(141) Definition and Validation of Case Detection Algorithms Using Quebec Medico-Administrative Databases (Med-Echo and RAMQ) for Depression Among Seniors

Saturday, August 27, 2022

8:00 AM – 6:00 PM CEST

Location: Hall C

Publication Number: 815

Presenting Author(s)

Giraud Ekanmian
– Centre de Recherche du CHU de Qu�bec-Universit� Laval, Qu�bec, Quebec, Canada

Co-Author(s)

Carlotta Lunghi
– Pharmacology Unit � Department of Medical and Surgical Sciences � University of Bologna, Pharmacology Unit, Department of Medical and Surgical Sciences, University of Bologna, Bologna, Bologna, Italy
Helen-Maria Vasiliadis, MSc, PhD
– Département des sciences de la santé communautaire, Université de Sherbrooke, Canada
Line Guénette, BPharm, MSc, PhD
– Centre de Recherche du CHU de Québec, Université Laval, Québec city, QC, Canada

Background: Health administrative databases can be used in epidemiological surveillance of depression, provided that a validated and efficient algorithm is used

Objectives: Our objective was to compare how well the Quebec medico-administrative databases for identification of depression cases match clinical diagnostic assessments of depression in a sample of primary care patients aged 65 years and above participating in the ESA-Services study

Methods: We developed and tested different algorithms to identify cases of depression using a variation in the application of the following criteria: (1) hospitalization claims with International classification of diseases 9th or 10th revision (ICD-9/10) depression codes, (2) psychiatric or family physician visits with ICD-9 depression codes, and (3)  claims for antidepressant drugs. We varied the time frame allowed between claims from 12 to 24 months. The gold standard used was depression assessed through a structured interview based on self-reported symptoms aligned with DSM-IV criteria during the ESA-Services study. The interview data were linked to the RAMQ health databases. We computed many algorithms' sensibility, specificity, and positive and negative predictive values.

Results: The prevalence of depression was 14.73% based on the ESA-Services definition and 22.21% according to our most efficient algorithm, defined as follows: 1 ICD-10 hospitalization code for depression (F320, F321, F322, F323, F328, F329, F330, F328, F329, F331, F332, F333, F334, F338, F339), or 1 ICD-9 diagnostic code for depression ( 311, 3090, 3091, 3119 , 3004, 3011, 296, 3092), or 1 American Hospital Formulary Service (AHFS) code for antidepressant claims (AHF281604, AHF283632 ). Sensibility was 44.3% (95% Confident interval (CI): 38.0-50.5), specificity was 78.6% (95% CI: 76.4-80.7), positive predictive value was 26.3% (95% CI: 22.0-30.5), and negative predictive value was 89.1% (95% CI: 87.4-90.8). The agreement between the algorithm and the structured interview was low, with a kappa of 0.18 (95% CI: 0.13-0.22).

Conclusions: There are some limitations of using administrative claims data to identify patients with depression in the Quebec elderly population. This may be due to the high level of undiagnosed depression among older adults. Using health administrative databases to identify patients with depression may lead to an important underestimation of depression prevalence