Comparing respiratory illness surveillance case definitions to detect Bordetella pertussis in children aged <5 years with respiratory illness in South Africa, 2017-2023

Abstract

Background Pertussis is vaccine-preventable and surveillance can guide interventions. Assessing the performance of syndromic surveillance and the World Health Organization (WHO) pertussis case definitions can identify improvements to enhance detection and monitoring of Bordetella pertussis. Methods We analysed respiratory illness sentinel surveillance data among children aged &lt;5 years from January 2017 through December 2023. Participants were enrolled as outpatients with influenza-like illness (ILI), or hospitalised patients with severe respiratory illness (SRI) surveillance. Nasopharyngeal swabs were tested for B. pertussis using polymerase chain reaction (PCR). Sensitivity, specificity, and performance indicators of case definitions were evaluated against PCR results. Results Of 23,642 participants with PCR results, B. pertussis was detected in 0.7% (30/4,125; median age: 6.2 months; interquartile range [IQR]: 2.1–15.8) from ILI and 1.6% (314/19,517; median age: 1.7 months; IQR: 1.2–2.4) from SRI surveillance. Compared to the WHO pertussis case definition, a modified pertussis case definition (including apnoea, omitting cough duration) improved sensitivity (ILI: 30.0% vs. 43.3%; SRI: 55.7% vs. 60.2%), but reduced specificity (ILI: 90.5% vs. 75.8%; SRI: 88.3% vs. 80.9%). Negative predictive values were high for both definitions (&gt;99%), while positive predictive values were low (&lt;15%), reflecting low prevalence. Both WHO- and modified pertussis case definitions missed a large proportion of true pertussis cases (ILI: 70.0% vs. 56.7%; SRI: 44.3% vs. 39.8%). Conclusion Both WHO and modified pertussis case definitions missed many laboratory-confirmed pertussis cases, underestimating disease burden. Revising the WHO pertussis case definition and integrating pertussis into syndromic surveillance could improve detection while leveraging existing resources.