Background: The World Health Organization’s new ICD-10 code for lab-confirmed SARS-CoV-2 was adopted by Canada in Feb. 2020. Demonstrating that these codes reliably identify COVID-19 infection within administrative health data could greatly expedite research and surveillance activities.
Objectives: To assess the operating characteristics of ICD code U07.1 within administrative data.
Methods: We used administrative data from 5 provinces (access to each province’s holdings was facilitated by Canada’s Health Data Research Network). Analyses were restricted to adults (18+) with at least one SARS-CoV-2 polymerase chain reaction, PCR test (from Feb. 2020 onward). We analyzed data from those with at least one healthcare encounter captured within hospital or emergency department (ED) administrative data. Data spanned Alberta, British Columbia, Manitoba (MB), Ontario (ON), and Newfoundland/Labrador (NL). A positive PCR was the reference standard for SARS-CoV-2 detection. We estimated sensitivity, specificity and positive predictive value (PPV) with 95% confidence intervals (CI). Stratified analyses included demographics/timing of first PCR.
Results: The sensitivity of code U07.1 for PCR-confirmed SARS-CoV-2 in hospital data was consistently high and comparable across provinces, the lowest in NL (the smallest sample) at 82.8% (95% CI 76.6-87.9%) and the highest in ON (the largest sample) at 90.9% (95% CI 90.8-91.0%). PPV estimates were high across all provinces. The sensitivity of code U07.1 was lower in ED settings, and there was also more variability, ranging from 39.8% (95% CI 38.0%-41.6%) in MB to 71.8% (95% CI 71.7%-71.9%) in ON. We estimated lower sensitivity for ICD code U07.1 among hospitalized females versus males (sex at birth) in most provinces. The sensitivity of code U07.1 was better in urban versus rural areas in most provinces (for both hospital and ED data). Across most provinces, code U07.1 was more sensitive in older individuals (both in hospital and ED data). Sensitivity, specificity, and PPV were higher if PCR testing was done within a 24–48-hour period before hospital admission or ED presentation.
Conclusions: Across provinces, ICD-10 code U07.1 had high sensitivity in hospitalization data but varied for ED data (where diagnostic coding is likely assigned based on presenting symptoms rather than PCR results). Differential operating characteristics across provinces may represent regional/calendar variations in the spectrum of COVID presentations over the time period under study. Lower sensitivity in females may reflect incidental SARS-CoV-2 detection for mothers giving birth (most common reason for hospitalization in 2020-2021), remaining a hypothesis to be tested.
