Bulletin of the World Health Organization

The global burden of cholera

Mohammad Ali a, Anna Lena Lopez a, Young Ae You a, Young Eun Kim a, Binod Sah a, Brian Maskery a & John Clemens a

a. International Vaccine Institute, SNU Research Park, San 4-8 Nakseongdae-dong Gwanak-gu, Seoul, Republic of Korea 151-919.

Correspondence to Mohammad Ali (e-mail: mali@ivi.int).

(Submitted: 14 July 2011 – Revised version received: 14 November 2011 – Accepted: 21 November 2011 – Published online: 24 January 2012.)

Bulletin of the World Health Organization 2012;90:209-218A. doi: 10.2471/BLT.11.093427

Introduction

Cholera is an important public health problem worldwide. Although most cholera infections are not detected, large cholera outbreaks, such as those seen in Haiti,1 Viet Nam2 and Zimbabwe3 in recent years, can occur. Industrialized countries have seen practically no cholera cases for over a century because of their good water and sewage treatment infrastructure. However, the causative agents (Vibrio cholerae O1 and O139) continue to thrive wherever crowded housing conditions exist and water and sanitation facilities are suboptimal.

The World Health Organization (WHO) maintains a public database of cholera cases and provides outbreak updates and an annual summary of aggregate national data in the Weekly Epidemiological Record. Owing to its rapid spread, cholera was one of three diseases requiring notification to the WHO in the 1969 International Health Regulations (IHR), and today, after the 2005 revision of the IHR, cholera outbreaks still require notification. Despite this, cholera notification remains incomplete because of inadequate laboratory and epidemiological surveillance systems and economic, social and political disincentives to case reporting.47 WHO estimates that the officially reported cases represent only 5–10% of the actual number occurring annually worldwide. Of the estimated 3 to 5 million cases that occur globally every year, about 100 000 to 120 000 die.8,9 In 2002, Lanata et al. calculated, using the fraction of diarrhoea cases estimated to be caused by cholera (0.05%), that 11 million cholera cases occur globally every year among children under 5 years of age.10 However, adults and older children can also get cholera, and mortality can be high in all age groups.1113

One new generation oral cholera vaccine is pre-qualified by WHO for purchase by the United Nations (UN), and another, less expensive oral vaccine has recently been licensed and is undergoing evaluation for WHO pre-qualification.14 WHO recommends instituting cholera immunization, in conjunction with other prevention and control strategies, in endemic areas, and perhaps in areas at risk for cholera outbreaks.15 To plan interventions appropriately, policy-makers at all levels need to know the burden of disease and the population at risk.

In this paper we present age-specific estimates of the cholera burden in endemic and non-endemic countries and a global estimate. To derive our figures we used population-based incidence data and reports from WHO and supplementary sources.

Methods

Data sources and framework

To supplement the WHO cholera reports, we developed a framework based on literature searches and alternative disease monitoring systems, including the Program for Monitoring Emerging Diseases (ProMED), an online forum for infectious disease specialists, microbiologists and public health officials established in 1994 and administered by the International Society for Infectious Diseases since 1999.16 Although it is a passive reporting system, it complements the WHO reports by providing detailed data from media reports describing outbreaks. In an analysis of ProMED data, the number of cholera cases reported was found to be about five times higher than the number reported by the WHO.8,17 We used population-based cholera incidence data from a multi-country study that included Asian and African countries.1820 In addition to these sources, we reviewed data from the Global Infectious Disease and Epidemiology Network (GIDEON) database, published articles listed in PubMed, and online government reports.

Country classification

In classifying countries, we used WHO’s division into five mortality strata, as follows: A, very low child and very low adult mortality; B, low child and low adult mortality; C, low child and high adult mortality; D, high child and high adult mortality; E, high child and very high adult mortality. Most developed countries in North America, Europe and the Western Pacific, which primarily report imported cholera cases, belong to stratum A. The Region of the Americas (AMR) and the European Region (ER) have very little cholera. Of the 35 countries in the AMR, only Brazil and Paraguay have reported non-imported cases over the last nine years (5 in Brazil in 2005 and 5 in Paraguay in 2009). Of the 52 countries in the ER, only Poland, the Russian Federation and the Ukraine have reported non-imported cholera cases (one case each) since 2005. We therefore classified the AMR and the ER as cholera-free. For the 148 remaining countries in the five mortality strata, we reviewed reported cholera incidence data for 2000–2008 to determine whether cholera was: (i) endemic; (ii) observed but not endemic, or (iii) not observed.

Endemic and non-endemic countries

In classifying countries as endemic or non-endemic, we used the definitions recently applied by the WHO Strategic Advisory Group of Experts on Vaccines and Immunization (SAGE).15 Countries were classified as endemic if they had reported cholera cases in at least three of the five most recent years (either 2003–2007 or 2004–2008), as published in the Weekly Epidemiological Record. If a country had no cholera reports in the Weekly Epidemiological Record, we searched PubMed and the ProMED and GIDEON reporting systems using the terms acute watery diarrhoea (or diarrhoea), cholera (or cholerae) to identify endemic countries. Of the 148 countries, 51 were classified as cholera-endemic (Fig. 1). The reported cases, by country and year, are listed in Appendix A (available at: https://trdrdcsys.ivi.int/cic/appendixA.pdf); details on the references and methods used to obtain the reports are in Appendix B (available at: https://trdrdcsys.ivi.int/cic/appendixB.pdf).

Fig. 1. Geographical patterns of the estimated cholera incidences for entire population in endemic countries
Fig. 1. Geographical patterns of the estimated cholera incidences for entire population in endemic countries

We identified another 18 countries that did not meet the criterion for endemic cholera but that had reported cholera cases sometime between 2000 and 2008. We classified these countries as non-endemic. Of these countries, those that had reported only imported cases (e.g. developed countries in North America, Western Europe and the Pacific) or very sporadic non-imported cases (e.g. countries of Eastern Europe or South America with five cases or fewer in non-consecutive years) were considered cholera-free and not included in the calculation.

Population at risk

We used the United Nations Development Programme population data21 for 2005 to estimate the population at risk of cholera in each country. Since not all the population of a cholera-endemic country is at risk for cholera, we used instead the percentage of the population of each country lacking access to improved sanitation, according to UN data from 2008.22 We made one exception: We considered the entire population of Bangladesh at risk, on advice from the country’s cholera experts, because of frequent and widespread flooding. For the three largest cholera-endemic countries – China, India and Indonesia – we only included the fraction of the population lacking access to improved sanitation in states and provinces that had reported cholera (2000–2007 in China23 and Indonesia24 and 1997–2006 in India25).

Disease burden estimation

In estimating cholera caseloads for both endemic and non-endemic countries, we focused on cholera episodes severe enough to require medical treatment.

Endemic countries

We obtained incidence rates from the Diseases of the Most Impoverished (DOMI) cholera surveillance programmes in Kolkata, India,18 Jakarta, Indonesia,19 and Beira, Mozambique20 (Table 1), since these population-based surveillance figures included both inpatient and outpatient cases of laboratory-confirmed cholera. We applied incidence rates from these studies to the at-risk populations of cholera-endemic countries by WHO mortality stratum. To be conservative, we assumed an incidence of zero in the population not at risk (i.e. the population with access to improved sanitation in countries other than Bangladesh). We grouped countries according to WHO region and WHO mortality stratum.26 An alphabetical list of the countries we included in our calculation, by WHO stratum, is given in Appendix C (available at: https://trdrdcsys.ivi.int/cic/appendixC.pdf).

We assumed that the incidence rates observed in the DOMI studies were representative of the at-risk populations in the WHO mortality strata. This assumption was based on perceived similarities in cholera risk between the DOMI sites and the at-risk populations of the strata as a whole. For example, in the South-East Asia Region (SEAR), we applied Jakarta’s incidence rates to the at-risk population of Indonesia and other endemic countries in stratum B (SEAR-B), and we applied Kolkata’s incidence rates to India and all other cholera-endemic countries in stratum D (SEAR-D), except for Bangladesh. We also applied Kolkata’s incidence rates to countries of the Eastern Mediterranean Region (EMR) in WHO mortality stratum D (EMR-D), which includes Pakistan, Somalia and other countries with a large cholera burden. For Bangladesh we used a cholera incidence rate based on in-country expert interviews conducted for the Cholera Vaccine Investment Country Case Study: Bangladesh (B Maskery, personal communication).

In the African Region (AFR), we applied Beira’s incidence rates to the at-risk populations of cholera-endemic countries belonging to WHO’s mortality stratum E (AFR-E). In the absence of representative data for African countries in stratum D (AFR-D), we compared the mean and median estimates of cholera incidence reported in the Weekly Epidemiological Record for countries in AFR-D and AFR-E from 2000 up to 2008, since countries in the AFR report cholera cases to WHO more consistently than countries in other WHO regions. The average incidence rates reported in endemic countries in AFR-E were about 1.7 times higher than the rates reported in endemic countries in AFR-D. Thus, we assumed that incidence among the at-risk populations of countries in AFR-D would be approximately half as high as in Beira. In the absence of data for countries in the Western Pacific Region (WPR) and EMR-B, we assumed that the incidence of cholera in the at-risk populations would be low, about 0.1 cases per 1000 persons (Table 2).

Non-endemic countries

The number of cholera cases in non-endemic countries was estimated from the reported numbers of cases identified through the Weekly Epidemiological Record, ProMED, GIDEON and published articles in PubMed for the period from 2000 to 2008. The estimated number of cases is shown in Appendix B. We estimated the average number of cholera cases per year by adding all the reported cases identified in 2000–2008, careful to not repeat data from different articles for the same year, and dividing the total by nine (for the nine-year interval). Because most of the data included in this report were obtained from WHO’s Weekly Epidemiological Record, whose reporting efficiency WHO estimates as ranging from 5–10%,8 we assumed that the annual average numbers of reported cases represent approximately 10% of the cholera burden in non-endemic countries. We divided the average annual number of cases in 2000–2008 by this reporting efficiency rate to estimate the cholera burden in these countries.

Age-specific incidence rates

We estimated age-specific cholera incidence rates for each endemic country using WHO-region-specific overall incidence rates (Table 2) and the incidence age distribution from the Kolkata study (Table 1). We used the Kolkata age distribution because population-based age-specific incidence is generally unavailable by WHO mortality stratum. To account for differences between each country’s population age distribution and that of Kolkata, we normalized age-specific incidence rates such that the overall incidence for a given WHO mortality stratum matches the estimates in Table 2.

Estimation of cholera deaths

Cholera is distinctive among diarrhoeal diseases in that mortality is high among patients of all ages. Estimates of cholera mortality reported to WHO by Member States may be inaccurate. For example, during a cholera outbreak in several Kenyan villages in 2008, active, community-based case-finding showed much higher numbers of cases and deaths than reported: 37% more surviving cases and 200% more fatal cases were found, raising the estimated case-fatality rate (CFR) from 5.5% to 11.4%.27 Because in less-developed countries many deaths from cholera probably occur before patients reach the hospital for treatment, facility-based cholera mortality figures are very likely to be underestimates.

Drawing on reported CFRs from several published studies (Table 3, available at: http://www.who.int/bulletin/volumes/90/3/11-093427), we computed variance-weighted46 average cholera CFRs by WHO region. We took the sum of the product of CFR and weight and divided it by the sum of the weights (Appendix D, available at: https://trdrdcsys.ivi.int/cic/appendixD.pdf). No CFR estimates for AFR-D were available from peer-reviewed sources for the past 10 years, so instead we used the estimated CFRs for AFR-E, which we found to be about 20% lower.

Cholera CFR estimates were not available for any of the cholera-endemic countries in SEAR-B or WPR-B. We assigned these countries a CFR of 1.0% on the assumption that they had lower CFRs because their health expenditure per capita was higher than in AFR and SEAR-D countries (Table 3). Using these CFRs, we computed the expected annual number of deaths from cholera for all selected endemic and non-endemic countries except Bangladesh, for which we used a CFR of 1.5% provided by an expert (B Maskery, personal communication). Since age-specific mortality data were not available, we assumed that CFRs were the same across all age groups in each stratum.

Results

At-risk population

Table 4 shows the at-risk populations of endemic countries, aggregated by WHO mortality strata. An estimated 1.4 billion of the world’s population is at risk for cholera, and SEAR-D, which includes Bangladesh and India, has the largest populations at risk, followed in descending order by AFR-E and AFR-D. For country-specific at-risk population data, see Appendix C.

Cholera cases

An estimated 2.8 million cholera cases occur each year in endemic countries (Table 4), and the average global annual incidence rate is 2.0 cases per 1000 people at risk (range: 0.10–4.0). If the population not at risk is counted, the estimated average incidence in cholera-endemic countries drops to about 1.15 cases per 1000 population. The countries with the highest incidence rates are in Africa and southern Asia (Fig. 1). Lower incidence rates were estimated for South-East Asia. Within Africa, western countries (primarily those in stratum D) were estimated to have lower incidence rates than countries in eastern Africa (primarily those in stratum E) because cholera incidence among at-risk populations in countries in AFR-D was assumed to be about 50% of the rate reported in Beira, based on the cholera incidence rates reported to WHO and other data sources. Only 1.2% of the estimated cases occur in SEAR-B, WPR-B and EMR-B, which is not surprising since these countries belong to lower mortality strata and have better infrastructure. Incidence rates are highest among children under 5. Overall, we estimate that about half of all cholera cases occur in this age group, and that the expected annual number of cholera cases in non-endemic countries is about 87 000 (Table 5). Estimates of the expected number of cholera cases by country and by age group are given in Appendix E (available at: https://trdrdcsys.ivi.int/cic/appendixE.pdf.).

Cholera deaths

The estimated numbers of annual cholera deaths in endemic countries, by age and by WHO mortality stratum, are summarized in Table 6. Estimated numbers of country-specific cholera deaths are available in Appendix F (available at: https://trdrdcsys.ivi.int/cic/appendixF.pdf). Using population data from 2005, we estimated that cholera kills about 91 000 people annually, on average, in endemic countries, with about half of the deaths occurring in children under 5. This corresponds to a rate of 6.3 deaths per 100 000 people at risk. Mortality rates vary from 0.1 deaths per 100 000 people at risk in EMR-B and WPR-B to 15.2 deaths per 100 000 in AFR-E. Since we assumed that CFRs were the same across age groups, age-specific mortality rates, like age-specific incidence rates, were highest among young children (Fig. 2). In total, we estimated an average of about 2500 cholera deaths per year in a non-endemic country (Table 5).

Fig. 2. Age-specific mortality rates for at-risk populations by WHO stratum in cholera-endemic countries, 2005
Fig. 2. Age-specific mortality rates for at-risk populations by WHO stratum in cholera-endemic countries, 2005

Sensitivity analyses

We performed sensitivity analyses using different criteria to estimate: (i) the population at risk in endemic countries, (ii) cholera incidence rates, and (iii) CFRs. In the base-case analysis we assumed the population at risk in endemic countries to be the fraction lacking access to improved sanitation. We also conducted the analysis under a more restrictive assumption, namely, that the population at risk was the fraction lacking access to an improved water supply (smaller in most countries than the fraction lacking improved sanitation). When we applied the incidence rates from Table 1 to UN data on the population without an improved water source,47 the annual cholera burden decreased to 1.17 million cases and 41 000 deaths. We also applied the less restrictive assumption that in endemic countries (including all states and provinces of China, India and Indonesia) the entire population was at risk, whereupon the global cholera burden estimates increased to 5.1 million cases and 156 000 deaths.

When we assumed that incidence rate estimates were 50–150% of the base values for WHO mortality strata, the estimated annual number of cases and deaths ranged from 1.42 to 4.26 million and from 46 000 to 137 000, respectively. Finally, we found that CFRs generally vary between 1% and 5%, depending on access to appropriate treatment (Table 3). When these CFRs were applied across all mortality strata, the expected number of cholera deaths ranged from 28 000 to 142 000.

Discussion

Our study shows that the global burden of cholera is large, particularly in developing countries. Every year an estimated 2.8 million cases of cholera (uncertainty range: 1.2–4.3 million) and about 91 000 deaths (uncertainty range: 28 000–142 000) occur in endemic countries, and another 87 000 cases and 2500 deaths occur in non-endemic countries. The burden of cholera is greatest in Africa and southern Asia, with about 99% of the cases occurring in AFR-D, AFR-E, SEAR-D and EMR-D. About half of the cases and deaths occur in children under 5. Because most cholera cases occur in poor communities, few systematic studies of endemic cholera incidence have been conducted. In addition, since embargos from trading partners and tourism losses usually follow confirmation of cholera outbreaks,48,49 many countries are reluctant to conduct cholera surveillance or to officially report cholera cases or deaths. Finally, many cholera deaths occur before patients can reach treatment facilities, which further complicates disease burden estimates.27

Our estimates are based on a framework using available endemic cholera incidence rates, identification of endemic countries based on WHO criteria and, for Bangladesh, interviews with in-country experts. Our estimated number of cases and cholera-attributable deaths were consistent with previous WHO estimates of underreporting of 3 to 5 million cases and 100 000 to 120 000 deaths every year.9,17 These estimates are much higher than the 317 534 cases and 7543 deaths reported worldwide in 2010, suggesting that cholera is substantially underreported. Even if the 500 000 to 700 000 cases of acute watery diarrhoea reported annually in some parts of Asia are counted as cholera cases,50 the resulting numbers are still lower than our estimates. However, our estimated burden is substantially lower than the 11 million cases found among children under 5 in 2002 by Lanata et al., who used different methods and included published studies from 1990–2002.10

There are limitations to our study. First, we obtained some incidence rates from systematic surveillance over a two-year period in three research settings and applied them to cholera-endemic countries by WHO mortality stratum. Such settings may not be representative of each region. In addition, cholera incidence varies from year to year, even in endemic countries. Thus, two years of surveillance data may not be enough to estimate the average cholera burden. Second, supplemental data from WHO and other sources were also included to adjust incidence rates within the AFR; however, few countries outside this region report cholera consistently. To overcome the data limitations, we carefully evaluated every country setting and used our best judgment to attribute incidence rates by WHO mortality stratum. Third, we generalized data from Kolkata, in SEAR-D, to estimate the age distribution of cases in regions other than SEAR-D. We did this because Kolkata was the only site for which current data on the age-specific incidence of cholera in all age groups were available.

A further limitation of our study is that we treated Bangladesh, China, India and Indonesia differently from the way we treated other countries. We assumed the entire population of Bangladesh to be at risk because, unlike other cholera-endemic countries, Bangladesh has a low elevation and experiences tropical monsoons that undermine the benefits of improved sanitation. China, India and Indonesia were treated differently for three reasons: (i) they had large populations capable of greatly influencing estimates; (ii) their risk of cholera is heterogeneous, and (iii) they report data by state or province. In addition, we excluded certain parts of these countries after carefully reviewing the literature on cholera in 2003–2008. The inclusion of the entire populations of these countries could artificially increase the estimated at-risk population and cholera incidence.

We tried to be conservative in our estimates by defining the population at risk of cholera using the UN Millennium Development Goals Indicators database on country-specific access to improved sanitation and assuming zero incidence in the population not at risk. To address the impact of these limitations, we performed sensitivity analyses and calculated ranges for the total numbers of cholera cases and deaths.

On the positive side, we made transparent assumptions for generalizing data to the global population, and we estimated the global burden of cholera based on cholera incidence rates obtained from population-based surveillance studies. Our cholera burden estimates are in line with WHO’s lesser estimate. We did not include the recent major epidemic of cholera in Haiti in our study because it occurred after the time frame for our analysis and cholera had not been reported in Haiti for more than 100 years. If, as some predict, cholera becomes endemic in Haiti, then Haiti’s cholera burden should be included in future analyses.

Epidemic cholera is a disease of great local importance, as evidenced by large, recent outbreaks in Haiti and Zimbabwe. However, the burden of cholera in endemic areas, which appears to dwarf the burden in non-endemic areas, is often overlooked. At two district hospitals in rural Bangladesh, cholera was the most common cause of severe dehydration from acute watery diarrhoea among children less than 5 years of age; it was 1.8 times more common than rotavirus diarrhoea as a cause of severe dehydration.51 In another study from a hospital in Kolkata, V. cholerae O1 was the pathogen most commonly isolated from admitted patients who had diarrhoea and the one most often identified in fatal cases over a two-year period.52

Our estimate of the global burden of cholera may direct efforts to develop strategies for mitigating endemic and epidemic cholera in the face of climatic and environmental changes and slow progress towards improving water and sanitation in endemic countries. We need adequate prevention, preparedness and control measures to mitigate the impact of cholera.6 Poor case reporting is problematic because it can lead to insufficient allocation of resources to effectively deal with cholera. The data we present may help policy-makers and the WHO’s Global Task Force on Cholera Control to determine how much investment future cholera control interventions will require. They also underscore the need to improve cholera surveillance, especially among at-risk populations in endemic countries.


Acknowledgements

We are grateful to Deok Ryun Kim and to the members of the Cholera Vaccine Investment Case Technical Advisory Committee for their valuable contributions.

Funding:

Financial support was provided by the Bill and Melinda Gates Foundation through the Cholera Vaccine Initiative (CHOVI) programme, administered by the International Vaccine Institute (IVI), Seoul, Republic of Korea. Current donors providing unrestricted support to IVI include the Governments of Kuwait, Republic of Korea and Sweden.

Competing interests:

None declared.

References

Share