Toward the use of proxy reports for estimating long-term patient-reported outcomes in multiple sclerosis.

ABSTRACT

BACKGROUND: Assessment of disease impact in multiple sclerosis (MS) is usually driven by information obtained directly from patients using patient-reported outcomes. However, when patients' response in longitudinal studies is less reliable or missing, proxy respondents may be used. OBJECTIVE: The objective of this paper is to evaluate whether long-term patient scores can be reliably estimated using scores obtained from proxies. METHODS: Baseline, six-month and two-year data were collected from 155 patients and proxies on the physical scale of the Multiple Sclerosis Impact Scale (MSIS-29). Linear regression analyses were performed with the patient two-year scores as outcome, proxy two-year scores as predictor and other variables that could contribute to a better prediction of the patient follow-up score. RESULTS: The patient follow-up score could be predicted rather accurately (R(2) = 0.74) using the patient baseline score and the proxy follow-up score. The correlation between observed and predicted scores was 0.86. The model performed well in different follow-up durations and even better in an external cohort. CONCLUSION: A simple model of a constant value (intercept), the patient baseline score and the proxy follow-up score can predict patients' follow-up score on the physical impact of MS.