Evidence from clinical trials should contribute to informed decision making and a learning health care system. People frequently, however, find participating in clinical trials meaningless or disempowering. Moreover, people often do not incorporate trial results directly into their decision making. The lack of patient centeredness in clinical trials may be partially addressed through trial design. For example, Bayesian adaptive trials designed to adjust in a pre-specified manner to changes in clinical practice could motivate people and their health care providers to view clinical trials as more applicable to real-world clinical decisions. The way in which clinical trials are designed can transform the evidence generation process to be more patient centered, providing people with an incentive to participate or continue participating in clinical trials. In order to achieve the transformation to patient-centeredness in clinical trial decisions, however, there is a need for transparent and reliable methods and education of trial investigators and site personnel.
patient-centered; trial design; pragmatic; Bayesian; adaptive; RCT
The value of the information that genetic testing services provide can be questioned for insurance-based health systems. The results of genetic tests oftentimes may not lead to well-defined clinical interventions; however, Lynch syndrome, a genetic mutation for which carriers are at an increased risk for colorectal cancer, can be identified through genetic testing, and meaningful health interventions are available via increased colonoscopic surveillance. Valuations of test information for such conditions ought to account for the full impact of interventions and contingent outcomes.
To conduct a discrete-choice experiment to elicit individuals’ preferences for genetic test information.
A Web-enabled discrete-choice experiment survey was administered to a representative sample of US residents aged 50 years and older. In addition to specifying expenditures on colonoscopies, respondents were asked to make a series of nine selections between two hypothetical genetic tests or a no-test option under the premise that a relative had Lynch syndrome. The hypothetical genetic tests were defined by the probability of developing colorectal cancer, the probability of a false-negative test result, privacy of the result, and out-of-pocket cost. A model specification identifying necessary interactions was derived from assumptions of risk behavior and the decision context and was estimated using random-parameters logit.
A total of 650 respondents were contacted, and 385 completed the survey. The monetary equivalent of test information was approximately $1800. Expenditures on colonoscopies to reduce mortality risks affected valuations. Respondents with lower income or who reported being employed significantly valued genetic tests more.
Genetic testing may confer benefits through the impact of subsequent interventions on private individuals.
colorectal cancer; discrete choice experiment; genetic testing; Lynch syndrome
Variation in care within and across geographic areas remains poorly understood. The goal of this paper is to examine whether physician social networks—as defined by shared patients—are associated with rates of complications following radical prostatectomy.
In five cities, we constructed networks of physicians based on their shared patients in 2004–2005 SEER-Medicare data. From these networks, we identified subgroups of urologists who most frequently shared patients with one another. Among men with localized prostate cancer who underwent radical prostatectomy, we used multilevel analysis with generalized linear mixed effect models to examine whether physician network structure—along with specific characteristics of the network subgroups—was associated with rates of 30-day and late urinary complications, and long term incontinence after accounting for patient-level sociodemographic, clinical factors, and urologist patient volume.
Networks included 2677 men in 5 cities who underwent radical prostatectomy. The unadjusted rate of 30-day surgical complications varied across network subgroups from an 18.8 percentage point difference in the rate of complications across network subgroups in City 1 to 26.9 percentage point difference in City 5. Large differences in unadjusted rates of late urinary complications and long term incontinence across subgroups were similarly found. Network subgroup characteristics—average urologist centrality and patient racial composition—were significantly associated with rates of surgical complications.
Analysis of physician networks of SEER-Medicare data provides insight into observed variation in rates of complications for localized prostate cancer. If validated, such approaches may be used to target future quality improvement interventions.
cancer; claims data; health services; outcomes research
Computerized provider order entry (CPOE) is the process of entering physician orders directly into an electronic health record. Although CPOE has been shown to improve medication safety and reduce health care costs, these improvements have been demonstrated largely in the inpatient setting; the cost-effectiveness in the ambulatory setting remains uncertain.
The objective was to estimate the cost-effectiveness of CPOE in reducing medication errors and adverse drug events (ADEs) in the ambulatory setting.
We created a decision-analytic model to estimate the cost-effectiveness of CPOE in a midsized (400 providers) multidisciplinary medical group over a 5-year time horizon— 2010 to 2014— the time frame during which health systems are implementing CPOE to meet Meaningful Use criteria. We adopted the medical group’s perspective and utilized their costs, changes in efficiency, and actual number of medication errors and ADEs. One-way and probabilistic sensitivity analyses were conducted. Scenario analyses were explored.
In the base case, CPOE dominated paper prescribing, that is, CPOE cost $18 million less than paper prescribing, and was associated with 1.5 million and 14,500 fewer medication errors and ADEs, respectively, over 5 years. In the scenario that reflected a practice group of five providers, CPOE cost $265,000 less than paper prescribing, was associated with 3875 and 39 fewer medication errors and ADEs, respectively, over 5 years, and was dominant in 80% of the simulations.
Our model suggests that the adoption of CPOE in the ambulatory setting provides excellent value for the investment, and is a cost-effective strategy to improve medication safety over a wide range of practice sizes.
adverse drug events; ambulatory care; computerized physician order entry system; cost-benefit analysis (cost-effectiveness); medication errors
To understand how older adults perceive their risk of Alzheimer’s Disease (AD) and how this may shape their medical care decisions, we examined whether presence of established risk factors of AD is associated with individuals’ perceived risk of AD, and with preference for preventing AD.
Participants: Data came from the US Health and Retirement Study participants who were asked questions on AD risk perception (N = 778). Measurements: Perceived risk of AD was measured by respondents’ estimate of their percent chance (0–100) developing AD in the next 10 years. Preference for AD prevention was measured with questions eliciting willingness to pay for a drug to prevent AD. Analysis: Multivariate linear regressions were used to estimate correlates of perceived risk and preference for prevention.
Better cognitive functioning and physical activity are associated with decreased perceived risk. Neither age nor cardiovascular disease is associated with perceived risk. African Americans have lower perceived risk than non-Latino whites; the difference is wider among people age 65 and above. Only 4% to 7% of the variation in perceived risk was explained by the model. Preference for prevention is stronger with increased perceived risk, but not with the presence of risk factors. Persons with better cognitive functioning, physical functioning, or wealth status have a stronger preference for prevention.
Some known risk factors appear to inform, but only modestly, individuals’ perceived risk of AD. Furthermore, decisions about AD prevention may not be determined by objective needs alone, suggesting a potential discrepancy between need and demand for AD preventive care.
Alzheimer’s disease; prevention preference; risk perception; willingness to pay
Inverse probability of treatment weighting (IPTW) has been used in observational studies to reduce selection bias. To obtain estimates of the main effects, a pseudo data set is created by weighting each subject by IPTW and analyzed with conventional regression models. Currently variance estimation requires additional work depending on type of outcomes. Our goal is to demonstrate a statistical approach to directly obtain appropriate estimates of variance of the main effects in regression models.
We carried out theoretical and simulation studies to show that the variance of the main effects estimated directly from regressions using IPTW is underestimated, and that the type I error rate is higher due to the inflated sample size in the pseudo data. The robust variance estimator using IPTW often slightly overestimates the variance of the main effects. We propose to use the stabilized weights to directly estimate both the main effect and its variance from conventional regression models.
We applied the approach to a study examining the effectiveness of serum potassium monitoring in reducing hyperkalemia-associated adverse events among 27,355 diabetic patients newly-prescribed a renin-angiotensin-aldosterone system (RAAS) inhibitor. The incidence rate ratio (with monitoring versus without monitoring) and confidence intervals were 0.46 (0.34, 0.61) using the stabilized weights compared to 0.46 (0.38, 0.55) using typical inverse probability of treatment weighting.
Our theoretical, simulation results and real data example demonstrate that the use of the stabilized weights in the pseudo data preserves the sample size of the original data, produces appropriate estimation of the variance of main effect, and maintains an appropriate type I error rate.
Inverse probability of treatment weighting; stabilized weights; type I error rates; incidence rate ratio; confidence intervals
Profile instruments are frequently used to assess health-related quality of life and other patient-reported outcomes. Preference-based measures are required for health-economic cost-utility evaluations. Although regression-based approaches are commonly used to map from profile measures to preference measures, we show that this results in biased estimates because of regression to the mean. Linking (scale-aligning) is proposed as an alternative.
Mapping outcomes; Linking scales; Test equating; Scale aligning; Profile instruments; Preference-based measures; Patient-reported outcomes
Patient-reported outcomes; Comparative effectiveness research; Patient-centered outcomes research; PRO; CER; PCOR
To link pediatric health-related quality of life (HRQOL) and health conditions by establishing clinically meaningful cutoff scores for a HRQOL instrument, the PedsQL.
We conducted telephone interviews with 1745 parents whose children were between 2–18 years old and enrolled in the Florida KidCare program and Children’s Medical Services Network in 2006. Two anchors, the Children with Special Health Care Needs (CSHCN) Screener and the Clinical Risk Groups (CRGs), were used to identify children with special health care needs or chronic conditions. We established cutoff scores for the PedsQL’s physical, emotional, social, school, and total functioning using the areas under the curves (AUCs) to determine the discriminative property of the PedsQL referring to the anchors.
The discriminative property of the PedsQL was superior, especially in total functioning (AUC > 0.7), between children with special health care needs (based on the CSHCN Screener) and with moderate and major chronic conditions (based on the CRGs) as compared to healthy children. For children < 8 years, the recommended cutoff scores for using total functioning to identify CSHCN were 83, 79 for moderate and 77 for major chronic conditions. For children ≥ 8 years, the cutoff scores were 78, 76 and 70, respectively.
Pediatric HRQOL varied with health conditions. Establishing cutoff scores for the PedsQL’s total functioning is a valid and convenient means to potentially identify children with special health care needs or chronic conditions. The cutoff sores can help clinicians to conduct further in-depth clinical assessments.
Children; clinically meaningful difference; cutoff; health-related quality of life; PedsQL
Generic preference-based measures were criticized for being inappropriate in some conditions. One solution is to include “bolt-on” dimensions describing additional specific health problems.
This study aimed to develop bolt-on dimensions to the EuroQol five-dimensional questionnaire (EQ-5D) and assess their impact on health state values.
Bolt-on dimensions were developed for vision problems, hearing problems, and tiredness. Each bolt-on dimension had three severity levels to match the EQ-5D. Three “core” EQ-5D states across a range of severity were selected, and each level of a bolt-on item was added, resulting in nine states in each condition. Health states with and without the bolt-on dimensions were valued by 300 members of the UK general public using time trade-off in face-to-face interviews, and mean health state values were compared using t tests. Regression analysis examined the impact of the bolt-on variants and the level of the bolt-on items after controlling for sociodemographic characteristics.
Bolt-on dimensions had an impact on health state values of the EQ-5D; however, the size, direction, and significance of the impact depend on the severity of the core EQ-5D state and of the bolt-on dimension. Regression analysis demonstrated that after controlling for possible differences in sociodemographic characteristics between the groups, there were no significant differences in health state values between the three bolt-on dimensions but confirmed that the impact depended on the severity of the EQ-5D health state and the levels of bolt-on dimensions.
The impact of a bolt-on dimension on the EQ-5D depends on the core health state and the level of the bolt-on dimension. Further research in this area is encouraged.
bolt-on; EQ-5D; health state valuation; hearing; tiredness; vision
Evidence-based healthcare decisions are best informed by comparisons of all relevant interventions used to treat conditions in specific patient populations. Observational studies are being performed to help fill evidence gaps. However, widespread adoption of evidence from observational studies has been limited due to a variety of factors, including the lack of consensus regarding accepted principles for their evaluation and interpretation. Two Task Forces were formed to develop questionnaires to assist decision makers in evaluating observational studies, with one Task Force addressing retrospective research and the other prospective research. The intent was to promote a structured approach to reduce the potential for subjective interpretation of evidence and drive consistency in decision-making. Separately developed questionnaires were combined into a single questionnaire consisting of 33 items. These were divided into two domains: relevance and credibility. Relevance addresses the extent to which findings, if accurate, apply to the setting of interest to the decision maker. Credibility addresses the extent to which the study findings accurately answer the study question. The questionnaire provides a guide for assessing the degree of confidence that should be placed from observational studies and promotes awareness of the subtleties involved in evaluating those.
bias; checklist; comparative effectiveness research; confounding; consensus; credibility; decision-making; epidemiologic research design; observational study methods; prospective observational study; publishing standards; quality; questionnaire; relevance; retrospective observational study; validity
The appropriate development of a model begins with understanding the problem that is being represented. The aim of this article was to provide a series of consensus-based best practices regarding the process of model conceptualization. For the purpose of this series of articles, we consider the development of models whose purpose is to inform medical decisions and health-related resource allocation questions. We specifically divide the conceptualization process into two distinct components: the conceptualization of the problem, which converts knowledge of the health care process or decision into a representation of the problem, followed by the conceptualization of the model itself, which matches the attributes and characteristics of a particular modeling type with the needs of the problem being represented. Recommendations are made regarding the structure of the modeling team, agreement on the statement of the problem, the structure, perspective, and target population of the model, and the interventions and outcomes represented. Best practices relating to the specific characteristics of model structure and which characteristics of the problem might be most easily represented in a specific modeling method are presented. Each section contains a number of recommendations that were iterated among the authors, as well as among the wider modeling taskforce, jointly set up by the International Society for Pharmacoeconomics and Outcomes Research and the Society for Medical Decision Making.
conceptualization; best practices; methods; modeling
The phase 3 trial, RESPOND-2, demonstrated that the addition of boceprevir (BOC) to peginterferon-ribavirin (PR) resulted in significantly higher rates of sustained virologic response (SVR) in previously treated patients with chronic hepatitis C virus (HCV) genotype-1 infection as compared with PR alone. We evaluated the cost-effectiveness of treatment with boceprevir in previously treated chronic hepatitis C patients in the United States utilizing treatment related data from RESPOND-2 and PROVIDE.
We developed a Markov cohort model to project the burden of HCV disease, lifetime costs and quality-adjusted life years (QALY) associated with PR and two BOC-based therapies—response-guided therapy (BOC/RGT) and fixed-duration therapy for 48 weeks (BOC/PR48). We estimated treatment related inputs (efficacy, adverse events, and discontinuations) from clinical trials and obtained disease progression rates, costs and quality-of-life data from published studies. We estimated the incremental cost-effectiveness ratio (ICER) for BOC-based regimens as studied in RESPOND-2, as well as by patient’s prior response to treatment and IL-28B genotype.
Boceprevir-based regimens were projected to reduce the lifetime incidence of liver-related complications by 43–53% in comparison with treatment with PR. The ICER of BOC/RGT in comparison with PR was $30,200, and the ICER of BOC/PR48 in comparison with BOC/RGT was $91,500. At $50,000 willingness-to-pay, the probabilities BOC/RGT and BOC/PR48 being the preferred option were 0.74 and 0.25, respectively.
In patients previously treated for chronic HCV genotype-1 infection, boceprevir was projected to increase QALYs and reduce the lifetime incidence of liver complications. In addition, boceprevir-based therapies were projected to be cost-effective in comparison with PR alone at commonly used willingness-to-pay thresholds.
protease inhibitor; Hepatitis C; Markov model; decision analytic
To estimate the cost-effectiveness of a trial of labor after one previous cesarean (TOLAC) when incorporating long-term events and outcomes.
A Markov model comparing TOLAC with elective repeat cesarean delivery (ERCD) was developed for a hypothetical cohort with no contraindication to a TOLAC. Women were selected from a prospective study to derive probability estimates for potential events through three subsequent pregnancies. Probabilities for cerebral palsy and stress urinary incontinence, cost data, and quality adjusted life years (QALYs) were obtained from the literature. The primary outcome was cost-effectiveness measured as the marginal cost per QALY gained, with a $50,000 threshold per QALY used to define cost-effectiveness.
The TOLAC strategy dominated the ERCD strategy at baseline, with $164.2 million saved and 500 QALYs gained per 100,000 women. The model was sensitive to six variables: the probability of uterine rupture and successful TOLAC among women with no prior vaginal delivery, the frequency of stress urinary incontinence, and the costs of failed TOLAC, successful TOLAC, and ERCD. When the probability of TOLAC success was at the base value, 67.2%, TOLAC was preferred if the probability of uterine rupture was 3.1% or less. When the probability of uterine rupture was at the base value, 0.8%, the TOLAC strategy was preferred as long as the probability of success was 47.2% or more. Probabilistic sensitivity analysis confirmed the base-case analysis.
Under baseline circumstances, TOLAC is less expensive and more effective than an ERCD when considering long-term consequences when the likelihood of success is 47.2% or more.
cost-effectiveness; trial of labor; elective repeat; accreta
Radical cystectomy (RC) is the standard treatment for muscle-invasive Urothelial carcinoma of the bladder (UCB). Tri-modality bladder preserving therapy (BPT) is an alternative to RC, but randomized comparisons of RC versus BPT have proven infeasible. To compare RC versus BPT, we undertook an observational cohort study using registry and administrative claims data from the SEER-Medicare database.
We identified patients age 65 years or older diagnosed between 1995 and 2005 who received RC (n=1,426) or BPT (n=417). We examined confounding and stage misclassification in the comparison of RC and BPT using multivariable adjustment, propensity score-based adjustment, instrumental variable (IV) analysis and simulations.
Patients who received BPT were older and more likely to have comorbid disease. After propensity score adjustment, BPT was associated with an increased hazard of death from any cause (HR 1.26; 95% CI, 1.05 – 1.53) and from bladder cancer (HR 1.31; 95% CI, 0.97 – 1.77). Using the local area cystectomy rate as an instrument, IV analysis demonstrated no differences in survival between BPT and RC (death from any cause HR 1.06; 95% CI, 0.78 – 1.31; death from bladder cancer HR 0.94; 95% CI, 0.55 – 1.18). Simulation studies for stage misclassification yielded results consistent with the IV analysis.
Survival estimates in an observational cohort of patients who underwent RC versus BPT differ by analytic method. Multivariable and propensity score adjustment revealed greater mortality associated with BPT relative to RC, while IV analysis and simulation studies suggest that the two treatments are associated with similar survival outcomes.
Comparative Effectiveness Research; Urinary bladder neoplasms; Cystectomy; Radiotherapy; Chemotherapy; SEER Program
The SF-6D preference-based scoring system was developed several years after the SF-12 and SF-36 instruments. A method to predict SF-6D scores from information in previous reports would facilitate backwards comparisons and the use of these reports in cost-effectiveness analyses.
This report uses data from the 2001–2003 Medical Expenditures Panel Survey (MEPS), the Beaver Dam Health Outcomes Survey, and the National Health Measurement Study. SF-6D scores were modeled using age, sex, mental component summary (MCS) score, and physical component summary (PCS) score from the 2002 MEPS. The resulting SF-6D prediction equation was tested with the other datasets for groups of different sizes and groups stratified by age, MCS score, PCS score, sum of MCS and PCS scores, and SF-6D score.
The equation can be used to predict an average SF-6D score using average age, proportion female, average MCS score, and average PCS score. Mean differences between actual and predicted average SF-6D scores in out-of-sample tests was −0.001 (SF-12 version 1), −0.013 (SF-12 version 2), −0.007 (SF-36 version 1), and −0.010 (SF-36 version 2). Ninety-five percent credible intervals around these point estimates range from ±0.045 for groups with 10 subjects to ±0.008 for groups with more than 300 subjects. These results were consistent for a wide range of ages, MCS scores, PCS scores, sum of MCS and PCS scores, and SF-6D scores. SF-6D scores from the SF-36 and SF-12 from the same data set were found to be substantially different.
Simple equation predicts an average SF-6D preference-based score from widely published information.
MCS; PCS; prediction; SF-6D; SF-36; SF-12
Invasive pneumococcal disease is a major cause of preventable morbidity and mortality in the United States, particularly among the elderly (>65 years). There are large racial disparities in pneumococcal vaccination rates in this population. Here, we estimate the cost-effectiveness of a hypothetical national vaccination intervention program designed to eliminate racial disparities in pneumococcal vaccination in the elderly.
In an exploratory analysis, a Markov decision-analysis model was developed, taking a societal perspective and assuming a 1-year cycle length, 10-year vaccination program duration, and lifetime time horizon. In the base-case analysis, it was conservatively assumed that vaccination program promotion costs were $10 per targeted minority elder per year, regardless of prior vaccination status and resulted in the elderly African American and Hispanic pneumococcal vaccination rate matching the elderly Caucasian vaccination rate (65%) in year 10 of the program.
The incremental cost-effectiveness of the vaccination program relative to no program was $45,161 per quality-adjusted life-year gained in the base-case analysis. In probabilistic sensitivity analyses, the likelihood of the vaccination program being cost-effective at willingness-to-pay thresholds of $50,000 and $100,000 per quality-adjusted life-year gained was 64% and 100%, respectively.
In a conservative analysis biased against the vaccination program, a national vaccination intervention program to ameliorate racial disparities in pneumococcal vaccination would be cost-effective.
cost-effectiveness; disparities; invasive pneumococcal disease; vaccination
A new method is presented for both synthesizing treatment effects on multiple outcomes subject to measurement error and estimating coherent mapping coefficients between all outcomes. It can be applied to sets of trials reporting different combinations of patient- or clinician-reported outcomes, including both disease-specific measures and generic health-related quality-of-life measures. It is underpinned by a structural equation model that includes measurement error and latent common treatment effect factor. Treatment effects can be expressed on any of the test instruments that have been used.
This is illustrated in a synthesis of eight placebo-controlled trials of TNF-α inhibitors in ankylosing spondylitis, each reporting treatment effects on between two and five of a total six test instruments.
The method has advantages over other methods for synthesis of multiple outcome data, including standardization and multivariate normal synthesis. Unlike standardization, it allows synthesis of treatment effect information from test instruments sensitive to different underlying constructs. It represents a special case of previously proposed multivariate normal models for evidence synthesis, but unlike the former, it also estimates mappings. Combining synthesis and mapping as a single operation makes more efficient use of available data than do current mapping methods and generates treatment effects that are consistent with the mappings. A limitation, however, is that it can only generate mappings to and from those instruments on which some trial data exist.
The method should be assessed in a wide range of data sets on different clinical conditions, before it can be used routinely in health technology assessment.
congeneric tests; cross-walking; mapping; multioutcome synthesis
Patient reported outcomes (PRO) assessing multiple gastrointestinal symptoms are central to characterizing the therapeutic benefit of novel agents for irritable bowel syndrome (IBS). Common approaches that sum or average responses across different illness components must be unidimensional and have small unique variances to avoid aggregation bias and misinterpretation of clinical data. This study sought to evaluate the unidimensionality of the IBS Symptom Severity Scale (IBS-SSS) and to explore person centered cluster analytic methods for characterizing multivariate-based patient profiles.
Ninety-eight Rome-diagnosed IBS patients completed the IBS-SSS and a single, global item of symptom severity (UCLA Symptom Severity Scale) at pretreatment baseline of an NIH funded clinical trial. A k-means cluster analyses were performed on participants symptom severity scores.
The IBS-SSS was not unidimensional. Exploratory cluster analyses revealed four common symptom profiles across five items of the IBS-SSS. One cluster of patients (25%) had elevated scores on pain frequency and bowel dissatisfaction, with less elevated but still high scores on life interference and low pain severity ratings. A second cluster (19%) was characterized by intermediate scores on both pain dimensions, but more elevated scores on bowel dissatisfaction. A third cluster (18%) was elevated across all IBS-SSS sub-components. The fourth and most common cluster (37%) had relatively low scores on all dimensions except bowel dissatisfaction and life interference due to IBS symptoms.
PRO endpoints and research on IBS more generally relying on multicomponent assessments of symptom severity should take into account the multidimensional structure of symptoms to avoid aggregation bias and to optimize the sensitivity of detecting treatment effects.
disease severity; questionnaire development; outcome research; health status indicators; rating scale; psychometric properties; global assessment
An article by Lu et al. in this issue of Value in Health addresses the mapping of treatment or group differences in disease-specific measures (DSMs) of health-related quality of life onto differences in generic health-related quality-of-life scores, with special emphasis on how the mapping is affected by the reliability of the DSM. In the proposed mapping, a factor analytic model defines a conversion factor between the scores as the ratio of factor loadings. Hence, the mapping applies to convert true underlying scales and has desirable properties facilitating the alignment of instruments and understanding their relationship in a coherent manner. It is important to note, however, that when DSM means or differences in mean DSMs are estimated, their mapping is still of a measurement error–prone predictor, and the correct conversion coefficient is the true mapping multiplied by the reliability of the DSM in the relevant sample. In addition, the proposed strategy for estimating the factor analytic mapping in practice requires assumptions that may not hold. We discuss these assumptions and how they may be the reason we obtain disparate estimates of the mapping factor in an application of the proposed methods to groups of patients.
cross-walk; HRQOL; mapping; reliability
In health technology assessment, decisions about reimbursement for new health technologies are largely based on effectiveness estimates. Sometimes, however, the target effectiveness estimates are not readily available. This may be because many alternative instruments measuring these outcomes are being used (and not all always reported) or an extended follow-up time of clinical trials is needed to evaluate long-term end points, leading to the limited data on the target clinical outcome. In the areas of highest priority in health care, decisions are required to be made on a short time scale. Therefore, alternative clinical outcomes, including surrogate end points, are increasingly being considered for use in evidence synthesis as part of economic evaluation.
To illustrate the potential effect of reduced uncertainty around the clinical outcome on the utility when estimating it from a multivariate meta-analysis.
Bayesian multivariate meta-analysis has been used to synthesize data on correlated outcomes in rheumatoid arthritis and to incorporate external data in the model in the form of informative prior distributions. Estimates of Health Assessment Questionnaire were then mapped onto the health-related quality-of-life measure EuroQol five-dimensional questionnaire, and the effect was compared with mapping the Health Assessment Questionnaire obtained from the univariate approach.
The use of multivariate meta-analysis can lead to reduced uncertainty around the effectiveness parameter and ultimately uncertainty around the utility.
By allowing all the relevant data to be incorporated in estimating clinical effectiveness outcomes, multivariate meta-analysis can improve the estimation of health utilities estimated through mapping methods. While reduced uncertainty may have an effect on decisions based on economic evaluation of new health technologies, the use of short-term surrogate end points can allow for early decisions. More research is needed to determine the circumstances under which uncertainty is reduced.
Bayesian analysis; health technology assessment; meta-analysis; multiple end points; rheumatoid arthritis; surrogate end points
To examine cost responsiveness and total costs associated with a simulated “value based” insurance design (VBID) for statin therapy in a Medicare population with diabetes.
Four-year panels constructed from the 1997–2005 Medicare Current Beneficiary Survey selected by self-report or claims-based diagnoses of diabetes in Year 1, and use of statins in Year 2 (N= 899). We computed number of 30-day statin prescription fills, out-of-pocket (OOP) and third party drug costs, and Medicare Part A and B spending.
Multivariate ordinary least squares regression models predicted statins fills as a function of OOP costs, and a generalized linear model with log link predicted Medicare spending as a function of number of fills, controlling for baseline characteristics. Estimated coefficients were used to simulate changes in fills associated with copayment caps from $25 to $1, and to compute changes in 3rd-party payments and Medicare cost offsets associated with incremental fills. Analyses were stratified by patient cardiovascular event risk.
A simulated OOP price of $25[$1] increased plan drug spending by $340[$794], and generated Medicare Part A/B savings of $262[$531]. Medicare Part A/B savings were greater for higher risk patients, generating a net savings for the plans.
Reducing statin copayments for Medicare beneficiaries with diabetes resulted in modestly increased use, and reduced medical spending. The VBID simulation strategy met financial feasibility criteria, but only for higher risk patients.
diabetes; Medicare; medication adherence; cost offsets
To describe rank reversal as a source of inconsistent interpretation intrinsic to indirect comparison (Bucher et al., 1997) of treatments and to propose best practice.
We prove our main points with intuition, examples, graphs, and mathematical proofs. We also provide software and discuss implications for research and policy.
When comparing treatments by indirect means and sorting them by effect size, three common measures of comparison (risk ratio, risk difference, and odds ratio) may lead to vastly different rankings.
The choice of risk measure matters when making indirect comparisons of treatments. The choice should depend primarily on the study design and conceptual framework for that study.
risk ratio; risk difference; odds ratio; indirect comparisons; risk
To develop a set of EQ-5D health state values for the Argentine general population.
Consecutive subjects attending six primary care centers in Argentina were selected based on quota sampling and interviewed using the EuroQol Group protocol for measurement and valuation of health studies. Initially respondents were randomly assigned a unique card set; however, to improve efficiency, subjects were later randomly assigned to one of three fixed sets of EQ-5D states. Using the VAS and TTO responses for these states, we estimated a valuation model using ordinary least squares regression clustered by respondent. Predicted values for EQ-5D health states are compared to published values for the United States.
Six hundred eleven subjects were interviewed by 14 trained interviewers, rendering 6,887 TTO and 6,892 VAS responses. The model had an R2 of 0.897 and 0.928 for TTO and VAS respectively. The mean absolute difference between observed and predicted values was 0.039 for TTO and 0.020 for VAS, each showing a Lin’s concordance coefficient above 0.98. United States and Argentine TTO predicted values were highly correlated (Pearson’s rho=0.963), though the average absolute difference was clinically meaningful (0.06), rejecting the US values for nearly two thirds of the states (62.8%). The Argentine population placed lower values on mild states and higher values on severe states.
This study provides an Argentine value set that could be used locally or regionally, with meaningful and significant differences with that of the US. Health policy in Latin America must incorporate local values for sovereignty and validity.
utility measurement; HRQOL; patient preference; cost effectiveness analysis