Practice characteristics influencing variation in provision of depression care in general practice in Norway; a registry-based cohort study (The Norwegian GP-DEP study)

Background: There is growing evidence of variation in treatment for patients with depression, not only across patient characteristics, but also with respect to the organizational and structural framework of general practitioners' (GPs') practice. However, the reasons for these variations are sparsely examined. This study aimed to investigate associations of practice characteristics with provision of depression care in general practices in Norway. Methods: A nationwide cohort study of residents aged ≥ 18 years with a new depression episode in general practice during 2009-2015, based on linked registry data. Exposures were characteristics of GP practice: geographical location, practice list size, and duration of GP-patient relationship. Outcomes were talking therapy, antidepressant medication and sick listing provided by GP during 12 months from date of diagnosis. Associations between exposure and outcome were estimated using generalized linear models, adjusted for patients' age, gender, education and immigrant status, and characteristics of GP practice. Results: The study population comprised 285 113 patients, mean age 43.5 years, 61.6% women. They were registered with 5 574 GPs. Of the patients, 52.5% received talking therapy, 34.1% antidepressant drugs and 54.1% were sick listed, while 17.3% received none of the above treatments. Patients in rural practices were less likely to receive talking therapy (adjusted relative risk (adj RR) = 0.68; 95% confidence interval (CI) = 0.64-0.73) and more likely to receive antidepressants (adj RR = 1.09; 95% CI = 1.04-1.14) compared to those in urban practices. Patients on short practice lists were more likely to receive medication (adj RR = 1.08; 95% CI = 1.05-1.12) than those on long practice lists. Patients with short GP-patient relationship were more likely to receive talking therapy (adj RR = 1.20; 95% CI = 1.17-1.23) and medication (adj RR = 1.08; 95% CI = 1.04-1.12), and less likely to be sick-listed (RR = 0.88; 95% CI = 0.87-0.89), than patients with long GP-patient relationship. Conclusions: Provision of GP depression care varied with practice characteristics. Talking therapy was less commonly provided in rural practices and among those with long-lasting GP-patient relationship. These differences may indicate some variation, and therefore, its reasons and clinical consequences need further investigation.