RÉSUMÉ
A study sample of 7,010 episodes of invasive Streptococcus pneumoniae disease was obtained by combining 13 existing datasets. Disease episodes due to each of 12 pneumococcal serogroups (1, 3-9, 14, 18, 19, and 23) were then compared with episodes in a constant internal control group to describe serogroup-specific variations in disease frequency by age, sex, and geographic origin. The results are presented as odds ratios (with 95% confidence intervals) derived by logistic regression, with adjustment for the major confounders, including dataset of origin. Variation in the male:female ratios between serogroups is small, suggesting that capsular characteristics are an unlikely explanation for the male preference of S. pneumoniae. Serogroups associated with higher nasopharyngeal prevalence (e.g., 19 and 24) are relatively more common in Europe and North American, while the invasive serotypes 1 and 5 are much more common in South America. The custom of reporting serogroup frequencies in two age groups, children and adults, conceals much of the variation in the age distributions across the whole span of life. The reduction of risk associated with serogroups 6, 14, 18, 19, and 23 beyond childhood follows different gradients, being most abrupt in serogroups 14 and most gradual in serogroup 18. The relative risk of disease with serotype 1 declines steadily throughout life, while with serotypes 3 and 8 it increases over middle age. Serogroups 7 and 23 are found unusually frequently in the third decade of life. Because of the wide differences in the epidemiology of individual serogroups of S. pneumoniae, it is questionable whether pneumococcal infection should continue to be classified as a single disease entity.