Impact of Missing Data on Bias and Precision When Estimating Change in Patient-Reported Outcomes from a Clinical Registry

• Activity Number: 259 - SPEED: Missing Data and Causal Inference Methods, Part 2
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 3:05 PM to 3:50 PM
• Sponsor: Health Policy Statistics Section
• Author(s): Olawale Fatai Ayilara* and Lixia Zhang and Tolulope T Sajobi and Richard Sawatzky and Eric Bohm and Lisa M Lix
• Companies: University of Manitoba and University of Manitoba and University of Calgary and School of Nursing, Trinity Western University and University of Manitoba and University of Manitoba
• Keywords: auxiliary variables; maximum likelihood estimation; missing data; mixed-effects model ; clinical registry

Abstract

Clinical registries, which capture information about the health and healthcare use of patients with a specific health condition, often contain patient-reported outcomes (PROs) that provide insights about patient’s perspectives on their own health. PROs are prone to missing data problem, which can result in a loss of statistical power to detect the effect of healthcare interventions. This paper compares the precision and bias of missing data methods when estimating longitudinal change in PRO scores. Analyses were conducted using a population-based joint replacement registry and simulation. Mixed-effects models based on complete case analysis (CCA), maximum likelihood (ML) and multiple imputation (MI) without and with an auxiliary variable (MI-Aux) were used to estimate change in PRO scores. Models based on CCA resulted in a wider 95% CIs than ML and MI methods. Simulation results showed that MI-Aux method, in which a single simulated auxiliary variable was strongly correlated (i.e., 0.8) with the outcome, increased the precision, when compared with the MI method. Inclusion of auxiliary variable in MI models reduces the bias, but identifying the optimal variable may be challenging.