Completeness of repeated patient-reported outcome measures in adult rehabilitation: a randomized controlled trial in a diverse clinical population.

Abstract

Background: Data collection through patient-reported outcome measures (PROMs) is essential for the purpose of rehabilitation research and registries. Existing problems with incomplete PROM data may relate to the patient burden and data set length. This study aimed to analyse response patterns and degree of data completeness in systematic outcome assessments conducted within a clinical study in a multidisciplinary rehabilitation setting, comparing completeness of a brief and a longer set of PROMs.

Methods: The Norwegian RehabNytte Cohort was developed to monitor patients' long-term benefit of rehabilitation and progress on PROMs. Adults admitted to one of 17 institutions providing mostly inpatient rehabilitation in secondary healthcare were included between January 2019 and March 2020, and followed for one year. For the purpose of the current randomized controlled trial, the Cohort-patients in 16/17 institutions were randomized to complete either a brief or a longer set of PROMs at admission, discharge, and after 3, 6 and 12 months. The brief set comprised the EQ-5D-5L and additional generic PROMs commonly used in rehabilitation settings. The longer data set expanded upon the brief set by including the Patient-Specific Functional Scale and the 29-item version of the PROMIS Profile instruments. Completeness was measured as the extent of present applicable PROM data at each time point. In addition, we assessed response patterns in terms of dropout rates and intermittently missed assessment events. The RehabNytte study is registered under ClinicalTrial.gov (NCT03764982, first posted 05.12.2018).

Results: Of the 2904 patients included, 1455 were assigned to the brief data set and 1449 to the longer data set. The majority of patients were referred to rehabilitation for rheumatic and musculoskeletal diseases (39.3%) or cancer (26.9%). The data set completeness was significantly higher in the brief set compared to the longer (p < 0.001). From admission to 12 months follow-up, differences in completeness between groups ranged from 6.5 to 12.6 percentage points, consistently favouring the brief set. Completeness was highest at admission, reaching 96.8% (95% CI 0.96-0.98) for the brief set and 84.2% (95% CI 0.82-0.86) for the longer set. The lowest completeness was observed at discharge, with 46.0% (95% CI 0.43-0.49) for the brief set and 39.5% (95% CI 0.37-0.42) for the longer one. Discharge was the only time point without automatic reminders to non-responders from the digital data collection system. Patients responding to the longer data set exhibited the highest dropout rates, while degree of intermittent missing data was comparable between groups. In both groups, only one-third of patients provided complete or partly responses at all five time points.

Conclusions: This study demonstrated that a brief set of PROMs achieved higher data completeness compared to a longer set, when used for repeated measurements in a rehabilitation research setting.