Medical Decision Making

This paper examines the economic attractiveness of universal vaccination of infants with hepatitis B virus (HBV) vaccine by calculating the incremental cost-effectiveness of this strategy when compared with the currently recommended strategy of screening all pregnant women and vaccinating only infants born to HBsAg+ mothers. A decision-analytic model involving a Markov process to model the long-term sequelae of HBV infection was constructed to estimate the expected costs and life expectancies for a cohort of newborns under two strategies: the current screening policy (SELECTIVE), which involves active and passive vac cination of infants born to carrier mothers, and a policy that combines the current screening strategy (including active and passive vaccination of infants born to carriers) with active vaccination alone for children of non-carriers (UNIVERSAL). A hypothetical cohort of children born in either Canada or the United States in 1991 was examined. Cost estimates were derived for Ontario. From a societal perspective, the incremental cost required to achieve one extra life year was found to be $30,347, comparable to the cost-effectivenesses of other health care strategies commonly used in North America. The result is sensitive to the duration of vaccine effectiveness and particularly to the price of the vaccine. Universal vaccination results in net cost saving at a vaccine price of approximately $7 per dose, from a societal perspective. It is concluded that universal vaccination against HBV in infancy is economically attractive, comparable in cost-effectiveness to existing health care interven tions. Lower vaccine prices would substantially improve the attractiveness of such a program. Implementation of universal vaccination should be considered in North America, contingent on vaccine price reduction. A monitoring program to ensure the long-term efficacy of the vaccine should be part of such a program. Key words: vaccination; hepatitis B; cost-effec tiveness; prevention; chronic hepatitis. (Med Decis Making 1993;13:4-20)

Thinking-aloud protocols provided by Joseph and Patel were reanalyzed to determine the extent to which their conclusions could be replicated by independently developed coding schemes. The data set consisted of protocols from four cardiologists (low domain knowledge = LDK) and four endocrinologists (high domain knowledge = HDK), individually working on a diagnostic problem in endocrinology. The two analyses agree that the HDK physicians related data to potential diagnoses more than did the LDK group and were more focused on the correct diagnostic components. However, the reanalysis found no meaningful differ ence between the groups in diagnostic accuracy, speed of diagnosis, or the breadth of the search space used to seek a solution. In the reanalysis, the HDK physicians employed more single-cue inference and less multiple-cue inference. The generalizability of results of pro tocol-analysis studies can be assessed by using several complementary coding schemes. Key words: domain knowledge; protocol analysis; cognition; reasoning; diagnostic process. (Med Decis Making 1993;13:21-29)

Substantial uncertainty often remains at the time that important diagnostic or therapeutic decisions must be made, despite the availability of multiple clinical indicators. Multiple in dicators may be used to define observation patterns that are associated with the presence or absence of disease. Clinical prediction rules based on groups of observation patterns have been used to quantify probabilities and reduce error rates for some medical problems, but efficient use of multiple indicators remains a major challenge in medical practice. Medical outcomes and clinical observations are frequently categorical. Two statistical techniques appropriate for generating prediction rules from categorical data are logit analysis (LA) and recursive partitioning analysis (RPA). LA and RPA were compared in evaluating observation patterns for fractures among 666 upper-extremity injuries in children, and in developing prediction rules for selective radiographic assessment. Fracture estimates and error reduc tions provided by RPA and LA were very similar. Each technique generated a set of prediction rules with a range of misclassification probabilities, and evaluated the probabilities of fracture for all observation patterns. LA used more information than RPA in observation pattern evaluations, however, and provided fracture estimates specific to each pattern. With currently available statistical software, RPA output provides better statistical guidance in generating prediction rules, whereas LA provides more statistical information of use in evaluating ob servation patterns. LA warrants attention similar to that conferred on RPA. It appears that complementary use of LA and RPA would be valuable in developing clinical guidelines. Key words: predictive models; observation patterns; prediction rules; logit analysis; recursive partitioning analysis. (Med Decis Making 1993;13:30-42)

The authors designed a decision support system to assist mental health professionals to perform differential diagnoses of psychotic, mood, and organic mental disorders in accor dance with the American Psychiatric Association's revised third edition of the Diagnostic and Statistical Manual of Mental Disorders. A prototype system arrived at through a rigorous methodology illustrates a style of development that attempts to ensure system maintainability, correctness, and consistency of deduction and promotes high quality in software. Key words: mental disorders; decision support system; computer-assisted diagnosis; expert systems; artificial intelligence. (Med Decis Making 1993;13:43-48)

Current advances in high-speed computing and increased availability of statistical software have led to widespread use of statistical methods for the development of computerized protocols predictive of binary health outcomes. If these predictive algorithms are to be used in settings other than those for which they were developed, e.g., applied in a different geographic setting or extrapolated for use in a slightly different population, then they should be carefully validated to ensure appropriate application. Miller et al. (Stat Med. 1991) provided a comprehensive methodology for external validation of logistic prediction models, and ap plied these methods in a temporal validation setting. In this article, the authors emphasize how these methods can be applied to general forms of probabilistic predictions and provide several SAS® macros for computation of the desired statistics. Key words: validation; logistic regression; predictive probabilities. (Med Decis Making 1993;13:49-58)

To assess how patients' and physicians' treatment preferences are influenced by graphic data displays (five-year survival curves), a cross-sectional survey of patients, physicians, and medical students was done in a university-based Department of Veterans Affairs Medical Center. Participants in the study were 119 patients seen in a general medicine clinic, 43 physicians, and 67 medical students. Three five-year survival graphs were used. Each graph contained survival curves for two alternative unidentified treatments for an unidentified med ical condition. Graph 1 was a baseline graph used in previous studies of framing effects. Graph 2 contained one survival curve having an area under the curve that was 24% greater than that in graph 1. Graph 3 contained one survival curve that had an area under the curve that was 42% greater than that in graph 1. Respondents were asked to indicate which treatment they preferred for each graph and which aspects of the five-year survival curves most influenced their choices. Respondents did not receive numerical data about the dif ference between the areas under the two curves. Most patients did not change their pref erences across the three graphs. A significantly larger (p ≤ 0.0001) proportion of physicians and medical students than of patients changed their preferences across the three graphs. Patients appeared to use information contained in five-year survival graphs in a different way than physicians and medical students, minimizing or neglecting the importance of ele ments of the five-year survival curve other than immediate (year 0) and long-term (year 5) survival data when presented with curve areas of differing sizes, while the physicians reported using intermediate (year 2 to year 4) data as well as curve shape. Key words: cognitive biases; framing effect; informed consent; medical decision making; preference; five-year survival; life expectancy. (Med Decis Making 1993;13:59-63)

The authors used a decision-analytic approach to develop a Maternal Transport Index (MTI) from ACOG guidelines for maternal transport. Data were obtained from three questionnaires administered to five perinatologists, practicing in facilities with various casemixes. Each questionnaire was based on a given level of hospital and contained scenarios describing indications for maternal transport. The MTls, ratios of the logs of the proportions with given outcomes in Level III hospitals relative to Level I (or II) hospitals, ranged from 1.0 to 26.3 for newborn outcomes. They were greater for Level I hospitals (than Level II) and when newborn outcomes included severe disability as well as death. Within gestational age cat egories, the MTI was generally greatest for active preterm labor and, within complication categories, for 24-26 or 27-31 weeks' gestation. It was large for maternal outcomes only for two rare acute medical conditions. The MTI has potential use in setting priorities for maternal transport. Key words: Maternal Transport Index; obstetrics; outcomes; newborns. (Med Decis Making 1993;13:64-73)