The association between general practitioner participation in joint teleconsultations and rates of referral: a discrete choice experiment

BACKGROUND Joint consultations - such as teleconsultations - provide opportunities for continuing education of general practitioners (GPs). It has been reported this form of interactive case-based learning may lead to fewer GP referrals, yet these studies have relied on expert opinion and simple frequencies, without accounting for other factors known to influence referrals. We use a survey-based discrete choice experiment of GPs' referral preferences to estimate how referral rates are associated with participation in joint teleconsultations, explicitly controlling for a number of potentially confounding variables. METHODS We distributed questionnaires at two meetings of the Portuguese Association of General Practice. GPs were presented with descriptions of patients with dermatological lesions and asked whether they would refer based on the waiting time, the distance to appointment, and pressure from patients for a referral. We analysed GPs' responses to multiple combinations of these factors, coupled with information on GP and practice characteristics, using a binary logit model. We estimated the probabilities of referral of different lesions using marginal effects. RESULTS Questionnaires were returned by 44 GPs, giving a total of 721 referral choices. The average referral rate for the 11 GPs (25%) who had participated in teleconsultations was 68.1% (range 53-88%), compared to 74.4% (range 47-100%) for the remaining physicians. Participation in teleconsultations was associated with reductions in the probabilities of referral of 17.6% for patients presenting with keratosis (p = 0.02), 42.3% for psoriasis (p < 0.001), 8.4% for melanoma (p = 0.14), and 5.4% for naevus (p = 0.19). CONCLUSIONS The results indicate that GP participation in teleconsultations is associated with overall reductions in referral rates and in variation across GPs, and that these effects are robust to the inclusion of other factors known to influence referrals. The reduction in range, coupled with different effects for different clinical presentations, may suggest an educational effect. However, more research is needed to establish whether there are causal relationships between participation in teleconsultations, continuing education, and referral rates.