results have been seen in tests designed to detect neonatal sepsis.) Soluble antigens may be distributed systemically from carriage strains, allowing absorption of polysaccharides detectable in serum and urine, when in fact the underlying infection is viral.21 Accuracy may also vary when a diagnostic kit tested in a controlled setting, such as a hospital, is evaluated in a community setting where pneumonia incidence is relatively low but carriage is high. Finding a way to preserve and share epidemiologically and bacteriologically characterized specimens would aid efforts to evaluate newly developed diagnostics.
Although saddled with less than perfect diagnostic tools, researchers nevertheless are evaluating in humans vaccines against a number of ARI pathogens, including S. pneumoniae and Hib. Unfortunately, the classic double-blind randomized placebo-controlled efficacy trial is not ideally suited to vaccine studies. For one thing, it may sometimes be difficult to include a placebo group in such a way that is acceptable to the participants. And while recordkeeping in controlled clinical trials is often problematic no matter what the setting, it frequently poses a much greater challenge in the developing world. Elaborate, large-scale clinical studies are also expensive and time consuming.
While randomized clinical trials are central to evaluating vaccine efficacy, some information about vaccine effectiveness may be obtained via hospital-based surveillance of pneumonia in already-vaccinated populations. Using this approach, all children admitted to a hospital with severe pneumonia or other severe disease would have blood cultures taken. The type and number of pathogens identified during this process would be compared with their expected distributions, based on previous blood-culture studies. By looking specifically at the incidence of infection with the vaccine-type organism, one could begin to get a picture of vaccine efficacy, even in less than ideal circumstances (Box 4). A number of useful comparisons might be made, including between pre- and postvaccination disease rates. Because it does not require vaccination records to provide useful information, surveillance may prove more useful in the developing-world setting than the case-control paradigm.
A recent postlicensure Hib trial in Chile illustrates another way developing or transitional countries can obtain useful information about vaccine effectiveness without investing in traditional double-blind, placebo-controlled studies. After epidemiologic research implicated Hib as the cause of meningitis outbreaks, Chilean health officials began to consider adding the vaccine to the group of vaccines administered under the Expanded Programme on Immunization (EPI). Before making a long-term commitment to purchase the conjugate, however, they wanted proof that the vaccine would reduce disease incidence. Since Hib had already been licensed, it was the public health impact (effectiveness) rather than the biological activity (efficacy) that was at issue.