Our review comprised studies identified through Medline. The initial studies were identified by a computer search of the multilingual scientific literature published since 1990. Articles derived by using the keywords Shigella, dysentery, bacillary, and shigellosis were linked with a set of other articles obtained by using the keywords incidence, prevalence, public health, death rate, mortality, surveillance, burden, distribution, and permutations of the root word epidemiol-. We conducted searches for each Asian country, except Japan and Israel. The resulting cross-linked set contained 319 articles, which we culled to 164 articles that were relevant to the goal of the search. Additional sources were located through consultations with experts in the field, proceedings of expert meetings, and the ongoing Diarrheal Diseases Surveillance Programme of the International Centre for Diarrheal Disease Research, Bangladesh (ICDDR,B). To enable comparison over time, we adopted the methods of the previous review on shigellosis (1).

We created an algorithm to estimate the number of Shigella infections that occurred each year in Asia. In a preliminary step, the world’s population was divided into 4 age strata (0–11 months, 1–4 years, 5–14 years, and >15 years). Published rates of diarrhea for each of the 4 strata were used to estimate the diarrheal disease incidence. An increase in the severity of a patient’s illness influences the proportion of diarrheal episodes attributable to Shigella spp. This correlation can be presumed to strengthen as the proportion of Shigella infections increases because sampling progresses from cases of diarrhea detected by household surveillance to those among outpatients to persons admitted to hospital (14). Thus, we subdivided the total diarrheal disease incidence into these 3 settings: estimates of mild cases in persons who stayed at home; more severe cases needing care at a clinic but not hospitalization; and cases requiring hospitalization.

The total number of diarrhea cases attributable to shigellosis was calculated for the <1-year and 1–4-year age groups by multiplying the number of diarrhea cases in the 2 settings (community and treatment facilities) by the percentage of diarrhea cases from which Shigella spp. were isolated (Table 1). For older children and adults, we calculated total cases by multiplying the median percentages of diarrhea cases attributable to shigellosis in persons 5–14 years and >15 years of age by the number of diarrhea cases in these 2 age groups according to clinical setting (Table 2).

We adopted the estimates of Kosek et al. (15) to calculate the number of diarrhea episodes per person per year within countries in Asia (Table 1). These estimates were based on the review of 13 longitudinal studies of stable populations in 8 countries in Asia, where active surveillance was conducted during 1984–1995.

We estimated the proportion of diarrheal episodes in each stratum that can be attributed to shigellosis by analyzing only studies in which surveillance was conducted since 1990 and that used microbiologic confirmations to report the percentage of Shigella spp.–related diarrhea cases for the specified age group. An overall median percentage of shigellosis was then calculated for each stratum and multiplied by the total number of diarrheal cases in the stratum to derive the number of shigellosis cases in each stratum. In addition to the median, a weighted mean with 95% confidence intervals (CIs) was calculated for these analyses by using Freeman-Tukey transformed proportions. Weights used were equal to the inverse standard errors of these transformed proportions (18,19). The numbers of shigellosis cases were added to give an overall estimate of shigellosis-related illness. Case-fatality rates (CFRs) for persons hospitalized with Shigella infection at the ICDDR,B hospital were used to calculate age-specific rates of Shigella spp.–associated death. This hospital treats >100,000 diarrhea patients annually and is the same hospital used for CFRs in the earlier study.

Illness was expressed as episodes of diarrhea per person-year from which shigellae were recovered. Studies were included in the death estimates if deaths caused by Shigella spp. could be ascertained through active surveillance. The review comprised prospective and retrospective studies but not studies based on vital statistics only. Death was considered to have been caused by diarrhea only if diarrhea was listed as the primary cause.

Approximately 3,938,020,000 persons resided in Asia during 2005. This estimate included 78,533,000 infants <1 year of age and 361,252,000 children 1–4 years of age (20,21).

Our review calculated that ≈125 million cases of endemic shigellosis occur annually in Asia, of which ≈14,000 (0.011%) cases result in death. Children <5 years of age are at highest risk for Shigella spp.–related illness and death. Although this estimate suggests that shigellosis incidence is substantial and similar to the earlier estimate, the updated death estimate is 98% lower than the estimate by Kotloff et al. (1) that used data primarily from the 1980s. Assuming that the population of Asia is ≈80% of the total population of the developing countries, ≈130 million Shigella infections and ≈880,000 deaths occurred in Asia according to the earlier estimate.

With such a large difference in estimated incidences, one estimate may be more accurate than the other. Alternatively, Shigella spp.–related deaths may have decreased substantially since the 1980s, even in the absence of specific interventions against shigellosis. We believe the latter explanation best explains the large difference in estimates of deaths.

The major variable that was lower in our calculations was the CFR for hospitalized patients, especially children. In the earlier estimate, a CFR of 11% was used from the ICDDR,B hospital (24). Recent data from the same hospital indicate the rate is now ≈0.01% overall and only 0.89% for the youngest age group. A recent estimate from Africa found a CFR of <1% during an outbreak associated with S. dysenteriae, suggesting that this low CFR may not be limited to Asia (25).

The decrease in CFRs could be explained by >1 factor. Case management might have improved, strains might be less virulent, or children might be healthier when they become infected and therefore have less severe complications. Case management in the hospital is unlikely to have changed substantially, and in fact the increasing resistance of current strains to antimicrobial drugs makes case management more difficult. Case management in the home may have changed, however, because antimicrobial drugs are widely available, and families may purchase effective antimicrobial drugs, e.g., ciprofloxacin, and begin treatment earlier in the course of the illness (12). Virulence of infecting strains could be lower; infections with S. dysenteriae type 1 are unusual. During past epidemics with this serotype, however, the CFR for S. flexneri was as high as it was with S. dysenteriae type 1 (24). Thus, virulence is unlikely to explain the decrease in the number of deaths.

Improved health of children who become infected appears to best explain the decreased CFR. Nutritional status of children in Bangladesh has continued to improve slowly (26). Perhaps more essential is the high proportion receiving measles vaccine and vitamin A (27). Anecdotally, in children dying of shigellosis during the 1980s, postmeasles dysentery was often diagnosed, and measles increased the severity of diarrhea, including shigellosis (9,10). Measles with dysentery is rarely seen now in Bangladesh.

Our review has some limitations. Although we reviewed all available published data on shigellosis in Asia since 1990, few sites conduct active surveillance for this infection, and only one estimates CFRs. With this large population, extrapolating accurately to the entire continent might not be possible. Nevertheless, the same methods were used in this and the earlier review. The large multicenter study on shigellosis in Asia did not record any deaths, suggesting that fatalities from shigellosis are not common (22).

Second, the review included only Asia, and the situation in Africa is possibly (even likely) different (28). The higher rates of HIV infection and AIDS and malaria, different nutritional deficiencies, different rates of measles vaccination, and different health systems and civil disturbance might suggest higher Shigella spp.–related deaths in Africa. Unfortunately, until recently, no long-term surveillance for diarrhea has existed in Africa on which to base estimates.

Third, the data in the review were based on microbiologic diagnosis of Shigella infections. Although isolation of Shigella spp. from fecal samples is the most specific diagnostic test for shigellosis, the culture method has limited sensitivity because of the relatively fastidious nature of the organism. Adoption of improved specimen transport methods and newer and more sensitive molecular laboratory diagnostic methods (e.g., PCR) reportedly having high sensitivities may detect more infections (29) but is unlikely to alter the death estimates.

The remarkable 98% decrease in deaths from shigellosis in the absence of a Shigella spp.–specific intervention suggests that other nonspecific interventions have helped to lower Shigella spp.–specific deaths. These, we believe, include measles vaccine, vitamin A supplements, and overall improvement in nutrition. Although the ready availability of antimicrobial drugs encourages the development of antibiotic drug resistance because of frequent abuse, we cannot rule out the possibility that the rapid availability of these antibiotics (especially fluoroquinolones) also have benefited children with dysentery who may be receiving treatment more quickly than they previously did.

The findings from our review may provide lessons regarding other infectious diseases. Approximately 50% of deaths among children <5 years of age have malnutrition as an underlying cause (30). Also, malnutrition and infection are clearly related, with one leading to the other. By reducing rates of other common infections, e.g., measles, and improving the nutritional status, including micronutrient nutrition, of children, diseases from other infections, such as Shigella spp., may decrease.

The 4 species and numerous serotypes of Shigella spp. are a challenge for vaccine developers, but shigellosis remains high, and increasing resistance to antibiotic drugs continues to make treatment difficult. An effective Shigella spp. vaccine may have substantial benefits, but our study suggests that Shigella spp.–related deaths can be, and have been, substantially reduced with currently available interventions and that such interventions do not need to be Shigella spp. specific.