handicaps Year bomb survivors (IQ-DQ <70) defects defect Present Blind Percen t Number 1968-74 (18) . . 1971-72 (21) . . 1948-56 (38) . . 1966-71 (37) . . 1970-72 (44) . . 1964-69 (52) . . 1973-74 (58) . . 1947-53 (74,75). . . 1969-70 (1 10) . . 1966-69 (105) . . 1965-70 (124) . . 1940-57 (35) . . 1952-56 (132) . . 1951-53 (81) . . pre-1960 (96) . . 1961-65 (59) . . 1966-75 (61) 205 16 49 88 232 42 42 133 17 68 176 100 60 65 240 44 49 94 10 0 12 9 6 ( d ) 3 25 5 32 7 3 7 7 44 8 0 32 0 3 20 5 17 22 10 2 6 14 33 9 1 11 16 8% 19 43 e 19 e 2 31 f e 7.4 20 18 67 26 9.2 23 14.3 12.8 17 3 21 e 44 e 1 28 f e 5 35 18 40 e e e 7 12 = not specified. aUnless otherwise noted bNumbers in parentheses refer to references in the list that appears at the end of this case study cScarring of the internal eye. dIncludes IQ 80 as major handlcap. Serious handicap rates among survivors Pre-1960: 24% (156/649) Post-1965: 13.6%(1 16/850) eCould not be calculated from repel ted data fof 91 who were examined
PAGE 43
Year bomb 1965-70 (2) 1974-76 (17) 1968-74 (18) 1940-52 (36) 1968-70 (40) 1948-54 (38) 1968-72 (51 ) 1974 (95). . 1969-70 (110) 1966-74 (1 18) 1965-70 (124) 1947-50 (75) 1960-66 (44) 1961-68 (47) Table 20. Serious Handicaps, Birthweight <= 1,000 Grams a Number of survivors 20 16 27 42 14 7 27 43 3 27 40 7 5 12 Serious retardation (IQ-DQ <70) 2 5 0 4 0 7 0 1 6 Major Retrolental neurological fibroplasia c defects Present Blind 1 3 3 3 2 1 2 0 1 1 4 2 6 7 2 0 1 1 6 Survivors with 1 or more serious handicaps Percent Number 20 /0 4 19 3 7 2 14.3 6 14.3 2 57 4 18.5 5 21 9 0 0 7 2 20 8 100 7 40 2 17 2 = not specified Serious handicap rates aUnless otherwise noted bNumbers in parentheses refer to references in the Ii S t that appears at the end of this case study among survivors Pre-1965: 29% (21/73) c Scarring of the internal eye the various defects, and have studied survivors at varying ages, thus making the findings difficult to compare. In general, the incidence of serious handicap is inversely proportional to birthweight. The most serious and frequent problems are reported in infants with very low birthweight, that is, less than 1,500 g. Tables 19 and 20 summarize a large number of reports on the incidence of serious handicap in very low birthweight infants. Throughout this analysis, serious handicaps are defined as the following: severe mental retardation (IQ or developmental quotient below 70); cerebral palsy of significant degree (spastic diplegia, paraplegia, tetraplegia, hemiplegia); major seizure disorders; and retrolental fibroplasia (scarring of the internal eye) with blindness or significant residual impairment of vision. Figures 4 and 5 illustrate the findings of these studies over time. It appears there has been some narrowing of the range of reported handicaps, as well as a reduction in the incidence of severely affected infants, in the most recent years. For infants weighing 1,500 g or less, there appears to have been approximately a 40-percent decline; before 1960, about one-fourth suffered severe neurological and other handicaps, while recent reports average under 14 percent. Post-1965: 16 % (35/217) Figure 4. Serious Handicaps, Birthweight <1,500 Grams ~~ 90 t 85 80 75 70 65 60 55 50 45 40 35 30 25 20 15 10 5 0 Ref. 35 r Ref. 38 Ref. 74,75 Ref. 132 m Ref. 81 Ref. 124 Ref~ 44 Ref. 24 Ref. 18 R~5 1 I 1 I IRef.l 1940 1945 1950 1955 1960 1965 1970 1975 SOURCE See table 19
PAGE 44
Figure 5. Serious Handicaps, Birthweight <1,000 Grams 90L 85 80 75 k 70 65 60 55 50 45 40 35 30 25 20 15 10 5 0 Ref .38 Ref.44 Ref. 2, 124 ef.~ Ref. 47 ~fr Re~ Re~40 Ref: 109 Ref. 118~ I 1 1 111Ref. 110 ; tionalized or seriously afflicted (35). Second, the total number of reports and infants studied is small. Infants weighing 1,500 g or less totaled about 1,400 patients, and the entire literature on infants weighing 1,000 g or less includes only about 300 individuals over nearly 40 years. Finally, the reports do not always use standardized criteria or ages for the followup analyses. These data from the literature cannot be relied on with full confidence for the reasons listed above. They do, however, correlate reasonably well with, and thus are supported by, the results now appearing from the first large population-based study of morbidity by birthweights. Sam Shapiro and others at Johns Hopkins University have studied the low birthweight population of eight regions in the United States (115). In their first report, they noted that for infants with birthweights of 1,500 g or less and born in the first half of 1976, the incidence of severe impairment was 14.3 percent, very nearly the same as the figure we calculated from the literature. These data constitute a pre1940 1945 1950 1955 1960 1965 1970 1975 liminary baseline for an evaluation of the impact of further perinatal regionalization, and as such, they are not necessarily representative of the outcomes expected for the maximum level of intensive care. Nevertheless, the findings are rel-SOURCE See table 20 The results for infants weighing 1,000 g or less are comparable. They suggest a decline from about 30 percent of those affected to some 16 percent. Several serious problems limit the conclusions which can be drawn from these reports. First, it is not clear how the populations being studied may have been biased toward either favorable or unfavorable outcomes. The condition of infants entering a given nursery for immediate neonatal care and the proportion of surviving infants returning for followup studies are dependent on a wide variety of uncontrolled factors. For example, one author published a followup study which tracked down data on most of the survivors who had not returned to be studied and found that many were institu evant for this analysis, since newborn intensive care centers were in operation in all regions studied, and infants in the very low birthweight groups are very likely to have been cared for within the existing centers. Neurological and Intellectual Defects Although the numbers for each reported complication become increasingly small as one separates out each adverse outcome, major neurological defects such as cerebral palsy, seizure disorders, and hydrocephalus are among the most commonly noted complications. The reports that allow one to identify infants with neurological defects suggest a reduction in the incidence of such problems, from an average of 16 to 18 percent prior to 1965, down to approximately 10 to 11 percent among infants weighing 1,500 g or less in recent years (see table 19). The results for infants weighing 1,000 g or less sug-
PAGE 45
gest an apparent increase over time, but the absolute number of infants reported on in this birthweight category was too low to allow any conclusions (see table 20). Infants may suffer severe mental retardation whether or not serious neurological damage is also present. In the studies before 1960, about one-fourth of all infants weighing 1,500 g or less were seriously retarded. In the reports since 1965, there has been an apparent decline of severely retarded infants to less than 10 percent (see table 19). Parallel figures have been reported for infants weighing 1,000 g or less, but, again, the number of infants reported in this birthweight category prior to 1960 is quite small (see table 20). In sum, it appears that the incidence of severe neurological damage has declined, and the current incidence is on the order of 10 percent of very low birthweight survivors. Similarly, the incidence of severe mental retardation has probably declined, and is less than 10 percent. In addition to concerns that neonatal intensive care would produce severely impaired infants, there have also been widespread fears that the relatively normal survivors would have a high incidence of moderate impairment. This category would include infants with IQs or developmental quotients between 70 and 80, infants who have soft necrologic signs, and infants who develop behavior, learning, and language disorders. These infants comprise a group with moderate or minor intellectual or neurologic damage. It is not possible to estimate changes in this group over time because of the limited data. Present data indicate that on the order of 10 to 12 percent of infants weighing under 1,500 g will suffer intellectual or neurologic problems that do not constitute major handicaps. Data on the incidence of learning, language, and behavior disorders, along with all information on other minor handicaps, are in many ways less reliable than those reported for severe handicaps. That is because the definition of severe handicap is more precise and because the recognition of problems such as hydrocephalus or cerebral palsy is much more pressing and more likely to cause parents to seek medical attention than is the recognition of learning and language disorders or mild neurological impairment. Moreover, great caution is needed in interpreting data on minor problems because there are no reliable standards for the incidence of behavior, language, and mild necrologic problems in the general population of children. General Health of Low Birthweight Survivors The Johns Hopkins study referred to above (115, 136) is an important new source of information which allows one to compare the overall problems of low birthweight infants with those suffered by full-term infants. Among survivors of birthweight 1,500 g or less, nearly 40 percent were hospitalized at least once during the first year of life, compared to just over 8 percent of infants weighing 2,500 g or more. Even when infants with severe impairments and moderate or minor congenital anomalies are excluded, still more than one-fourth of the otherwise normal, very low birthweight infants were hospitalized, compared to only about 7 percent of all infants. When all serious illnesses were taken into account, less than 40 percent of the infants weighing 1,500 g or less were free from morbidity during their first year of life, compared to over 70 percent of all infants free. It is clear that extremely low birth weight survivors carry a very high burden of illness. These findings are consistent with the isolated reports of morbidity in the literature. Survivors of respirator therapy are known to have a significant incidence of chronic pulmonary problems, at least during the first year of life (45). Gastrointestinal problems, particularly failure to thrive, are quite common among extremely low birthweight infants. Other indicators of morbidity which have been documented include visual impairment, particularly severe myopia, and minor hearing defects. Some of these problems will disappear over time. For example, most recent studies have demonstrated that all but a very small proportion of low birthweight infants do attain relatively normal stature by early childhood (18,25).
PAGE 46
Conclusion Over the past 10 to 15 years, since neonatal intensive care methods have been applied, there has been a dramatic improvement in survival of low birthweight newborns, and most of the survivors are normal (see figures 6 and 7). While it is impossible to say precisely what proportion of this improved outlook is due to each of the possible factors mentioned above, the evidence strongly suggests that medical care of the newborn deserves much of the credit. One finding of some concern is that the absolute number of seriously handicapped inFigure 6. Survival of Newborns, Outcomes per 1,000 live births Birthweight <= 1,500 Grams 178 235 1960 1971-75 486 Years A Normal u Abnormal, severe Abnormal, q \ ,..,. Dead .: moderate SOURCE See tables 17 and 19 dividuals may be increasing. For example, with birth rates at the 1978 level, intensive care nationwide produced about 350 severely handicapped individuals of birthweight 1,500 g or less who would have died in 1960 (see table 21). This figure must be balanced against the over 16,000 net increase in normal individuals at present surviving. It appears that the worst predictionsthat most of the survivors of neonatal intensive care would be seriously handicapped are by no means verifiable. The overwhelming majority of survivors are normal; a number have minor handicaps; and a smaller number are severely handicapped. In this sense, we conclude that neonatal intensive care is effective. Figure 7. Survival of Newborns, Birthweight <= 1,000 Grams II Outcomes per 1,000 live births 1960 Normal c 1 Abnormal, moderate c 1 SOURCE See tables 18 and 20 n 369 H 73 1971-75 1976Years Abnormal, ,,.:,. c 1 severe ~ a g.::. Dead :$$:
PAGE 47
Case Study #10 The Costs and Effectiveness of Neonatal lntensive Care l 39 Table 21 .Total Number and Distribution of Survivors, Birthweight <= 1,500 Grams, 1978 Expected number a Estimated actual number b Net change Normal ., ., 2,790 18,833 + 16,043 Abnormal moderate 5,425 4,069 1,356 Abnormal severe 2,596 2,945 + 349 Dead ~ ~ ~ 27,979 12,904 15,075 a Expected number total births ( 1 500 g or less 1978 x outcomes estimated for births in 1960 (from flg. 6) bEstimated actual number = total births (1,500 g or less 1978) x outcomes estimated for broths since 1976 (from fig. 6.) SOURCE Text and previous tables ECONOMIC ANALYSIS OF NEONATAL INTENSIVE CARE CBA and CEA can provide some useful information on the economic implications of neonatal intensive care. With the expansion of NICUs, an increasing number of infants are being treated in Level 11 and III units. The high financial and human costs associated with neonatal technology and long hospital stays for the newborn make it necessary to examine and compare information on costs and effectiveness. Any CEA, however, is only as good as the data to be analyzed. The Imitations of the available information on neonatal intensive care have already been noted. In the following discussion, we tie together what is presently known about the costs incurred by and the outcomes expected for infants receiving neonatal intensive care. We summarize the characteristics of CEA and review existing studies, and then we discuss the implications of applying a specific cost-effectiveness approach to the results of this case study. In the strictest sense, neither CEA nor CBA is applicable to neonatal intensive care. CEA is primarily intended to measure and compare the costs of more than one way of arriving at similar outcomes. Ideally, such a method could be applied to comparisons between neonatal intensive care and prenatal prevention-oriented strategies, but the data needed to make such comparisons are lacking. Not nearly enough is known about causal relationships between changes in socioeconomic and nutritional status on the one hand and perinatal mortality on the other to predict the outcomes of targeted interventions. Even for risk factors (such as smoking and teenage pregnancy) that are causally related to neonatal disorders, the marginal costs and effectiveness of prevention strategies have not been carefully studied. FinaIly, since there are no alternative postpartum procedures to arrive at outcomes similar to those of modern neonatal intensive care, strict CEA does not apply. CBA also has severe limitations. In particular, placing dollar values on different human outcomes such as interrupted pregnancies, abnormal survivors, and even normal individuals is philosophically controversialand in any case, it is difficult to arrive at satisfactory figures. Nevertheless, CBAs and CEAs are useful for illustrating the economic implications of present intensive care methods for the newborn. Unfortunately, most of the analyses that have been done to date are simplistic. Many reports in the literature contain statements that the costs of hospitalization in an intensive care nursery are far less than the costs of life-long institutionalization of a severely defective survivor. Such analyses presume fully beneficial outcomes with treatment and totally unavoidable, severe handicaps without treatment. These assumptions are not justified. A passage from the Governor of Pennsylvanias Health Task Force Report of the Committee on Infant Intensive Care (1974) illustrates the type of CEA of neonatal intensive care that is most often encountered (3): In the current economic climate, one must consider the relative cost-benefits of an approximate 20-day hospital stay receiving intensive care (about $2,000 to $3,000) versus a lifetime institutional stay receiving custodial care
PAGE 48
40 Background Paper #2: Case Studies of Medical Technologies (about $200,000 to $300,000), whose cost is usually borne by the State. A similar one-dimensional view of cost effectiveness is found in Ross Laboratories Planning and Design for Perinatal and Pediatric Facilities (109). That publication states that neonatal intensive care is expensive to provide but more expensive not to (109). Institutionalization costs of $4,000 to $6,000 per year for 40 to 45 years for a brain-damaged child are contrasted to an average of 20 days of intensive care at $250 per day, or $5,000 total (109). Neonatal intensive care may, in fact, be both cost effective and cost beneficial, but it is impossible to make that determination from such limited efforts which assume the effectiveness of neonatal intensive care and do not recognize all the possible outcomes, including death and morbidity. A few thoughtful analyses have been conducted. The French Ministry of Health faced the question of how to reduce perinatal mortality before an extensive system of neonatal intensive care was in place (27). The Ministry applied a CEA to a variety of medical care options, including antirubella inoculation of youn g women, improved prenatal care, and NICUs. On the basis of its admittedly tenuous outcome data, the Ministry designed a blend of the various approaches that was calculated to be cost effective in terms of real governmental costs. The French analysis indicated that, as of nearly 10 years ago, NICUs decreased mortality but increased the incidence of handicap, making them relatively cost ineffective. In addition, two British investigators, Akehurst and Holtermann, have discussed the data collection and economic analysis basic to a thorough CEA of intensive newborn care (1). Application of One Method of Economic Analysis to the Results of This Case Study As detailed in the sections above, the existing literature and data on costs and effectiveness required extensive secondary analysis in order to generate a few tentative conclusions. The information needed for a complete CEA or CBA of neonatal intensive care includes a number of items not at present available (see table 22). For example, the amount of additional care required by survivors is not well established, let alone the costs of such care. Calculations would have to take into consideration such diverse elements as hospital and medical care needs of survivors after the neonatal period and the projected costs of treating such poorly understood complications as behavioral and learning disorders. Table 22.lnformation Required for a CEA or CBA of Neonatal Intensive Care costs: l Dollar costs of neonatal intensive care l Dollar costs of additional care required by survivors Dollar costs of alternatives to intensive medical care l Dollar value of intangible costs Benefits/effectiveness: l Likelihood of specific outcomes with intensive medical care and with alternatives to intensive care l Dollar benefits over lifetime of survivors with each possible outcome l Dollar value of intangible benefits These limitations put the performance of an elaborate CEA or CBA beyond the scope of the present study. In order to discuss the findings of this study in an economic framework, we apply our results to a model previously developed by economist Marcia Kramer (70). Kramers model was presented at a conference on Ethics of Newborn Intensive Care, cosponsored by the Health Policy Program, of UCSF in 1974. Her formula balances the actual dollar costs against the outcomes to be expected from different levels of intensity of newborn medical care. The net marginal costs of replacing less sophisticated newborn care with neonatal intensive care can thus be measured, making this a useful approach for the purposes of this study (see table 23). The dollar values assigned by Kramer reflected the belief that normal survivors are cheap and economically productive; nonsurvivors are relatively inexpensive; and seriously defective survivors are both expensive and not productive. The cost of neonatal medical care is only the first cost to enter into the calculation. Postnatal direct costs included food, clothing, and education, and indirect costs included the mothers forgone earnings. After maturity, the
PAGE 49
Case Study #10: The Costs and Effectiveness of Neonatal intensive Care l 41 expenditure flow reverses for a normal survivor as earnings outpace consumption. The severely abnormal survivor, however, is assumed never to generate income in excess of consumption. In addition, the abnormal survivor incurs costs exceeding normal consumption, including institutional custodial care, special education, and extraordinary medical care. Once the cost and benefit assumptions within categories are arrived at, one can examine the net costs of a variety of situations with different patterns of care and outcomes. Table 23.Kramers Cost-Effectiveness Formula (1) E = i= 1 where: E = N j = C it = C iw = B j = i = i = i = P it P iw = 3 P iw (C iw -B j ), P it (N l + C it -B i )-? expected net incremental dollar cost incurred by providing neonatal intensive care. the medical care cost of neonatal intensive care itself for a child with eventual outcome i (usually N 3 > N 2 > N 1 ); other direct and indirect dollar costs incurred during the lifetime of a child who is given intensive care at birth (t = treated) and whose eventual outcome is i (usually C 3 > C 2 > C 1 ); direct and indirect dollar costs incurred during the lifetime of a child who is denied intensive care (w = treatment withheld) and whose eventual outcome is i(C 1w =0 if euthanasia is the alternative to care, C jw = C it otherwise); dollar benefits yielded over the lifetime of a child with eventual outcome i(B 2 > B 3 > B 1 = o); 1: an infant death 2: a normal survivor 3: an abnormal survivor probability of outcome i given treatment decision t or w, respectively, where ; P it = : P iw = 1 i= 1 i= 1 (For this case study, Ps are taken from figs. 6 and 7) SOURCE: M.J. Kramer, Ethical lssues in Neonatal Care An Economic Perspective, i n Ethics 0f Newborn Intensive Care, A R Jensen and M. Garland (eds) (Berkeley, Calif.: Universlty of California, Institute of Governmental Studies, 1976) One hypothetical pattern Kramer analyzed illustrated the outcome most feared: Mortality rates fell with intensive medical care, but most of the survivors were severely defective. That situation did not prove cost effective. Kramer also analyzed a second hypothetical pattern. It considered that medical care would lower mortality rates from one relatively low level to an even lower level, and the number of abnormal survivors would decline slightly. That hypothetical situation was found to be cost effective. Based on the effectiveness material discussed in the previous part of this case study, it appears that the present situation is different from either of the two hypothetical patterns that Kramer presented. Our analysis shows mortality rates have fallen significantly, thereby improving the chances of survival in the very low birthweight groups. At the same time, the incidence of normal survivors has increased. Thus, the chance of a normal survivor now is many times greater than it was in 1960 (see figures 6 and 7). Although the incidence of serious handicaps has fallen, the rapid increase in survival means there may be a slightly increased chance of a very low birthweight infant becoming an abnormal survivor. Applying Kramers methodology (see table 23) to the data collected in this case study yields the finding that neonatal intensive care of infants weighing 1,500 g or less is marginally cost effective. Treatment of the subgroup weighing 1,000 g or less is not yet cost effective if the reported outcomes for 1971 to 1975 are used. When the most recent reports are used to estimate present outcomes, however, treatment of this subgroup becomes cost effective. The criterion for cost effectiveness is simply a negative net incremental dollar cost incurred by providing neonatal intensive care. Data limitations are an inherent weakness of these cost-effectiveness estimates. For example, Kramers cost and expenditure estimates were simply updated to 1978 dollar figures in proportion to changes in the consumer price index. More current dollar figures would improve the accuracy of the calculation. Furthermore, Kramers model incorporated institutionalization costs for each abnormal survivor, but the recent trend favors care at home. A more careful examination of what percentage of infants are institutionalized, for how long, and for what
PAGE 50
42 Background Paper #2: Case Studies of Medical Technologies price would improve and perhaps change the net values. In addition to the data limitations, there are a number of methodological factors that could be examined and included to provide a more accurate and comprehensive estimate of the net incremental dollar cost of providing neonatal intensive care. In particular, use of a range of discount rates with a sensitivity analysis would provide a more sophisticated set of figures. Even with the limitations discussed above, this illustration of a CEA highlights certain important aspects of the present return on the investment in neonatal intensive care. The characteristics of each birthweight group that led to a finding of cost effective or not cost effective raise considerations that go well beyond the financial ones. For the group weighing 1,000 g or less, the main reason that intensive care with outcomes such as those reported for the period 1971 to 1975 (see figure 7) does not turn out to be cost effective compared with outcomes reported for 1960 is that there was a small increase in the chance that a severely abnormal individual in this weight group would survive, That is, the costs of producing and caring for even a few severely abnormal survivors appear to be greater than the economic benefits gained by an eightfold increase in the rate of normal survivors (from 17 per 1,000 live births to 135 per 1,000 live births in that birthweight group) during the same period. Thus, in this illustrative calculation, the increased number of normal survivors in the group weighing 1,000 g or less would not quite offset the high costs of the severely abnormal ones. However, even if a more sophisticated and precise cost-effectiveness calculation were to confirm this result, the legal and ethical issues that would be raised by any attempt to withhold care from all newborns of 1,000 g or less simply to avert the exceptional costs associated with the severely abnormal survivors could not be ignored. Because the lives of many healthy babies would be lost without intensive care, such a decision would never be made on cost-effectiveness grounds alone. This situation contrasts sharply with the hypothetical situation noted above, in which neonatal intensive care resulted in increased survival but was not cost effective because it uniformly led to defective individuals. In that hypothetical situationwhich it appears we will not have to confront after all in the real worldthe tradeoff was not between normal and abnormal survivors, but between fewer or greater numbers of defective survivors. Neither situation would be financially cost effective for society, but the factors to be weighed are quite different. Neonatal intensive care for the group of infants weighing 1,500 g or less, although arguably cost effective, raises equally complex policy questions. The proportion of normal survivors in this group increased somewhat between 1960 and 1971 to 1975, while the ratio of normal to severely abnormal survivors increased dramatically (see figure 6). However, the data suggest that there may have been an increase in the absolute number of survivors with minor abnormalities. These survivors are productive individuals but are afflicted with minor degrees of handicap or illness. Thus, even though it might be cost effective to provide newborn care which results in this kind of outcome, the analysis is less than satisfying, because there is no adequate measure of the quality of life these individuals would attain. Our analysis deals only with the question of whether neonatal intensive care is cost effective when compared with less intensive care of ill newborns. It does not address the larger question of whether such care is cost effective when compared with alternative programs to reduce the levels of prematurity and other risk factors in the population. That separate and even more complex question would require a thorough analysis of the costs and effectiveness of socioeconomic initiatives and prenatal medical care, followed by a comparison with the costs and outcomes of neonatal intensive care.
PAGE 51
Case Study #10: The Costs andd Effectiveness of Neonatal Intensive Care l 43 FEDERAL POLICIES RELATED TO NEONATAL INTENSIVE CARE 12 Should the United States expand or limit the technology of neonatal intensive care? What Federal, State, and local policies encourage or inhibit rational planning of neonatal intensive care? These broad policy questions are being asked by a growing number of decisionmakers across the country. Public policies to date have emphasized both cost-based and charge-based reimbursement as well as regionalization of neonatal and obstetric services. Areawide planning assumes that the resources are scarce and should be allocated in the most efficient manner. With the present mix of reimbursement and regulatory controls, however, this policy may not be enforceable. A major Federal policy regarding neonatal intensive care today is the National Guidelines for Health Planning promulgated under the National Health Planning and Resources Development Act of 1974 (Public Law 93-641). The neonatal resource standards set in the guidelines address regionalization by establishing a maximum number of beds per live births (4 beds per 1,000 births) and a minimum size of special care units (15 beds) (see table 24). These standards were largely based on the 1976 report of the Committee on Perinatal Health (32), the 1977 report of the American Academy of Pediatrics (4), and a review of Health Systems Plans. Our analysis of utilization and supply data strongly suggests that closer examination of these figures by DHHS is needed. The present standard of 4 beds per 1,000 live births does not reflect adequate population-based need, only crude upper limits. Our best estimate is that only about half that number, or approximately 2.3 neonatal intensive care beds per 1,000 live births, probably are present in the United States today. The combination of a generous per-birth bed figure and the 15-bed minimum unit size may result in an incentive toward marked oversupply. To meet the 15-bed minimum standard, many hospitals would have to choose among closing, consolidating, or expanding their units. With most of the United States well below the 4 bed per 1,000 live births minimum, hospitals in most health service areas could justify enlarging their units up to 15 beds. It would be very difficult for health systems agencies (HSAs) to deny requests to enlarge NICUs in areas that are well under the 4-bed maximum. Since at least 6 percent of births are probably receiving intensive care at the current bed level, a major increase in NICU supply would very likely represent excess capacity and promote increased utilization. This result would contradict the intended effect of the Federal policy behind the neonatal resource standard. Medicaid funding, as described earlier, 13 i s already exerting relatively tight controls over reimbursement of neonatal intensive care. By reimbursing for costs and not for charges, medicaid programs put a lid on unit expenditures for neonatal intensive care. The extensive crosssubsidization that results, however, suggests that further examination of medicaid reimbursement practices is needed. Certain State medicaid limitations on reimbursement of neonatal intensive care create additional problems in providing care to newborns in low-income families. Medicaid reimbursement policies in Missouri, for example, cover only the first 21 days of newborn intensive care (48). Medicaid insurance benefits for teenage and unwed mothers further exacerbate reimbursement problems. In January 1978, Butterfield (20) surveyed 30 States to determine the extent of their authorizations or State funding for perinatal services for the medically indigent. H e found that only about half of the States provided coverage for perinatal care for the medically indigent. Those States with authorizations ranged from a high of $6 million in California to a low of $400,000 in Louisiana. The most common response was that there were no statutes providing State medical assistance for perinatal ~w~ltl~~ in OCtO~er IQ7Q. 1See sectl~ln ab~~ve {~n reimbursement l(~r neonatal Intensiv e care.
PAGE 52
44 Background Paper #2: Case Studies of Medical Technologies Table 24.National Guidelines for Health Planning: Obstetrical and Neonatal Resource Standards Standard-specific Definitional Standard Purpose adjustment considerations Special considerations Obstetrical services Planning on a Developing regional Level / unitprovides Special moral and ethiregional basis with linkages among OB and neonatal services. l 1,500 births per year in Level II and Ill units. l 75% occupancy in units with greater than 1,500 births per year (Levels II and Ill). systems of care for services for uncomplical preferences may maternal and perinatal cated maternity and necessitate adjusthealth services newborn cases. ments. reduces maternal, Level II unitprovides The degree of complex. fetal, and neonatal services for uncompliity of patient needs mortality rates and imcated cases, the majorshould determine proves development of ity of complicated scarce resources. problems, and specialized neonatal services. Level /// unitprovides services for all serious illness and abnormalities. where, and by whom, the care should be provided. Established arrangements should provide for early access of high-risk cases and prompt referral among levels of care. In keeping with the national priority, consolidation of multiple, small OB units with low occupancy rates should be undertaken unless such action is undesirable because of needs to assure access and sensitive care. Neonatal special care units l l l Planning on a The regionalized apAreas with unusually regional basis with preach reduces mortalhigh rates of high-risk linkages with obstetriity rates and improves pregnancy allows for cal services. the development of adjustment upward. No more than 4 beds/ scarce resources. For a Level II unit, 1,000 live births in a where travel time due defined neonatal servto geographic remoteice area. ness is a hardship A minimum of 15 allows for adjustment beds for Level II or Ill downward. units. Level I unitprovides recovery care. Level // unitprovides intermediate and recovery care and some specialized services. Level III unitprovides intermediate, recovery, and intensive care. Bed includes heated units and bassinettes. Unit = a distinct and separate physical facility in an institution Because centers often serve a patient load of more than 1 million, a defined neonatal service area should be identified by the relevant HSAs and State agency. Adequate communication and transportation systems including joint transfers of mother and child and maintenance of family contact should be developed. HSA = health systems agency SOURCE National Guidelines for Health Plannning, 42 CFR, sees 121203 and 121204, Health Resources Administration. DHEW, Mar 28, 1978 care for the medically indigent and that no such ensuring that (1) Medicaid reimbursement legislation was under consideration. rates for obstetrical and well-baby care are sufficient to encourage private providers to accept These medicaid policies result in the refusal of Medicaid patients, (2) more low-income women some physicians and hospitals to accept medicbecome eligible for Medicaid, and (3) States inaid and low-income patients. To alleviate presclude coverage of at least HEW-specified minient inequities and to reduce infant mortality, the mum prenatal care services under their Medicaid General Accounting Office has made the followprogram. ing recommendations for changing medicaid Maternal and child health services and crippolicies (48): pled childrens services operating under title V
PAGE 53
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care l 45 of the Social Security Act have encouraged regionalization and improved access to newborn intensive care. Federal funding for maternal and child health services under title V significantly increased from $85 million in 1974 to $206 million in 1977. Distribution of title V funds to States is based on a uniform allotment of $70,000 plus additional funding according to the number of live births. These funds must be matched by the State. The remaining half of maternal and child health funds goes for: 1) programs for mentally retarded children, 2) special programs of national or State importance, and 3) formula grants based on number of live births and per capita income (48). In order to receive title V funds, each State submits a Maternal and Child Health Plan that describes the States needs and programs in five areas: 1) family planning, 2) maternity and infant intensive care, 3) women and infant care nutrition services, 4) children and youth, and 5) childrens dental health (see table 25 for the proportion of Federal maternal and child health formula grant funds spent on newborn intensive care). Title V also authorizes funds for crippled childrens services. In some States, such as California, these funds have been used to cover intensive care costs. Unfortunately, no information exists on the prevalence of this practice. However, 37 out of 55 State Health Authorities reported purchasing inpatient care for mothers, infants, and children from the maternal and child health services program authorities (10). Since payments for neonatal intensive care are not reported as a separate category, it is impossible to collect total neonatal intensive care expenditures from the maternal and child health services program. The improved pregnancy outcome (IPO) program, which began in 1976 as part of an overall HEW child health strategy, is a recent addition to title V programs. The IPO program was established to improve maternal care and pregnancy outcomes in States with high infant mortality rates. A requirement for funding is that a regionalized concept of prenatal and perinatal care be included in State plans. Funds may then be used to provide secondary and tertiary care referral systems, outreach systems, transport, provision of basic maternity care, identification of high risk pregnancies and infants, and outreach programs for pregnant teenagers. According to the recent General Accounting Office report on Federal efforts to improve pregnancy outcome, HEW awarded $9 million in IPO grants to 22 States, the District of Columbia, and Puerto Rico for 1978 (48). Each IPO State grant can total up to $400,000 annually for up to 5 years. Another DHHS initiative to reduce infant morbidity and mortality is the improved child health program. In 1978, $3 million of maternal and child health funds and $1 million in title X family planning funds and National Health Service Corps personnel were made available to some 31 high-risk areas in 11 States. This money (up to $300,000 annually for up to 4 years) is to be used for coordinating comprehensive care to high-risk mothers and infants in areas with excessive morbidity and mortality (48). Although DHHSs support for improving pregnancy outcome has increased significantly in the last 10 years, many complain that the programs are still grossly underfunded and fragmented. Some complain, for example, that money comes through too many different sources, sometimes bypassing State and local maternal and child health agencies. Others question whether the money is actually going to areas with the greatest need for reducing infant mortality (48). Planning efforts by State and local maternal and child health agencies, crippled childrens services agencies, and health planning agencies are increasingly overlapping. Each of the three agencies in each State is responsible for identifying infant health status needs, available services, and program alternatives to meet unmet needs. In some States, the three agencies have worked closely together to ensure consistent maternal and child health plans. For the most part, however, such coordinated planning is lacking, Whether this problem can be solved through informal, State-by-State means of communication or through specific mandates for integrated and consistent plans is a question that needs to be further addressed by both State and Federal agencies.
PAGE 54
46 l Background Paper #2: Case Studies of Medical Technologies Table 25.Budgeted Use of Federal Maternal and Child Health (MCH) Formula Grant Funds for Infant Intensive Care, by State, Fiscal Year 1978 Funds for infant intensive care Total MCH formula Funds for program of Percent of funds for State grant funds projects a Amount program of proiects a Alabama. . . . $ 4,807,268 $ 3,119,125 $ 200,000 Alaska . . . 380,200 273,700 67,000 Arizona. . . . 3,180,901 1,174,315 103,611 Arkansas . . . 3,080,400 1,421,300 221,000 California. . . 11,254,203 5,516,191 372,683 Colorado . . . 3,554,659 2,791,797 92,502 Connecticut . . 1,674,012 1,050,639 51,610 Delaware ... . . 869,282 329,900 50,000 District of Columbia. 5,650,431 4,006,810 Florida. . . . 6,297,410 4,836,598 100,000 60/0 24 9 16 7 3 5 15 2 Georgia . . . 6,476,053 3,897,400 884,196 Hawaii . . . 1,087,320 937,112 25,000 Idaho . . . ?,181 ,847 796,992 70,000 Illinois b . . . Indiana. . . . 3,766,399 1,521,256 266,459 lowa . . . . 2,969,480 1,723,511 192,532 Kansas. . . . 1,838,237 1,014,343 122,061 Kentucky. . . 4,860,198 1,669,946 356,475 Louisiana. . . 5,606,425 380,631 147,402 Maine . . . 1,447,000 697,000 115,000 23 3 9 18 11 12 21 39 17 Maryland . . . 6,924,932 5,909,932 Massachusetts . 4,793,097 3,717,338 Michigan . . . 7,586,233 5,347,400 Minnesota. . . 3,036,262 2,612,594 Mississippi . . 4,456,906 1,159,473 18,000 72,324 35,132 179,312 Missouri. . . . 4,397,076 2,280,276 Montana. . . . 1,267,960 494,222 Nebraska. . . 1,860,220 1,432,880 Nevada. . . . 429,721 205,744 New Hampshire . 759,000 659,000 98,985 20,000 61,477 10,657 55,000 0.3 2 1 15 4 4 4 5 8 New Jersey . . 3,475,146 1,415,685 New Mexico. . . 1,299,429 810,179 66,583 8 New York. . . . 19,230,467 14,420,828 212,486 1 North Carolina . . 6,638,695 1,074,802 North Dakota. . . 644,900 194,000 16,000 8 Ohio . . . . 7,960,502 3,996,444 184,539 5 Oklahoma . . . 2,602,000 1,358,891 15,000 1 Oregon. . . . 2,571,356 998,268 275,000 28 Pennsylvania. . . 9,292,751 5,679,083 119,740 2 Rhode Island. . . 673,183 220,000 South Carolina . . 4,985,462 1,414,871 150,125 11 South Dakota . . 654,375 490,251 187,145 38 Tennessee. . . 4,707,280 1,602,250 197,240 12 Texas . . . . 11,620,397 4,651,632 153,000 3 Utah . . . . 1,684,592 1,109,266 281,457 25 Vermont. . . . 423,927 423,927 36,542 9 Virginia . . . 4,806,000 2,211,711 97,777 4 Washington. . . 2,840,000 1,358,000 212,000 16 West Virginia. . . 2,965,558 1,202,085 160,000 13 Wisconsin . . . 5,309,900 2,364,900 768,500 33 Wyoming. . . . 472,876 106,500 Total. . . . $200,351,998 $108,083,998 $7,121,552 1170 aSubtotal of MCH formula grant funds allocated to programs in the following five areas. 1 ) family planning, 2) maternity and infant intensive care, 3) women and infant care nutrition services, 4) chidren and youth, and 5) childrens dental health blllinois has not returned questionnaire SOURCE General Accounting Office, U S Congress, Better Management and More Resources Needed To Strengthen Efforts To Improve Pregnancy Outcome, HRD. 80-24 (Washington, D C GAO, Jan 21, 1980)
PAGE 55
Case Study #10: The Costs and Effectiveness of Neonatal lntensive Care l 47 REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Akehurst, R. L., and Holtermann, S., Application of Cost-Benefit Analysis to Programmed for the Prevention of Mental Handicap, in Major Mentul FItmdicap: Methods and Costs of Prevention, Ciba Foundation Symposium 59 (Amsterdam: Elsevier Scientific Publishing Co., 1978). Alden, E. R., et al., Morbidity and Mortality of Infants Weighing Less Than 1,000 Grams in an Intensive Care Nursery, Pediatrics 50:40, 1972. Allan, G., et al., Report of the Cornrnittee on Infant Zntensive Cure (Harrisburg, Pa.: Commonwealth of Pennsylvania, Governors Health Task Force, 1974). American Academy of Pediatrics, Committee on Fetus and Newborn, Standards and Recommendations for Hospital Care of Newborn Infants, 2974-2977, 6th ed. (Evanston, Ill.: AAP, 1977). ,Committee on Fetus and Newborn and the Section on Perinatal Pediatrics, Estimates of Need and Recommendations for Personnel in Neonatal Pediatrics, Pediatrics 65:850, 1980. American Board of Pediatrics, Chapel Hill, N. C., personal communication, Apr. 23, 1980, American Hospital Association, Hospital Statistics (Chicago: AHA, 1977). Hospital Statistics (Chicago: AHA, 1978). Hospital Statistics (Chicago: AHA, 1979). Association of State and Territorial Health Officials, Services, Expenditures and Programs of State and Territorial Health Agencies NPHPRS publication No. 42 (Silver Spring, Md.: ASTHO, May 1979). Banta, H. D., and Thacker, S., Costs and Benefits of Electronic Fetal Monitoring: A Review of tlze I.iterature, HEW publication No. (PHS) 79-3245 (Hyattsville, Md.: National Center for Health Services Research, 1979). Barnett, H. L. (cd.), Pediatrics (New York: Appleton-Century-Crof ts, Meredith Corp., 1972). Basler, D. S., and Hager, D. E,, Z%nning and Design for Perinatal and Pediatric Facilities (Columbus, Ohio: R OSS Laboratories, 1977). Ross Planning Associates, unpublished study performed in conjunction with Guide to Referral Centers Providing Perinatal and Neo15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. natal Care (Columbus, Ohio: Ross Laboratories, 1979). Beauchamp, T. L., Macroallocation and Policy Decisions, lecture presented at the Conference on Life Prolonging Technology and Public Policy, Washington, D. C., Oct. 20, 1977. Belenky, D. A., et al., IS Continuous Transpulmonary Pressure Better Than Conventional Respiratory Management of Hyaline Membrane Disease? A Controlled Study, Pediatrics 58:800, 1976. Bhat, R., et al., Immediate and Long-Term Outcome of Infants Less Than 1,000 Grams, Crit, Care Med. 6:147, 1978. BIake, A., et al., Perinatal Intensive Care, ]. Psychosom. Res. 21:261, 1977. Budetti, P., Social Considerations of Health Policy for Perinatal Care, speech delivered at Symposium on Progress and Problems in Perinatal Medicine, sponsored by the March of Dimes and University of California, San Francisco, Calif., Sept. 22, 1978. Butterfield, L. J., The Childrens Hospital, Denver, Colo., unpublished data from survey of Maternal and Child Health Directors, January 1978. Calame, A., et al., Psychological and Neurodevelopmental Outcome of High Risk Newborn Infants, Helv. Paed. A. 31:287, 1976. Calabresi, G. Tragic Choices (New Haven, Corm.: Yale University Press, 1977). California Administrative Code, title 22, Health Facilities and Referral Agencies, register 75 (No. 24), sec. 70481Intensive Care Newborn Nursery Service Definition, June 14, 1975. California State Department of Health Services, Maternal and Infant Health Section, Berkeley, Calif., California perinatal data, 1977. Office of Statewide Health Planning and D&elopment, Sacramento, Calif., Annual Report of Hospitals, 1977, revised March 1979. Chance, G. W., et al., Neonatal Transport: A Controlled Study of Skilled Assistance, ]. Pediat. 93:662, 1978. Chapalain, M. T., Perinatality: French CostBenefit Studies and Decisions on Handicap and Prevention, in Major Mental Handicap: Methods and Costs of Preventiotl, Ciba Foundation Symposium 59 (Amsterdam: Elsevier Scientific Publishing Co., 1978).
PAGE 56
48 l Background Paper #2: Case Studies of Medical Technologies 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. Chez, R. A., et al., High Risk Pregnancies: Obstetric and Perinatal Factors, in Prevention Of Ernbryoruc, Fetal and Perinatal Disease, R. L. Brent and M. I. Harris (eds. ), Fogarty International %ries on Preventive Medicine, vol. 3, publication No. (NIH) 76-853 (Bethesda, Md.: National Institutes of Health, 1976). Childrens Hospitals Automated Medical Programs, Columbus, Ohio, unpublished data for 1975-77. Childrens Hospital Medical Center of Northern California, report series A, fiscal period Jan. 1, 1976, to Dec. 31, 1976 (Oakland, Calif., Apr. 7, 1978) Childress, J., Risk, in Encyclopedia of Bioethics, vol. 4 ~New York: Free Press, 1978). Committee on Perinatal Health, Toward Zrnproving the Outcome of Pregnancy: Recommendations for the Regional Development o f Maternal and Perinatal Health Services (White Plains, N. Y.: National Foundation of the March of Dimes, 1977), Cotton, R., Nashville, Term., Fiscal Performance of Vanderbilt NICU, unpublished mimeo, 1976. CT Head Scans Urged for Low Weight Premies, Med. World News 20:13, 1979. Dann, M., et al., A Long-Term Follow-Up Study of Small Premature Infants, Pediatrics 33:945, 1964. The Development of Prematurely Born Children With Birth Weights or Minimal Postnatal Weights of 1,000 Grams or Less, Pediatrics 22: 1958 Drillien, C. M., Aetiology and Outcome in Low-Birthwei,~ht Infants, Develop. Med Child. Neuro. 14:563, 1972. The Incidence of Mental and Physical Handicaps in School-Age Children of Very Low Birth Weight, Pediatrics 27:452, 1961. Duff, R. S., and Campbell, A. G. M., Moral and Ethical Dilemmas in the Special Care Nursery, N. Eng, ,1. Med. 289:890, 1973. Dweck, H. S et al., Early Development of the Tiny Premature Infant, Am. ]. Dis. Child. 126:28, 1973. Eisner, V., et al., Improvement in Infant and Perinatal Mortality in the United States, 1965-1973, Am. J. Public Health 68;359, 1978. The Risk of Low Birthweight, Am. ], Public Health 69:887, 1979. Ferrara, A,, Planning Specialized Neonatal Beds in an Urban Regionalized Perinatal Center, in A Review of Planning Methods and 44. 45. 46. 47. 48. 49. 50. 51. 52. 53. 54. 55. .56. 57. Criteria for NZCL.JS, vol. II, HRA contract No. 231-77-0108, study prepared by the Boston University Center for Health Planning for the Health Resources Administration, Hyattsville, Md., 1979. Fitzhardinge, P, M., Follow-Up Studies on the Low Birth Weight Infant, Clin. Per. 3:5o3, 1976. Fitzhardinge, P. M., et al., Mechanical Ventilation of Infants of Less Than 1,.501 gm Birth Weight: Health, Growth and Necrologic Sequelae, }. Pediatrics 88:531, 1976. Fletcher, J., Pediatric Euthanasia, Decision Making and the Defective Newborn, C. A Swinyard (cd. ) (Springfield, Ill.: Charles C. Thomas, 1978). Francis-Williams, J., and Davies, P. A., Very Low Birthweight and Later Intelligence, Develop. Med. Child. Neuro. 16:709, 1974. General Accounting Office, U.S. Congress, Better Management and More Resources Needed To Strengthen Federal Efforts To improve Pregnancy Outcome, HRD-80-24 (Washington, D. C.: GAO, Jan. 21, 1980). Gordon, R. R., Neonatal and Perinatal Mortality Rates by Birth Weight, Br. Med. }. 2:1202, 1977, Gortmaker, S. L., The Effects of Prenatal Care Upon the Health of the Newborn, Am. ]. Public Health 69:653, 1979. Grassy, R. G., et al., The Growth and Development of Low Birth Weight Infants Receiving Intensive Neonatal Care, Clin, Pediat. 15:549, 1976. Hatt, J. J., et al., L~levage et lAvenir des Primatur
PAGE 57
58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. Hemmers, M., and Kendall, A. C., The Prognosis of the Very Low-Birthweight Infant, Develop, Med. Child. Neurol. 18:745, 1976. Hughes-Davies, T. H., Conservative Care of the Newborn Baby, Arch. Dis. Child. 54:59, 1979. Jackson, N. S., Division of Neonatal Biology, Department of Pediatrics, University of Washington, Seattle, Wash., unpublished data, 1979. Jones, R. A., et al., Infants of Very Low Birthweight, Lancet 1 :1332, 1979. Jensen, A. R., and Garland, M., ~tlzics oflVewborn Zntensive care (Berkeley, Calif.: Uni versity of California, Institute of Governmental Studies, 1976). Jensen, A. R., and Lister, G., Newborn Intensive Care: The Ethical Problems, FZast, Cent. Rpt. 8:15, 1978. Kaufman, S. L., and Shepard, D. S., Actual and Model Budgets for Perinatal Centers, in A Review of Planning Methods for Neonatal intensive Care Units VOI. 11, HRA contract No. 231-77-0108, study prepared by the Boston University Center for Health Planning for the Health Resources Administration, Hyattsville, Md., 1979. Cost by Day of Neonatal Intensive Care, paper presented at the 108th Annual Meeting of the American Public Health Association, Detroit, Mich. Oct. 19-23, 1980. Reimbursement of Neonatal Intensive Care: A Descriptive Overview, in A Review of Planning Metkods for Neonatal Intensive Care Units, vol II, HRA contract No 231-77-0108, study prepared by the Boston University Center for Health Planning for the Health Resources Administration, Hyattsville, Md., 1979. Kitchen, W., and Campbell, D., Controlled Trial of Intensive Care for Very Low Birthweight Infants, Pediatrics 48:711, 1971. Kitchen, W. H., et al., A Longitudinal Study of Very Low-Birthweight InfantsI. Study Design and Mortality Rates, Develop, Med. Child. Neurol. 20:605, 1978. Kleinman, J. C., et al., A Comparison of 1960 and 1973-1974 Early Neonatal Mortality in Selected States, Am. ]. Epidern. 108:454, 1978. Kramer, M., Ethical Issues in Neonatal Intensive Care: An Economic Perspective, in Ethics of Newborn of Zntensizle Car-e, A. R. Jensen and M. Garland (eds. ) (Berkeley, Calif.: University of California, Institute of Governmental Studies, 1976). 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. Lee, K. S., et al., Determinants of the Neonatal Mortality, Am. ], Dis. Child. 130:842, 1976. Neonatal Mortality: An Analysis of the Recent Improvement in the United States, Am. J. Public Health 70:15, 1980. The Very Low Birthweight Rate: Principal P;edictor of Neonatal Mortality in Industrialized Populations, J. Pediat, 97:759, 1980. Lubchenco, L., et al., Long-Term Follow-Up Studies of Prematurely Born Infants: Influence of Birthweight and Gestational Age on Sequelae, }. Pediat. 80:509, 1972. Long-Term Follow-Up Studies of Premature~y Born Infants: Relationship of Handicaps to Nursery Routines, ]. Pediat. 80:501, 1972. Machtinger, Y., United Hospital Fund of New York, New York City, unpublished data for 1977. Maisels, M. J., et al., Elective Delivery of the Term Fetus: An Obstetrical Hazard, /. A.M.A. 238:2036, 1977. Maryland State Advisory Committee on Perinatal Care, Standards of Care in Maryland Nursery Units (Baltimore: Department of Health and Mental Hygiene, 1978). Mathis, E., Chief, Health Facilities Statistics Branch, National Center for Health Statistics, Hyattsville, Md., personal communication 1979. McCarthy, J. T., et al., Who Pays the Bill for Neonatal Intensive Care? ]. Pediat. 95:755, 1979. McDonald, A. D., Cerebral Palsy in Children of Very Low Birthweight, Arch. Dis. Child. 38:579, 1963. McManus, M. A., and Butterfield, L. J., Health Resources Administration, Hyattsville, Md., Report on Local and State Practices Respecting Obstetrical and Neonatal Services: An Analysis of the National Guidelines for Health Planning, unpublished manuscript, Aug. 25, 1978. Meier, R. C., Seattle, Wash., Costs for Treating HMD Infants in the Neonatal Intensive Care Unit of the University of Washington Hospital, unpublished manuscript, 1978. Morrison, I., The Elderly Primigravida, Amer. ]. Obst, G. 121:465, 1975. National Center for Health Statistics, Vital Statistics of the United States, 1967 (Hyattsville, Md.: NCHS, 1968).
PAGE 58
50 Background Paper #2: Case Studies of Medical Technologies 86. 87. 88. 89. 90. 91. 92. 93. 94. 95. 96. 97. Monthly Vital Statistics Report, Nata/it~ Statistics, Teenage Childbearing: United States, 1966-75, Natality Statistics, HEW publication No. (PHS) 77-1120, vol. 26, No. 5 (suppl. ) (Hyattsville, Md.: NCHS, 1977). Viial Statistics of the United States, 1976 (Hyattwille, Md.: NCHS, 1977). Vital Statistics of the United States 197.5, Vol. Z, Natality, HEW publication No. (pHS) 78-1113 (Hyattsville, Md.: NCHS, 1978). Prenatal Care United States, 1969-7S, in Vital and Health Statistics, series 21, No. 33, HEW publication No. (PHS) 78-1911 (Hyattsville, Md.: NCHS, 1978) Vital Statistics of the United States 2975, Vol. 1/, Mortality, Part A, HEW publication No. (.PHS) 79-1114 (Hyattsville, Md.: NCHS, 1979). Monthly Vital Statistics Report, Advance Report, Final Mortality Statistics, 1977, HEW publication No. (PHS) 79-1120, vol. 28, No. 1, Ma> 1979 (Hyattsville, Md.: NCHS, 1979). Monthly Vital Statistics Report, Advance Report, Final Natality Statistics, 1978, HEW publication No. (PHS) 80-1120, vol. 29, No. 1, (suppl .), Apr. 28, 1980 (Hyattsville, Md,: NCHS, 1980), Monthly Vital Statistics Report, Provi, sional Statistics, Annual Summary for th e United States, 1979, DHHS publication No. (PHS) 81-1120, VO1. 28, No. 13, Nov. 13, 1980 (Hyattsville, Md.: NCHS, 1980). Neonatal Intensive Care Standards and Measures: Report of the State and Regional Task Force on Neonatal Intensive Care and the Technical Adviscry Group of the Perinatal Welfare Committee (Boston: Office of State Health Planning, Massachusetts Department of Public Health, 1976, revised March 1979). Pape, K., et al., Status at Two Years of LowBirth-Weigh! Infants Born in 1974 With Birth Weights of Less Than 1,001 gin, ]. Pediat. 92: 253, 1978. Parmelee, A. H., et al., Neurological Evaluation of the Premature Infant: A Follow-Up Study, Bio. Neonat. 15:65, 1970. Pennsylvani,~ Department of Health, Health Data Center, Harrisburg, Pa., unpublished data on special nursery care units available in Pennsylvania hospitals, July 1, 1977-June 30, 1978. 98. 99. 100. 101. 102. 103. 104. 105. 106. 107. 108. 109. 110. 111. 112. 113. Phibbs, C. S., et al. Analysis of Factors Associated With Costs of Neonatal Intensive Care, Pediat. Res, 14:438, 1980 (abstract) ; complete article forthcoming, Newborn Risk Factors and the Costs of Neonatal Intensive Care, Pediatrics (in press), 1981. Phibbs, R. H., University of California, San Francisco, Calif., unpublished data, 1979. Placek, P. J., and Taffel, S. M., Trends in Cesarean Section Rates for the United States, 1970-78. Public Health Reports 95:540, 1980. Pomerance, J. J., Neonatal Intensive Care UnitBasic Equipment Needs: Symposium on Organization for Perinatal Care, C/in. Per. 3:353, 1976. Pomerance, J. J., et al., Cost of Living for Infants Weighing 1,000 Grams or Less at Birth, Pediatrics 61:908, 1978. Pomerantz, L., Division of Medical Facilities, Ohio State Department of Health, Columbus, Ohio, unpublished data for 1977. Ramsay, P., Ethics at the Edges of Life (New Haven, Corm.: Yale University Press, 1978). Rawlings, J., et al., Changing Prognosis for Infants of Very Low Birth Weight, L.ancet 1:516, 1971. Reed, D. M., et al., The Epidemiology of Respiratory Distress Syndrome in Norway, Am. J. Epid. 107:299, 1978. Robertson, N. R. C., letter, Lancet 2:362, 1979. Ross Laboratories, Zatrogenic Problems in Neonatal Intensive Care (Columbus, Ohio, 1976). Planning and Design for Perinatal and Pediatric Facilities (Columbus, Ohio, 1977). Sabel, K. G., et al., Remaining Sequelae With Modern Perinatal Care, Pediatrics 57:652, 1976. Schlesinger, E. R., Neonatal Intensive Care: Planning for Services and Outcomes Following Care, J. Pediat. 82:916, 1973. Schroeder, S. A., et al., Frequency and Clinical Descriptions of High-Cost Patients in 17 Acute-Care Hospitals, N. Eng, ]. Med. 300:1306, 1979. Science Management Corp., Report on State Practices, Part 11: State Practices Regarding OB and Neonatal Definitions and Resource Requirements, contract report No. HRA-23278-0131, report to the Health Resources Administration, Hyattsville, Md., 1979.
PAGE 59
Case Study #10: The Costs and Effectiveness of Neonatal Intensive Care l 51 114. 115. 116. 117. 118. 119. 120. 121. 122. 123. Shannon, D. C., et al., Boston, Mass., Survival, Cost of Hospitalization and Prognosis in Infants Critically 111 With Respiratory Distress Syndrome: Cost of Intensive Care in Newborns, unpublished manuscript, 1979. Shapiro, S., et al., Relevance of Correlates of Infant Deaths for Significant Morbidity at 1 Year of Age, Am, j. Obst. G. 136:363, 1980. Shepard, D. S., and Ghanotakis, A. J., Boston, Mass., Hospital Costs in Massachusetts, mimeo, June 1979. Silverman, W. A., Incubator-Baby Sideshows, Pediatrics 64:127, 1979. Stewart, A., The Survival of Low Birthweight Infants, Br. ]. Hosp. Med. 18:182, 1977. Stewart, A. L. et al., Prognosis for Infants Weighing 1,000 g or Less at Birth, Arch. Dis. Child. 52:97, 1977. Swinyard, C. A. (cd.), Decision Making and the Defective Newborn (Springfield, Ill.: Charles C. Thomas, 1978). Swyer, P. R., and Goodwin, J. W. (eds. ), Regional Services in Reproductiz?e Medicin e (Toronto: The Joint Committee of the Society of Obstetricians and Gynecologists of Canada and the Canadian Pediatric Society, 1973). Special Care Units for Newborns: Report of the Joint Committee of the Society of Obstetricians and Gynecologists of Canada and the Canadian Pediatric Society on the Regionalization of Reproductive Care in Canada, in Regional Services in Reproductive Medicine (Toronto: The Joint Committee of the Society of Obstetricians and Gynecologists of Canada and the Canadian Pediatric Society, 1973). Taffel, S., Factors Associated With Low Birthweight, United States, 1976, Series 21, No. 37 (Hyattsville, Md.: (National Center for Health Statistics, April 1980). 124. 125. 126. 127. 128. 129. 130. 131. 132. 133. 134. 135. 136. Teberg, A., et al., Recent Improvement in Outcome for the Small Premature Infant, C/in. Pediat. 16:307, 1977. Thompson, T., and Reynolds, J., The Results of Intensive Care Therapy for Neonates, Per. Med. 5:59, 1977. Thompson, R. E., et al., Impact of Intensive Newborn Care on Neonatal Mortality in a Community Hospital, South. Med. 1. 69:688, 1976. Usher, R,, Changing Mortality Rates With Perinatal Intensive Care and Regionalization, Sern. Perinat. 1:309, 1977. Williams, R. L., Appendix to Intrauterin e Growth and Neonatal Risk in California G. C. Cunningham, et al. (eds. ) (Santa Barbara, Calif.: University of California, 1976). Measuring the Effectiveness of Perinatal Medical Care, Med. Care, 17:95, 1979. University of California, Santa Barbara, Calif., unpublished data, 1979. Williams, R. L., and Hawes, W., Cesarean Section, Fetal Monitoring, and Perinatal Mortality in California, Am. J. Public Health 69:864, 1979. Wright, F. H., et al., A Co~trolled Follow-Up Study of Small Premature Born From 1952 Through 1956, Am. ]. Dis, Child. 124:506, 1972. Wung, J. T., et al., CDP: A Major Breakthrough! Pediatrics 58:783, 1976. Zachman, R. D., and Graven, S. N., A Neonatal Intensive Care Unit: A Four-Year Summary, Am. J. Dis. Child. 128:165, 1974. Brown, E. R., and Taeusch, H. W., letter, Lancet 2:362, 1979. McCormick, M. C., et al., Rehospitalization in the First Year of Life for High Risk Survivors, Pediatrics 66:991, 1980. 0 U.S. GOVERNMENT PRINTING OFFICE 1981 341-844/1016
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EN92EYLFR_7I239Y INGEST_TIME 2017-05-18T17:31:42Z PACKAGE AA00055059_00015
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES
|