Reliability, validity and discriminability of patient reported outcomes for non-specific low back pain in a nationwide physical therapy registry: A retrospective observational cohort study.

Abstract

A national clinical registry was established in the Netherlands containing data directly sampled from electronic health record systems of physical therapists (PTs). This registry aims to evaluate the potential of patient reported outcome measures (PROMs) to develop quality indicators (QIs) in physical therapy care. To test to what extent the collected PROM data are reliable, valid and discriminatory between practices in measuring outcomes of patients with non-specific low back pain (NSLBP). In this retrospective cohort study 865 PT practices with 6,560 PTs voluntarily collected PROM data of patients with NSLBP, using the Quebec Back Pain Disability Scale (QBPDS), the Numeric Pain Rating Scale (NPRS) and the Patient Specific Functioning Scale (PSFS). Reliability was determined by analysing the completeness of the dataset, the comparability by using national reference data, and through checking selection bias in the included patients. Validity was tested using the known-groups contrast between patients with (sub)acute vs. chronic NSLBP. To determine discriminative ability of outcomes between PT practices, case-mix corrected hierarchical multilevel analyses were performed. Reliability was sufficient by confirming fifteen of the sixteen hypotheses: 59% of all patients opted in for data analysis, 42% of these included patients showed repeated measurement, comparing with reference data and potential selection bias showed < 5% between group differences, while differences between (sub)acute and chronic NSLB-groups were significantly larger than 5% (less treatment sessions, lager differences in outcomes in (sub)acute NSLB patients). In addition, all nine adjusted hierarchical multilevel models confirm that the collected dataset on outcomes in PT care is able to discriminate between practices using PROM results of patients with NSLBP (ICC-scores range 0.11-0.21). Although we have shown the reliability, validity and discriminative ability of the dataset in the quest to develop QIs, we are aware that reducing missing values in patient records and the selective participation of PTs that belong to the innovators needs attention in the next stages of implementation to avoid bias in the results. PROMs of patients with NSLBP collected in the national clinical registry of KNGF are reliable, valid and able to discriminate between primary care PT practices.