Evaluating Outcomes and Self-Reported Change in Low Back Pain

Abstract

Background LBP is the most common musculoskeletal disorder worldwide. There are several outcome measures for assessing effectiveness of interventions. In this thesis, we conducted studies on two common outcome measures: Global Perceived Effect (GPE) and Impact PROMIS, a newly proposed short form based on a general health measure from the Patient-Reported Outcome Measures Information System (i.e. PROMIS) suite of tools. GPE quantifies patients’ overall perception of improvement after an intervention. However, factors that patients with LBP and physiotherapists consider while rating GPE are not clear. Moreover, psychometric properties of the Impact PROMIS questionnaire have yet to be explored, highlighting a gap in understanding the suitability of the measure in assessing outcome domains in LBP. Objectives 1) To examine the agreement between patients and physiotherapists, and factors that are associated with their GPE rating. 2) To estimate the psychometric properties (convergent validity and responsiveness) and minimal clinically important difference (MCID) of the 9-item Impact PROMIS questionnaire in individuals with chronic non-specific low back pain. Methods Study one: We assessed correlation and agreement between GPE rated by patients and physiotherapists. Furthermore, logistic regression analysis was used to build two models for individuals with LBP and physiotherapists. Lastly, we compared models according to model fit statistics. Study two: To assess convergent validity, we used Spearman correlation analysis and hypothesis testing of the Impact PROMIS domains of pain, disability, function and quality of life. For responsiveness we used three methods: distribution-based (effect size), construct (correlation analysis and hypothesis testing) and criterion (Receiver Operator Characteristic curves with GPE) for the Impact PROMIS questionnaire with aforementioned domains. Results The first study showed a moderate agreement on GPE ratings between patients and physiotherapist. Also, regression analysis demonstrated the most informative model for patients with the highest explanatory ability includes pain, disability and psychosocial factors; however, physiotherapist’s model (pseudo-R²=10) comprises pain, disability and function. The second study showed the Impact PROMIS questionnaire has high convergent validity as 80% of hypotheses were met. For responsiveness, using a distribution-based methods, the effect size was large (0.8); construct approach, 60% of hypotheses were met; and criterion approach, the Area Under Curve (AUC) was 0.76, indicating moderate responsiveness. Moreover, the MCID was estimated to be 7 points. Conclusion In the first study, we found a moderate agreement, and a discrepancy between patients and physiotherapists. In fact, patients’ improvement is associated with changes in psychosocial factors, while physiotherapists may not be sufficiently emphasized on these factors. However, there are other factors that we could not identify in this study to explain the rest of GPE response variance. The second study suggested the Impact PROMIS questionnaire has high convergent validity and acceptable level of responsiveness. Thus, it could be used in research and clinical practice for patients with LBP. However, we strongly recommend using a legacy or personalized measure for assessing function along with the Impact PROMIS questionnaire.