Medical Decision Making

To evaluate the relative impact of clinical factors and health care environment resources on the expert ratings of appropriateness of preoperative erythropoietin in elective orthopedic surgery, the authors analyzed 6905 individual votes on 496 clinical scenarios from 14 experts, applying a multivariate logistic model. Sixty-six percent of the indications were appropriate when resource constraints (RC) were not considered and 53% when they were, resulting in a drop in the median vote of 2 points on a 9-point scale (P < 0.05). Initial hemoglobin level, expected perioperative blood loss, and RC were by far the most significant contributors to the model (P < 0.01), but other factors (i.e., clinical specialty of the expert, prior history of transfusion reactions, patient age, cardiovascular disease, anemia of chronic disease) also contributed significantly (P< 0.01). For assessing appropriateness of care, this study confirms the need for detailed clinical scenarios and a multidisciplinary panel carefully selected to reflect those involved in the interventions under consideration.

Background The mean survival time (MS) has acquired increasing importance as an outcome indicator for patient care and technology assessment. The authors use lifelong followup data from gastric cancer patients to study whether the MS is predictable from 5-year follow-up information based on 2 parametric models. Methods . The authors used 3597 gastric cancer patients operated on between 1950 and 1969 to create 50 groups. For each group, the disease-related survival curve (DRSC) was estimated from the 5-year follow-up data using the Boag model. The MS for the group was then estimated by combining the DRSC with the survival curve for the age and sex-matched contemporaries (control group) based on the competing risk model. Alternatively, it was estimated by using the DRSC and the MS for the control group (the survival limit model). These predicted MS values were compared with the full follow-up observations. Results . Although individual prediction errors varied depending on the group size (63 to 3597 patients) and the length of MS (0.3 to 20.2 years), the mean prediction errors were reasonably small; the survival limit model overestimated MS by 4.7% (95% confidence interval [CI], 1.6 to 7.8) and the competing risk model by 3.2% (95% CI, 0.1 to 6.5). Conclusions . MS for gastric cancer patients is parametrically predictable from 5-year follow-up data. This analysis should be applicable to other diseases showing log-normal failure time distributions.

Choices of health care providers can become inconsistent when people lack sufficient context to assess the value of available information. In a series of surveys, general population samples were randomized to read descriptions of either 2 possible health care providers or a single provider. Some information about providers was easy to consider (e.g., travel time), but some was difficult to interpret without additional context (e.g., success rates). Ratings of the described health care providers varied significantly by whether options were evaluated independently or concurrently. For example, one fertility clinic (33% success rate, 15 min away) was rated higher than a 2nd (40% success rate, 45 min away) when each clinic was considered separately (7.1 v. 6.2, P = 0.046), but preferences reversed in joint evaluation (5.9 v. 6.7, P = 0.051). The results suggest that clinicians and developers of patient information materials alike should consider information evaluability when deciding how to present health care options to patients.

Objective . When making decisions, people are known to try to minimize the regret that would be provoked by unwanted consequences of these decisions. The authors explored the strength and determinants of such anticipated regret in a study of physicians’ decisions to order prostate-specific antigen (PSA) tests. Methods . 32 US and 33 French primary care physicians indicated the likelihood they would order a PSA for 32 hypothetical men presenting for routine physical exams. They then indicated how much regret they would feel if they found advanced prostate cancer in 12 other patients for whom they had chosen not to order PSAs several years before. The latter patients differed according to age (55, 65, or 75 years), a prior request or not for PSA testing, and no or some irregularity of the prostate on the earlier rectal exam. Results . ANOVA found that regret was higher when the patient had requested a PSA, the prostate was irregular, and the patient was younger. Shape had less effect when the patient had requested a PSA. US physicians had more regret than the French, patient request had a greater impact on the Americans, and increasing patient age reduced regret more among the French. In a 1-way correlation, the regret score was associated with the likelihood of ordering PSAs for both the French (r = 0.64, P < 0.005) and the Americans (r = 0.42, P< 0.02). In a regression analysis too, the regret score was the most important predictor of the likelihood of ordering a PSA (β = 0.37, P < 0.0001). Conclusions . Regret over failing to diagnose aggressive prostate cancer is associated with a policy of ordering PSAs. This regret appears to be culturally sensitive.

Purpose . To predict the EuroQoL EQ-5D utility index from the SF-12 Health Survey for a US national sample of adults. Methods . The authors used the 2000 Medical Expenditure Panel Survey to examine the relationship between instruments. Linear regression was used to predict EQ-5D scores from Physical Component Summary (PCS) and Mental Component Summary (MCS) scores of the SF-12. A prediction model was derived in one half of the sample and validated in the other half. Results . Complete responses to both measures were available for 14,580 adults; 7313 (50.2%) surveys were used for the derivation set. The 2-variable model predicted 61% of the variance in EQ-5D scores and provided reasonable ability to predict mean EQ-5D scores from mean PCS and MCS scores. Confidence intervals are dependent on sample size and variance of PCS and MCS scores. Conclusions . EQ-5D scores can be reasonably predicted from the SF-12. This model allows researchers to estimate utility data for use in decision and cost-utility analyses.

Can person tradeoff (PTO) value judgments be elicited by a computer, or is a face-to-face interview needed? The authors randomly assigned 95 subjects to interview or computer methods for the PTO, a valuation measure that is often difficult for subjects. They measured relative values of foot numbness, leg paralysis, and quadriplegia (all 3 pairs) at 2 reference group sizes (10 or 100). Relative values did not differ between computer and interview. Overall, 21% of responses were equality responses, 13% were high extreme values, and 5% violated ordinal criteria. The groups did not differ in these measures. The authors also assessed consistency across reference group size (10 v. 100). Although relative values were significantly lower for 100 than for 10, mode did not influence the size of this effect. Subjects made, on average, equally consistent judgments for the 3 comparisons. A computerized PTO elicitation protocol produced results of similar quality to that of a face-to-face interview.

In this methodological article, the author reviews 1) the disproportionate impact that HIV/AIDS is having on communities of color in the United States, 2) what is known about the cost-effectiveness of HIV prevention interventions for racial/ethnic minority communities (including the methods used in these studies), and 3) the relative lack of methodological guidance in the field for conducting economic evaluation studies specifically for communities of color. The author finds that race/ethnicity affects cost-utility analyses in several heretofore unrecognized ways. In this article, methodological techniques to address these concerns are proposed. In particular, the author recommends cost-utility analytic strategies that maximize comparability among studies and avoid the introduction of methodological discrimination.

Objectives . The authors sought to develop a conceptual framework for evaluating whether existing information technologies and decision support systems (IT/DSSs) would assist the key decisions faced by clinicians and public health officials preparing for and responding to bioterrorism. Methods. They reviewed reports of natural and bioterrorism related infectious outbreaks, bioterrorism preparedness exercises, and advice from experts to identify the key decisions, tasks, and information needs of clinicians and public health officials during a bioterrorism response. The authors used task decomposition to identify the subtasks and data requirements of IT/DSSs designed to facilitate a bioterrorism response. They used the results of the task decomposition to develop evaluation criteria for IT/DSSs for bioterrorism preparedness. They then applied these evaluation criteria to 341 reports of 217 existing IT/DSSs that could be used to support a bioterrorism response. Main Results . In response to bioterrorism, clinicians must make decisions in 4 critical domains (diagnosis, management, prevention, and reporting to public health), and public health officials must make decisions in 4 other domains (interpretation of bioterrorism surveillance data, outbreak investigation, outbreak control, and communication). The time horizons and utility functions for these decisions differ. From the task decomposition, the authors identified critical subtasks for each of the 8 decisions. For example, interpretation of diagnostic tests is an important subtask of diagnostic decision making that requires an understanding of the tests’ sensitivity and specificity. Therefore, an evaluation criterion applied to reports of diagnostic IT/DSSs for bioterrorism asked whether the reports described the systems’ sensitivity and specificity. Of the 217 existing IT/DSSs that could be used to respond to bioterrorism, 79 studies evaluated 58 systems for at least 1 performance metric. Conclusions. The authors identified 8 key decisions that clinicians and public health officials must make in response to bioterrorism. When applying the evaluation system to 217 currently available IT/DSSs that could potentially support the decisions of clinicians and public health officials, the authors found that the literature provides little information about the accuracy of these systems.

There has been an increasing interest in using expected value of information (EVI) theory in medical decision making, to identify the need for further research to reduce uncertainty in decision and as a tool for sensitivity analysis. Expected value of sample information (EVSI) has been proposed for determination of optimum sample size and allocation rates in randomized clinical trials. This article derives simple Monte Carlo, or nested Monte Carlo, methods that extend the use of EVSI calculations to medical decision applications with multiple sources of uncertainty, with particular attention to the form in which epidemiological data and research findings are structured. In particular, information on key decision parameters such as treatment efficacy are invariably available on measures of relative efficacy such as risk differences or odds ratios, but not on model parameters themselves. In addition, estimates of model parameters and of relative effect measures in the literature may be heterogeneous, reflecting additional sources of variation besides statistical sampling error. The authors describe Monte Carlo procedures for calculating EVSI for probability, rate, or continuous variable parameters in multi parameter decision models and approximate methods for relative measures such as risk differences, odds ratios, risk ratios, and hazard ratios. Where prior evidence is based on a random effects meta-analysis, the authors describe different ESVI calculations, one relevant for decisions concerning a specific patient group and the other for decisions concerning the entire population of patient groups. They also consider EVSI methods for new studies intended to update information on both baseline treatment efficacy and the relative efficacy of 2 treatments. Although there are restrictions regarding models with prior correlation between parameters, these methods can be applied to the majority of probabilistic decision models. Illustrative worked examples of EVSI calculations are given in an appendix.