The Patient Goal Priority Questionnaire Is Moderately Reproducible in People With Persistent Musculoskeletal Pain

Abstract

BackgroundThe Patient Goal Priority Questionnaire (PGPQ) is a patient-specific measure for identification of behavioral goals and evaluation of clinically significant changes. The use of such a measure in clinical settings and research requires that identified goals be consistent over time. Self-reports of behaviors related to the goals should be reliably estimated.ObjectiveThe purpose of this study was to estimate chance-corrected agreement and test-retest reliability of the PGPQ. Chance-corrected agreement between the PGPQ and a similar therapist-guided goal identification tool, the Patient Goal Priority List (PGPL), also was estimated.DesignA correlative and prospective design with 3 measurement points (M1, M2, and M3) was used in the study.MethodsFifty-four people who consulted physical therapists in primary care for persistent musculoskeletal pain were included in the study. Analyses of chance-corrected agreement and test-retest reliability of the PGPQ were done at M1 and M2. Chance-corrected agreement between procedures (PGPQ and PGPL) also was analyzed at M1 and M3.ResultsThe percentage of agreement on content of the priority lists of the PGPQ at M1 and M2 was 52%. Cohen kappa values for agreement of rankings ranged between .47 and .64. Test-retest reliability coefficients for the self-report scales of the PGPQ ranged from .35 to .81. Chance-corrected agreement decreased when physical therapists were involved in the goal identification process using the PGPL (kappa=.08–.46).LimitationsVarying item content and a small, heterogeneous sample possibly increased variation and the standard error of measurements. The feasibility of using traditional approaches to psychometric evaluation of patient-specific measures is questionable.ConclusionsChance-corrected agreement and test-retest reliability of the PGPQ were moderate. Involving a physical therapist in the goal identification procedure possibly introduced further bias. The size of the measurement error must be taken into account when using the PGPQ for estimations of clinically important changes.