Bayesian analysis of longitudinal dyadic data with informative missing data using a dyadic shared-parameter model

Analyzing longitudinal dyadic data is a challenging task due to the complicated correlations from repeated measurements and within-dyad interdependence, as well as potentially informative (or non-ignorable) missing data. We propose a dyadic shared-parameter model to analyze longitudinal dyadic data with ordinal outcomes and informative intermittent missing data and dropouts. We model the longitudinal measurement process using a proportional odds model, which accommodates the within-dyad interdependence using the concept of the actor-partner interdependence effects, as well as dyad-specific random effects. We model informative dropouts and intermittent missing data using a transition model, which shares the same set of random effects as the longitudinal measurement model. We evaluate the performance of the proposed method through extensive simulation studies. As our approach relies on some untestable assumptions on the missing data mechanism, we perform sensitivity analyses to evaluate how the analysis results change when the missing data mechanism is misspecified. We demonstrate our method using a longitudinal dyadic study of metastatic breast cancer.

A post-hoc health-related quality of life (HRQoL) analysis of patients with metastatic colorectal cancer (mCRC) in the phase III CONCUR trial.

667 Background: CONCUR (NCT01584830) showed that regorafenib (REG) significantly improves overall survival (OS) and progression-free survival (PFS) vs. placebo (PBO) in Asian patients with mCRC who progressed after standard therapy (J Li, et al. WCGI 2014). Post hoc exploratory analyses were conducted to assess the effect of treatment on HRQoL. Methods: Patients were randomly assigned 2:1 to treatment with either REG (n=136) or PBO (n=68). The HRQoL analyses included all 204 patients and were selected a priori based on clinical relevance; the global health status/QoL (QL) and the physical functioning (PF) scales of the EORTC QLQ-C30 questionnaire were used. A linear mixed-effects model (LMM) was used to examine the treatment effect on HRQoL and trends over time, assuming that data were missing at random. A pattern-mixture model (PMM) was applied to assess the treatment effect while accounting for potentially informative missing data. Time-to-deterioration (TTD) of HRQoL and responder analyses were conducted to determine the treatment effect based on timing and proportion of patients reaching a minimal important difference (MID) change in QL/PF (≥10 points). Results: The QL and PF changes over time were numerically similar between REG and PBO based on the LMM. The PMM grouped patients based on timing of last HRQoL assessment (<3 or ≥3 cycles) and had results similar to the LMM, demonstrating little impact of informative missing data. For the TTD analysis, when an event was defined as the earliest MID decrease in QL/PF, disease progression, or death, REG showed significantly different TTD curves from PBO (QL: median 8.0 vs. 7.0 weeks, hazard ratio (HR)=0.54; PF: median 7.9 vs. 7.0 weeks, HR=0.59, respectively; all p<0.01). Median TTD was comparable between treatments after removing progression/death from the definition. The responder analyses showed that a similar proportion of patients achieved an improvement in MID in REG vs. PBO (QL: 27.2% vs. 29.4%; PF: 14.0% vs.16.2%, respectively). Conclusions: The findings of this exploratory analysis demonstrate that HRQoL is similar for the REG and PBO groups, indicating that REG prolongs OS and PFS vs. PBO while maintaining a comparable HRQoL. Clinical trial information: NCT01584830.