Reducing survey burden: feasibility and validity of PROMIS measures in multiple sclerosis

Abstract

Background: Patient-reported outcomes are important for clinical research and care, yet administering and scoring the questionnaires requires considerable effort and time. The Patient Reported Outcomes Measurement Information System (PROMIS) could considerably reduce administrative obstacles and lessen survey burden for participants.

Objective: Assess the feasibility and validity of PROMIS, compared to commonly-used legacy measures for multiple sclerosis (MS).

Methods: In this cross-sectional survey, 133 participants with confirmed MS completed legacy surveys and PROMIS Computerized Adaptive Tests (CATs) for depression, anxiety, pain, fatigue and physical function. We conducted a multi-trait, multi-method analysis and verified results with confirmatory factor analysis.

Results: The correlations between PROMIS and the corresponding legacy measures were large (0.67 to 0.87). The multi-trait, multi-method criteria were generally well met, providing good evidence of the validity of PROMIS measures. PROMIS surveys asked fewer questions and required substantially less time to complete than the legacy scales.

Conclusions: Our results provide evidence of the construct validity of PROMIS for use with MS patients. Several aspects of the PROMIS CATs made them an important resource, including: (a) less time was required to complete them; (b) missing data was reduced; and (c) the automatic scoring referenced the general population. Our findings support the use of PROMIS in MS research and may have broader implications for clinical care, as well.

Keywords: Computerized Adaptive Test; PROMIS; multiple sclerosis; outcome measurement; self-reporting; survey design; validation.

© The Author(s) 2013.

Figures

Figure 1

Study flow diagram.

Figure 2

Mean PROMIS T scores for depression, anxiety, fatigue, pain, and physical function by MS subtype in a cross-sectional cohort (n=133). T scores are anchored to a general population mean of 50 with a standard deviation of 10. Higher scores indicate more of the trait being measured (e.g. depression scores above 50 indicate more depression than the average general population, physical function scores above 50 indicate more/better physical function than the average general population). RRMS: relapsing remitting MS; SPMS: secondary progressive MS; PPMS: primary progressive MS; Error bars: 95% confidence intervals.

Figure 3

Mean PROMIS T scores for fatigue, pain, and physical function by self-reported disability in a cross-sectional cohort (n=133). T scores are anchored to a general population mean of 50 with a standard deviation of 10. Higher scores indicate more of the trait being measured (e.g. fatigue scores above 50 indicate more fatigue than the average general population, physical function scores above 50 indicate more/better physical function than the average general population). Error bars: 95% confidence intervals.