Severe underestimation of pertussis related hospitalizations and deaths in the Netherlands: A capture-recapture analysis

Ojective Despite vaccination, pertussis has remained endemic, sometimes leading to severe disease. We aimed to quantify the completeness of reporting (CoR) of pertussis hospitalizations and deaths in the Netherlands. Study design CoR was estimated using capture-recapture analyses. Hospitalizations (2007–2014) from the National Registration Hospital Care (hospital data) were matched to the notifiable Infectious Disease case registry (notifications) providing (month and) year of birth, gender and postal code. Deaths (1996–2014) from Statistics Netherlands (death registry) were matched to notifications using gender, age, year of death and notification date. Cases <2 years (y) and ≥2y were analysed separately. Chao's estimator estimated the total population, which was used to calculate CoR. Results Using strict matching criteria, we found 461 matches among 876 (hospital data) and 757 (notifications) hospitalizations <2y. The population estimate of hospitalized infants was 1446, resulting in CoR between 52% and 61%. For hospitalizations ≥2y (246; hospital data and 264; notifications) 43 matches were found, with a population estimate of 1512 and CoR between 16.5% and 22%. Among thirteen (death registry) and eight (notifications) deaths <2y, seven cases overlapped. The population estimate was 16. CoR of the two sources was 50–81%. With two (death registry) and eight (notifications) deaths ≥2y without overlap, the population estimate was 26 and CoR 8–31%. Conclusion Results showed substantial underestimation of pertussis hospitalizations and deaths. This has to be taken into account in evaluation of current and future immunization programs.