Reliability of COVID-19 case definitions in administrative and clinical databases (Q21-04)

Reliability of COVID-19 case definitions in administrative and clinical databases (Q21-04)

Reliability of COVID-19 case definitions in administrative and clinical databases (Q21-04)

Q21-04

Overview

What it the issue?
  • Coronavirus disease 2019 (COVID-19) has had a substantial effect on the Canadian population leading to interest in investigating disease trends, risk factors and treatment effectiveness, and in understanding variations in patient attributes associated with COVID-19 infection and outcomes across patient populations.
  • There is little information about the accuracy of Canadian administrative health data for COVID-19 case ascertainment.
What was the aim of the study?
  • Our main objective was to assess the validity (i.e., accuracy) of COVID-19 diagnosis coding in hospital discharge abstracts, emergency department (ED) records and outpatient healthcare provider service claims in three Canadian provinces.
How was the study conducted?
  • Population-based inpatient, ED, and outpatient service records were linked to results of SARS-CoV-2 polymerase chain reaction (PCR) tests, which are used to diagnose COVID-19. The study data were from British Columbia, Manitoba, and Ontario for April 1, 2020 to March 31, 2021.
  • Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of COVID-19 diagnosis codes (ICD-10 U07.1) and fee codes were estimated for each quarter in the study period, overall and by province, age group, and sex.
What did they study find?
  • We studied over 19 million Canadian residents and more than 13 million SARS-CoV-2 PCR tests.
  • Validity estimates varied across the three study provinces, but were generally highest for inpatient hospital data.
  • Specificity and NPV were consistently high (i.e., most estimates were > 95%).
  • Hospital discharge abstracts:
    • Overall sensitivity was 86% in the first three months (Apr to Jun 2020), but decreased to 66% in the last three months (Jan to Mar 2021).
    • Overall PPV was relatively stable, from 50% and 66% across quarters.
  • ED records:
    • Overall sensitivity was 60% (first three months) and 48% (last three months).
    • Overall PPV was 77% (first three months) and 68% (last three months).
  • Outpatient service claims:
    • Overall sensitivity was 20% (first three months) and 25% (last three months).
    • Overall PPV was 7% (first three months) and increased to 29% (last three months).
Implications
  • This multi-province validation study supports the use of inpatient and ED records as alternatives to population- based laboratory tests for identification of COVID-19 cases during the first year of the pandemic.
  • Our findings do not support the use of outpatient service claims to identify people infected with COVID-19.
Key message
  • This study provides new information regarding the validity of COVID-19 diagnoses captured in Canadian inpatient hospital records, ED records and outpatient healthcare provider service claims during the first year of the pandemic.

Manuscripts

Lix LM, Renoux C, Moriello C, Choi KL, Dormuth CR, Fisher A, Dahl M, Wu F, Asaf A, Paterson JM, for the Canadian Network for Observational Drug Effect Studies (CNODES) Investigators. Validity of diagnoses of SARS-CoV-2 infection in Canadian administrative health data: a multiprovince, population-based cohort study. CMAJ Open. 2023;11(5):E790-E798.

Presentations

Project Team

Project Co-Lead
Michael Paterson MSc
Content Expert
Christel Renoux PhD
Research Assistant
Carolina Moriello MSc
Site Investigator
Michael Paterson MSc
Ontario
Site Investigator
Colin R. Dormuth ScD
British Columbia
Site Investigator
Matthew Dahl BSc
Manitoba
Analyst
Fangyun Wu MSc
Ontario
Analyst
Anat Fisher MD, PhD
British Columbia
Analyst
Matthew Dahl BSc
Manitoba