Deciding to consult the general practitioner for joint pain: a choice-based conjoint analysis study

Abstract

A substantial proportion of older adults with non inflammatory joint pain do not consult their general practitioner (GP) despite apparent clinical need. This thesis describes the development, execution, and interpretation of an original study using conjoint analysis – a fairly novel approach with some advantages over conventional observational and qualitative studies - to understand the relative importance of need-related and service-related factors on the decision to consult the GP.

Background reading, a systematic review of previously published conjoint analysis studies, and a series of developmental studies involving patients and members of the public informed the design of the main study.

A partial-profile choice-based conjoint (PPCBC) questionnaire was chosen, comprising 10 choice tasks using a combination of selected attributes (pain characteristics, pain disruption to everyday life, comorbidity, assessment and investigations available, available treatment options, and perceived GP attitude). The PPCBC questionnaire was postally-administered to 1170 adults aged 50 years and over with hip, knee, or hand pain identified from an existing population cohort study in North Staffordshire.

863 questionnaires were returned (adjusted response rate 74%; mean age: 70 years; 55% female) and well-completed (<5% missing data). The extent to which pain disrupted everyday life (1.10 logits) and perceived GP attitude (0.86 logits) were the most important determinants of the decision to consult the GP. Service factors were highly influential with a ‘negative’ GP attitude potentially outweighing the perceived value of optimal assessment and management. Latent class analysis identified possible subgroups with differing strengths of preference.

Conjoint analysis is feasible and offers unique insights into the relative importance of actual and hypothetical services. While it presents many challenges - extensive developmental testing, complex design and analysis procedures, ability to integrate findings from a range of different methods – it can provide important information on patients’ preferences for existing and emerging treatments and models of care.