Team Members: Bui, A., de los Reyes, F., Foster, J., Gentry, J., Limayem, A., Lopez, J., Stott, R.
Disease Outbreak in Natural Disaster Zones
With the increasing awareness toward worldwide food or waterborne diseases and concern about overpopulation along with potential natural disasters and outbreaks, considerable efforts have been devoted by epidemiologists and researchers toward determining new solutions and possible pragmatic remedies to address a rapid intervention for people affected from epidemic disease in an attempt to mitigate risk to a tolerable level and ensure public health safety all over the globe (World Health Organization, 2010). Cholera pandemic caused by Vibrio cholerae microorganism has always been associated to natural disasters including primarily earthquakes and famines occurring in a poor environmental sanitation conditions and aggravated by refugees’ movements (Albert et al., 1993). From the year 1817 till date there have been seventh pandemics of cholera ravaging primarily developing countries including mainly Indian subcontinent and river delta of Bengal. In Haiti, the O1 strain, biotype El Tor and serotype Ogawa was responsible of 252, 640 hospitalizations and approximately 6000 death cases (Promed, 2011).
Although there are approximately 206 serogroups of V. cholerae, only O1 and O139 serogroups carry a set of virulence genes including primarily cholera toxin (CT or Ctx) responsible of the massive watery diarrhea characteristic of cholera infection (Nair, 1994). Serogroup O1 includes two biotypes, classical and El Tor. While these two biotypes are both associated to cholera disease, biotype El Tor is considered to be the most predominant in the seventh worldwide pandemics. In 1998 an outbreak of cholera in India and Bangladesh which subsequently spread into several parts of the subcontinent was caused by an O139 Bengal (novel non-O1 strain). However, several pieces of evidence suggested that strain O139 Bengal closely resembles biotype El Tor of the serogroup O1 (Chatterjee et al., 1998).
In January 2010, 10 months after the earthquake in Haiti, an outbreak of cholera occurred despite there having been no cholera outbreaks reported in Haiti for over a century (Piarroux et al 2011). Twelve months later, around 92,000 cases of cholera had been reported in Haiti from all 10 departments and the city of Port-au-Prince. This included around 43,000 hospitalizations and 2,000 deaths. The case fatality was 2.3% overall and 3.3% among persons hospitalized. In July 2011, the latest reports indicated that the outbreak had increased to 360,000 cholera cases with 5,500 fatalities representing an attack rate of around 3% for a total population of 10,032,619 (World Bank 2011).
Although substantial efforts has been deployed to subside the devastating microbial pandemic caused by serogroup O1 and O139 cholera strains, natural disasters and environmental conditions due to poor sanitation and the rapid growth of V. cholerae remain the potential issues to achieve a rapid eradication of the disease. Given to these mentioned barriers and variabilities, microbial risk assessment pragmatic approach has been implemented in an effort to evaluate the risk estimate along with the level of uncertainty and variability to ensure public health safety greater control.
This research study encompasses what is currently known about cholera disease pandemic and its various consequences. Specific focus is directed toward investigating a current case study associated to Haiti earthquake via the most up-to-date scientific-based systematic approach, quantitative microbial risk assessment QMRA. QMRA preventive model is addressed to provide the risk estimate based on response-dose and exposure assessment through statistical tools. Intervention scenarios and remedies are also suggested to add an effective point-of-care approach to ensure worldwide public health safety.
Hazard Identification is the first step qualitative process aiming to identify pathogenic microbial hazard present in a situation which can be detrimental to human health. In light of the above information regarding the cholera outbreak in Haiti, the scope of this QMRA is to investigate the relative effects of drinking water treatment and sanitation in reducing the risk of acquiring cholera disease from contaminated water from the Aribonite river in Haiti following the outbreak. Vibrio cholerae is a gram negative bacterium, facultative anaerobic, rod-shaped, non-spore forming and motile with a polar flagellum. The major epidemics or outbreaks of cholera around the world have originated in coastal regions associated with blue-green algae as a planktonic organism in water column (Colwell & Huq, 2001). Cholera epidemics occur seasonally in endemic areas, occurring during the spring and fall months. Outbreaks of cholera in noncholera epidemic areas have been attributed to travel and shipping activities, however there is evidence that V. cholerae is always present in the aquatic environment associated with the reservoir zooplankton. Crustacean copepods, a major member of zooplankton along coastal areas are considered to be the host for V. cholera in the aquatic environment (Colwell & Huq, 2001). There are approximately 206 serogroups of V. cholerae. Only toxigenic O1 and O139 are the causative agents of cholera with a pandemic and epidemic potential. Strains other than O1 are called non-O1; are sporadic infections and do not cause epidemics (Desmarchelier, 1997). In Haiti V. cholerae, serogoup O1, biotype El Tor and serotype Ogawa caused 252, 640 hospitalizations and approximately 6000 death cases (Promed, 2011). The largest outbreak was registered in Peru with approximately 400, 000 cases and 4000 death caused by the El Tor biotype. V. cholerae enters the intestine, adhere to the small intestine and releases the cholera toxin (CT) or Ctx. This enterotoxin Ctx extracts electrolytes and water from the body. After 24 to 48 hours symptoms can occur with a profuse and painless “rice-watery” diarrhea and vomiting of clear fluid subsequent to rapid and substantial dehydration, loss of 1 gallon proteins and electrolyte imbalance. Optimal conditions for growth include pH 6-10, temparature 18-37 OC and Nacl addition (1%). The identification of V.cholerae is well defined via biochemical, serotyping and rapid molecular tests.
Transmission of V.cholerae is via the faecal-oral route with transmission primarily via contaminated food or drinking water. Exposed populations are therefore generally those with poor sanitation in developing countries. Susceptible populations are reported to be children, blood group o and individuals with reduced gastric acidity.
Symptoms ranging from mild to severe can develop with substantial dehydration and protein loss as well as severe electrolyte imbalance accompanied with reduced urine production, low blood pressure. Untreated, infection from V. cholerae can result in severe dehydration which can rapidly lead to shock, coma and death within 2-3 hours.
The most common exposure pathway is through ingestion of drinking water and food contaminated with V. cholerae from the feces of an infected person (Figure 1). Person to person transmission is extremely rare. Therefore, the basis of this risk assessment will include the following exposure scenarios:
1. Residential: Someone living near the river drinking contaminated surface water
2. Occupational: Rice-field workers accidentally ingesting contaminated surface water
3. Residential + Occupational: Ingestion through drinking water + Accidental ingestion through occupation
As the majority of people in developing countries do not have access to conventional piped water supplies, the amount of water ingested every day was assumed to be 1 liter per person per day
The World Health Organization’s Guidelines for Drinking-Water Quality, 4th edition,
Dose-response relationships can be developed from epidemiological investigations of outbreaks and sporadic case series, human feeding trials or animal models of a particular pathogen and related (surrogate) pathogens. In this instance, human feeding trial data were available for V. cholerae and were assessed for inclusion in the Haiti QMRA on the basis of specificity and endpoint diagnosis (Table 1).
Human feeding trials have used healthy adult volunteers so the response for children is not known although they represent a particularly high susceptible subpopulation for cholera (United Nations, 2011). These studies have also administered the challenge doses with NaHCO3 to neutralise gastric acid with the consequence that lower innocula were required to induce diarrhoea (Cash et al 1974) although lower innocula were associated with diminished severity (Levine et al 1981). In addition to gastric acidity, other host factors such as blood group O may also enhance disease susceptibility although the mechanism responsible for this is not known (Glass et al 1985).
Many of the available dose response studies report the endpoint as infection for which evidence of Vibrio cholerae excretion is looked for in stool samples. Using this data, the dose response is therefore modelled on the total number of cases (infections) as a function of the dose . However, the excretion of Vibrio cholerae may be a condition of asymptomatic carriers. The endpoint of this risk assessment was considered to be gastrointestinal illness which is typical of cholera and therefore infection endpoint -based DR models were not considered for inclusion within this QMRA.
Currently, dose response models based on the illness endpoint are limited. Haas et al (1999) present information on a morbidity dose-response distribution based on the Hornick et al (1991) trial with doses administered with a pH buffer. However the Hornick et al 1971 data is based on experimental data with the classical Vibrio cholerae strain Inaba 569A. ,Recent studies have shown that the classical biotype strains are rarely isolated globally (Sack et al 2003). The causative agent for the Haiti outbreak was the El Tor so the classical biotype model was not considered to be the most appropriate model relevant to current exposure to choleragenic V.cholerae in Haiti.
Using human feeding trial data sourced from Levine et al (1981, 1988) and Black et al (1987), an illness based dose response model was developed for volunteer ingestion with Inaba N16961 strain which is a Vibrio cholerae El Tor biotype (Stott et al 2011). These doses were all administered with simultaneous ingestion of bicarbonate. A beta-Poisson distribution provided the best fit. From the comparison with Haas et al 1999, it can be seen that the Classical and El Tor curves indicate a difference in the dose response curve for each biotype. This may indicate a difference in virulence between the two biotypes or an artifact of the studies.
It should be noted that dose administration with bicarbonate results in substantial increase in infectivity and pathogenicity. Thus in the absence of bicarbonate in the exposed Haiti population, the use of this model may overestimate the probability of illness given the exposure dose since the absence of bicarbonate would result in less chance of illness. However, Levine et al (1981) reported that similar dose response curves were obtained from human volunteer studies with El Tor V.cholerae where doses were administered with acid-neutralizing solutions or with a standard meal of fish, rice, custard and skim milk. Assumptions:
- Effect of water/food matrix on infectivity similar to DR model based on buffered studies
- Similar dose response for children and adults
- Validity for combining El Tor studies for Stott et al 2011 model
- Similar susceptibility for all blood groups
Table 1. Dose response models available for Vibrio cholera and comments regarding selection for the Haiti QMRA
|Reference||Host Type/Pathogen Strain||Biotype||Route/# of Doses||Response||Best Fit Model||parameters||N50 (ID50)||Comments|
|Hornick et al.(1971)||Human / Inaba 569B||Classical||Ingestion / 7||Infection||Beta Poisson||α = 0.198||6.36E+08||Actually illness (without pH buffering )|
|Hornick et al.(1971)||Human / Inaba 569B||Classical||Ingestion||Infection||Beta Poisson||α = 0.318,||6,816||Actually illness (with pH buffering)|
|Hornick et al.(1971)||Human / Inaba 569B||Classical||Ingestion /||Infection||Beta Poisson||α = 0.109,||3.88E+07|
|Cash et al.(1974)||Human / Inaba 569B||Classical||Ingestion /||Infection||Beta Poisson||α = 0.131||2.91E+09|
|Teunis et al 1996 (based on Cash et al 1974)||Human / Inaba 569B||Classical||Ingestion/7||Infection||Beta Poisson||α = 0.508 (β=7.52E+07)||2.19E+08||Without pH buffer (Infection endpoint =V.c excretion)|
|Ingestion/3||Beta Poisson||α = 0.164(β=0.136)||9.18||WITH pH buffer|
|Haas et al 1999 (based on Hornick et al 1971)||Human / Inaba 569B||Classical||Ingestion||Infection||Beta Poisson||α = 0.25||243||WITH pH buffer|
|Illness||Beta Poisson||α = 0.49||3365||Diarrhoea (mild to severe) WITH pH buffer|
|Stott et al 2011 (Levine et al 1981, 1988, Black et al 1987)||Human / Inaba N16961||El Tor||Ingestion/7||Illness||Beta Poisson||α = 0.169||137||Diarrhoea WITH pH buffer|
Pathogen concentrations in water No data was available for the concentration of V.cholerae in surface waters in Haiti despite an extensive literature search. Therefore, to obtain estimates for river water concentration, scholarly papers were referenced and a backforecasting method based on dose response models was used to generate potential concentration values with which to generate an environmental exposure dose.
Spira et. al 1980 reported the concentration of classic and El Tor biotype V.cholerae in surface river water in rural Bangladesh in an endemic situation. From this study, it was concluded that “the spectrum of concentration is skewed greatly toward low concentrations of cholera vibrios in all water types”. Additionally, “it was clear, however, that high concentrations of V.cholerae (i.e., >104 mL) were extremely uncommon and that persons that became infected during the course of this study were unlikely to have ingested more than 105 viable organisms per day” . The results showed that a majority of the sampled water yielding V.cholerae were less than 5 CFU/mL i.e 5000/L. In another study by Kaper et. al 1995, it was reported that “while conditions will obviously vary widely from one community to another”, it was considered that the inoculum in nature would likely be within the range of 102 to 103.
Backforecasting using dose response models was also used to derive potential environmental exposure dose based on outbreak data. The mechanistic beta-Poisson dose response model used for V.cholerae is listed below in Equation (1). Dose response models for both the classic and El Tor Inaba strains were used (see Table 1).
P(d), or probability of illness, was calculated using attack data from the Ministère de la Santé Publique et de la Population of Haiti, or MSPP. MSPP data has been gathered since the beginning of the outbreak to the present day and is given for every Haitian region as: cases seen, cases hospitalized, exited cases, hospitalized (institutional) deaths, outside (community) deaths, and total deaths. Assuming that the number of “cases seen” is the number of individuals infected by V.cholerae illness (mild to severe), the probability of illness was calculated by dividing the number of cases seen in the Artibonite region by the total population of the Artibonite region (1.57 million). December 15, 2010 outbreak data from MSPP was selected to represent the highest peak in cholera outbreak in the Artibonite region to date and was chosen as a worst case scenario. The same procedure was repeated using the 3% attack data for the Haiti population (see background information). Results are shown in Table 2 Table 2. Results used to determine potential concentrations of V.cholerae in surface drinking water (river)
|Spira et. al 1980: Surface river water samples||Majority < 5 CFU/mL|
|Kaper et. al 1995: inoculum in nature||Range of 100 – 10000 CFU|
|beta-Poisson dose response model (MSPP data): 0.03% (Artibonite attack rate)||0.68 CFUa 0.003 CFUa|
|beta-Poisson dose response model (article data): 3% (Haiti attack rate)||87 CFUb 0.5 CFUb|
- a: Using Haas et al 1999 DR model for classical strain
- b: Using Stott et al 2001 DR model for El Tor Strain
On the basis of this very limited data in Table 2 and using best judgment, the concentration of V.cholerae in Haiti’s river water was considered to range from 0 to 5000 CFU/ liter
Table 3. Variables, distributions, assumptions and scenarios used in the QMRA model for risk characterization
|Concentration of V.cholerae in river water||Triangular||Min=0, mode=1, max =5000||Based on backforecasting using DR model, attack rate for Haiti and literature|
|Log normal||Mean =1, SD=0.58||WHO guidelines,||Roseberry 1992 L/pp/day|
|Removal by drinking water treatment: chlorination||Triangular||Min (highest % inactivation)=0.00001, mode=0.001, max (lowest % inactivation) =0.2||Based on 3 log removal (mode) with Aquatab and least removal (max) =20%|
|Removal by drinking water : point of use: Solar disinfection||Triangular||Min =0.0001, mode =0.01, Max = 0.2||Typical 1 log (SODIS website, assumptions and EAWAG)|
|Removal by Sanitation||Triangular||Min=0.05, mode = 0.1, max =0.1||Assumption : 95% removal at best (min), 90% removal minimum (max)|
|Occupational / recreational exposure: Volume ingested||Log Normal||Mean=0.13, SD=0.14||Dufour, 2007 L/pp/day|
|2 Effect chlorination||12%, 20%, 30%, 50%||0%||0%|
|3 Effect chlorination + SODIS||50%||50%||0%|
|4 Effect sanitation||0%||0%||17%, 30%, 40%, 50%|
|5 Combinations||12% 50% 100% 30% 30% 20% 50%||0%||17% 50% 100% 20% 50% 30% 30%|
Now that you know the nature of the risk, you can start to explore the various methods of reducing that risk. While this appears to be a more traditional form of environmental health, pairing it with the risk assessment framework you will know have an well developed idea of what parts of the infection chain would be most appropriate to target. This would be either because a certain part contributes a considerable amount of risk or because it is easily addressed. Also part of risk management is mathematica cost and effectiveness models that guide you as to how to carry out these management strategies. More information available at the risk management home page
Albert MJ et al. (1993). Large outbreak of clinical cholera due to V. cholerae non-O1 in Bangladesh. Lancet, 341:704.
Black, R. E., Levine, M. M., Clements, M. L., Young, C. R., Svennerholm, A. M and Holmgren J (1987). Protective efficacy in humans of killed whole-vibrio oral cholera vaccine with and without the B subunit of Cholera Toxin. Infection and Immunity, 55, (5) p1116-1120
Cash, R. A., S. I. Music, J. P. Libonati, M. J. Snyder, R. P. Wenzel, and R. B. Hornick. 1974. Response of man to infection with Vibrio cholerae. I. Clinical, serologic, and bacteriologic responses to a known inoculum. J. Infect. Dis. 129:45–52.
Centers for Disease Control and Prevention (CDC). 2010. Update: outbreak of cholera – Haiti, 2010. Mor Mortal Wkly Rep. 59(48):1586-1590.
Colwell, R. & Huq, A. (2001). Marine ecosystems and cholera. Hydrobiologia, 460: 141 – 145.
Cravioto, A. et. al. 2010. Final Report of the Independent Panel of Experts on the Cholera Outbreak in Haiti. United Nations. 1 – 32.
Desmarchelier, P.M. 1997. Pathogenic Vibrios. In A.D. Hocking, G. Arnold, I. Jenson, K. Newton and P. Sutherland, eds. Foodborne Microorganisms of Public Health Significance 5th Edition, p 285 -312. North Sydney, Australian Institute of Food Science and Technology Inc.
Haas, C N., Rose, J. B. And Gerba, C. P (1999). Quantiative microbial risk assessment. John Wiley and Sons, ISBN 0-471-18397-0
Hornick, R.B., Music S.I., Wenzel R., Cash R.A., Libonati J.P., Snyder, M.J., Woodward, T.E. (1971) The Broad Street Pump Revisited: Response of Volunteers to Ingested Cholera Vibrios. Bulletin New York Academy of Medicine 47(10): 1181-1191.
Glass, R. I., J. Holmgren, C. E. Haley, M. R. Khan, A. M. Svennerholm, B. J. Stoll, K. M. Belayet Hossain, R. E. Black, M. Yunus, and D. Barua.1985. Predisposition for cholera of individuals with O blood group. Possible evolutionary significance. Am. J. Epidemiol. 121:791–796.
Kaper, J. B., Morris, J. G. Jr., and Levine, M. M. (1995). Cholera. Clinical Microbiology Reviews, 8, (1), p48-86
Levine, M. M., R. E. Black, M. L. Clements, D. R. Nalin, L. Cisneros, and R. A. Finkelstein. 1981. Volunteer studies in development of vaccines against cholera and enterotoxigenic Escherichia coli: a review, p. 443–459. In T. Holme, J. Holmgren, M. H. Merson, and R. Mollby (ed.), Acute enteric infections in children. New prospects for treatment and prevention. Elsevier/North-Holland Biomedical Press, Amsterdam.
Levine, M. M., Kaper, J. B., Herrington, D., Losonsky, G., Morris, J. G., Clements, M. L., Black, R. E., Tall, B. and Hall, R. (1988). Volunteer studies of deletion mutants of Vibrio cholerae O1 prepared by recombinant techniques.Infect. Immun. 56:161–167.
Manning, P. A., U. H. Stroeher, and R. Morona. 1994. Molecular basis for O-antigen biosynthesis in Vibrio cholerae O1: Ogawa-Inaba switching, p.77–94. In I. K. Wachsmuth, P. Blake, and Ø. Olsvik (ed.), Vibrio cholerae and cholera: molecular to global perspectives. ASM Press, Washington,D.C. Ministère de la Santé Publique et de la Population of Haiti, MSPP Daily Case Reports (2010). <http://mspp.gouv.ht/site/index.php?option=com_content&view=article&id=57&itemid=1> .
Nair GB et al. (1994). Laboratory diagnosis of Vibrio cholerae O139 Bengal, the new pandemic strain of cholera. LabMedica International, XI: 8–11.
Piarroux, R., et al., Understanding the cholera epidemic, Haiti. Emerg Infect Dis, 2011. 17: p. 1161-8.
Promed. International Society for Infectious Diseases. Available: <http://www.promedmail.org/pls/otn/f?p=2400:1001:80787::NO::F2400 _P1001_BACK_PAGE,F2400_P1001_PUB_MAIL_ID:1000,89503ences>
Sack, R. B., and C. E. Miller. 1969. Progressive changes of vibrio serotypes in germ-free mice infected with Vibrio cholerae. J. Bacteriol. 99:688–695.
Sack, R.B., Siddique, A.K., Longini, I.M. Jr, Nizam, A., Yunus, M., Islam, M.S., Morris, J.G. Jr, Ali, A.,Huq, A., Nair, G.B., Qadri, F., Faruque, S.M., Sack, D.A. & Colwell, R.R.. 2003. A 4-year study of the epidemiology of Vibrio cholerae in four rural areas of Bangladesh. Journal of InfectiousDiseases, 187: 96–101
Sack, D. A. et.al. (2004). Cholera. Lancet, 363:223-33.
Spira W.M., Khan M.U., Saeed Y.A., Sattar M.A., Microbiological surveillance of intra-neighbourhood El Tor cholera transmission in rural Bangladesh (1980) Bulletin of World Health Organization, 58(5): 731-7409.
United Nations (2011). Final report of the independent panel of experts on the cholera outbreak in Haiti. (accessed August 23 2011 http://www.haiti.mphise.net/node/1055)
Weir, Mark. “Vibrio cholerae.” Chapter 13 Center for Microbial Risk Assessment Wikipedia. <http://wiki.camra.msu.edu/>.
World Bank (2011). World Bank Indicators, www. Google.com/publicdata (accessed 25 August 2011).
World Health Organization (WHO). 2010. About microbial risk assessment in food
Available: http://www. Who.int/foodsafety/micro/about_mra/en