Medical Decision Making

Health economic decision-analytic models are used to estimate the expected net benefits of competing decision options. The true values of the input parameters of such models are rarely known with certainty, and it is often useful to quantify the value to the decision maker of reducing uncertainty through collecting new data. In the context of a particular decision problem, the value of a proposed research design can be quantified by its expected value of sample information (EVSI). EVSI is commonly estimated via a 2-level Monte Carlo procedure in which plausible data sets are generated in an outer loop, and then, conditional on these, the parameters of the decision model are updated via Bayes rule and sampled in an inner loop. At each iteration of the inner loop, the decision model is evaluated. This is computationally demanding and may be difficult if the posterior distribution of the model parameters conditional on sampled data is hard to sample from. We describe a fast nonparametric regression-based method for estimating per-patient EVSI that requires only the probabilistic sensitivity analysis sample (i.e., the set of samples drawn from the joint distribution of the parameters and the corresponding net benefits). The method avoids the need to sample from the posterior distributions of the parameters and avoids the need to rerun the model. The only requirement is that sample data sets can be generated. The method is applicable with a model of any complexity and with any specification of model parameter distribution. We demonstrate in a case study the superior efficiency of the regression method over the 2-level Monte Carlo method.

Decision makers often desire both guidance on the most cost-effective interventions given current knowledge and also the value of collecting additional information to improve the decisions made (i.e., from value of information [VOI] analysis). Unfortunately, VOI analysis remains underused due to the conceptual, mathematical, and computational challenges of implementing Bayesian decision-theoretic approaches in models of sufficient complexity for real-world decision making. In this study, we propose a novel practical approach for conducting VOI analysis using a combination of probabilistic sensitivity analysis, linear regression metamodeling, and unit normal loss integral function—a parametric approach to VOI analysis. We adopt a linear approximation and leverage a fundamental assumption of VOI analysis, which requires that all sources of prior uncertainties be accurately specified. We provide examples of the approach and show that the assumptions we make do not induce substantial bias but greatly reduce the computational time needed to perform VOI analysis. Our approach avoids the need to analytically solve or approximate joint Bayesian updating, requires only one set of probabilistic sensitivity analysis simulations, and can be applied in models with correlated input parameters.

Objectives. To compare model input influence on incremental net monetary benefit (INMB) across 3 uncertainty methods: 1) 1-way sensitivity analysis; 2) probabilistic analysis of covariance (ANCOVA); and 3) expected value of partial perfect information (EVPPI). Methods. In a preliminary model, we used a published cost-effectiveness model and assumed £20,000 per quality-adjusted life-year (QALY) willingness-to-pay (Case 1: lower decision uncertainty) and £8000/QALY willingness-to-pay (Case 2: higher decision uncertainty). We conducted 1-way sensitivity, ANCOVA (10,000 Monte Carlo draws), and EVPPI for each model input (1000 inner and 1000 outer draws). We ranked inputs based on influence of INMB and compared input ranks across methods within case using Spearman’s rank correlation. We replicated this approach in 3 follow-up models: an additional linear model, a less linear model with uncorrelated inputs, and a less linear model with correlated inputs. Results. In the preliminary model, lower and higher decision uncertainty cases had the same top 3 influential parameters across uncertainty methods. The 2 most influential inputs contributed 78% and 49% of variation in outcome based on ANCOVA for lower decision uncertainty and higher decision uncertainty cases, respectively. In the follow-up models, input rank order correlations were higher across uncertainty methods in the linear model compared with both of the less linear models. Conclusions. Evidence across models suggests influential input rank agreement between 1-way and more advanced uncertainty analyses for relatively linear models with uncorrelated parameters but less agreement for less linear models. Although each method provides unique information, the additional resources needed to generate and communicate advanced analyses should be weighed, especially when outcome decision uncertainty is low. For less linear models or those with correlated inputs, performing and reporting deterministic and probabilistic uncertainty analyses appear prudent and conservative.

Cost-effectiveness analysis (CEA) models are routinely used to inform health care policy. Key model inputs include relative effectiveness of competing treatments, typically informed by meta-analysis. Heterogeneity is ubiquitous in meta-analysis, and random effects models are usually used when there is variability in effects across studies. In the absence of observed treatment effect modifiers, various summaries from the random effects distribution (random effects mean, predictive distribution, random effects distribution, or study-specific estimate [shrunken or independent of other studies]) can be used depending on the relationship between the setting for the decision (population characteristics, treatment definitions, and other contextual factors) and the included studies. If covariates have been measured that could potentially explain the heterogeneity, then these can be included in a meta-regression model. We describe how covariates can be included in a network meta-analysis model and how the output from such an analysis can be used in a CEA model. We outline a model selection procedure to help choose between competing models and stress the importance of clinical input. We illustrate the approach with a health technology assessment of intravenous immunoglobulin for the management of adult patients with severe sepsis in an intensive care setting, which exemplifies how risk of bias information can be incorporated into CEA models. We show that the results of the CEA and value-of-information analyses are sensitive to the model and highlight the importance of sensitivity analyses when conducting CEA in the presence of heterogeneity. The methods presented extend naturally to heterogeneity in other model inputs, such as baseline risk.

Introduction. In the United States, more than 10% of national health expenditures are for prescription drugs. Assessing drug costs in US economic evaluation studies is not consistent, as the true acquisition cost of a drug is not known by decision modelers. Current US practice focuses on identifying one reasonable drug cost and imposing some distributional assumption to assess uncertainty. Methods. We propose a set of Rules based on current pharmacy practice that account for the heterogeneity of drug product costs. The set of products derived from our Rules, and their associated costs, form an empirical distribution that can be used for more realistic sensitivity analyses and create transparency in drug cost parameter computation. The Rules specify an algorithmic process to select clinically equivalent drug products that reduce pill burden, use an appropriate package size, and assume uniform weighting of substitutable products. Three diverse examples show derived empirical distributions and are compared with previously reported cost estimates. Results. The shapes of the empirical distributions among the 3 drugs differ dramatically, including multiple modes and different variation. Previously published estimates differed from the means of the empirical distributions. Published ranges for sensitivity analyses did not cover the ranges of the empirical distributions. In one example using lisinopril, the empirical mean cost of substitutable products was $444 (range = $23–$953) as compared with a published estimate of $305 (range = $51–$523). Conclusions. Our Rules create a simple and transparent approach to creating cost estimates of drug products and assessing their variability. The approach is easily modified to include a subset of, or different weighting for, substitutable products. The derived empirical distribution is easily incorporated into 1-way or probabilistic sensitivity analyses.

Background. Two vaccines (Rotarix and RotaTeq) are highly effective at preventing severe rotavirus disease. Rotavirus vaccination has been introduced in the United Kingdom and other countries partly based on modeling and cost-effectiveness results. However, most of these models fail to account for the uncertainty about several vaccine characteristics and the mechanism of vaccine action. Methods. A deterministic dynamic transmission model of rotavirus vaccination in the United Kingdom was developed. This improves on previous models by 1) allowing for 2 different mechanisms of action for Rotarix and RotaTeq, 2) using clinical trial data to understand these mechanisms, and 3) accounting for uncertainty by using Markov Chain Monte Carlo. Results. In the long run, Rotarix and RotaTeq are predicted to reduce the overall rotavirus incidence by 50% (39%−63%) and 44% (30%−62%), respectively but with an increase in incidence in primary school children and adults up to 25 y of age. The vaccines are estimated to give more protection than 1 or 2 natural infections. The duration of protection is highly uncertain but has only impact on the predicted reduction in rotavirus burden for values lower than 10 y. The 2 vaccine mechanism structures fit equally well with the clinical trial data. Long-term postvaccination dynamics cannot be predicted reliably with the data available. Conclusion. Accounting for the joint uncertainty of several vaccine characteristics resulted in more insight into which of these are crucial for determining the impact of rotavirus vaccination. Data for up to at least 10 y postvaccination and covering older children and adults are crucial to address remaining questions on the impact of widespread rotavirus vaccination.

The smallpox antiviral tecovirimat has recently been purchased by the U.S. Strategic National Stockpile. Given significant uncertainty regarding both the contagiousness of smallpox in a contemporary outbreak and the efficiency of a mass vaccination campaign, vaccine prophylaxis alone may be unable to control a smallpox outbreak following a bioterror attack. Here, we present the results of a compartmental epidemiological model that identifies conditions under which tecovirimat is required to curtail the epidemic by exploring how the interaction between contagiousness and prophylaxis coverage of the affected population affects the ability of the public health response to control a large-scale smallpox outbreak. Each parameter value in the model is based on published empirical data. We describe contagiousness parametrically using a novel method of distributing an assumed R-value over the disease course based on the relative rates of daily viral shedding from human and animal studies of cognate orthopoxvirus infections. Our results suggest that vaccination prophylaxis is sufficient to control the outbreak when caused either by a minimally contagious virus or when a very high percentage of the population receives prophylaxis. As vaccination coverage of the affected population decreases below 70%, vaccine prophylaxis alone is progressively less capable of controlling outbreaks, even those caused by a less contagious virus (R₀ less than 4). In these scenarios, tecovirimat treatment is required to control the outbreak (total number of cases under an order of magnitude more than the number of initial infections). The first study to determine the relative importance of smallpox prophylaxis and treatment under a range of highly uncertain epidemiological parameters, this work provides public health decision-makers with an evidence-based guide for responding to a large-scale smallpox outbreak.

Objectives. Type 2 diabetes mellitus (T2DM) clinical guidelines focus on optimizing glucose control, with therapy escalation classically initiated within a “failure-based” regimen. Within many diabetes models, HbA1c therapy escalation thresholds play a pivotal role, controlling duration of therapy and, consequently, incremental costs and benefits. The objective of this study was to assess the relationship between therapy escalation threshold and time to therapy escalation on predicted cost-effectiveness of T2DM treatments. Methods. This study used the Cardiff Diabetes Model to illustrate the relationship between costs and health outcomes associated with first-, second-, and third-line therapy as a function of time on each. Data from routine clinical practice were used to contrast predicted costs and health outcomes associated with guideline therapy escalation thresholds compared with clinical practice. The impact of baseline HbA1c and therapy escalation thresholds on cost-effectiveness was investigated, comparing a sodium/glucose cotransporter 2 inhibitor v. sulfonylurea added to metformin monotherapy. Results. Lower thresholds are associated with a shorter time spent on monotherapy, ranging from 1.1 years (escalation at 6.5%) to 13 years (escalation at 9.0%) and an increase in total lifetime cost of therapy. Treatment-related disutility is minimized with higher thresholds because progression to insulin is delayed. Using metformin combined with either dapagliflozin or a sulfonylurea to illustrate lower baseline HbA1c and/or higher therapy escalation thresholds was associated with increased cost-effectiveness ratios, driven by a longer duration of therapy. Discussion. A marked difference in treatment cost-effectiveness was demonstrated when comparing routine clinical practice with guideline-advocated therapy escalation. This is important to both health care professionals and the wider health economic community with respect to understanding the true cost-effectiveness profile of any particular T2DM therapy option.

Background. There is increasing evidence highlighting the effectiveness of patient decision aids (PtDAs), but evidence supporting their cost-effectiveness is lacking. We consider patients with obstructive sleep apnea (OSA), in whom a PtDA may decrease nonadherence to treatment by empowering patients to receive the option that is most congruent with their own values. Objective. To determine the potential costs and benefits of delivering a PtDA to patients with moderate OSA. Methods. A Markov cohort decision-analytic model was developed for patients with moderate OSA, comparing a PtDA to usual care over 5 years from a societal perspective. Data for patient preference for treatment options was taken from a recent randomized crossover trial, event data (cardiovascular, motor vehicle accidents) came from national databases and published literature. Potential improvements in adherence are unknown, so we considered a realistic range of values. Outcome measures were 5-year costs (in 2010 Canadian dollars), quality-adjusted life years (QALYs), and the incremental cost-effectiveness ratio (ICER). Results. When adherence to treatment was unchanged, the PtDA strategy was dominated by incurring lower QALYs and higher costs. When nonadherence was decreased by 20% in the PtDA arm (corresponding to an increase in adherence from 63% to 70% for continuous positive airway pressure and from 77% to 82% for mandibular advancement splints in year 1), the ICER fell to $62,414/QALY. Costs associated with the treatment devices and delivering the PtDA had the greatest effect on cost-effectiveness. Limitations. The model relies on surrogate measures and opinions for key parameters. Conclusions: The cost-effectiveness of PtDAs will depend on contextual factors, but a framework is described for properly considering their long-term cost-effectiveness. A number of important questions around the appropriateness of benefit measurement for PtDA trials are highlighted.