The use of a responder analysis to identify differences in patient outcomes following a self-care intervention to improve cancer pain management

Abstract

Previously, we demonstrated, in a randomized clinical trial, the effectiveness of a psychoeducational intervention to decrease pain intensity scores and increase patients' knowledge of cancer pain management with a sample of oncology patients with pain from bone metastasis. In the present study, we evaluated for changes in mood states (measured using the Profile of Mood States), quality of life (QOL; measured using the Medical Outcomes Study Short Form-36 (SF-36)), and pain's level of interference with function (measured using the Brief Pain Inventory (BPI)) from baseline to the end of the intervention first between the intervention and the standard care groups and then within the intervention group based on the patients' level of response to the intervention (i.e., patients were classified as non-responders, partial responders, or responders). No differences were found in any of these outcome measures between patients in the standard care and intervention groups. However, when patients in the intervention group were categorized using a responder analysis approach, significant differences in the various outcome measures were found among the three respondent groups. Differences in the physical and mental component summary scores on the SF-36 and the interference items on the BPI, among the three respondent groups, were not only statistically significant but also clinically significant. The use of responder analysis in analgesic trials may help to identify unique subgroups of patients and lead to the development of more effective psychoeducational interventions.

Determination of Cutpoints for Low and High Number of Symptoms in Patients with Advanced Cancer

“…These scores are similar to those reported in one study, 60 lower than scores reported in three studies of patients with advanced cancer, [61][62][63] and higher than scores reported in another study 64 of patients with advanced cancer. Reasons for these inconsistencies may include differences in studies' definition of advanced cancer, their inclusion and exclusion criteria, and timing of the assessments.…”

When more data steer us wrong: replications with the wrong dependent measure perpetuate erroneous conclusions

“…This analysis put interpretations of localized brain function at risk. To move further afield, Tidwell, Dougherty, Chrabaszcz, Thomas, and Mendoza (2014) reviewed the measures used in brain-training studies (e.g., Jaeggi, Buschkuehl, Jonides, & Shah, 2011;Miaskowski et al, 2007). In one common analysis, participants whose performance was affected by training ("responders"), and those whose performance does not change ("nonresponders") are treated as separate groups.…”

Patient-reported outcome measurement in drug discovery: a tool to improve accuracy and completeness of efficacy and safety data

“…One successful approach to the statistical analysis of PROs which directly integrates clinical meaningfulness is to prespecify the clinically meaningful score change in the PRO measure, and plan to quantify the proportion of study participants who experience that change at a prespecified time point(s) of interest (i.e. responders) [22]. The approach for handling missing data should be specified for both missing questions within a scale and completely missing questionnaires, including at least two different sensitivity analyses involving missing data methods such as multiple imputation [1].…”