Medical Decision Making recent issues

Bivariate Random Effects Meta-Analysis of ROC Curves

2008-10-03

Meta-analysis of receiver operating characteristic (ROC)-curve data is often done with fixed-effects models, which suffer many shortcomings. Some random-effects models have been proposed to execute a meta-analysis of ROC-curve data, but these models are not often used in practice. Straightforward modeling techniques for multivariate random-effects meta-analysis of ROC-curve data are needed. The 1st aim of this article is to present a practical method that addresses the drawbacks of the fixedeffects summary ROC (SROC) method of Littenberg and Moses. Sensitivities and specificities are analyzed simultaneously using a bivariate random-effects model. The 2nd aim is to show that other SROC curves can also be derived from the bivariate model through different characterizations of the estimated bivariate normal distribution. Thereby the authors show that the bivariate random-effects approach not only extends the SROC approach but also provides a unifying framework for other approaches. The authors bring the statistical meta-analysis of ROC-curve data back into a framework of relatively standard multivariate meta-analysis with random effects. The analyses were carried out using the software package SAS (Proc NLMIXED).

Meta-Analysis of Diagnostic Studies: A Comparison of Random Intercept, Normal-Normal, and Binomial-Normal Bivariate Summary ROC Approaches

Hamza, T. H., Reitsma, J. B., Stijnen, T. — 2008-10-03

Background . The authors compared 3 recently introduced refinements of the Littenberg and Moses summary receiver operating characteristic (ROC) method for pooling studies of a diagnostic test: the random intercept (RI) linear meta-regression model, the approximate normal distribution (normal-normal [NN] model), and the binomial distribution (binomial-normal [BN] model). Methods . Using data from a published meta-analysis of magnetic resonance imaging of the menisci and cruciate ligaments, the authors varied the overall sensitivity and specificity, the between-studies variance, the within-study sample size, and the number of studies to evaluate the performances of the 3 methods in a simulation study. The parameters to be compared are the associated intercept, slope, and residual variance, using bias, mean squared error, and coverage probabilities. Results. The BN method always gave unbiased estimates of the intercept and slope parameter. The coverage probabilities were also reasonably acceptable, unless the number of studies was very small. In contrast, the RI and NN methods could produce large biases with poor coverage probabilities, especially when sample sizes of individual studies were small or when sensitivities or specificities were close to 1. Although this was rare in the simulations, the bivariate methods can suffer from nonconvergence mostly due to the correlation being close to ± 1. Conclusion. The binomial-normal model performed better than the other recently introduced methods for meta-analysis of data from studies of test performance.

Integration of Meta-analysis and Economic Decision Modeling for Evaluating Diagnostic Tests

Sutton, A. J., Cooper, N. J., Goodacre, S., Stevenson, M. — 2008-10-03

Meta-analysis of diagnostic test accuracy data is more difficult than of effectiveness data because of 1) statistical challenges of dealing with multiple measures of accuracy (e.g., sensitivity and specificity) simultaneously and 2) incorporating threshold effects. A number of meta-analysis models are in use, ranging from naïve synthesis of independent sensitivity and specificity to optimization of a hierarchical summary receiver operating characteristic (SROC) curve. Little work has been done on how such analyses should inform decision models. This article aims to present a unified framework for the synthesis of primary data and economic evaluation of alternative diagnostic testing strategies using Bayesian Markov Chain Monte Carlo simulation methods. The authors extend this previous work by using systematic review to derive model parameters, fully allowing for uncertainty in their estimation, and formally incorporating variability between study results into the decision analysis. Using a simple decision model comparing alternative testing strategies for suspected deep vein thrombosis as an example, the authors consider how to use outputs of different alternative meta-analysis models in decision models. They also explore the limitations of diagnostic test studies, particularly when there is no obvious threshold value. To correct some of the limitations of diagnostic test studies, they propose that tests with implicit and explicit thresholds should be studied using distinctly different frameworks. Specifically, when a threshold exists, quantitative threshold information should be included in meta-analysis models to aid interpretation of SROCs. Setting policy to relate to a specific point may be much more difficult for studies with implicit thresholds.

Predictors of Diagnostic Accuracy and Safe Management in Difficult Diagnostic Problems in Family Medicine

2008-10-03

Objective. To investigate the role of information gathering and clinical experience on the diagnosis and management of difficult diagnostic problems in family medicine. Method. Seven diagnostic scenarios including 1 to 4 predetermined features of difficulty were constructed and presented on a computer to 84 physicians: 21 residents in family medicine, 21 family physicians with 1 to 3 y in practice, and 42 family physicians with ≥10 y in practice. Following the Active Information Search process tracing approach, participants were initially presented with a patient description and presenting complaint and were subsequently able to request further information to diagnose and manage the patient. Evidence-based scoring criteria for information gathering, diagnosis, and management were derived from the literature and a separate study of expert opinion. Results. Rates of misdiagnosis were in accordance with the number of features of difficulty. Seventy-eight percent of incorrect diagnoses were followed by inappropriate management and 92% of correct diagnoses by appropriate management. Number of critical cues requested (cues diagnostic of any relevant differential diagnoses in a scenario) was a significant predictor of accuracy in 6 scenarios: 1 additional critical cue increased the odds of obtaining the correct diagnosis by between 1.3 (95% confidence interval [CI], 1.0—1.8) and 7.5 (95% CI, 3.2— 17.7), depending on the scenario. No effect of experience was detected on either diagnostic accuracy or management. Residents requested significantly more cues than experienced family physicians did. Conclusions. Supporting the gathering of critical information has the potential to improve the diagnosis and management of difficult problems in family medicine.

Effect of Guidelines on Primary Care Physician Use of PSA Screening: Results from the Community Tracking Study Physician Survey

2008-10-03

Background. Little is known about the effect of guidelines that recommend shared decision making on physician practice patterns. The objective of this study was to determine the association between physicians' perceived effect of guidelines on clinical practice and self-reported prostate-specific antigen (PSA) screening patterns. Methods. This was a cross-sectional study using a nationally representative sample of 3914 primary care physicians participating in the 1998—1999 Community Tracking Study Physician Survey. Responses to a case vignette that asked physicians what proportion of asymptomatic 60-year-old white men they would screen with a PSA were divided into 3 distinct groups: consistent PSA screeners (screen all), variable screeners (screen 1%— 99%), and consistent nonscreeners (screen none). Logistic regression was used to determine the association between PSA screening patterns and physician-reported effect of guidelines (no effect v. any magnitude effect). Results. Only 27% of physicians were variable PSA screeners; the rest were consistent screeners (60%) and consistent nonscreeners (13%). Only 8% of physicians perceived guidelines to have no effect on their practice. After adjustment for demographic and practice characteristics, variable screeners were more likely to report any magnitude effect of guidelines on their practice when compared with physicians in the other 2 groups (adjusted odds ratio= 1.73; 95% confidence interval=1:25–2:38;P=0:001). Conclusions. Physicians who perceive an effect of guidelines on their practice are almost twice as likely to exhibit screening PSA practice variability, whereas physicians who do not perceive an effect of guidelines on their practice are more likely to be consistent PSA screeners or consistent PSA nonscreeners.

Lessons Learned by (from?) an Economist Working in Medical Decision Making

Wakker, P. P. — 2008-10-03

This article is a personal account of the author's experiences as an economist working in medical decision making. He discusses the differences between economic decision theory and medical decision making and gives examples of the mutual benefits resulting from interactions. In particular, he discusses the pros and cons of different methods for measuring quality of life (or, as economists would call it, utility), including the standard gamble, the time tradeoff, and the healthy-years equivalent methods.

Beyond Shared Decision Making: An Expanded Typology of Medical Decisions

2008-10-03

The most popular current models of medical decision making, identified by names such as shared decision making, informed decision making, and evidence-based patient choice, portray an empowered patient actively involved in his or her medical choices and generally assume that patient and physician reach agreement. These models are limited to a specific type of decision (in which there is more than one choice) and a specific process (in which agreement is reached). The authors extend the model of medical decision making beyond shared decisions in 2 dimensions. First, the authors incorporate a class of medical decisions in which there is only one medically reasonable treatment option, such as the removal of a primary melanoma. Patient preferences are irrelevant to whether or not the melanoma should be removed, so there is no treatment choice in which the patient can share. When there is only one realistic treatment option, the clinician's job is not to offer alternatives but to explain why there is only one viable choice and move the decision-making process forward. The physician does not thereby abridge the patient's autonomy; rather, the disease process itself constrains both patient and physician. Second, the authors include decisions in which patient and physician do not reach agreement. Sometimes the patient insists on a particular treatment and the physician reluctantly yields, sometimes it is the other way around, but disagreement is commonplace in clinical medicine and its presence deserves inclusion in the way we think about medical decisions. Conflict resolution requires acknowledging the potential for conflict.

The Half-Cycle Correction Explained: Two Alternative Pedagogical Approaches

Naimark, D. M. J., Bott, M., Krahn, M. — 2008-10-03

Students of Markov decision models are often taught to add a half-cycle's worth of incremental utility to the cumulative total for each health state. The reason for this ``half-cycle correction'' is often illustrated by a graph of the proportion of the hypothetical Markov cohort remaining in a given state. The ideal graph is shown as a smooth, declining, curve that represents the transition of patients randomly throughout each cycle. On the same graph, the effect of the accounting of state membership at the end of each cycle in discrete, computer-based approximations of the ideal Markov process is shown. Students are able to clearly see that the cumulative incremental utility in the discrete case underestimates the desired quantity. Likewise, they find the concept of shifting the ideal curve to the right by one-half cycle to reduce the latter discrepancy to be intuitive. However, students often find the approximate equivalence of shifting the ideal state membership curve and adding a half-cycle's worth of incremental utility to the total for the state at the beginning of a discrete Markov process to be a difficult cognitive leap. This article describes 2 pedagogical devices, algebraic and intuitive/visual approaches, that may assist the instructor of Markov theory to convey the latter concept. Elements of adult learning theory are discussed, which may help the instructor to choose which approach to employ. Implementation of the half-cycle correction in commonly used decision-analytic software is also discussed.

The Role of Value for Money in Public Insurance Coverage Decisions for Drugs in Australia: A Retrospective Analysis 1994-2004

Harris, A. H., Hill, S. R., Chin, G., Li, J. J., Walkom, E. — 2008-10-03

Objective . To analyze the relative influence of factors in decisions for public insurance coverage of new drugs in Australia. Data Sources. Evidence presented at meetings of the Australian Pharmaceutical Benefits Advisory Committee (PBAC) that makes recommendations on coverage of drugs under Pharmaceutical Benefits Scheme. Study Selection. All major submissions to the PBAC between February 1994 and December 2004 (n = 858) if one of the outcomes measured was life year gained (n=138) or quality-adjusted life years (QALYs) gained (n=116). Results. Clinical significance, cost-effectiveness, cost to government, and severity of disease were significant influences on decisions. Compared to the average submission, clinical significance increased the probability of recommending coverage by 0.21 (95% confidence interval [CI] 0.02 to 0.40), whereas a drug in a life-threatening condition had an increased probability of being recommended for coverage of 0.38 (0.06 to 0.69). An increase in $A10,000 from a mean incremental cost per QALY of $A46,400 reduced the probability of listing by 0.06 (95% CI 0.04 to 0.1). Conclusions. The PBAC provides an example of the long-term stability and coherence of evidence-based coverage and pricing decisions for drugs that weighs up the evidence on clinical effectiveness, clinical need, and value for money. There is no evidence of a fixed public threshold value of life years or QALYs, but willingness to pay is clearly related to the characteristics of the clinical condition, perceived confidence in the evidence of effectiveness and its relevance, as well as total cost to government.

Assessing the Influence of Gestalt-Type Characteristics on Preferences Over Lifetime Health Profiles

Oliver, A. — 2008-10-03

Introduction . In contrast to the basic tenets of economic theory, there is substantial evidence that people's remembered and predicted utility of events systematically differs from the utility that they experience. These systematic differences are caused by ``gestalt characteristics.'' The objective of this study was to test whether people maximize quality-adjusted life years (QALYs), or whether QALY maximization is compromised by their being influenced by factors that resemble the gestalt characteristics when choosing between lifetime health profiles. Methods. Time trade-off values were elicited from 50 respondents, who were also presented with a series of hypothetical questions that each depicted 2 lifetime health profiles. The respondents were asked to choose which of the 2 profiles in each question they would prefer to experience. By inputting the values that the respondents placed on the health states into the lifetime health profiles, it was possible to observe whether their answers were consistent with QALY maximization or with various hypothesized gestalt-type effects. Results. Across decisions that involve a simple trade-off between the length of life and the quality of the health state, choices consistent with QALY maximizing were relatively common, although even here approximately half of the respondents violated this rule. Consistency with QALY maximization was lower in most of the other tests and indicated that many people might, for example, prefer to trade off some lifetime health to experience a good end to life, or to avoid highly unstable lifetime health profiles. Conclusion . The respondents' answers were often consistent with the hypothesized gestalt-type effects, but it is probable that for some of the questions the characteristics themselves were not driving the respondents' answers and that factors such as complex rates of discounting might have played a role. However, whatever the driving motivation behind the respondents' answers, the important point to note from this study is that QALY maximization is often substantially and systematically violated when people are offered a choice over the lifetime health profiles that they would prefer to experience.

Obtaining Utility Estimates of the Health Value of Commonly Prescribed Treatments for Asthma and Depression

2008-10-03

Background. Comparing the costs and health value associated with alternative quality improvement efforts is useful. This study employs expert panel methodology to elicit numerical estimates based on a 0 to 1 utility scale of the health benefit of usual treatment patterns for 2 medical conditions. Method. The approach includes development of clinical profiles and derivation of treatment benefit estimates via the elicitation of utility ratings before and after treatment. Clinical profiles specified characteristics of patient groups, treatments to be rated, and their combinations. A panel of 13 asthma and depression experts made a series of utility ratings (before any new treatment, 1 or 3 mo later with no treatment, 1 or 3 mo after initiating various common treatments) for adult patient groups with depression or asthma. The panel convened to discuss discrepancies and subsequently made final ratings. Treatment benefit estimates were derived from the ratings made by the panelists after the panel meeting. Results. The treatment benefit estimates had face validity and minimal variability, indicating considerable consensus among experts. Treatment benefit estimates ranged from –0.03 to 0.25 for depression and from –0.04 to 0.24 for asthma. There was minimal variation in the estimates for both conditions (the estimates' standard deviations ranged from 0.01 to 0.06). Comparisons of the treatment benefit estimates before and after the expert panel meeting indicated substantial convergence, and evidence suggests that the benefit estimates are comparable across the 2 health conditions. Conclusion. Comparable estimates of treatment benefit for distinct health conditions can be obtained from experts using the expert panel methodology.

The Impact of Individualized Evidence-Based Decision Support on Aneurysm Patients' Decision Making, Ideals of Autonomy, and Quality of Life

2008-10-03

Background . A major challenge in surgery is the integration of evidence-based medicine and patient autonomy. The authors present a randomized trial studying the effect of an individualized evidence-based brochure (IB) on patients' autonomous behavior, patients' ideals of autonomy, and quality of life. Method . Patients with an asymptomatic abdominal aneurysm and their surgeon were randomized to receive a general brochure (GB) or an IB presenting survival information and a ranking of the treatment strategies. Before and after receiving the brochure, patients filled out questionnaires on their behavior during the consultation, ideals of patient autonomy, and quality of life. Surgeons answered a short checklist evaluating the consultation. Results . One hundred patients participated, 49 in the intervention, 51 in the control group. The IB group had a better understanding of important issues in the treatment decision, had prepared more questions, and was less satisfied with the duration of the consultation. Their impression that the surgeon perceived them more as a medical problem than a patient with a problem increased. They agreed less with the surgeon's advice and lost some of their belief in ``the doctor knows best.'' Beforehand, the IB group had a stronger preference for patient-based decisions, but afterward they displayed more surgeon-based decisions. No effects were seen on patients' quality of life. Conclusions. Individualized evidence-based information stimulated patients' active involvement but in the context of our study led to less patient-based decisions. Patient-made decisions and patient autonomy should, however, not be equated.

Beyond Utilitarianism: A Method for Analyzing Competing Ethical Principles in a Decision Analysis of Liver Transplantation

Volk, M. L., Lok, A. S. F., Ubel, P. A., Vijan, S. — 2008-10-03

Background: The utilitarian foundation of decision analysis limits its usefulness for many social policy decisions. In this study, the authors examine a method to incorporate competing ethical principles in a decision analysis of liver transplantation for a patient with acute liver failure (ALF). Methods. A Markov model was constructed to compare the benefit of transplantation for a patient with ALF versus the harm caused to other patients on the waiting list and to determine the lowest acceptable 5-y posttransplant survival for the ALF patient. The weighting of the ALF patient and other patients was then adjusted using a multiattribute variable incorporating utilitarianism, urgency, and other principles such as fair chances. Results. In the base-case analysis, the strategy of transplanting the ALF patient resulted in a 0.8% increase in the risk of death and a utility loss of 7.8 quality-adjusted days of life for each of the other patients on the waiting list. These harms cumulatively outweighed the benefit of transplantation for an ALF patient having a posttransplant survival of less than 48% at 5 y. However, the threshold for an acceptable posttransplant survival for the ALF patient ranged from 25% to 56% at 5 y, depending on the ethical principles involved. Discussion. The results of the decision analysis vary depending on the ethical perspective. This study demonstrates how competing ethical principles can be numerically incorporated in a decision analysis.

``Not Everyone Who Needs One Is Going to Get One'': The Influence of Medical Brokering on Patient Candidacy for Total Joint Arthroplasty

2008-10-03

Background. Many patients in Ontario, despite being appropriate candidates for total joint arthroplasty (TJA), are not offered surgery. To understand this discrepancy, the authors sought to explore the process by which physicians determine patient candidacy for TJA. Methods. Six focus groups (2 each of orthopedic surgeons, of rheumatologists, and of family physicians) and subsequent in-depth interviews were conducted with 50 practicing clinicians in Ontario. Results. Health care system constraints, including extensive waiting lists, lack of homecare and postoperative support, and, for surgeons, access to operating rooms and resources, are perceived by physicians to routinely influence the ultimate choice of candidates for TJA. Medical brokering, defined as strategies used by physicians in a constrained health system to prioritize patients and to negotiate relationships with other physicians, was an important factor in determining candidacy for TJA. Because individual physicians and surgeons appear to use their own criteria for making these decisions, and because these criteria are modified from time to time in response to specific institutional and system conditions, brokering results in varied decisions about candidacy regardless of patient suitability. Conclusions. Lack of consensus on the necessary patient characteristics for TJA candidacy does not in and of itself account for the discrepancy between the number of patients who are suitable candidates for TJA and those who receive the procedure. Until the process by which health care system constraints affect and complicate the decision-making process around TJA candidacy is more fully explored, patients may not receive appropriate and timely access to this procedure.

Applying Social Marketing in Health Care: Communicating Evidence to Change Consumer Behavior

Evans, W. D., McCormack, L. — 2008-10-03

Social marketing uses commercial marketing strategies to change individual and organizational behavior and policies. It has been effective on a population level across a wide range of public health and health care domains. There is limited evidence of the effectiveness of social marketing in changing health care consumer behavior through its impact on patient-provider interaction or provider behavior. Social marketers need to identify translatable strategies (e.g., competition analysis, branding, and tailored messages) that can be applied to health care provider and consumer behavior. Three case studies from social marketing illustrate potential strategies to change provider and consumer behavior. Countermarketing is a rapidly growing social marketing strategy that has been effective in tobacco control and may be effective in countering pharmaceutical marketing using specific message strategies. Informed decision making is a useful strategy when there is medical uncertainty, such as in prostate cancer screening and treatment. Pharmaceutical industry marketing practices offer valuable lessons for developing competing messages to reach providers and consumers. Social marketing is an effective population-based behavior change strategy that can be applied in individual clinical settings and as a complement to reinforce messages communicated on a population level. There is a need for more research on message strategies that work in health care and population-level effectiveness studies.

Budget Impact Analysis and Its Rational Basis

Hilden, J. — 2008-07-25

Quality Performance Measurement Using the Text of Electronic Medical Records

Pakhomov, S., Bjornsen, S., Hanson, P., Smith, S. — 2008-07-25

Background. Annual foot examinations (FE) constitute a critical component of care for diabetes. Documented evidence of FE is central to quality-of-care reporting; however, manual abstraction of electronic medical records (EMR) is slow, expensive, and subject to error. The objective of this study was to test the hypothesis that text mining of the EMR results in ascertaining FE evidence with accuracy comparable to manual abstraction. Methods. The text of inpatient and outpatient clinical reports was searched with natural-language (NL) queries for evidence of neurological, vascular, and structural components of FE. A manual medical records audit was used for validation. The reference standard consisted of 3 independent sets used for development (n=200 ), validation (n=118), and reliability (n=80). Results. The reliability of manual auditing was 91% (95% confidence interval [CI]= 85—97) and was determined by comparing the results of an additional audit to the original audit using the records in the reliability set. The accuracy of the NL query requiring 1 of 3 FE components was 89% (95% CI=83—95). The accuracy of the query requiring any 2 of 3 components was 88% (95% CI=82—94). The accuracy of the query requiring all 3 components was 75% (95% CI= 68— 83). Conclusions. The free text of the EMR is a viable source of information necessary for quality of health care reporting on the evidence of FE for patients with diabetes. The low-cost methodology is scalable to monitoring large numbers of patients and can be used to streamline quality-of-care reporting.

Development and Validation of a Risk Scoring Tool to Predict Respiratory Syncytial Virus Hospitalization in Premature Infants Born at 33 through 35 Completed Weeks of Gestation

2008-07-25

Objective. The purpose of the study was to develop and validate a clinical instrument predicting the risk of respiratory syncytial virus (RSV)-associated hospitalization (RSV-H) in premature infants born at 33 through 35 completed weeks of gestation (33—35GA). Design. An RSV risk scoring tool (RSV-RS) was developed by entering risk factors for RSV-H, determined in a Canadian prospective study, into a multiple logistic regression model. The scoring tool was then validated externally with data from a Spanish case-control study (FLIP). The Canadian cohort comprised 1758 RSV-positive infants born 33—35GA, of whom 66 (3.7%) had confirmed RSV-H. The FLIP data set comprised 186 (33.4%) RSV-H cases and 371 (66.7%) controls. Method. The primary outcome measure was RSV-H. The RSV-RS score was the sum of the weighted probabilities for each included risk factor multiplied by 100 and ranged from 0 to 100. Receiver operator characteristic curve analyses determined cutoff points to predict subjects at low, moderate, or high RSV-H risk. Results. The RSV-RS included 7 risk factors and cutoff scores of 0—48, 49—64, and 65— 100 for low-, moderate-, and high-risk subjects, respectively. For the Canadian cohort, RSV-RS sensitivity in predicting RSV-H cases was 68.2%, with 71.9% specificity. With the FLIP data set, the RSV-RS had lower accuracy (61.3% sensitivity; 65.8% specificity) but showed significant positive association with increased risk for RSV-H. Conclusion. The RSV-RS accurately identified 33—35GA infants at increased risk for RSV-H in a Canadian cohort. External validation with Spanish case-control study data further confirmed that the scoring tool is appropriate for the estimation of RSV-H risk.

Calculation of Prevalence with Markov Models: Budget Impact Analysis of Thrombolysis for Stroke

Mar, J., Sainz-Ezkerra, M., Miranda-Serrano, E. — 2008-07-25

Objectives . The objective is to develop a method for calculating the prevalence of stroke based on Markov models and to apply it to the assessment of the budget impact analysis of thrombolytic treatment. Methods. A Markov model was used to reproduce the natural history of stroke. The first step was to run the model to build the sojourn matrix from the initial population vector. The second step was to ascertain the number of individuals in the origin of each annual cohort. Finally, prevalence figures were obtained, validated, and used to calculate the impact of treatment with thrombolysis in 10% of patients with stroke in the Basque Country as if thrombolysis had begun in 2000 and would continue to 2015. Results. Stroke prevalence rates per 100,000 for the entire population are 898 for men and 686 for women, with a combined rate of 774 for men and women. Rates for stroke-related disability are 358 per 100,000 for men, 275 for women, and 309 for men and women combined. If 10% of the stroke patients would have received thrombolytic treatment from 2000 to 2015, the number of disabled in 2015 would be reduced by 328, and the net savings for the Basque society (2,100,000 inhabitants) would be 1.7 million. Conclusions. The budget impact analysis of thrombolysis for stroke starting in 2000 shows a positive impact on the health budget because it saves costs after 2006 and produces a net benefit in health from the beginning of treatment.

Estimating EuroQol EQ-5D Scores from Population Healthy Days Data

Jia, H., Lubetkin, E. I. — 2008-07-25

Background. Preference-based assessments of population health, which may be used for cost-utility analyses, are lacking for most states and communities. With adequate population data, preference-based values can be estimated from non-preference-based health-related quality of life (HRQOL) data. This study estimates scores on the EuroQol EQ-5D, a preference-based measure, from the Healthy Days Measures. Methods. No data set from the US population asks both the Healthy Days and EQ-5D questions for the same respondents. Therefore, estimates for EQ-5D scores were obtained indirectly by matching cumulative distributions of the 2 measures. These distributions were estimated from the 2000— 2002 Behavioral Risk Factor Surveillance System (BRFSS) and the Medical Expenditure Panel Survey (MEPS). The validity of estimates was examined by comparing the mean estimated and observed scores across particular population subgroups. A simulation study was conducted to compare the performance of the proposed method to the regression method. Results. The overall mean observed EQ-5D index was 0.871 and the mean estimated EQ-5D index was 0.872. In the majority of examined subgroups, the mean scores demonstrated a good match according to sociodemographic variables and health-related conditions and, with the exception of the most impaired health states, the differences tended to be less than 0.04. Conclusions. This study provided preliminary estimates of EQ-5D scores from the Healthy Days Measures and demonstrated acceptable validity of the estimates. Because the Healthy Days Measures have been included in many state and local surveys, preliminary cost-utility analyses and determination of burden of disease might be able to be conducted at the national, state, and community levels as well as over time. Key words: health-related quality of life; EQ-5D; Healthy Days Measures; cost-effective analysis. (Med Decis Mak ing 2008;28:491—499)

Feasibility and Reliability of the Annual Profile Method for Deriving QALYs for Short-Term Health Conditions

Janssen, M. F., Birnie, E., Bonsel, G. — 2008-07-25

Introduction . When health varies over time, the standard quality-adjusted life year model operates under the assumptions of time utility independence within each health state and additive independence between health states. These assumptions can be relaxed by an integral assessment of disease severity over time. The authors present the annual profile method (APM), which values health profiles on a 1-year base, and test the APM for feasibility, consistency, and test-retest reliability. Methods. A population panel, general practitioners, medical advisers, and a panel of the Dutch Consumers Association valued vignettes for 46 disease stages using the visual analog scale (VAS) and time tradeoff (TTO) methods. Vignettes contained disease-specific information, a generic description (EQ-6D5L), a description of the disease course over time, and a visual representation of the disease. Feasibility was tested by missing and inconsistent responses. Consistency between and within panels was tested with a generalizability study, analysis of variance, and standard correlation coefficients. Test-retest reliability was tested with a generalizability study and intra-class correlation coefficients. Results. Missing and inconsistent responses were < 2.6%. The valuations were consistent across panels, with generalizability coefficients of 0.78 (VAS) and 0.64 (TTO). Within the main population panel, internal consistency was satisfactory and the influence of background characteristics negligible. Test-retest reliability was high, with generalizability coefficients of 0.90 (VAS) and 0.72 (TTO). Conclusion. Feasibility and reliability of the APM for realistic health profiles are good to excellent. The APM is a promising step to bridge the gap between the quality-adjusted life year methodology and clinical reality.

End-of-Life Medical Treatment Choices: Do Survival Chances and Out-of-Pocket Costs Matter?

Chao, L.-W., Pagan, J. A., Soldo, B. J. — 2008-07-25

Background . Out-of-pocket medical expenditures incurred prior to the death of a spouse could deplete savings and impoverish the surviving spouse. Little is known about the public's opinion as to whether spouses should forego such end-of-life (EOL) medical care to prevent asset depletion. Objectives . To analyze how elderly and near elderly adults assess hypothetical EOL medical treatment choices under different survival probabilities and out-of-pocket treatment costs. Methods. Survey data on a total of 1143 adults, with 589 from the Asset and Health Dynamics Among the Oldest Old (AHEAD) and 554 from the Health and Retirement Study (HRS), were used to study EOL cancer treatment recommendations for a hypothetical anonymous married woman in her 80s. Results. Respondents were more likely to recommend treatment when it was financed by Medicare than by the patient's own savings and when it had 60% rather than 20% survival probability. Black and male respondents were more likely to recommend treatment regardless of survival probability or payment source. Treatment uptake was related to the order of presentation of treatment options, consistent with starting point bias and framing effects. Conclusions. Elderly and near elderly adults would recommend that the hypothetical married woman should forego costly EOL treatment when the costs of the treatment would deplete savings. When treatment costs are covered by Medicare, respondents would make the recommendation to opt for care even if the probability of survival is low, which is consistent with moral hazard. The sequence of presentation of treatment options seems to affect patient treatment choice.

Patient and Surrogate Disagreement in End-of-Life Decisions: Can Surrogates Accurately Predict Patients' Preferences?

Marks, M. A. Z., Arkes, H. R. — 2008-07-25

Background . When a patient is too incapacitated to make important end-of-life decisions, doctors may ask a preappointed surrogate to predict the patient's preferences and make decisions on the patient's behalf. The current study investigates whether surrogates project their own views onto what they predict the patients' preferences are. Methods. Using data from seriously ill patients and their surrogates, the authors created a ``projection'' variable that addresses the following question: When surrogates are asked to predict a patient's end-of-life preferences, do they mistakenly replace this prediction with what they would want the patient to do? The authors examined the 144 patient-surrogate pairs in which surrogates inaccurately predicted patients' CPR (cardiopulmonary resuscitation) v. DNR (do not resuscitate) decisions and the 294 pairs in which surrogates inaccurately predicted patients' extend life v. relieve pain preferences. Among these patient-surrogate pairs, the authors determined the extent to which surrogates' wishes for the patient matched their incorrect predictions of what the patient wanted. Results. Of the patient-surrogate pairs who disagreed on CPR v. DNR and extend life v. relieve pain preferences, 62.5% and 88.4% of surrogates demonstrated projection for CPR v. DNR decisions and extend life v. relieve pain preferences, respectively. Age-related and demographic variables did not predict cases in which projection did and did not occur. Conclusion. When surrogates inaccurately predict the CPR v. DNR and extend life v. relieve pain preferences of seriously ill, hospitalized loved ones, surrogates' prediction errors often represent surrogates' own wishes for the patient.

Do Decision Biases Predict Bad Decisions? Omission Bias, Naturalness Bias, and Influenza Vaccination

DiBonaventura, M. d., Chapman, G. B. — 2008-07-25

Purpose. Numerous studies using hypothetical vignettes have demonstrated decision biases or deviations from utility theory. Do people who commit biases in questionnaire studies make worse real-world decisions than do less biased people? Methods. Two hundred seventy university faculty and staff participated in a questionnaire study in which they reported whether they accepted a free influenza vaccine offered at their work place. Influenza vaccine acceptance was the measure of real-world decision making. Participants responded to 3 hypothetical scenarios. Two scenarios measured the omission bias and described a vaccine (scenario 1) and a medication (scenario 2) that prevented a negative health outcome but that itself could cause the negative health outcome. The omission bias is a preference for not vaccinating or medicating even when the vaccine/medication lowers the total risk of the negative outcome. A 3rd scenario measured the naturalness bias by presenting a choice between 2 chemically identical medications, one extracted from a natural herb and the other synthesized in a laboratory. Preference for the natural medication indicated the naturalness bias. Results. The results indicated that a substantial proportion of participants exhibited these biases and that participants who exhibited these biases were less likely to accept the flu vaccine. Conclusions. To the extent that declining a free flu vaccine is a worse real-world decision, people who demonstrate the naturalness and omission biases in hypothetical scenarios make worse real-world decisions. Key words: omission bias; naturalness bias; influenza; vaccination. (Med Decis Making 2008;28:532—539)

When Is Diagnostic Testing Inappropriate or Irrational? Acceptable Regret Approach

Hozo, I., Djulbegovic, B. — 2008-07-25

The authors provide a new model within the framework of theories of bounded rationality for the observed physicians' behavior that their ordering of diagnostic tests may not be rational. Contrary to the prevailing thinking, the authors find that physicians do not act irrationally or inappropriately when they order diagnostic tests in usual clinical practice. When acceptable regret (i.e., regret that a decision maker finds tolerable upon making a wrong decision) is taken into account, the authors show that physicians tend to order diagnostic tests at a higher level of pretest probability of disease than predicted by expected utility theory. They also show why physicians tend to overtest when regret about erroneous decisions is extremely small. Finally, they explain variations in the practice of medicine. They demonstrate that in the same clinical situation, different decision makers might have different acceptable regret thresholds for withholding treatment, for ordering a diagnostic test, or for administering treatment. This in turn means that for some decision makers, the most rational strategy is to do nothing, whereas for others, it may be to order a diagnostic test, and still for others, choosing treatment may be the most rational course of action.

Wide Social Participation in Prioritizing Patients on Waiting Lists for Joint Replacement: A Conjoint Analysis

2008-07-25

Objective. The aim was to develop a priority scoring system for patients on waiting lists for joint replacement based on a wide social participation, and to analyze the differences among participants. Methods. Conjoint analysis. Focus groups in combination with a nominal technique were employed to identify the priority criteria (N=36). A rank-ordered logit model was then applied for scoring estimations. Participants (N=860) represented: consultants, allied-health professionals, patients and their relatives, and the general population of Catalonia. Results. Clinical and social criteria were selected, and their relative importance (over 100 points) was: pain (33), difficulty in doing activities of daily living (21), disease severity (18), limitations on ability to work (10), having someone to look after the patient (9), being a caregiver (6), and recovery probability (4). Estimated criteria coefficients had the expected positive sign and all were statistically significant (P < 0.001). There were differences between groups; pain was rated higher by patients/relatives, and difficulty in doing activities was rated lower by patients/relatives and the general public. Most interaction terms for these criteria and groups were significant (P < 0.001). Consultants and allied-health professionals had the most similar prioritization pattern (r=0.97). Conclusion. Both clinical and social criteria are considered for prioritization of joint replacement surgery from a wide social perspective. The preference among professional and social groups varies and this might impact the result of patient prioritization. A wide social participation for obtaining adequate prioritizing systems for patients on waiting lists is desirable.

The Effect of Graphical and Numerical Presentation of Hypothetical Prenatal Diagnosis Results on Risk Perception

Siegrist, M., Orlow, P., Keller, C. — 2008-07-25

Objective. To evaluate various formats for the communication of prenatal test results. Design. In study 1 (N=400), female students completed a questionnaire assessing risk perception, affect, and perceived usefulness of prenatal test results. A randomized, 2 (risk level; low, high) x 4 (format; ratio with numerator 1, ratio with denominator 1000, Paling Perspective Scale, pictograms) design was used. Study 2 (N=200) employed a 2 (risk level; low, high) x 2 (format; Paling Perspective Scale, risk comparisons in numerical format) design. Results. In study 1, the Paling Perspective Scale resulted in a higher level of perceived risk across different risk levels compared with the other formats. Furthermore, participants in the low-risk group perceived the test results as less risky compared with participants in the high-risk group (P < 0.001) when the Paling Perspective Scale was used. No significant differences between low and high risks were observed for the other 3 formats. In study 2, the Paling Perspective Scale evoked higher levels of perceived risks relative to the numerical presentation of risk comparisons. For both formats, we found that participants confronted with a high risk perceived test results as more risky compared with participants confronted with a low risk. Conclusion. The Paling Perspective Scale resulted in a higher level of perceived risk compared with the other formats. This effect must be taken into account when choosing a graphical or numerical format for risk communication.

Expectations of Benefit in Early-Phase Clinical Trials: Implications for Assessing the Adequacy of Informed Consent

2008-07-25

Background. Participants in early-phase clinical trials have reported high expectations of benefit from their participation. There is concern that participants misunderstand the trials to which they have consented, which is based on assumptions about what patients mean when responding to questions about likelihood of benefit. Methods. Participants were 27 women and 18 men in early-phase oncology trials at 2 academic medical centers in the United States. To determine whether expectations of benefit differ depending on how patients are queried, the authors randomly assigned participants to 1 of 3 interviews corresponding to 3 questions about likelihood of benefit: frequency type, belief type, and vague. In semistructured interviews, participants were queried about how they understood and answered the question. Participants then answered and discussed 1 of the other questions. Results. Expectations of benefit in response to the belief-type question were significantly greater than expectations in response to the frequency-type and vague questions (P=0:02). The most common justifications involved positive attitude (n=27 [60%]) and references to physical health (n=23 [51%]). References to positive attitude were most common among participants with higher (> 70%) expectations (n = 11 [85%]) and least common among those with lower ( < 50%) expectations (n = 3 [27%]). Conclusions. The wording of questions about likelihood of benefit shapes the expectations that patients express. Patients who express high expectations may not do so to communicate understanding but rather to register optimism. Ongoing research will clarify the meaning of high expectations and examine methods for assessing understanding.

Systematic Review: Health-State Utilities in Liver Disease: A Systematic Review

McLernon, D. J., Dillon, J., Donnan, P. T. — 2008-07-25

Objectives. Health-state utilities are essential for cost-utility analysis. Few estimates exist for liver disease in the literature. The authors' aim was to conduct a systematic review of health-state utilities in liver disease, to look at the variation of study designs used, and to pool utilities for some liver disease states. Methods. A search of MED-LINE, EMBASE, and CINAHL from 1966 to September 2006 was conducted including key words related to liver disease and utility measuring tools. Articles were included if health-state utility tools or expert opinion were used. Variance-weighted mean utility estimates were pooled using metaregression adjusting for disease state and utility assessment method. Results. Thirty studies measured utilities of liver diseases/disease states. Half of these estimated utilities for hepatitis viruses: hepatitis A (n = 1), hepatitis B (n = 4), and hepatitis C (n = 10). Others included liver transplant (n= 6) and chronic liver disease (n= 5) populations. Twelve utility methods were used throughout. The EQ-5D (n = 10) was most popular method, followed by visual analogue scale (n = 9), time tradeoff (n = 6), and standard gamble (n = 4). Respondents were patients (n= 16), an expert panel (n = 10), non—liver diseases adults ( n=2), patient and expert (n = 1), and patient and healthy adult (n = 1). Type of perspective included community (n=21), patient (n=4), and both (n = 5). The pooled mean estimates in hepatitis C with moderate disease, compensated cirrhosis, decompensated cirrhosis, and post—liver transplant using the EQ-5D were 0.75, 0.75, 0.67, and 0.71, respectively. The change in these utilities using different methods were -0.07 (visual analogue scale), -0.01 (health utilities index version 3), +0.04 (standard gamble), + 0.08 (health utilities index version 2), + 0.12 (time tradeoff), and + 0.15 (standard gamble—transformed visual analogue scale). Conclusions. The authors have created a valuable liver disease— based utility resource from which researchers and policy makers can easily view all available utility estimates from the literature. They have also estimated health-state utilities for major states of hepatitis C.

Modeling the Incubation Period of Inhalational Anthrax

Wilkening, D. A. — 2008-07-25

Ever since the pioneering work of Philip Sartwell, the incubation period distribution for infectious diseases is most often modeled using a lognormal distribution. Theoretical models based on underlying disease mechanisms in the host are less well developed. This article modifies a theoretical model originally developed by Brookmeyer and others for the inhalational anthrax incubation period distribution in humans by using a more accurate distribution to represent the in vivo bacterial growth phase and by extending the model to represent the time from exposure to death, thereby allowing the model to be fit to nonhuman primate time-to-death data. The resulting incubation period distribution and the dose dependence of the median incubation period are in good agreement with human data from the 1979 accidental atmospheric anthrax release in Sverdlovsk, Russia, and limited nonhuman primate data. The median incubation period for the Sverdlovsk victims is 9.05 (95% confidence interval = 8.0-10.3) days, shorter than previous estimates, and it is predicted to drop to less than 2.5 days at doses above 10⁶ spores. The incubation period distribution is important because the left tail determines the time at which clinical diagnosis or syndromic surveillance systems might first detect an anthrax outbreak based on early symptomatic cases, the entire distribution determines the efficacy of medical intervention—which is determined by the speed of the prophylaxis campaign relative to the incubation period—and the right tail of the distribution influences the recommended duration for antibiotic treatment.

Willingness to Pay for a Cure in Patients with Chronic Gout

Khanna, D., Ahmed, M., Yontz, D., Ginsburg, S. S., Tsevat, J. — 2008-07-25

Introduction. Gout is a chronic painful inflammatory arthritis. The authors interviewed patients with chronic stable gout to assess their hypothetical willingness to pay (WTP) to be cured of their gout. Patients and Methods. Patients with gout were asked how much money they would be willing to pay every month out of pocket or as a co-pay to cure their gout. To assess determinants of WTP amounts, the authors performed stepwise multivariable linear regression analysis, controlling for demographics, health status, and relative concern about gout. Results. Of the 78 patients, 70 (90%) were male, 54 (69%) were Caucasian, 21 (27%) were African American, and 32 (41%) had annual incomes < $25,000. The median WTP amount was $25 ($0, $75) per month, and the mean (s) was $52 ($74) per month (range, $0-$350); 23 (30%) patients were unwilling to pay any amount. Patients who rated their gout as their top health concern were willing to pay a median of $63 ($25, $100) per month. In multivariable analysis, gout as the top health concern, greater frequency of gouty attacks over the past 1 y, and younger age were significantly associated with WTP amounts (R² =0:19 ). Conclusion. Many patients with chronic gout would be willing to pay money every month in perpetuity to be cured of their gout. Younger patients, patients whose main health concern is gout, and patients with frequent attacks are willing to pay the most.

Errata

2008-07-25

Thank You to Our Reviewers

2008-06-02

The Half-Life of Truth: What Are Appropriate Time Horizons for Research Decisions?

Philips, Z., Claxton, K., Palmer, S. — 2008-06-02

Purpose. To evaluate alternative approaches taken to estimate the population that could benefit from research and to demonstrate that explicitly modeling future change leads to more appropriate estimates of the expected value of information (EVI). Methods. Existing approaches to estimating the population typically focus on the time horizon for decisions, employing seemingly arbitrary estimates of the appropriate horizon. These approaches implicitly use the time horizon as a proxy for future changes in technologies, prices, and information. Different approaches to quantifying the time horizon are explored, in the context of a stylized model, to demonstrate the impact of uncertainty in this estimate on EVI. An alternative approach is developed that explicitly models future changes in technologies, prices, and information and that demonstrates the impact on EVI estimates. Results. Explicitly modeling future changes means that the EVI for the decision problem may increase or decrease over time, but the EVI for the group of parameters that can be evaluated by current research tends to decline. The finite and infinite time horizons for the decision problem represent special cases (e.g., price shock or no changes, respectively). This type of analysis can be used to inform policy decisions relating to the timing of research. Conclusions. The value of information depends on future changes in technologies, prices, and evidence. Finite time horizons for decision problems can be seen as a proxy for the complex and uncertain process of future change. A more explicit approach to modeling these changes could provide a more appropriate basis for calculating EVI, but this raises a number of significant methodological and technical challenges.

The Option Value of Delay in Health Technology Assessment

Eckermann, S., Willan, A. R. — 2008-06-02

Processes of health technology assessment (HTA) inform decisions under uncertainty about whether to invest in new technologies based on evidence of incremental effects, incremental cost, and incremental net benefit monetary (INMB). An option value to delaying such decisions to wait for further evidence is suggested in the usual case of interest, in which the prior distribution of INMB is positive but uncertain. Methods of estimating the option value of delaying decisions to invest have previously been developed when investments are irreversible with an uncertain payoff over time and information is assumed fixed. However, in HTA decision uncertainty relates to information (evidence) on the distribution of INMB. This article demonstrates that the option value of delaying decisions to allow collection of further evidence can be estimated as the expected value of sample of information (EVSI). For irreversible decisions, delay and trial (DT) is demonstrated to be preferred to adopt and no trial (AN) when the EVSI exceeds expected costs of information, including expected opportunity costs of not treating patients with the new therapy. For reversible decisions, adopt and trial (AT) becomes a potentially optimal strategy, but costs of reversal are shown to reduce the EVSI of this strategy due to both a lower probability of reversal being optimal and lower payoffs when reversal is optimal. Hence, decision makers are generally shown to face joint research and reimbursement decisions (AN, DT and AT), with the optimal choice dependent on costs of reversal as well as opportunity costs of delay and the distribution of prior INMB.

Technical Note: Acceptability Curves Could Be Misleading When Correlated Strategies Are Compared

Sadatsafavi, M., Najafzadeh, M., Marra, C. — 2008-06-02

Discriminating Quality of Hospital Care in the United States

Normand, S.-L. T., Wolf, R. E., McNeil, B. J. — 2008-06-02

Background and objective. The Centers for Medicare and Medicaid Services (CMS) report quality of care for patients hospitalized with acute myocardial infarction (AMI), congestive heart failure (CHF), and community-acquired pneumonia (CAP) with the intention of rewarding superior performing hospitals. The aim of the study was to compare identification of superior hospitals for providing financial rewards using 2 different scoring systems: a latent score that weights individual clinical performance measures according to how well each discriminated hospital quality and a raw sum score (the system adopted by CMS). Methods. This observational cohort study used 2761 acute care hospitals in the United States reporting AMI clinical performance measures, 3271 reporting CHF measures, and 3714 hospitals reporting CAP measures. For each clinical condition, the main outcome measures included the average raw sum score, the latent score estimated from an item response theory (IRT) model, and the percentage of false negative superior designations made on the basis of raw sum scores relative to latent scores. Results. The average raw sum score was highest for AMI (88.8%) and lower for CHF (73.1%) and CAP (76.3%). AMI measures were equally nondiscriminating of hospital quality; hospital discharge instruction was most discriminating of CHF quality; pneumococcal vaccination was most discriminating of CAP quality. False negative rates varied 2-fold: AMI (10%), CHF (16%), and CAP (24%). Conclusions. Neither the AMI raw sum score nor latent score discriminates hospital quality due to ceiling effects. Current methods for aggregating measures result in different hospital superior designations than those based on the latent score. Organizations that financially reward hospitals on the basis of such scores need to assess predictive validity of scores and determine a minimum level of classification accuracy.

Impact of PSA Screening on the Incidence of Advanced Stage Prostate Cancer in the United States: A Surveillance Modeling Approach

Etzioni, R., Gulati, R., Falcon, S., Penson, D. F. — 2008-06-02

Background and objective. Both the detection and the treatment of prostate cancer have undergone important clinical advances. Simultaneously, both distant stage incidence and disease-specific mortality have fallen in the United States. A recent study suggests that if prostate-specific antigen (PSA) testing explains the decline in distant stage incidence, then it may be largely responsible for the decline in mortality. The objective was to quantify this link between PSA screening and the decline in distant stage incidence. Methods. A fixed-cohort simulation model of prostate cancer progression and screening was adapted to a population-based model that integrates new data on trends in testing and biopsy practices. The model was calibrated to pre-PSA incidence and then screening was superimposed, obtaining incidence projections in the absence and presence of testing. This approach permits calculation of clinically relevant measures for model validation and direct assessment of the role of testing in the distant stage incidence decline. Results. The model validated well with prior studies of natural history, and the sensitivity analysis indicated that the findings were robust to variation in model parameters. Model results indicate that PSA screening accounts for approximately 80% of the observed decline in distant stage incidence. Conclusions. PSA screening contributed to the observed declines in distant stage incidence and mortality in the 1990s. However, additional factors, such as increasing awareness of prostate cancer and advances in treatment, have probably also played a role in these trends.

Modeling the Logistics of Response to Anthrax Bioterrorism

2008-06-02

Background. A bioterrorism attack with an agent such as anthrax will require rapid deployment of medical and pharmaceutical supplies to exposed individuals. How should such a logistical system be organized? How much capacity should be built into each element of the bioterrorism response supply chain? Methods. The authors developed a compartmental model to evaluate the costs and benefits of various strategies for preattack stockpiling and postattack distribution and dispensing of medical and pharmaceutical supplies, as well as the benefits of rapid attack detection. Results. The authors show how the model can be used to address a broad range of logistical questions as well as related, nonlogistical questions (e.g., the cost-effectiveness of strategies to improve patient adherence to antibiotic regimens). They generate several key insights about appropriate strategies for local communities. First, stockpiling large local inventories of medical and pharmaceutical supplies is unlikely to be the most effective means of reducing mortality from an attack, given the availability of national and regional supplies. Instead, communities should create sufficient capacity for dispensing prophylactic antibiotics in the event of a large-scale bioterror attack. Second, improved surveillance systems can significantly reduce deaths from such an attack but only if the local community has sufficient antibiotic-dispensing capacity. Third, mortality from such an attack is significantly affected by the number of unexposed individuals seeking prophylaxis and treatment. Fourth, full adherence to treatment regimens is critical for reducing expected mortality. Conclusions. Effective preparation for response to potential bioterror attacks can avert deaths in the event of an attack. Models such as this one can help communities more effectively prepare for response to potential bioterror attacks.

Time-Tradeoff Utilities for Identifying and Evaluating a Minimum Data Set for Time-Critical Biosurveillance

2008-06-02

Background. Researchers and policy makers are interested in identifying, implementing, and evaluating a national minimum data set for biosurveillance. However, work remains to be done to establish methods for measuring the value of such data. Purpose. The purpose of this article is to establish and evaluate a method for measuring the utility of biosurveillance data. Method. The authors derive an expected utility model in which the value of data may be determined by trading data relevance for time delay in receiving data. In a sample of 23 disease surveillance practitioners, the authors test if such tradeoffs are sensitive to the types of data elements involved (chief complaint v. emergency department [ED] log of visit) and proportional changes to the time horizon needed for receiving data (24 v. 48 h). In addition, they evaluate the logical error rate: the proportion of responses that scored less relevant data as having higher utility. Results. Utilities of chief complaints were significantly higher than ED log of visit, F(1, 21)= 5.60, P < 0.05, suggesting the method is sensitive. Further utilities did not depend on time horizon used in the exercise, F(1, 21) = 0.00, P = ns. Of 92 time tradeoffs elicited, there were 5 logical errors (i.e., 5% logical error rate). Conclusions. In this article, the authors establish a time-tradeoff exercise for valuing biosurveillance data. Empirically, the method shows initial promise for evaluating a minimum data set for biosurveillance. Future applications of this approach may prove useful in disease surveillance planning and evaluation.

The Cost-Effectiveness of Counseling Strategies to Improve Adherence to Highly Active Antiretroviral Therapy among Men Who Have Sex with Men

Zaric, G. S., Bayoumi, A. M., Brandeau, M. L., Owens, D. K. — 2008-06-02

Objective: Inadequate adherence to highly active antiretroviral therapy (HAART) may lead to poor health outcomes and the development of HIV strains that are resistant to HAART. The authors developed a model to evaluate the cost-effectiveness of counseling interventions to improve adherence to HAART among men who have sex with men (MSM). Methods. The authors developed a dynamic compartmental model that incorporates HIV treatment, adherence to treatment, and infection transmission and progression. All data estimates were obtained from secondary sources. The authors evaluated a counseling intervention given prior to initiation of HAART and before all changes in drug regimens, combined with phone-in support while on HAART. They considered a moderate-prevalence and a high-prevalence population of MSM. Results. If the impact of HIV transmission is ignored, the counseling intervention has a cost-effectiveness ratio of $25,500 per quality-adjusted life year (QALY) gained. When HIV transmission is included, the cost-effectiveness ratio is much lower: $7400 and $8700 per QALY gained in the moderate- and high-prevalence populations, respectively. When the intervention is twice as costly per counseling session and half as effective as estimated in the base case (in terms of the number of individuals who become highly adherent, and who remain highly adherent), then the intervention costs $17,100 and $19,600 per QALY gained in the 2 populations, respectively. Conclusions. Counseling to improve adherence to HAART increased length of life, modestly reduced HIV transmission, and cost substantially less than $50,000 per QALY gained over a wide range of assumptions but did not reduce the proportion of drug-resistant strains. Such counseling provides only modest benefit as a tool for HIV prevention but can provide significant benefit for individual patients at an affordable cost.

A Test of Numeric Formats for Communicating Risk Probabilities

Cuite, C. L., Weinstein, N. D., Emmons, K., Colditz, G. — 2008-06-02

Background. Because people frequently encounter information about the probability of health risks, there is a need for research to help identify the best formats for presenting these probabilities. Methods. Three waves of participants were recruited from visitors to a cancer-related Internet site. Participants were presented with a hypothetical scenario that required them to perform 2 mathematical operations of the types that might be encountered in discussions of risk. Each wave encountered different operations. The operations used were compare, halve, triple, add, sequence, and tradeoff. Three numeric formats for communicating risk likelihoods were tested: percentages (e.g., 12%), frequencies (e.g., 12 in 100), and 1 in n (e.g., 1 in 8), and many levels of risk magnitude were crossed with the 3 formats. Results. The total sample of 16,133 individuals represented an overall participation rate of 36.1%. Although the relative performance of the formats varied by operation, aggregated across operations, the percentage and frequency formats had higher overall accuracy rates than the 1-in-n format (57% and 55% v. 45%, respectively). Participants with less education, African Americans, Hispanics, and women had more difficulty with the mathematical operations. Discussion. Percentage and frequency formats facilitate performance of simple operations on risk probabilities compared with the 1-in-n format, which should usually be avoided.

Do Patients' Communication Behaviors Provide Insight into Their Preferences for Participation in Decision Making?

2008-06-02

Background. The Institute of Medicine report "Crossing the Quality Chasm'' encourages physicians to tailor their approaches to care according to each patient's individual preferences for participation in decision making. How physicians should determine these preferences is unclear. Objective. The objective of this study is to assess whether judgments of patient communication behaviors, either globally or individually, can yield insight into patient preferences for participation in decision making. Methods. Using questionnaire responses to 3 items about the desired level of participation in decision making from a communication study involving 886 audiotaped visits between older patients and surgeons, the authors purposively selected 25 patients who preferred a large role and 25 who preferred a small role in decision making. Two independent raters listened to the audiotapes and coded them for the presence of 7 communication behaviors (question asking, information behavior, initiating, statements of preference, processing, resistance, deference). On the basis of their listening and coding, raters judged patient preferences for participation in decision making. Results. Neither rater accurately judged preferences for participation in decision making beyond chance agreement (kappa statistics: rater 1 = 0.16, rater 2 = 0.20). Inter-rater reliability for the communication behaviors was also generally poor. Area-under-the-curve values for all communication behaviors hovered around 0.50, indicating that none of the behaviors had adequate power to discriminate between patients preferring large versus small roles. Conclusion. Patient preferences for participation in decision making cannot be reliably judged during routine visits based on judgments of patient communication behaviors. Engaging patients in a discussion of preferences for decision making may be the best way to determine the role each wants to play in any given decision.

What Factors Influence Case Managers' Resource Allocation Decisions? A Systematic Review of the Literature

Fraser, K. D., Estabrooks, C. — 2008-06-02

Objective. Case managers' decisions directly affect the amount and type of services individual clients receive, as well as overall home care program available resources. We know little about the resource allocation decision-making processes of case managers. The question guiding this review was, ``What factors influence case managers' resource allocation decisions in home care?'' Methods. The authors did a systematic literature review to answer the above question. After assessing the articles for inclusion, they assessed the quality (internal validity) of each included study. They described the characteristics of the studies and provided a synthesis of the findings of the primary studies. Results. Five qualitative and 6 quantitative articles met the inclusion criteria for this review. The findings of these studies are equivocal. Despite this, the authors were able to create a preliminary taxonomy of the factors that influence case manager resource allocation decisions. Despite evidence-based decision making receiving so much attention in contemporary health care literature, the authors found a near absence of reference to research use in the context of case manager decision making. Conclusions. Currently, there are relatively few studies in the literature on the factors that influence, and how they are used in, case manager resource allocation decisions. Studies are often lacking in terms of conceptual clarity and theoretical framing. They are often not guided by theoretical frameworks and are not situated within the larger field of decision making or even within the clinical decision-making literature. These issues are impeding progress in this area.

Factors Associated with Obstetrician-Gynecologists' Response to the Women's Health Initiative Trial of Combined Hormone Therapy

Power, M. L., Baron, J., Schulkin, J. — 2008-06-02

The Women's Health Initiative trial of combined estrogen and progestin (WHI E+P) ended prematurely after preliminary evidence indicated that harms exceeded benefits, with no cardiovascular benefit. There was controversy over the results and the decision to end the trial early, with many obstetrician-gynecologists expressing reservations about the evidence. The Research Department of the American College of Obstetricians and Gynecologists conducted a study regarding the WHI E+P, sending questionnaires to 2500 randomly selected Fellows; 703 Fellows returned usable surveys (28.1%). Despite almost universal awareness of the results of the WHI E+P (> 97%), almost half of the responding physicians did not find the results convincing and disagreed with the decision to stop the trial. In this further examination of the data, we identified characteristics of the respondents who were associated with either accepting or rejecting the WHI E+P. The year residency was completed, the relative importance a respondent attributed to randomized clinical trials (RCTs), concern about harms of action, and opinion of alternative therapies were significant factors. One of 5 respondents found the results convincing and agreed with the decision to end the trial (acceptors). One of 3 respondents did not find the results convincing and disagreed with the decision to end the trial (rejectors). Acceptors had completed residency more recently (1991 v. 1985, P = 0.001), rated evidence from RCTs as more important (P = 0.006), were more concerned with harms of action (22.4% v. 10.6%, P = 0.004), and were more likely to have a favorable opinion of alternative therapies to hormone therapy (64.1% v. 44.4%, P < 0.001).

A Cost-Effectiveness Framework for Profiling the Value of Hospital Care

Timbie, J. W., Newhouse, J. P., Rosenthal, M. B., Normand, S.-L. T. — 2008-06-02

Provider profiling and performance-based incentive programs have expanded in recent years but need a theoretical framework for measuring and comparing the ``value'' of clinical care across medical providers. Cost-effectiveness analysis provides such a framework but has rarely been used outside of the treatment choice context. The authors present a profiling framework based on cost-effectiveness methods and illustrate their approach using data on in-hospital survival and the cost of care for a heart attack from a sample of Massachusetts hospitals during fiscal year 2003. They model each outcome using hierarchical models that allow performance to vary across hospitals as a function of a latent quality effect and an effect of case mix. They also estimate incremental outcomes by conditioning on each hospital's pair of random effects, using indirect standardization to estimate ``expected'' outcomes, and then taking their difference. Incremental cost and effectiveness outcomes are combined using incremental net monetary benefits. Using cost-effectiveness methods to profile hospital ``value'' permits the comparison of the benefit of a service relative to the cost using existing societal weights.

Evaluation of the GIDEON Expert Computer Program for the Diagnosis of Imported Febrile Illnesses

2008-06-02

Objective. The authors evaluate the performance of the expert system Global Infectious Diseases and Epidemiology Network (GIDEON) in diagnosing febrile illnesses occurring after a stay in the tropics. Methods. One investigator (E.B.) entered into the program the collected characteristics of 161 febrile travelers randomly extracted from a database of 1842 cases prospectively included during a study on imported fever. Accuracy was considered acceptable if the correct diagnosis appeared in the top 5 GIDEON ranking list. Interuser agreement was assessed by J.V.d.E. and J.M., who also entered the data of the first 50 sample cases with an established diagnosis. Results. The sample was epidemiologically and clinically representative of the whole cohort. An infectious etiology had been established in 129 cases; diagnosis was unknown in 31 cases and non-infectious in 1 case. GIDEON generated a median of 29 diagnoses per case, including 23 with a probability lower than 1%. Accuracy was acceptable in 64% of the 129 fevers with infectious etiology. It tended to decrease when more than 3 findings were entered per case. Eleven (8%) severe conditions were rejected by GIDEON because non-disease-related characteristics had been introduced. In other cases, the posttest probability was inadequately affected by the insufficient weight of absent relevant findings. Interuser agreement was good for acceptable accuracy and final ranking (kappa=0.83 and 0.72, respectively). Conclusion. The performance of GIDEON in diagnosing imported fever is relatively good and reproducible but is impaired by some conceptual weaknesses. Its use might be hazardous for inexperienced physicians.

Incorporation of Process Preferences within the QALY Framework: A Study of Alternative Methods

McNamee, P., Seymour, J. — 2008-06-02

Objective. This article explores the implications of incorporating process preferences using time tradeoff and standard gamble methods to assess the benefits of health care. Methods. Data were derived from 2 sources: a randomized controlled trial of alternative palliative care treatments (plastic stents, thermal ablation, or brachytherapy) for esophageal cancer, and a valuation survey conducted among individuals who had previously undergone curative treatment for such cancer. Costs and quality-adjusted life years (QALYs) associated with different palliative treatments in terms of health outcome values were compared to costs and QALYs based on process values derived from 3 different treatment allocation methods: 1) receipt of most preferred treatment; 2) receipt of least preferred treatment; and 3) mean process values. Results. Process values produced a different number of QALYs and QALY gains compared to those derived from health outcome values. However, treatment recommendations based on process values corresponded with those based on health outcome values: brachytherapy was identified as the more cost-effective treatment in terms of the incremental cost-per-QALY ratio by both the standard health outcome values approach and methods based on process values. These findings were supported by probabilistic analysis using the net monetary benefit framework. Conclusions. Estimation of process preferences provides additional information to policy makers in judgments over the cost-effectiveness of health care programs. These methods offer a promising alternative to standard cost-per-QALY estimation using health outcomes. However, further research examining the role of process preferences in decision making in other clinical applications appears warranted.

Highlights of This Issue

2008-03-26

Policy Rounds: A New Series and a Call for Papers

Helfand, M., Sanders, G. D. — 2008-03-26