Evaluation of case definitions to detect Bordetella pertussis infections in hospitalised individuals in South Africa, 2015-2017

Abstract

Background

Despite routine vaccination, pertussis remains a significant public health concern. Robust data on global pertussis burden is lacking. Pertussis surveillance relies on clinical suspicion in cases with classic symptoms as per World Health Organization (WHO) or Centers for Disease Control (CDC) case definitions. Atypical presentation may result in underestimation of disease burden. We evaluated performance of different case definitions to detect Bordetella pertussis (B. pertussis) illness among children enrolled as part of syndromic, sentinel pneumonia surveillance programme (PSP) in South Africa.

Objectives

To describe the demographic characteristics of all enrolled participants at selected PSP from January 2015 to December 2017. To identify demographic/ clinical predictors of a positive B. pertussis polymerase chain reaction (PCR) and to evaluate the sensitivity, specificity, positive (PPV) - and negative predictive value (NPV) of selected individual symptoms and signs and existing pertussis and WHO SARI case definitions for detecting PCR-confirmed B. pertussis in patients <5 and ≥5 years of age enrolled in the PSP during this time period.

Methods

This cross-sectional study used secondary data collected from hospitalised patients of all ages enrolled in the PSP at 7 sentinel sites in 5 provinces, January 2015 through December 2017. Patients were systematically enrolled based on broad case definitions (including suspected sepsis [infants aged 2 days - <3 months], physician diagnosed LRTI [<5 years] and acute respiratory infection with fever and cough [≥5 years]). Patients with suspected pulmonary TB and children ≤10 years with suspected pertussis were included at 2 and 1 site, respectively. Enrolled patients were tested for B. pertussis and other respiratory pathogens. STATA 15 was used to calculate sensitivity, specificity, PPV and NPV for classic pertussis symptoms and signs and standard pertussis and WHO Severe Acute Respiratory Illness (SARI) (acute respiratory illness with fever, cough and ≤10 days onset) case definitions in the <5 year and ≥5 year age groups, using B. pertussis PCR as the gold standard. Binomial logistic regression investigated demographic/ clinical features predictive of a positive pertussis laboratory result.

Results

We identified B. pertussis in 2.5% (188/7504) and 1.1% (54/4866) patients < and ≥ 5 years of age enrolled in PSP respectively. In the < 5 year group, classic symptoms were uncommon (chronic cough 13% (25/187); paroxysmal cough 34% (57/170); whoop 10% (17/170); post-tussive emesis 26% (44/170)) but highly specific (88%-99%) for B. pertussis. Forty-five percent of children and 57% of the ≥5 year group had fever. Multivariate analysis of the <5 year age group, revealed age, whoop, paroxysmal cough, post-tussive emesis and cyanosis as significant predictors of a positive pertussis laboratory result. Due to small case numbers, results for the ≥ 5-year-old group should be treated with reserve. However, when controlling for age and race, whoop and HIV infection were significantly associated with a positive pertussis test on multivariate logistic regression. The 2003 WHO pertussis case definitions had low sensitivity (8-14%), high specificity (>97%); and low PPV (3-17%); while WHO SARI case definition had low sensitivity (35-39%) and specificity (51-65%) with low PPV (<2%) for B. pertussis across age groups. The updated 2018 WHO pertussis case definitions performed well in infants <6 months; sensitivity increased to 47.5% (95%CI38.4-56.8), while maintaining high specificity (81.2%, [95% CI 79.9-82.5]), PPV remains low (7.7% [95%CI 5.9-9.8]). Sensitivity decreases while specificity and PPV increase with increasing number of case-defined symptoms. Thirty-eight percent (65/170) of pertussis-positive and 14.2% (977/6874) pertussis-negative children <5 years met the 2018 WHO pertussis case definition (cough and at least one symptom). In the ≥5 year group, 3/40 (7.5%) and 2.4% (104/4276) pertussis-positive and negative individuals respectively fulfilled the criteria.

Conclusion

WHO/CDC pertussis case definitions could miss 50-90% of pertussis cases underestimating disease burden, while WHO SARI case definition would systematically exclude afebrile patients. More sensitive case definitions and heightened awareness of atypical presentation are needed for diagnosis of pertussis in children. The amended WHO (2018) pertussis case definitions improved sensitivity while maintaining specificity in the <6 months age group. Existing syndromic surveillance platforms could provide for pertussis surveillance where routine laboratory testing is unavailable, providing critical information for health policy decisions.