Construct Validity and Precision of Different Patient-reported Outcome Measures During Recovery After Upper Extremity Fractures.

Abstract

BACKGROUND Patient perceptions of their limitations after illness and injury can be quantified using patient-reported outcome measures (PROMs). Few studies have assessed construct validity (using correlations and factor analysis) and precision (floor and ceiling effects) of a range of frequently used PROMs longitudinally in a population of patients recovering from common upper extremity fractures according to area (general health, region-specific, or joint-specific measures) and mode of administration (fixed-scale or computer adaptive test). QUESTIONS/PURPOSES (1) What is the strength of the correlation between different PROMs within 1 week, 2 to 4 weeks and 6 to 9 months after shoulder, elbow, and wrist fractures? (2) Using a factor analysis, what underlying constructs are being measured by these PROMs? (3) Are there strong floor and ceiling effects with these instruments? METHODS Between January 2016 and August 2016, 734 patients recovering from an isolated shoulder, elbow, or wrist fracture completed physical-limitation PROMs at baseline (the initial office visit after diagnosis in the emergency department), 2 to 4 weeks after injury, and at the final assessment 6 to 9 months after injury. In all, 775 patients were originally approached; 31 patients (4%) declined to participate due to time constraints, four patients died of unrelated illness, and six patients were lost to follow-up. The PROMs included the PROMIS Physical Function (PF, a computer adaptive, general measure of physical function), the PROMIS Upper Extremity (UE, a computer adaptive measure of upper extremity physical function), the QuickDASH (a fixed-scale, region-specific measure), the Oxford Shoulder Score (OSS), the Oxford Elbow Score (OES) and the Patient-rated Wrist Evaluation (PRWE) (a fixed-scale, joint-specific measure), and the EQ-5D-3L (a fixed-scale measure of general health). PROMs were evaluated during recovery for construct validity (using correlations and factor analysis) and precision (using floor and ceiling effects). RESULTS Physical-limitation PROMs were intercorrelated at all time points, and the correlation strengthened over time (for example, PROMIS UE and QuickDASH at 1 week, r = -0.4665; at 2 to 4 weeks, r = -0.7763; at 6 to 9 months, r = -0.8326; p < 0.001). Factor analysis generated two factors or groupings of PROMs that could be described as capability (perceived ability to perform or engage in activities), and quality of life (an overall sense of health and wellbeing) that varied by time point and fracture type, Joint-specific and general-health PROMs demonstrated high ceiling effects 6 to 9 months after injury and PROMIS PF, PROMIS UE and QuickDASH had no floor or ceiling effects at any time points. CONCLUSIONS There is a substantial correlation between PROMs that assess physical limitations (based on anatomic region) and general health after upper extremity fractures, and these relationships strengthen during recovery. Regardless of the delivery mode or area of focus, PROMs largely appear to represent two underlying constructs: capability and quality of life. Computer adaptive tests may be favored over fixed-scale measures for their efficiency and limited censoring. LEVEL OF EVIDENCE Level II, therapeutic study.