Pneumococcal serotypes and virulence.

To the Editor—Pneumococcus is a venerable microorganism with an amazingly rich history. Those of us who study this organism should constantly examine our present findings in light of the wealth of previous literature. I am concerned when investigators draw conclusions from their own excellent but potentially limited data without referring to or noting discrepancies with that earlier body of knowledge. A case in point is the recent article by Sandgren et al. [1], on the virulence of pneumococci. The authors state that type 1 pneumococcal infection in humans is associated with a low mortality. Of the 4 references that they cite to support this statement [2–5], only 1 [4] actually provides data, and it bases this conclusion on a study of 12 isolates, the source of which was not stated. In contrast, Heffron [6] cites 11 studies from the preantibiotic era in which the overall mortality due to type 1 lobar pneumonia was 58% (range, 14%–82%) in 351 bacteremic cases and 15% (range, 3%– 23%) in 457 cases of nonbacteremic disease. The variability from one report to the next is interesting, because it shows the importance of citing broadly rather than narrowly, but one can hardly conclude that type 1 pneumococcus is not highly virulent in humans. Similarly, in the antibiotic era, Austrian and Gold [7] found a mortality of only 8% in bacteremic type 1 pneumonia, but Calder et al. [8] recorded 2 deaths in 7 cases, and Mufson et al. [9] noted 6 deaths in 21 cases. The characterization of these cases as being type 1 came about because this was the first strain that the Klemperers systematically studied; it could have been an accident but was not. In the recent epidemic of type 1 pneumococcal meningitis [10], the mortality was 44%. Nearly all the references cited in a recent article by Sandgren et al. [1] were from the preceding decade, most from the preceding 5 years, and discrepancies between their findings and those reported in the earlier literature were simply not addressed. I am not proposing that we weigh down our articles with early references, but I do think that it is important for investigators to maintain a close familiarity with older literature and to either cite similarities or try to explain or resolve discordant results.