In randomized trials, the per-protocol effect is defined as the effect of being assigned a treatment strategy and receiving treatment according to the assigned strategy; therefore, it encompasses the effect of the treatment strategy and any effect of assignment that does not operate through the received treatment. However, the per-protocol effect is sometimes interpreted as reflecting the effect of the treatment strategy itself, regardless of assignment. Here, we argue by example that this is not necessarily the case. We examine causal structures for randomized trials where these two causal estimands - the per-protocol effect and the effect of the treatment strategy - are not equal, and where their corresponding identifying observed data functionals are not the same, but both require information on assignment for identification. Our examples clarify the conceptual difference between these estimands and the conditions that guarantee their equality. Our examples also highlight that in some cases identification of these estimands requires information on assignment, even when assignment is randomized, unless one makes additional assumptions - informally, that assignment does not affect the outcome except through treatment (i.e., an exclusion-restriction assumption), and that assignment is not a confounder of the treatment-outcome association conditional on other variables in the analysis. These assumptions may not be plausible, particularly when the treatment assignment process in the trial differs materially from routine practice in outcome-relevant ways (e.g., "blinding" investigators and participants to treatment assignment). Our examination emphasizes the important role of assignment in defining and estimating causal effects in randomized trials, with implications for statistical analysis and the interpretation of trial results.
When individuals participating in a randomized trial differ with respect to the distribution of effect modifiers compared compared with the target population where the trial results will be used, treatment effect estimates from the trial may not directly apply to target population. Methods for extending -- generalizing or transporting -- causal inferences from the trial to the target population rely on conditional exchangeability assumptions between randomized and non-randomized individuals. The validity of these assumptions is often uncertain or controversial and investigators need to examine how violation of the assumptions would impact study conclusions. We describe methods for global sensitivity analysis that directly parameterize violations of the assumptions in terms of potential (counterfactual) outcome distributions. Our approach does not require detailed knowledge about the distribution of specific unmeasured effect modifiers or their relationship with the observed variables. We illustrate the methods using data from a trial nested within a cohort of trial-eligible individuals to compare coronary artery surgery plus medical therapy versus medical therapy alone for stable ischemic heart disease.
It has long been accepted that observational analyses have an important role in evaluating use patterns and assessing the safety of different treatments, including cardiovascular devices, in clinical practice. With the proliferation of large electronic databases, there has been increasing interest in using observational analyses to also examine the comparative effectiveness of devices. However, these analyses are often met with skepticism because of concerns about whether they can generate credible evidence about causal effects. This is in part a result of the difficulty in meeting the assumptions necessary to interpret observational associations as causal effects and of the wide variability in analytic rigor. In this review, we outline frameworks and review methods for using observational analyses to answer questions about the effectiveness and safety of cardiovascular devices. We highlight the target trial framework as a practical tool for guiding observational comparative effectiveness analyses. We illustrate how the framework allows investigators planning and conducting observational analyses to organize their activities as responses to 3 prompting questions. First, what is the research question of the study (ie, "What do we want?")? Second, what are the resources-including background knowledge, research concepts, principles and methods, and available data-that can be brought to bear on the research question (ie, "What do we have?")? And third, what specific steps should be taken to use the available resources to answer the research question (ie, "What do we do?")? We focus our exposition on the evaluation of cardiovascular devices, for which randomized trial data are often limited and there is a strong need for real-world evidence. In this setting, real-world evidence is usually derived from observational comparisons of the treatment of interest with relevant comparator groups using data captured during routine care. A principled approach to the planning and conduct of observational analyses can improve the quality of real-world evidence generation and ensure that the results of observational studies on medical devices can support meaningful conclusions about the risks and benefits of new devices.
BACKGROUND:Limited evidence is available from randomized trials to guide clinical decision-making for cardiovascular disease (CVD) prevention with statins among women with breast cancer. We leveraged real-world data to estimate the effect of statin therapy vs usual care on the 5-year risk of CVD among breast cancer survivors. METHODS:We sequentially emulated target trials using the National Cancer Institute-Kaiser Permanente Breast Cancer Survivors Cohort. Women aged 40 - 84 years, diagnosed with stage 0-III breast cancer from 1993 - 2022, without CVD, no statins in the past 6 months, and low-density lipoprotein cholesterol from 100 - 160 mg/dL were included. The outcome was CVD (ischemic heart disease, stroke, and cardiomyopathy/heart failure). We compared statin therapy to usual care (no statins unless indicated) in intention-to-treat and per-protocol analyses. RESULTS:Statin initiators were more likely to be older with a higher prevalence of cardiovascular comorbidities. In the intention-to-treat analysis, the 5-year risk difference was 0.3% (95%CI: -0.9,1.5), and risk ratio was 1.04 (95%CI: 0.86,1.25) for statin vs usual care. At 5 years, the proportion adherent to their initial strategy was 48% for statin and 24% for usual care. In the per-protocol analysis, the risk difference was -0.9% (95%CI: -2.9,2.2) and the risk ratio was 0.85 (95%CI: 0.51,1.36) for statin vs usual care. CONCLUSIONS:Our findings suggest that statin therapy may reduce the risk of CVD for breast cancer survivors, but the estimates are too imprecise for specific recommendations. Future research is needed to understand reasons for statin discontinuation, and opportunities to encourage adherence in this high-risk population.
Evidence syntheses and meta-analyses are used to inform clinical practice guidelines and health economic evaluations. However, heterogeneity of treatment effects poses a significant challenge. Conventional meta-analysis addresses heterogeneity through random-effect assumptions, which are not supported by design and lead to estimates that may not apply to any real-world population. Causally-interpretable meta-analysis (CIMA) offers a rigorous framework for specification, identification, and estimation of causal effects when combining information from multiple randomized trials. Initial development of CIMA focused on using individual data from randomized trials, but such data are often unavailable in practice. Here, we propose a new version of CIMA that only requires aggregate data from trials, addressing the limitations of traditional meta-analysis methods while relying only on aggregate data. The method leverages the trials' reported estimates of marginal and one-at-a-time subgroup treatment effects and descriptive statistics for baseline covariates to build moment equations for identifying and estimating a parametric conditional average treatment effect (CATE) function. The average treatment effect in a new target population is obtained by marginalizing the CATE function over the individual covariate data that defines the target population. The method can also be used to obtain causally-interpretable indirect treatment comparisons in the target population. We establish the asymptotic properties of the method, assess its finite-sample performance in simulation studies, and illustrate the application of the method by re-analyzing a published meta-analysis for SGLT2 inhibitors in patients with heart failure.
BACKGROUND:When decision makers use evidence from a randomized trial to inform population-level decisions, the target population they envision rarely aligns with the population of individuals who enrolled in the trial. Here, we extend inferences from the VALIDATE-SWEDEHEART randomized trial (hereafter, the index trial), which compared the effects of bivalirudin and heparin during percutaneous coronary intervention on the risk of death, reinfarction, and bleeding, to two clinically relevant target populations: first, the trial-eligible population of individuals eligible for the index trial regardless of enrollment, and second, the treatment-candidate population of individuals who are considered candidates for bivalirudin and heparin under routine care, regardless of eligibility for the index trial. METHODS:Using data from the index trial, we fit logistic regression models for the outcome at 180 days in each group based on assigned treatment. We then standardized risk estimates to the baseline covariate distribution of the trial-eligible and treatment-candidate target populations, which were characterized using data from Swedish healthcare registries. RESULTS:The estimated risk difference comparing bivalirudin versus heparin was -1.1% (-3.1%, 0.9%) in the trial-eligible population and -1.0% (-3.0%, 1.0%) in the treatment-candidate population. The corresponding risk ratios were 0.92 (0.80, 1.07) and 0.93 (0.80, 1.07), respectively, aligning closely with estimates from the index trial. Absolute risks in each treatment group were, however, between 0.8 and 1.2 percentage points higher in comparison with the index trial. CONCLUSIONS:Estimated risk ratios for the broader trial-eligible and treatment-candidate populations generally align with the findings from the index trial. While trials provide essential evidence for healthcare, questions often arise about wider, clinically relevant populations beyond the population of trial participants. By leveraging data from trials and observational data sources, we can attempt to address questions in these wider target populations.
We describe how the target trial framework can be used to plan and report analyses that attempt to answer causal questions by combining information from multiple, diverse sources. Such analyses may involve comparisons of treatments evaluated in different populations, for example when an index trial is combined with other data sources in external comparator analyses, or when extending causal inferences from a randomized trial to a new target population in generalizability and transportability analyses. When planning such analyses, the specification of the target trial supports the explicit definition of the target population with an associated sampling model. We propose this as an additional component for the target trial framework, especially relevant for analyses that combine information, because it influences the choice of eligibility criteria, the specification of the causal model, the choice of causal contrasts, and reasoning about identification strategies. Furthermore, the framework encourages careful mapping of data elements from multiple data sources to a single target trial. This mapping process can highlight potentially irreconcilable misalignments between data sources with respect to specific components of the framework – for example, in the definitions of eligibility criteria, treatment assignment, and treatment receipt. Such misalignments can arise when attempts to specify a target trial that aligns with a specific data source introduce or worsen misalignments with other proposed data sources. The extent of such misalignments may warrant switching to other data sources, or prospectively obtaining data, to emulate the proposed target trial. We conclude that the target trial framework promotes transparent discussion about the design of and assumptions made in analyses that answer causal questions by combining information from diverse sources.
Descriptions of the target trial framework often appear to assume that investigators knew the protocol of the target trial from the start of the investigation. In practical applications, however, the protocol of the target trial is constrained by the observational data that investigators have access to. When using preexisting observational databases, the target trial protocol typically needs to be iteratively developed as investigators learn about the data. Here we examine the process by which investigators adapt their original causal question, operationalized into the protocol of a hypothetical trial (the index trial), to the available data. This adaptive process results in a final causal question, operationalized into the protocol of another hypothetical trial (the target trial), that depends both on the original causal question and on the available data. As a result, prespecification of an emulatable target trial protocol is not generally possible because adaptations are expected after inspecting the data. The adaptive nature of the specification of the target trial protocol raises important questions about the types of data examinations that are permissible to guide the adaptations, the procedures for transparent reporting of each adaptation and its rationale, and the possibility of prespecifying the rules that will govern the investigators' decisions to adapt the protocol to the data.
Randomized controlled trials often enroll participants whose characteristics differ from those of a target population, which can limit the generalizability of the estimated treatment effects when effect modifiers differ across populations. While existing generalizability methods primarily focus on estimating the average treatment effect (ATE) in the target population, such summaries may obscure important heterogeneity that is relevant for clinical and policy decision-making. In this work, we illustrate an approach for estimating the conditional average treatment effect (CATE) in a target population of trial-eligible individuals as a function of prespecified effect modifiers within a nested trial setting. Our approach combines semiparametric theory with flexible estimation: we first estimate nuisance functions using data-adaptive methods and construct pseudo-outcomes from conditional influence functions, then estimate the CATE function via local linear (kernel) regression. Sample splitting and cross-fitting are used to reduce overfitting bias and ensure asymptotic valid inference. Finite-sample performance is assessed via simulations and illustrated in the Coronary Artery Surgery Study (CASS).
BACKGROUND:Although mitral transcatheter edge-to-edge repair (MTEER) was approved for secondary mitral regurgitation after the COAPT trial, findings of other MTEER trials have been mixed, raising questions about the applicability of the COAPT results to contemporary clinical practice. OBJECTIVES:We used transportability methods to estimate the treatment effects of COAPT trial interventions applied to 2 target populations: 1) trial-eligible patients representative of U.S. clinical practice; and 2) treatment-candidate patients with secondary mitral regurgitation representative of U.S. clinical practice, regardless of trial eligibility. METHODS:We identified patients from the Society of Thoracic Surgeons/American College of Cardiology Transcatheter Valve Therapy (TVT) Registry who were treated with MTEER for secondary mitral regurgitation from March 14, 2019 to September 30, 2023. To select trial-eligible individuals, we applied COAPT trial eligibility criteria to the TVT Registry sample. We used inverse odds of participation weighting to standardize patient-level COAPT data to the data distribution of each target population sample and estimated treatment-specific outcomes. The primary outcome was heart failure hospitalization at 2 years. We also examined 10 secondary outcomes, including all-cause death. RESULTS:Our analyses included 614 COAPT trial patients and 15,275 TVT Registry patients, of which 7,289 were COAPT trial-eligible. Trial-eligible TVT Registry patients were less likely to have ischemic cardiomyopathy (34.1% vs 60.8%) and more likely to have 4+ mitral regurgitation (79.4% vs 47.9%) compared with trial patients. We estimated that compared with medical therapy alone, MTEER in conjunction with other COAPT interventions (eg, optimization of medical therapy) in the trial-eligible population would result in 2-year absolute risk reductions of 17.0% for heart failure hospitalizations (95% CI: -28.7% to -5.7%) and 15.4% for all-cause death (95% CI: -26.6% to -5.2%), effect sizes similar to those estimated in the trial (P for difference between the trial and target populations >0.05 for both outcomes). The estimated treatment effect for heart failure hospitalizations in the broader treatment-candidate target population was also similar to that in the COAPT trial (P for difference = 0.90). CONCLUSIONS:Although COAPT trial patients had different baseline characteristics than patients undergoing MTEER in contemporary U.S. practice, we estimated that treatment effects would be similar had real-world patients received COAPT trial interventions, under the assumptions required for transportability (eg, conditional exchangeability across data sources, positivity of trial participation).
Importance:The prevalence of obesity and cardiovascular-kidney-metabolic (CKM) syndrome continues to rise. Indications for novel CKM therapies, including glucagonlike peptide 1 receptor agonists (GLP-1RAs), sodium-glucose cotransporter-2 inhibitors (SGLT2is), and nonsteroidal mineralocorticoid antagonists (nsMRAs) continue to expand, yet the proportion of adults meeting expanded indications, including for multiple medications remains unclear. Objective:To examine proportion of adults meeting US Food and Drug Administration (FDA)-approved indications for GLP1-RAs, SGLT2is, and nsMRAs across national survey, community-based, and ambulatory health care samples. Design, Setting, and Participants:This study used a representative cross-sectional survey of US adults (National Health and Nutrition Examination Survey [NHANES], weighted 245 million; mean [SD] age, 47 [18] years; 126.8 million [52%] female), 5 pooled community-based cohort studies (the Framingham Heart Study, the Multi-Ethnic Study of Atherosclerosis, the Prevention of Renal and Vascular Endstage Disease Study, the Atherosclerosis Risk in Communities Study, and the Cardiovascular Health Study; n = 30 929; mean [SD] age, 63 [14] years; 16 749 [54%] female), and 2 ambulatory health care samples (the Beth Israel Deaconess Medical Center cohort [BIDMC], n = 84 714; mean [SD] age, 46 [17] years; 51 113 [60%] female] and the Mass General Brigham cohort [MGB], n = 362 485; mean [SD] age, 48 [17] years; 227 206 [61%] female). Data were analyzed from November 2024 to November 2025. Exposures:FDA-approved indications for GLP-1RAs, SGLT2is, and nsMRAs. Main Outcomes and Measures:Medication class eligibility within each study sample. Results:The proportion of individuals who met current FDA-approved indications for 1 or more CKM medication was 60% in NHANES (representing 148 million US adults), 61% in the pooled cohorts, 42% in the BIDMC ambulatory cohort, and 46% in the MGB ambulatory cohort. Eligibility for GLP-1RA therapy was most common, with 56% (representing 137.1 million US adults) in NHANES, 49% in the pooled cohorts, 41% in the BIDMC cohort, and 46% in the MGB cohort. This was followed by SGLT2i therapy (24% [57.9 million] in NHANES, 33% in the pooled cohorts, 14% for both BIDMC and MGB) and nsMRA (5% [11.7 million] in NHANES, 5% in the pooled cohorts, and 1% to 2% in ambulatory samples). Overlapping eligibility for multiple classes was common, with 12% to 17% for GLP1-RA and SGLT2i therapies and 1% to 5% for all 3 classes (an estimated 11.7 million US adults in NHANES). Conclusions and Relevance:This study found that up to 61% of adults met FDA-approved indications for at least 1 of 3 novel CKM therapy classes. This represents an estimated 148 million US adults, including 11.7 million US adults with potential FDA indications for triple therapy, highlighting the urgent need to optimize implementation and utilization of CKM syndrome therapies.
Introduction: Hospitalization for COVID-19 in the Pre-Omicron era was associated with an increased risk of major adverse cardiovascular events (MACE), but data from the Omicron era are sparse. Hypothesis: We investigated the hypothesis that the risk of MACE among older adults hospitalized for COVID-19 during the Omicron era is lower than in the Pre-Omicron era, and is similar to that in a historical influenza cohort. Methods: Retrospective cohort study using 100% of Medicare fee-for-service claims, including beneficiaries aged ≥ 65 years hospitalized for COVID-19 in the Omicron era (11/26/21–09/30/2022), for COVID-19 in the Pre-Omicron era (03/01/20 –11/25/21), or for influenza in the pre-COVID-19 era (03/01/16 – 09/30/18). Outcomes were measured at 1 year of follow-up after the index hospital admission date. The primary endpoint was the cumulative incidence proportion (risk) of MACE (composite of all-cause death, myocardial infarction, ischemic stroke/ transient ischemic attack, pulmonary embolism, deep vein thrombosis, heart failure hospitalization, and cardiac arrest). Secondary outcomes included individual components of MACE and hospitalization for atrial fibrillation. We used inverse probability weighting to allow comparisons to a common standard (Omicron cohort), the Kaplan-Meier method to estimate the cumulative incidence of outcomes including death, and the Aalen-Johansen method for outcomes that did not include death. Results The analysis included 363,719 patients in the Omicron cohort (mean age 79 years; 51% women; 18% dual enrollees), 724,657 in the Pre-Omicron cohort, and 149,572 in the influenza cohort. The risk of MACE at 1 year was lower in the Omicron than in the pre-Omicron cohort (48.6% vs. 49.3%, risk difference after standardization [RD], -0.7%; 95% CI -0.9% to -0.5%) but higher than in the influenza cohort (48.6% vs. 30.3%; RD, 18.3%; 18.1% – 18.6%). Compared with the pre-Omicron cohort, the Omicron cohort had a significantly lower risk of death at 1 year (34.2% vs. 40.1%, RD, -5.9%; -6.1% to -5.7%) but a higher risk of hospitalizations for myocardial infarction, stroke, heart failure, and atrial fibrillation (figure). Conclusions Among older adults, the risk of MACE within 1 year after COVID-19 hospitalization in the Omicron era has declined since the Pre-Omicron era, driven by all-cause death, but it remains elevated. Notably, during the Omicron period the risk of MACE is still higher than that observed in a historical influenza cohort.
BACKGROUND:Patients hospitalized with COVID-19 from socioeconomically vulnerable communities are at risk for in-hospital cardiovascular events. However, the association of socioeconomic vulnerability and outcomes after hospitalization is uncertain. METHODS AND RESULTS:American Heart Association COVID-19 Cardiovascular Disease Registry hospitalizations between March 1, 2020, and June 30, 2022, linked with Medicare fee-for-service claims, were analyzed. We used Centers for Disease Control and Prevention's Social Vulnerability Index to ascertain county-level and Medicare-Medicaid dual eligibility to ascertain patient-level social vulnerability. We evaluated the association between social vulnerability and a composite of myocardial infarction, stroke, heart failure, venous thromboembolism, cardiogenic shock, cardiac arrest, and death, following discharge, using Cox regression models. The study included 8565 patients (mean age 78 years, 50% female, 16% Black, 4% Hispanic, 25% dual eligible, 34% residing in the most vulnerable counties). Patients residing in the most vulnerable counties, and dual eligible patients, were more likely to be female, Black or Hispanic, and have increased comorbidities. A total of 3783 (52%) patients experienced a composite outcome. We found no association between the most vulnerable, compared with least vulnerable, counties and cardiovascular events (hazard ratio [HR], 0.97 [95% CI, 0.87-1.07]). Dual eligibility, compared with nondual eligibility, was associated with increased cardiovascular events (HR, 1.28 [95% CI, 1.19-1.37]), which was attenuated after adjusting for comorbidities (HR, 0.97 [95% CI, 0.89-1.04]). CONCLUSIONS:Among survivors of COVID-19 hospitalization, patient-level social vulnerability was associated with cardiovascular events, explained by increased comorbidities. County-level social vulnerability was not observed to be a risk for postdischarge events. Findings suggest targeting public health efforts toward dual eligible patients to mitigate poor outcomes.
We discuss generalizability analyses under a partially nested trial design, where part of the trial is nested within a cohort of trial-eligible individuals, while the rest of the trial is not nested. This design arises, for example, when only some centers participating in a trial are able to collect data on nonrandomized individuals, or when data on nonrandomized individuals cannot be collected for the full duration of the trial. Our work is motivated by the Necrotizing Enterocolitis Surgery Trial, which compared initial laparotomy versus peritoneal drain for infants with necrotizing enterocolitis or spontaneous intestinal perforation. During the first phase of the study, data were collected from randomized individuals as well as consenting nonrandomized individuals; during the second phase of the study, however, data were only collected from randomized individuals, resulting in a partially nested trial design. We propose methods for generalizability analyses with partially nested trial designs. We describe identification conditions and propose estimators for causal estimands in the target population of all trial-eligible individuals, both randomized and nonrandomized, in the part of the data where the trial is nested while using trial information spanning both parts. We evaluate the estimators in a simulation study and provide an illustration using the Necrotizing Enterocolitis Surgery Trial study.