What influences chronic pain management? A best-worst scaling experiment with final year medical students and general practitioners

Abstract

Background: Chronic pain education is an essential determinant for optimal chronic pain management. Given that attitudes and preferences are involved in making treatment decisions, identifying which factors are most influential to final year medical students' and general practitioners' (GPs) chronic pain management choices is of importance. This study investigates Swedish and Australian students' preferences with respect to a chronic pain condition, using a best-worst scaling (BWS) experiment, which is designed to rank alternatives.

Methods: BWS, a stated-preference method grounded in random utility theory, was used to explore the importance of factors influencing chronic pain management.

Results: All three cohorts considered the patients' pain description and previous treatment experience as the most important factors in making treatment decisions, whereas their demographics and voices or facial expressions while describing their pain were considered least important. Factors such as social support, patient preferences and treatment adherence were, however, disregarded by all cohorts in favour of pain assessment factors such as pain ratings, description and history. Swedish medical students and GPs show very high correlation in their choices, although the GPs consider their professional experience as more important compared to the students.

Conclusion: This study suggests that the relative importance of treatment factors is cemented early and thus underline the critical importance of improving pain curricula during undergraduate medical education.

Keywords: Pain management; best–worst scaling; general practitioners; medical students; pain education.

Figure 1.

Example best–worst scaling choice set for the task: ‘Imagine that you will choose treatment for John. Among these factors, select which one is the most important and least important by ticking the boxes’.

Figure 2.

Best, worst and no-choice responses (as percentage of the total response rate) for the Swedish (SWE) and Australian (AUS) students and the Swedish GPs (S-GP). The items have been ordered from the highest to lowest best scores for the Swedish student cohort. For two of the factors (‘c’ and ‘g’), the group distributions precluded chi-square analysis. For the other groups, however, factors ‘b’, ‘e’, ‘i’ and ‘j’ had chi-square p values that were smaller than the critical value of p of 0.022 assuming a 5% false discovery rate. Factor ‘a’ had a p value of 0.037 and the other factors had p values > 0.05.

Figure 3.

Standardized best–worst (B-W) scores for the Swedish and Australian students and the Swedish GPs. The items have been ordered from the highest to lowest best-minus-worst scores for the Swedish student cohort. The standardized scores are taken from Tables 2–4.

Figure 4.

Correlation between (a, b) the mean standardized B-W scores and (c, d) the exp logit coefficients for the Swedish students (N = 30) versus either the Australian students (N = 21; (a, c)) or the Swedish GPs (N = 16; (b, d)). The letters indicate the item in question (see Tables 2–4 for explanations). The 95% confidence bands for the regression lines are shown.