Validating pertussis data measures using electronic medical record data in Ontario, Canada 1986-2016

Area of Science:

• Medical Informatics
• Epidemiology
• Public Health

Background:

• Pertussis (whooping cough) is a reportable infectious disease.
• Ascertainment bias in reporting limits the accuracy of pertussis data.
• Existing data measures need validation against a comprehensive standard considering case severity.

Purpose of the Study:

• To validate pertussis data measures using a reference standard.
• To explore the impact of case severity on the accuracy and detection of pertussis.
• To assess the performance of various pertussis detection algorithms in electronic medical records.

Main Methods:

• Evaluated 25 pertussis detection algorithms in a primary care EMR database (1986-2016).
• Estimated sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
• Conducted sensitivity analyses and evaluated reasons for missed detections.

Main Results:

• The best algorithm achieved 20.6% sensitivity; reclassifying symptom-only cases improved sensitivity to 100% but lowered PPV.
• Age at first episode was associated with detection in 50% of scenarios.
• False negatives frequently had prior immunization history.

Conclusions:

• Improving pertussis detection requires multiple data sources and careful case definition.
• A trade-off exists between PPV and sensitivity in pertussis detection algorithms.
• Enhanced EMR data utilization and improved identification of vaccinated individuals are key to reducing ascertainment bias.