Association of Self‐Reported Clinical Behavior With Guideline Adherence in Schizophrenia and Major Depressive Disorder After Guideline Education
Kazuhiko Yamamuro, Hikaru Hori, Takashi Tsuboi, Shinichiro Ochi, Toru Horinouchi, Yasushi Kawamata, Keisuke Mori, Yusuke Arai, Kazutaka Ohi, Junya Matsumoto, Toshiaki Onitsuka, Koichiro Watanabe, Ken Inada, Norio Yasui‐Furukori, Ryota HashimotoABSTRACT
Background
The Effectiveness of Guidelines for Dissemination and Education in Psychiatric Treatment (EGUIDE) project is a nationwide Japanese educational program promoting guideline‐concordant pharmacotherapy for schizophrenia and major depressive disorder (MDD). Psychiatrists' self‐reported clinical behavior (CB) scores after EGUIDE training are partially associated with objective quality indicators of specific practices. However, their association with comprehensive objective measures of overall guideline adherence remains unclear.
Methods
This multicenter observational study included 1897 EGUIDE‐trained psychiatrists. CB was assessed online immediately after training. The Guideline Adherence Rate (GAR) was calculated for 5328 patients with schizophrenia and 3538 patients with MDD whose discharge records were linked to their psychiatrist's CB score. Associations were evaluated using unadjusted patient‐level Spearman correlations, without modeling clustering within psychiatrists or institutions. Sensitivity analyses aggregated the data to the psychiatrist‐year level, adjusted for institution type, psychiatrist experience, and patient age, and used generalized estimating equations (GEE) to correct the patient‐level inference for clustering.
Results
CB scores showed positive but very weak correlations with GAR (schizophrenia, ρ = 0.09–0.14; MDD, ρ = 0.08–0.11; all p < 0.001). Squared coefficients (approximately 0.006–0.020) indicated that CB scores accounted for < 2% of the rank‐order variation in GAR. These associations remained statistically significant in psychiatrist‐year‐level, covariate‐adjusted, and clustering‐corrected (GEE) sensitivity analyses.
Conclusions
Post‐training self‐reported CB scores were weakly associated with patient‐level GAR. Given the cross‐sectional, unadjusted analysis and the non‐independence of patients treated by the same psychiatrist, these findings do not establish CB as a reliable proxy for actual clinical practice.
Trial Registration
University Hospital Medical Information Network Clinical Trials Registry, UMIN000022645