IJPDS International Journal of Population Data Science IJPDS 2399-4908 Swansea University 10.23889/ijpds.v9i5.2628 9:5:144 Enhancing population-level research among people who inject drugs: a validation and retrospective cohort study using health administrative data in Ontario, Canada Greenwald Zoë R 1 Feld Jordan J 1 2 Werb Dan 1 Austin Peter C 3 1 Drover Samantha 3 Fridman Daniel 3 Bayoumi Ahmed 1 3 Gomes Tara 1 3 Kendall Claire E 3,4 4 Lapointe-Shaw Lauren 1 3 2 Scheim Ayden I 5 6 Bartlett Sofia R 7 Benchimol Eric I 3 1 Bouck Zachary 5 Greenaway Christina 8 Janjua Naveed Z 7 Leece Pamela 9 Wong William WL 10 Sander Beate 1 3 9 Kwong Jeffrey C 1 3 9 University of Toronto University Health Network ICES Bruyère Research Institute Centre on Drug Policy Evaluation, St. Michael's Hospital Drexel University British Columbia Centre for Disease Control Division of Infectious Diseases, Jewish General Hospital Public Health Ontario University of Waterloo 18 09 2024 2024 9 5 2628 This work is licenced under a Creative Commons Attribution 4.0 International License. This article is available from the IJPDS website at: https://ijpds.org/article/view/2628 Objective

Health administrative data can support population-level research among people who inject drugs (PWID), however these data sources remain underutilized due to the difficulty in identifying drug use in routinely collected data. We validated case-ascertainment algorithms to identify PWID and tested their application in a hepatitis C (HCV) cohort in Ontario, Canada.

 Approach

We conducted a validation study using reference standard cohorts of PWID recruited via community-based studies and population controls linked to health administrative data in Ontario, Canada (1992-2020). Tested case-ascertainment algorithms included combinations of hospitalizations/emergency department (ED) visits for drug use/poisoning, physician visits for drug use, opioid agonist treatment (OAT), or injecting-related infections. Sensitivity and specificity were estimated for lifetime history and recent injecting (past 1-5 years). We applied a high-performing algorithm among all Ontarians with laboratory-confirmed HCV between 1999-2018, to identify a sub-cohort of PWID with HCV.

 Results

An algorithm including ≥1 hospitalization/ED visit or ≥1 physician visit for drug use or ≥1 OAT record had high accuracy for identifying IDU history (91.6% sensitivity, 94.2% specificity) and recent IDU (using 3 years lookback: 80.4% sensitivity, 99% specificity). When applied to a provincial cohort of 112,947 Ontarians diagnosed with HCV, this algorithm estimated 46% (N=52,248) had a history of IDU, of whom 52% (N=27,246) had an indication of recent IDU (within the past 3 years).

 Conclusion

The methods developed in this study can enhance the capacity of population-level health research among people who inject drugs and support applied public health interventions towards hepatitis C elimination.