Real-World Data as External Controls: Practical Experience from Notable Marketing Applications of New Therapies.

INTRODUCTION: Real-world data (RWD) can contextualize findings from single-arm trials when randomized comparative trials are unethical or unfeasible. Findings from single-arm trials alone are difficult to interpret and a comparison, when feasible and meaningful, to patient-level information from RWD facilitates the evaluation. As such, there have been several recent regulatory applications including RWD or other external data to support the product's efficacy and safety. This paper summarizes some lessons learned from such contextualization from 20 notable new drug or biologic licensing applications in oncology and rare diseases. METHODS: This review focuses on 20 notable new drug or biologic licensing applications that included patient-level RWD or other external data for contextualization of trial results. Publicly available regulatory documents including clinical and statistical reviews, advisory committee briefing materials and minutes, and approved product labeling were retrieved for each application. The authors conducted independent assessments of these documents focusing on the regulatory evaluation, in each case. Three examples are presented in detail to illustrate the salient issues and themes identified across applications. RESULTS: Regulatory decisions were strongly influenced by the quality and usability of the RWD. Comparability of cohort attributes such as endpoints, populations, follow-up, index and censoring criteria, as well as data completeness and accuracy of key variables appeared to be essential to ensure the quality and relevance of the RWD. Given adequate sample size of the clinical trials or external control, the use of appropriate analytic methods to properly account for confounding, such as regression or matching, and pre-specification of these methods while blinded to patient outcomes seemed good strategies to address baseline differences. DISCUSSION: Contextualizing single-arm trials with patient-level RWD appears to be an advance in regulatory science; however, challenges remain. Statisticians and epidemiologists have long focused on analytical methods for comparative effectiveness but hurdles in use of RWD have often occurred upstream of the analyses. More specifically, we noted hurdles in evaluating data quality, justifying cohort selection or initiation of follow-up, and demonstrating comparability of cohorts and endpoints.

Impact of valve surgery on 6-month mortality in adults with complicated, left-sided native valve endocarditis: a propensity analysis.

CONTEXT: Complicated, left-sided native valve endocarditis causes significant morbidity and mortality in adults. The presumed benefits of valve surgery remain unproven due to lack of randomized controlled trials. OBJECTIVE: To determine whether valve surgery is associated with reduced mortality in adults with complicated, left-sided native valve endocarditis. DESIGN AND SETTING: Retrospective, observational cohort study conducted from January 1990 to January 2000 at 7 Connecticut hospitals. Propensity analyses were used to control for bias in treatment assignment and prognostic imbalances. PATIENTS: Of the 513 adults with complicated, left-sided native valve endocarditis, 230 (45%) underwent valve surgery and 283 (55%) received medical therapy alone. MAIN OUTCOME MEASURE: All-cause mortality at 6 months after baseline. RESULTS: In the 6-month period after baseline, 131 patients (26%) died. In unadjusted analyses, valve surgery was associated with reduced mortality (16% vs 33%; hazard ratio [HR], 0.43; 95% confidence interval [CI], 0.29-0.63; P<.001). After adjustment for baseline variables associated with mortality (including hospital site, comorbidity, congestive heart failure, microbial etiology, immunocompromised state, abnormal mental status, and refractory infection), valve surgery remained associated with reduced mortality (adjusted HR, 0.35; 95% CI, 0.23-0.54; P<.02). In further analyses of 218 patients matched by propensity scores, valve surgery remained associated with reduced mortality (15% vs 28%; HR, 0.45; 95% CI, 0.23-0.86; P =.01). After additional adjustment for variables that contribute to heterogeneity and confounding within the propensity-matched group, surgical therapy remained significantly associated with a lower mortality (HR, 0.40; 95% CI, 0.18-0.91; P =.03). In this propensity-matched group, patients with moderate to severe congestive heart failure showed the greatest reduction in mortality with valve surgery (14% vs 51%; HR, 0.22; 95% CI, 0.09-0.53; P =.001). CONCLUSIONS: Valve surgery for patients with complicated, left-sided native valve endocarditis was independently associated with reduced 6-month mortality after adjustment for both baseline variables associated with the propensity to undergo valve surgery and baseline variables associated with mortality. The reduced mortality was particularly evident among patients with moderate to severe congestive heart failure.

Complicated left-sided native valve endocarditis in adults: risk classification for mortality.

CONTEXT: Complicated left-sided native valve endocarditis causes significant morbidity and mortality in adults. Lack of valid data regarding estimation of prognosis makes management of this condition difficult. OBJECTIVE: To derive and externally validate a prognostic classification system for adults with complicated left-sided native valve endocarditis. DESIGN, SETTING, AND PATIENTS: Retrospective observational cohort study conducted from January 1990 to January 2000 at 7 Connecticut hospitals among 513 patients older than 16 years who experienced complicated left-sided native valve endocarditis and who were divided into derivation (n = 259) and validation (n = 254) cohorts. MAIN OUTCOME MEASURE: All-cause mortality at 6 months after baseline. RESULTS: In the derivation and validation cohorts, the 6-month mortality rates were 25% and 26%, respectively. Five baseline features were independently associated with 6-month mortality (comorbidity [P =.03], abnormal mental status [P =.02], moderate to severe congestive heart failure [P =.01], bacterial etiology other than viridans streptococci [P<.001 except Staphylococcus aureus, P =.004], and medical therapy without valve surgery [P =.002]) and were used to create a prognostic classification system. In the derivation cohort, patients were classified into 4 groups with increasing risk for 6-month mortality: 5%, 15%, 31%, and 59% (P<.001). In the validation cohort, a similar risk among the 4 groups was observed: 7%, 19%, 32%, and 69% (P<.001). CONCLUSIONS: Adults with complicated left-sided native valve endocarditis can be accurately risk stratified using baseline features into 4 groups of prognostic severity. This prognostic classification system might be useful for facilitating management decisions.