Reporting of snakebite—and particularly envenoming—by health authorities is generally very poor in most developing countries. To evaluate snakebite incidence and mortality, researchers therefore rely upon systematic reviews of the medical literature. Most of the current, accessible primary studies use the basic method of retrospective compilation of hospital registers or statistics from medical services. Primary data may also be obtained from prospective surveys, which can give better information on symptoms, complications, or effectiveness of treatment, but such surveys take longer and are more expensive.
However, both types of health centre surveys—retrospective and prospective—only account for a proportion of all snakebites, since some patients fail to attend health centres. And in developing countries, most patients (60%–80%) who do arrive at health centres with snakebite do so after a considerable delay (sometimes several days after the bite) because they first attend a traditional healer. Delay in attending health centres has been well documented in Africa [3
], and to a lesser extent in Asia [9
] and Latin America [11
One may assume that some snakebite victims die before reaching the health centre in due time (leading to underestimation of snakebite mortality), and others do not go to the health centre because they were cured (leading to underestimation of morbidity). Nevertheless, complications of snakebite leading to serious sequelae (amputations or neurologic deficits) are common. Certainly, snakebite morbidity is more likely to be underestimated than mortality because death is a less frequent outcome and probably better reported than envenoming.
An alternative study methodology uses household surveys to question a representative part of the population to estimate the incidence and mortality of snakebite in the community. This technique, recently validated by prospective follow-up of populations that confirmed its reliability [13
], is a good complement to hospital surveys. However, although household snakebite surveys can be valuable and informative in helping to plan the community's need for antivenom, this method is not yet well developed.
In the new study by de Silva and colleagues, the data were obtained from a limited number of studies, which were local and scattered. The main limitation of their study, as for similar types of evaluation, is the concern about how representative it is of the actual epidemiological situation. In my own 1998 study [5
], I estimated that these local surveys were fairly representative, but de Silva and colleagues believe that they are not. They argue that my assumption was undoubtedly too optimistic. Let's hope that de Silva and colleagues' study will encourage clinicians and health authorities to report snakebite cases and deaths more accurately.