ESTIMATING THE PREVALENCE OF TREATMENT RESISTANT DEPRESSION AND COMPARING HEALTHCARE RESOURCE UTILISATION USING LINKED PRIMARY CARE AND PSYCHIATRIC SECONDARY CARE ELECTRONIC HEALTH RECORDS IN ENGLAND
Author(s)
Georgina Ireland, PhD1, Maxime Taquet, MRCPsych1, Dalila Monica Moisescu, MPhil1, Adam Marsh, MSc2, Neva Eleangovan, Bsc2, David Price, MRCGP2, Benjamin Fell, PhD3.
1Akrivia Health, Oxford, United Kingdom, 2Observational and Pragmatic Research Institute, Singapore, Singapore, 3Akrivia Health, OXFORD, United Kingdom.
1Akrivia Health, Oxford, United Kingdom, 2Observational and Pragmatic Research Institute, Singapore, Singapore, 3Akrivia Health, OXFORD, United Kingdom.
OBJECTIVES: Treatment-resistant depression (TRD) is a severe and costly form of major depressive disorder (MDD), but prevalence is difficult to estimate due to fragmented care across primary and secondary settings. This study uses linked datasets to estimate TRD prevalence and evaluate the added value of data linkage.
METHODS: We analysed linked electronic health records from three NHS Trusts participating in a pilot linkage between the Akrivia Health Research Database (secondary care [SC]) and the Optimum Patient Care Research Database (primary care [PC]). Adults aged 18-65 with an MDD diagnosis between 2000 and 2024 and at least six weeks of antidepressant treatment were included. Prescribing data were organised into treatment episodes (treatment regimens ≥6 weeks) and treatment blocks (sequences of episodes <42 days apart within an MDD period). TRD was defined as ≥2 antidepressant switches within a block, expanded to include ketamine or electroconvulsive therapy, or clinical note documentation of TRD identified via natural language processing. Healthcare resource use (HCRU), including inpatient bed-days and primary/secondary care appointments, was measured per year of follow-up.
RESULTS: Among 31,237 eligible patients, 9.5% (n=2,954) met TRD criteria based on antidepressant switching, rising to 11.0% (n=3,441) when additional indicators were included. TRD prevalence using only PC data was 8.9% (2,787/31,237) and 1.3% (400/31,237) using SC data. Patients with TRD had higher annual HCRU than non-TRD patients, with 2.8 vs 1.7 inpatient bed-days, 9.4 vs 6.0 SC appointments, and 9.7 vs 5.1 PC appointments per year.
CONCLUSIONS: Linked data provided a more complete view of treatment pathways and outcomes, improving TRD prevalence and HCRU estimates compared with single-source data. These findings highlight the importance of cross-sector data linkage for accurately identifying TRD populations and informing research and service planning.
METHODS: We analysed linked electronic health records from three NHS Trusts participating in a pilot linkage between the Akrivia Health Research Database (secondary care [SC]) and the Optimum Patient Care Research Database (primary care [PC]). Adults aged 18-65 with an MDD diagnosis between 2000 and 2024 and at least six weeks of antidepressant treatment were included. Prescribing data were organised into treatment episodes (treatment regimens ≥6 weeks) and treatment blocks (sequences of episodes <42 days apart within an MDD period). TRD was defined as ≥2 antidepressant switches within a block, expanded to include ketamine or electroconvulsive therapy, or clinical note documentation of TRD identified via natural language processing. Healthcare resource use (HCRU), including inpatient bed-days and primary/secondary care appointments, was measured per year of follow-up.
RESULTS: Among 31,237 eligible patients, 9.5% (n=2,954) met TRD criteria based on antidepressant switching, rising to 11.0% (n=3,441) when additional indicators were included. TRD prevalence using only PC data was 8.9% (2,787/31,237) and 1.3% (400/31,237) using SC data. Patients with TRD had higher annual HCRU than non-TRD patients, with 2.8 vs 1.7 inpatient bed-days, 9.4 vs 6.0 SC appointments, and 9.7 vs 5.1 PC appointments per year.
CONCLUSIONS: Linked data provided a more complete view of treatment pathways and outcomes, improving TRD prevalence and HCRU estimates compared with single-source data. These findings highlight the importance of cross-sector data linkage for accurately identifying TRD populations and informing research and service planning.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
EPH123
Topic
Economic Evaluation, Epidemiology & Public Health
Disease
Mental Health (including addiction), No Additional Disease & Conditions/Specialized Treatment Areas