Health Services Administration. Cornell University. Ithaca, New York. REFERENCES. 1. Stone, D. A.Diagnosis and the Dole: The Functions of Illness in.
278
Health Services Research 17:3 (Fall 1982)
explanatory variables and levers for reform. The flaws in this book are minor, and I highly recommend it to those interested in the politics of health, the sociology of professions, health care cost containment, and comparative health systems. The book is an insightful and absorbing addition to the literature on a health system that has important lessons for our own. JAMES W. BEGUN Sloan Program of Hospital and Health Services Administration Cornell University Ithaca, New York
REFERENCES 1. Stone, D. A. Diagnosis and the Dole: The Functions of Illness in American Distributive Politics. Journal of Health Politics, Policy and Law 4:507, Fall 1979. 2. Battistella, R. M. and S. R. Eastaugh. Hospital Cost Containment: the Hidden Perils of Regulation. Bulletin of the New York Academy of Medicine 56:62, January-February 1980. 3. Battistella, R. M. and S. R. Eastaugh. Hospital Cost Containment. Proceedings of the Academy of Political Science 33, 4:192, 1980.
ON CHILD HEALTH SERVICES RESEARCH A Resource Allocation Model for Child Survival. Howard Barnum, Robin Barlow, Luis Fajardo, and Alberto Pradilla. Cambridge, MA: Olgeschlager, Gunn & Hain, Inc., 1980. xviii + 190 pp. $30.00. A Resource Allocation Model for Child Survival is a book which reports on a project undertaken for the United States Agency for International Development, Office of Health, by the Center for Research on Economic Development at the University of Michigan. The reader should be forewarned that the prose is that of an academic research group's report and does not make for lively reading for the casually interested. Large sections of the book are technical, highly mathematical appendices, describing in detail the methodology of the study. For those who are interested there is even a complete 16-page listing of the computer program used for the simulation model.
Book Reviews
279
The basic approach is conceptually simple. The authors' desire to describe the relationships between the population and environmental characteristics (including levels of health improvement activities) of a community and its morbidity rates. They attempt to specify mortality rates for young children resulting from the morbidity levels due to all important health problems. Reduction in deaths due to changes in health-enhancing activities can thus be predicted. The authors can then add information on predicted costs of each health-aiding activity and solve by computer search techniques for the most cost-effective ways to reduce death rates. Essentially, adding practical detail to the conceptual framework is what concerns the authors in the book. The book develops a non-linear optimization model designed to provide policymakers with a tool for analyzing alternative allocations of resources to health services in less developed countries. The approach is valuable (the "correct" one from most economists' point of view) because the competing programs for reduction of child health problems are considered in the context of binding constraints on various of the resources needed by such programs. The methodology is essentially a sophisticated way of helping policymakers achieve a situation in which additional scarce resources are always allocated to the use that provides the greatest increase in the society's desired output. Obviously, one problem with such an approach, and with this group's specific exercise, is that all possible outputs of alternative resource-using programs must be measured in comparable units. The authors do not choose to use the familiar output measure of the dollar value of society's willingness to pay for what is produced. Instead, they use reduction of deaths in children before the age of 60 months as the desired output measure. While reduction of deaths is certainly a reasonable goal, it is not obvious that a set of policies designed to minimize deaths for the level of resources used, is best for the society. There might, for example, be alternative equal-cost policies that allow slightly more death, but reduce suffering and physical impairment by a much greater degree. The question never dealt with is "How can varying outputs of health improvement programs be compared?" An obviously important specific variant on that question is, "Does reduced infant mortality as an objective lead to the best choices of health programs for children?" A second, positive aspect of the authors' approach (besides the constrained maximization objective) is that it strongly makes the points to policymakers that all good things cannot be done, and that the same things are not necessarily best in different situations. Out of the simulations come results suggesting that relatively poor communities should follow patterns of expenditure different from those of richer communities. For a community with few resources, latrines are an exceptionally
280
Health Services Research 17:3 (Fall 1982)
effective investment, but for richer communities, it probably makes economic sense to put in toilets and sewers. While the study is strong on its ability to make this point, it fails to suggest an optimal pattern of additions of services for a community as it grows richer. Time cannot be taken account of in this methodology, so the method cannot provide answers to questions such as, "If the community is now at the stage where latrines are the best investment, should latrines be bought or should the community wait for a period of time in the hope that incomes will grow and toilets will become optimal?" and, "Does it make economic sense to put in latrines temporarily?" A third positive aspect of the study is that it explicitly treats interactions between diseases, i.e., one disease causing another or a combination of diseases more likely to cause death than a simple disease. Numerous other disease-related interactions are not considered however: for example, severity of each disease, season in which the disease is contracted, and age and physical strength of the diseased child. The preferable research approach calls for a completely specified model of child mortality in which all socio-economic characteristics of both the child and the environment, all prevention inputs, and all interactions of these factors are related to morbidity and mortality outcomes. Without controlling well for all factors affecting mortality, it is a risky exercise to predict the effect of a change in any one factor on mortality outcomes. It must be noted on behalf of this effort, however, that the authors make a creditable attempt at controlling for interactions of diseases, and in so doing, make first practical steps in a direction which can be pursued in future efforts, both by themselves and by others. The most serious shortcoming to this study is that, as the authors themselves note and more or less apologize for, all parameters in the model, including estimates of deaths resulting from given diseases, links between diseases, and the expected reductions in each disease due to each analyzed intervention activity, are obtained by direct interviews with only sixteen health professionals. Confronted with a question such as, "What percentage of change in the child morbidity rate will be seen among 100 new recipients of hygienic and nutrition education information in Cali, Colombia?" what faith could one have in the answer? The parameters obviously suffer from the opinion-like responses on which they are based, and, perhaps more importantly, from the fact that each respondent must, in effect, control for all other characteristics of Cali, Colombia in formulating an answer. Does the respondent think of already poorly nourished children, of children who come to his or her clinic, or children he or she saw six years ago in Cali to suggest a parameter estimate for the number of children who will die of a certain disease? To expect usable parameters to be obtained from such an approach calls for a large leap of faith.
Book Reviews
281
At this point, it is perhaps slightly risky for this reviewer to summarize the results reported by the study. While perhaps unacceptable as strong scientific evidence, the results can be treated as suggestive, at the very least, suggestive of relationships to be tested carefully when better data become available. The study reports a general finding that any given health-improving activity has diminishing cost effectiveness as its level of activity increases. This result is so strongly expected, given past research and the form of the model, that it can almost certainly be believed. The simulation results also suggest that the most cost-effective activities for very poor communities are promotional and mass media activities (especially in encouragement of breast feeding and early prenatal care); examinations, nutrition programs and tetanus immunizations for pregnant women; water activities and latrines; well-baby clinics; and outpatient services for children in the first 28 days. Only at higher community income levels do the results suggest that such high cost activities as more general immunizations, institutional deliveries of low-birth-weight-risk mothers, and inpatient care for children up to 12 months of age may become costeffective. As levels of health expenditures increase, the results also change in that, for optimal resource usage, toilets replace latrines, water in homes replaces public fountains, midwife and institutional deliveries replace unattended deliveries, and inpatient care replaces outpatient care. The general pattern suggested by the results is that at fairly low resource levels, dramatic decreases in mortality can be brought about by small expenditures of resources. As the authors state, "At higher resource levels, the payoff to additional health resources is not as dramatic and should be compared carefully with other priority areas competing for the use of community resources" (p. 166). The results suggest that in a community such as Cali the specific added resources most needed to reduce infant mortality are registered nurses and supply budgets. Capital, hospital beds, medical doctors and auxiliary nurses appear already available in such quantities that these resources would only start to constrain behavior at high levels of health service provision. On the basis of admittedly questionable results, it would appear that past efforts have overemphasized hospitals and beds while underproviding registered nurses and simple supplies. Given all the caveats implied, that the overall approach can lead to such important policy findings is indicative of its strength. The book reports research that needs improvement in methodology and data base. It must, however, be given full credit as an outstanding piece of work that has set a precedent for future efforts using similar optimization techniques. To dwell too much on the shortcomings of a work that had only limited funds, and a limited period for carrying out
282
Health Services Research 17:3 (Fall 1982)
the analysis, and that more or less had to build its own data substitutes, would be a disservice. Improved attempts using this basic approach can be important additions to the research community's attempts to help answer the difficult questions relating to resource allocation in poor countries. A Resource Allocation Model for Child Survival should be read and absorbed by those who are interested in taking part in the future research effort.
JOHN S. AKIN Associate Professor of Economics University of North Carolina, Chapel Hill Research Results of the National Day Care Study. Jeffrey Travers, Barbara Dillon Goodson, with J. D. Singer and D. B. Connell. Cambridge, MA: ABT Books, 1981. 256 pp. $20.00.
Undertaking research to clarify directions for social policy is a sobering task. While questions addressed in this context appear straightforward, as they are explored from the perspective of making real life decisions, straightforwardness vanishes and one is confronted with a maze of concerns regarding what exactly the problem is in operational terms, what variables are relevant to making judgments about alternative solutions, what costs attend alternative solutions, etc. The complexity, while inherent in human services, also stems from the fact that services are already functioning, rising frequently from ad hoc responses to demands by recipients or providers. This reflection is not new to those engaged in social policy research, but serves to remind us that we should be well aware of and counsel others against excessive expectations regarding tidy solutions. The study under review is a case in point. The National Day Care Study (NDCS), commissioned in 1974 by the office of Child Development (now the Administration for Children, Youth and Families), was asked to identify the impact of federal day care regulations on children's experiences in day care centres. Although one might suppose that there would be some concept of the influence of federal regulations, in terms of good data, this seems not to have been the case. Indeed after the NDCS was initiated, a major policy confrontation highlighting this problem developed between the states and federal government. In 1974, amendments were made to the Federal Interagency Day Care Requirements (FIDCR). The major amendment involved an increase in staff/child ratios. The resulting controversy focused on the substantial cost implications of the new FIDCR amendments and raised the issue of whether in fact staff/child ratios or other day care characteristics ame-
