(224) Operating Characteristics of COVID-19 Diagnosis Code in Hospital Settings

Saturday, August 27, 2022

8:00 AM – 6:00 PM CEST

Location: Hall C

Publication Number: 898

Presenting Author(s)

Cristiano S. Moura
– Research Institute of the McGill University Health Centre, Montreal, Quebec, Canada

Co-Author(s)

Coraline Danieli, PhD
– Centre for Outcomes Research and Evaluation, The Research Institute of the McGill University Health Centre, Canada
Fangming Liao, MSc
– St. Michael’s Hospital, Unity Health Toronto, Canada
Surain Roberts, PhD
– St-Michael's Hospital, Unity Health Toronto, Canada
Fahad Razak, MD, MSc, FRCPC
– St. Michael’s Hospital, Unity Health Toronto, Canada
Amol A. Verma, MD, MPhil, FRCPC
– St. Michael’s Hospital, Unity Health Toronto, Canada
David L. Buckeridge, MD, PhD
– Department of Epidemiology, Biostatistics, and Occupational Health, School of Population and Global Health, McGill University, Canada
Sasha Bernatsky, MD, FRCPC, PhD
– Centre for Outcomes Research and Evaluation, The Research Institute of the McGill University Health Centre, QC, Canada

Background: The use of hospital-based data for assessing the safety and effectiveness of COVID-19 treatments depends on the reliability of diagnosis codes to correctly identify cases.

Objectives: To assess operating characteristics of the COVID-19 diagnosis code U07.1 in hospital databases.

Methods: We used hospital-based data from two provinces in Canada: the General Medicine Inpatient Initiative (GEMINI) in Ontario (ON) and the CODA-19 Databank in Quebec (QC). These databases contain data on hospital encounters (basic demographic variables, diagnosis codes, hospital procedures, medications, and lab data) from multiple hospitals in the two provinces. We studied adults (≥ 18 years old) admitted to the hospitals between January 2020 and June 2021. We identified patients with the ICD-10 code U07.1 (virus identified) and assessed if they were tested for SARS-CoV-2 RT-PCR immediately before admission or during a hospital stay. Assuming the test result as the gold standard, we calculated operating characteristics (sensitivity, specificity, positive predictive value [PPV] and negative predictive value [NPV]) of the U07.1 code. Analyses were stratified by sex, residence area (urban vs rural), calendar period, and the timing of the first test (within 48h or after 48h of hospital admission).

Results: We studied 15,018 hospitalizations in ON (57.9% female, and median age of 65 years; IQR: 52-78) and 15,170 in QC (55.1% female, median age of 63 years, IQR: 39-76). In ON, the sensitivity of code U07.1 was 93.9% (95% CI: 91.5%-95.8%), the specificity was 99.6% (95% CI: 99.5%-99.7%), the PPV was 90.4% (95% CI: 87.8%-92.5%), and the NPV was 99.8% (95% CI: 99.7%-99.8%). In QC, the sensitivity was 87.6% (95% CI: 87.1%-88.1%), the specificity was 99.4% (95% CI: 99.3%-99.5%), the PPV was 87.3% (95% CI: 86.8%-87.8%), and the NPV was 99.4% (95% CI: 99.3%-99.3%). Operating characteristics were similar in most stratified analysis but sensitivity and PPV were better when the first test was done within the 48h of the admission.

Conclusions: Sensitivity and PPV of code U07.1 were high, and our results support the use of this code to identify COVID-19 cases in hospitals using this code.