Association between outcome disparities and pragmatic features related to clinical trial and real-world settings in nasopharyngeal carcinoma: A population-based retrospective cohort study, 2006-2016.

ABSTRACT

BACKGROUND: Prognosis often differs between trial participants and nontrial (pragmatic) patients in similar clinical scenarios, raising a concern that results of trials may not represent those in real-world practice. METHODS AND MATERIALS Individual patient data were extracted from three phase III randomized controlled trials and a big-data real-world database (n = 10,126). Patients with nasopharyngeal carcinoma receiving concurrent chemoradiotherapy (CCRT [control] 2438 vs. 519) or induction chemotherapy plus CCRT (experimental) were included. Propensity score matching and correspondence analysis were used for data mining. RESULTS: Compared with the real-world CCRT cohort, clinical trials preferred to include cases with T4 (25.3-43.3% vs. 18.8%) and N2 (44.4-60.7% vs. 38.9%) categories. Real-world patients were more likely to undergo shorter irradiation time (44 vs. 46-49 days), inadequate chemotherapy cycles (70.6% vs. 25.2-43.9%), other chemotherapy (36.4% vs. 0.0%), and flexible regimens (≥3 vs. 1). Although real-world patients had better survival than trial participants, the survival disparities disappeared in the matched cohorts, except for in one trial with the lowest pragmatism assessment caused by stringent eligibility criteria and low flexibility of delivery. Stage specification, year of treatment, and Epstein-Barr virus DNA were related to survival disparities (all P ≤ 0.034). The influence of pragmatic features on survival mainly affected the control (all P ≤ 0.043) rather than the experimental group. CONCLUSION: Special attention should be paid to the control group when interpreting trial results. Assessing whether the pragmatic features of studies deviate from routine practice will lead to better conversion of trial findings into clinical guidelines.