Incidence and diagnosis of pertussis in South African children hospitalized with lower respiratory tract infection

Background: The incidence of pertussis in children in low-and middleincome countries is poorly described. This study aimed to prospectively investigate the incidence of pertussis in South African children hospitalized with lower respiratory tract infection (LRTI). Methods: Children hospitalized with LRTI in Cape Town, South Africa were enrolled over 1 year. Clinical data were collected. A nasopharyngeal (NP) swab and induced sputum (IS) were taken, and polymerase chain reaction specific for Bordetella pertussis (IS481+/hIS1001-) and Bordetella parapertussis (IS1001+) was performed. Results: A total of 460 children with median age 8 [interquartile range (IQR), 4-18] months were studied. B. pertussis was detected in 17 (3.7%) while total Bordetella spp. were identified on 23 (5.0 %) of 460 NP. Adding IS testing increased the identification of B. pertussis to 32 of 460 cases (7.0%; 95% confidence interval, 4.8%-9.7%); P = 0.028 and total Bordetella to 41 of 460 (8.9%; 95% confidence interval, 4-10%); P = 0.020. Shorter duration of symptoms [median 2 (IQR, 2-3) days versus 5 (IQR, 3-7) days; P = 0.0008] was associated with detection of B. pertussis on IS versus NP. B. pertussis was detected in 15.8% (n=3/19) of HIV-infected children, 10.9% (n = 10/92) of HIV exposed uninfected and 5.4% (n = 19/349) of HIV-unexposed uninfected children. Risk of B. pertussis decreased with each additional dose of diphtheria, tetanus and acellular pertussis vaccine [0 doses = 17.9%; 1 dose = 7.0%; 2 doses = 6.9%; and >3 doses = 6.2%]. Conclusions: Pertussis is common in South African children hospitalized with LRTI particularly if HIV exposed or infected but decreases sequentially with vaccination doses. Polymerase chain reaction on IS specimen provides confirmation earlier than NP while increasing overall diagnostic yield.