Medical Decision Making

Adherence to statins for the primary prevention of cardiovascular disease (CVD) is low but may be improved through shared decision making (SDM). CVD risk estimation is fundamental to SDM at statin initiation, and the absence of a CVD risk score may indicate that SDM has not taken place. This study explores statin initiation decisions by investigating whether CVD risk is associated with adherence to statins and CVD outcomes.

A cohort of statin-naïve patients aged 40 to 84 y initiated on statins for primary prevention between 2017 and 2020 was identified and categorized by the presence or absence of a CVD risk score at statin initiation. Multivariable modeling determined the association between CVD risk score and statin adherence and persistence. A secondary analysis investigated the relationship with CVD outcomes and death.

A total of 255,730 patients were included with a mean follow-up of 4.6 y. The presence of a CVD risk score (67.7% of patients) was associated with a 7% increase in adherence and an 8% reduction in discontinuations as well as a 15% reduction in CVD and all-cause mortality.

The presence of a CVD risk score is associated with improvements in both adherence and persistence, which could indicate that the quality of initiation consultations, and a focus on SDM, improves the utility of statins. In addition, CVD risk scoring is associated with a large decrease in CVD/death, which cannot fully be explained by improvements in statin adherence and persistence, an important finding necessitating further investigation.

Efforts to improve the integration of CVD risk scores and SDM in statin initiation consultations may significantly improve statin utilization and CVD outcomes.

This study qualitatively explored how bolt-ons affect the perception of EQ-5D-5L core dimensions in a valuation context.

Sixty Indonesian adults (aged 20–67 y, 50% female) each valued 10 health states using composite time tradeoff (cTTO). States were presented in either forward (EQ-5D-5L, then with 1 bolt-on, then 2) or backward (reversed sequence) order. Participants were assigned to 1 of 3 bolt-on dyads: vision and tiredness, cognition and social relationships, or skin irritation and self-confidence. In semistructured qualitative interviews, respondents described how adding or removing bolt-ons changed the perceived importance of the 5 core dimensions. We classified these changes as related either to measurement or to valuation, with the latter further categorized as a relative or absolute shift in importance.

Cognition and vision generated the most shifts in perceived importance in the forward and backward groups, respectively. Regardless of ordering group, most shifts occurred between the EQ-5D-5L alone and the version with 1 bolt-on (first presented in the dyad), with significantly fewer shifts observed between the 1-bolt-on and 2-bolt-on states. In the forward group, most shifts were classified as measurement (56%) or relative preference (29%), while the reverse was true in the backward group: relative preference (53%) or measurement (31%). Absolute preference was least common across both groups.

This is the first study to explore how individuals reason when valuing EQ-5D-5L+bolt-on health states. Our findings suggest that interactions between dimensions are complex and may be influenced by presentation order. Further qualitative research should directly investigate absolute preferential reasoning.

Physicians often fail to communicate important details about key risks and benefits of treatment in prostate cancer consultations, an information gap that may inhibit effective patient participation in care decisions. We developed a framework to assess detail and personalization of physician communication of key concepts, derived from observations of treatment consultations.

We recorded and transcribed treatment consultations of 50 men with newly diagnosed prostate cancer across 10 multidisciplinary providers. Using open coding, analysts identified statements related to guideline-endorsed SDM concepts and categorized the level of detail for each topic. Concept-specific hierarchies indicating increasing detail were integrated into a unified framework. We reported the highest observed level for each concept across consultations.

Across 19,388 statements, sentences addressing cancer severity (1,016; 5.2%), baseline function (186; 1.0%), oncologic endpoints (343; 1.8%), cancer prognosis (318; 1.6%), life expectancy (111; 0.57%), and treatment-related side effects (952; 4.9%) were extracted. Empirically derived frameworks for evaluating the level of detail and personalization for each concept consistently displayed parallel levels of detail. We created a unified framework, with each level providing more precise, patient-specific risk communication: (0) omission, (1) no quantification, (2) binary assessment, (3) imprecise quantification, (4) specific quantification, and (5) patient-specific estimate. The prevalence of the highest observed level of detail and personalization conveyed varied by concept, ranging from 4% to 84%, 0% to 47%, 0% to 42%, 2% to 8%, 0% to 48%, and 0% to 66% for levels 0 to 5, respectively.

Our framework introduces an empirically derived, structured foundation for characterizing the level of detail and personalization of physician risk communication of key concepts in prostate cancer consultations.

Assessment of long-term survival for health technology assessment often necessitates extrapolation beyond the duration of a clinical trial. Flexible Bayesian survival models that incorporate longer-term data sources, including registry data and population mortality, have been proposed as an alternative to using standard parametric models with trial data alone.

The performance of extrapolations from the survextrap Bayesian M-spline survival model was evaluated through simulation, with 5 years of trial data follow-up for our primary analysis. We assessed long-term survival and incremental effect estimates under a range of settings and assumed long-term real-world data informed the control arm. Comparisons were made with standard and flexible parametric models.

When relevant long-term external data on the control arm were available, a flexible Bayesian approach substantially improved the precision and accuracy of extrapolations for that arm. Improvements in estimates of extrapolated treatment effects were more variable and sometimes deteriorated, reflecting sensitivity to assumptions about how hazards relate across arms. Compared with exponential and Weibull models, the survextrap Bayesian model had better within-trial fit and more plausible extrapolations, although a Royston–Parmar spline often provided comparable accuracy of incremental effects in many settings. We noted further that the survextrap Bayesian approach has the potential strength of explicitly quantifying the structural uncertainty, making transparent the strong assumptions required by any survival modeling approach to extrapolate.

Flexible Bayesian survival models can potentially improve long-term survival extrapolation in the control arm when well-matched external data are available, but improvements in estimates of incremental treatment effects were often quite variable across scenarios and models. This highlights the need for careful consideration of appropriate modeling assumptions and real-world data quality when using this approach.

To compare an established benefit-estimating algorithm for recommending and prioritizing preventive services for a patient (Individualized Precision Prevention; IPP) with concordant rankings from primary care providers (PCPs).

We developed 12 realistic routine patient care scenarios focused on preventive services and recruited 40 PCPs to rank the priority of recommended preventive services. Our analysis compared the benefit-estimating algorithm’s rankings of preventive services for each of the 12 patient scenarios to the PCPs’ rankings using length-dependent rank-biased overlap (LDRBO) calculations. Moderate concordance would suggest that the computer algorithm presented an opportunity to improve preventive care, whereas very high or low concordance would call into question what the algorithm could contribute to clinical practice.

For all 12 patient care scenarios, comparing the benefit-estimating algorithm’s output to the combined priority rankings from all PCPs yields a mean value of 0.45, corresponding to a moderate level of concordance or agreement between the numeric rankings of the algorithm and the expert provider rankings. This study illustrates the potential importance of having computed IPP recommendations readily available for point-of-care decision making by PCPs.

We demonstrate that this approach aligned with the overall judgment of clinical experts and may help providers prioritize preventive services in time-constrained clinical contexts. The modest correlation between the benefit-estimating algorithm and expert providers suggests that, in some cases, the algorithm has the potential to provide useful advice about preventive services during care.

Survival extrapolation is crucial in estimating the lifetime survival benefit of a treatment for health technology assessment (HTA). Conventional extrapolation methods, which assume that the long-term treatment effect (hazard ratio between treatment and comparator) follows the same pattern as observed in the short-term trial, have been challenged by a wide range of immuno-oncology therapies, particularly those with administrative stopping rules that mandate treatment discontinuation at a prespecified time point. A gradual waning of their treatment effects has been considered plausible and received growing attention from HTA stakeholders over the past decade. However, existing statistical methods often rely on unnecessarily strong waning assumptions.

We demonstrate the blended hazard method as a flexible way to account for treatment effect waning while incorporating external evidence in survival extrapolation.

The blended hazard method fits separate parametric survival regression models to the observed randomized controlled trial data and external data that inform the common long-term hazard when there is no treatment effect. For each arm, the fitted internal and external hazard functions are blended based on a time-varying weight function, so that the blended hazard is initially dominated by the fitted internal hazard, then gradually approaches the fitted external hazard over a blending interval, and is finally dominated by the fitted external hazard. The time and rate of blending the internal and external information can be controlled by the weight function to allow for sensitivity analyses. NICE TA366 on pembrolizumab for advanced melanoma not previously treated with ipilimumab is used as a case study to demonstrate the practical implementation of this method.

Extrapolations and restricted mean survival times from the blended hazard method closely matched the updated 7-y trial follow-up and showed better consistency than the TA366 base case across all sensitivity analysis scenarios.

The method explicitly accounts for gradual treatment effect waning while incorporating external evidence and offers the flexibility to accommodate a broad range of waning scenarios, thereby effectively characterising uncertainty in extrapolation.

Researchers widely use discrete choice experiments (DCEs) to assess health preferences across subgroups. However, variations in decision consistency, rather than true differences in preferences, can drive observed utility differences. Despite the growing use of DCEs to assess health preference heterogeneity, recent studies highlight a persistent lack of methodological transparency in accounting for unobserved heterogeneity, underscoring the need for technically robust approaches to support credible and actionable comparisons across groups. This study improves health preference research methods by directly addressing scale heterogeneity and reducing bias when comparing subgroups.

A simulated DCE evaluated hypothetical cancer treatments across 2 imagined groups (patients, caregivers). Each task presented 3 alternatives (including a status quo), varying in months gained, survival rate, side-effect severity, and out-of-pocket cost. Mixed logit models were estimated. Scale heterogeneity was addressed using the Swait–Louviere 2-step procedure. Willingness to pay (WTP) was computed and compared across groups via the Poe et al. (2005) simulation-based test.

The Swait–Louviere test confirmed significant scale heterogeneity (P < 0.05) but no meaningful taste differences (P > 0.10). Once scale effects were accounted for, the analysis revealed a shared preference structure across patients and caregivers, with variability driven by inconsistent decision making rather than true preference divergence. Consistent with this, none of the between-group WTP differences were statistically significant, reinforcing the absence of meaningful subgroup contrasts and underscoring the importance of separating scale from taste to avoid biased inference.

Adjusting for scale heterogeneity strengthens DCE validity by reducing bias from decision noise and enabling accurate subgroup comparisons. Using simulated data, this study applied the Swait–Louviere 2-step and scale-invariant WTP contrasts to separate taste from scale; both methods converged, showing that heterogeneity reflected scale rather than true preference differences, with negligible WTP gaps. Routine scale diagnostics, taste (preference) tests under equalized scale, and welfare space reporting are recommended to ensure valid inference. However, as this study used simulated data with no real respondents, its findings are illustrative only and not intended for real-world inference; generalizability and external drivers of scale heterogeneity were not assessed.

To develop and validate a microsimulation model to estimate the health outcomes and costs of chronic kidney disease (CKD) in type 2 diabetes (T2D) to inform health policies and reduce the burden of CKD.

We developed a comprehensive model for CKD in type 2 diabetes that assesses the impact of risk factors on the progression of urine albumin-to-creatinine ratio and estimated glomerular filtration rate and their impact on stroke, myocardial infarction (MI), congestive heart failure (CHF), end-stage kidney disease (ESKD), and death without dialysis or transplant using individual-level longitudinal data for T2D populations and summary data from published literature. We internally validated the model using data from the Chronic Renal Insufficiency Cohort (CRIC) of patients with T2D and CKD over 7 y and externally validated the model using the Canagliflozin and Renal Events in Diabetes with Established Nephropathy Clinical Evaluation (CREDENCE) trial data over 3 y.

The simulated event rates of ESKD, stroke, MI, CHF, and total mortality and the related 95% confidence intervals included the observed event rates in both the internal and external validation cohorts. Using this new model, we showed that lowering the blood pressure target from 140/90 to 120/80 mm Hg in patients with T2D and CKD was cost-saving at the population level.

The Michigan Model for Diabetes–Chronic Kidney Disease (MMD-CKD) model provides accurate estimates of disease progression in patients with T2D and CKD. Modeling disease progression in this population will facilitate future assessments of the cost-effectiveness of systematic screening and interventions for CKD, which may reduce the health and economic burden of CKD in T2D. This model can also serve as a tool for predicting the clinical outcomes of individual patients with T2D and CKD. MMD-CKD 1.0 R Shiny app and is publicly available (https://michigandiabetesmodelinggroup.github.io/Software_App).

The vaginal microbiota test predicts the success of in vitro fertilization (IVF), but with no therapy available to improve a low profile, couples must decide whether to proceed or postpone treatment. We aim to examine how couples interpret vaginal microbiome results and make postponement decisions within a shared decision making (SDM) framework.

Women undergoing IVF or IVF–intracytoplasmic sperm injection (IVF-ICSI) treatment at 2 Dutch hospitals received the ReceptIVFity test™, which classified the vaginal microbiome as high (52.6% chance of conception), medium (23.6%), or low (5.9%) profile based on predicted implantation success after a fresh embryo transfer. Physicians discussed the results with couples using SDM, after which the couples decided whether to proceed or postpone treatment. The primary outcome was the patients’ perceived involvement in shared decision making, assessed with the SDM-Q-9 questionnaire. The secondary outcome was the proportion of couples postponing treatment after a low microbiome profile.

Between October 2018 and November 2020, 728 women were enrolled. SDM-Q-9 responses showed high perceived involvement overall but lower scores for “exploring options,” reflecting limited alternatives when the choice is to proceed or postpone treatment. A low profile was found in 35.4% (258/728). After the SDM consultation, 49.6% (128/258) chose to postpone treatment, with postponement rates increasing to over 80% among couples in later IVF cycles. Decisions were influenced by personal, emotional, and practical considerations, including the Dutch insurance reimbursement system (3 insured IVF or IVF-ICSI cycles regardless of postponement) and the absence of effective treatment to modify a low profile.

These findings demonstrate that couples can understand and use prognostic information when supported by SDM and that the ReceptIVFity test™ facilitated discussion about chances of success, timing of treatment, decisions to proceed or postpone, and personal values.