Diagnosis of pneumococcal pneumonia in epidemiological studies: evaluation in Kenyan adults of a serotype-specific urine latex agglutination assay.

A serotype-specific tube latex agglutination assay for 10 pneumococcal serotypes was evaluated with use of urine samples from 72 Kenyan adults with pneumonia whose blood or lung aspirate cultures were positive for Streptococcus pneumoniae, 203 patients with pneumonia whose cultures were negative for S. pneumoniae, and 101 afebrile controls. Detection thresholds for purified capsular polysaccharide in normal urine ranged from 0.33 to 10 ng/mL. The sensitivity of the assay for the 10 pneumococcal serotypes was 0.57 (95% confidence interval, 0.44-0.70) and was unaffected by human immunodeficiency virus seropositivity, prior antibiotic use, and bacteremic or nonbacteremic status but varied significantly by serotype. Of the pneumococci obtained by culture, 81% were of serotypes (1, 4, 5, 6, 7, 9, 12, 14, 19, and 22) that were included in the antigen assay. Strong simultaneous agglutinations against two different serotypes were found in urine samples from two patients. The specificity of the assay was 0.98 (lower 95% confidence limit, 0.95). Subjective reading of agglutination results introduced variation in specificity that may be inapparent if not formally measured. The assay extended the diagnostic yield in pneumococcal pneumonia by a factor of 2.2 (from 54 diagnoses established by blood culture to 119 established by both methods) and may therefore prove useful in reducing the sample size of epidemiological studies of pneumococcal pneumonia in adults.