Validity of the Dutch physiotherapy diagnosis system to classify nonspecific low back pain in primary care

Abstract

Background Efficiency and effectiveness are key measures in quality of healthcare. Stratified care can be helpful to provide insight into efficiency and effectiveness for subgroups on an aggregate level across therapists, practices or even across regions in the country. Dutch physiotherapists in primary care use the “Diagnose Classificatie Systeem Paramedische Hulp” (DCSPH) to classify patients. The question is whether the long-standing DCSPH is a valid system to allocate patients in relevant categories. To further investigate this, patients with nonspecific low back pain (NSLBP) were selected to study the added value of DCSPH. Aim The primary goal of this study is to evaluate the inter-rater reliability of the Dutch Diagnosis Classification System Paramedical Help (DCSPH) in comparison with an eligible alternative classification system. Method This study used a retrospective mixed method design. First, a literature search was performed to search for potential classification methods based on prognostic factors in NSLBP. Second, we used a database to explore how the DCSPH and an alternative classification method are used in classifying patients with NSLPB in current practice. Third, we explored the inter-rater reliability of the DCSPH and the alternative classification method using data of sixty patients to score the agreement between six physiotherapists. Finally, stakeholders in the field were asked to give their opinion about the alternative classification method. Results We designed an alternative classification system that consists of four prognostic profiles, based on several aspects of ten identified treatment guidelines and fourteen systematic reviews. The Cohen’s kappa tests for the DCSPH ranged between .006 and .133 and the overall Fleiss kappa test was .002. The Cohen’s kappa tests of the alternative classification ranged between .184 and .557 and the overall Fleiss was .291. Experts indicated the alternative classification system as useful in daily practice. Conclusion The inter-rater reliability was poor for the DCSPH and fair to moderate for the alternative classification system in classifying patients with NSLBP. It seems promising to classify patients in subgroups based on prognostic factors. Future research should focus on establishing a more accurate tool to define subgroups in NSLBP.