A large waterborne viral hepatitis E epidemic in Kanpur, India.

A large waterborne viral hepatitis E epidemic in Kanpur, India S.R. Naik,1 R. Aggarwal,2 P.N. Salunke,3 & N.N. Mehrotra4 In 1991 the largest epidemic of viral hepatitis E yet reported occurred in Kanpur (population, 2.1 mil- lion), India. The incidence of icteric hepatitis from December 1990 to April 1991 among the inhabitants of 420 randomly sampled houses in seven of the city's 50 wards was 3.76% (138 out of 3666 individuals), i.e., an estimated 79091 persons in the city as a whole were affected. The attack rate was higher for males than females (5.3% versus 3.3%; P = 0.013) and for adults than children aged <10 years (4.26% versus 1.29%; P = 0.0006). The incidence of hepatitis was higher in those city wards that were supplied with drinking-water consisting of a mixture of river Ganges and tubewell water than in those wards supplied only with tubewell water (5.6% versus 1.2%; P = 106). In the mixed-water areas, the incidence decreased as the drinking-water source changed from only tap to both tap and handpump, to only handpump (7.8%, 6.8%, and 4.3% respectively; P = 0.023). None of the sera collected from 41 hepatitis patients during the epidemic showed evidence of hepatitis virus A or B. There were two peaks in the epidemic (in February and April 1991). The first peak was probably caused by faecal contamination of river water, indicated by water analysis data, and the second, by in- adequate chlorination of water in a reservoir. There was no evidence of secondary intrafamilial spread. Epidemics of hepatitis occur frequently in the Indian subcontinent and are mostly due to the enterically transmitted hepatitis E virus (1-3). The largest pre- viously reported epidemic of viral hepatitis occurred in Delhi in 1955-56, when an estimated 29 300 persons were affected with jaundice (4). Recently, we carried out an epidemiological investigation of a much larger epidemic of hepatitis in the city of Kanpur and report here our results. Methods Pilot survey Following reports of widespread hepatitis in Kanpur in the second week of April 1991, we conducted a pilot survey on 19-20 April in one locality of the city using a systematic sampling technique, with I Professor and Head, Department of Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Post Box 375, Lucknow-226 001, India. Requests for reprints should be sent to this author. 2 Assistant Professor, Department of Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, India. 3 Senior Resident, Department of Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Lucknow, India. 4Scientist, Central Drug Research Institute, Lucknow, India. Reprint No. 5320 municipal house numbers as the unit for randomiza- tion. In those instances where a house was locked or was non-residential, the next-nearest house was sur- veyed. A total of 54 families in a heavily affected area were visited; the incidence of icteric hepatitis in the sampled population between December 1990 and the pilot survey in late April 1991 was 10.6% (35 of 331 family members). Widespread public and medi- cal opinion in the city suggested that those areas of Kanpur that were supplied with water from the river Ganges had higher incidences of hepatitis. Since most of the previously reported epidemics of hepati- tis in India have been caused by contaminated water supplies (3), we carried out a detailed field survey to determine the relationship between the water supply and the incidence of hepatitis in Kanpur and to quan- tify the magnitude of the problem presented by hepa- titis in the city. Kanpur city and its water supply system Kanpur city in the northern state of Uttar Pradesh is a major centre of industry and is the eighth most populous city in India. According to provisional data released by the Census of India, its population on 1 March 1991 was 2 103 483 (5). The city has a linear-type development along the right bank of the river Ganges. Kanpur has two major sources of water supply: surface water from the river Ganges, to the north of the city, and the Lower Ganges Canal, to the south; Bulletin of the World Health Organization, 70 (5): 597-604 (1992) © World Health Organization 1992 597 S.R. Naik et al. and deep tubewells. Kanpur is divided into the fol- lowing districts for the purposes of water supply; the city, south, east, and west service districts (Fig. 1). Municipal water for the first two of these districts is mainly drawn from the river Ganges. Approximately 150 million litres of water are pumped daily from the intake point at Bhaironghat to the Bainaj- haber water treatment plant, where it is mixed with approximately 70 million litres of water from the Lower Ganges Canal. The water then undergoes prechlorination, sedimentation, flocculation, and filtra- tion using slow sand and rapid gravity filters. The water is subsequently chlorinated to a free chlorine content of 1.2-1.5 mg/l and held in four under- ground storage reservoirs. From these reservoirs, the water is pumped to 26 zonal pumping stations, at each of which it is mixed with water obtained from deep tubewells. Secondary chlorination is performed before the resulting mixed water (MW) is supplied to consumers. In contrast, water for the east and west supply districts is obtained solely from deep tubewells; the tubewell water (TW) is chlorinated using bleaching powder. In addition, some families have access to hand-operated tubewells (handpumps) which draw subsoil water from various depths; this water is mainly used for drinking purposes. A small number of open wells also exist in the city and serve as a source of water for a small proportion of the population. The city is divided by the local health authorities into 50 wards, each with a population of approxi- mately 50 000; 40 of these wards are supplied with mixed water, and the remaining 10 wards use exclusively tubewell water. Fig. 1. Map of Kanpur city showing the municipal wards, water supply system, and the areas included in the sample survey (hatched). Broken hatching indicates wards which receive mixed water derived from both the river Ganges and tubewells (MW areas). Areas that receive only tubewell water are shown blank (L.G.C. = Lower Ganges Canal). Sample size and sampling techniques Based on the 10.6% incidence of jaundice in our pilot survey in a heavily affected area of Kanpur, we assumed an incidence of 5% in the city and the south service districts (MW areas) and an incidence of 2.5% in the east and the west districts (TW areas). Sample size calculations based on these estimates indicated that a minimum of 1212 subjects in both the TW and MW groups were needed for the study to have a power of 0.90 (probability of type II (beta) error = 0.10) to detect this difference in incidences at an alpha level of 0.05 (6). Assuming an average of eight persons per house, we calculated that a mini- mum of 152 houses in each of the MW and TW areas would have to be sampled. In order to take into account the possibility of exclusions (locked or non- residential buildings), we included 420 houses (60 houses in each of 7 wards). A multistage sampling technique and random number tables were used to select the sample (6). Four wards from the MW areas and three from the TW areas were first chosen and in each of these wards 60 municipal house numbers were then selected. Field survey Each house was visited between 25 April and 4 May 1991 by one of the survey teams, each of which included at least one medically qualified person. Many house numbers corresponded to more than one family unit, and in these instances all the families living in the house were included in the survey. Houses that were locked were excluded without any substitution. The survey proforma collected the following data for each house number: the number of family units; the number of family members in each unit; the source of drinking-water supply for each family unit; and the number of those who had had jaundice since December 1990, together with their age, sex and time of onset of the illness. Hepatitis was defined as a yellow discoloration of the conjunc- tivae or of a typical prodrome followed by deep- coloured urine. Individuals who were still affected at the time of the visit were examined clinically to exclude other causes of jaundice and to confirm the clinical diagnosis of hepatitis. Confirmation of the diagnosis for those who had already recovered was based on examination of clinical and laboratory records, if available. Patients who had features that suggested underlying chronic liver disease or extrahepatic biliary obstruction were excluded. Water analysis data Water analysis data were obtained from the water analysis laboratory at Kanpur Jal Sansthan, which is WHO Bulletin OMS. Vol 70 1992598 Viral hepatitis E epidemic in Kanpur, India responsible for supervising and maintaining the civic water supply for Kanpur. Serological data Blood samples were obtained from 41 persons (who were not seen in the survey) with hepatitis in Kanpur city. The sera were separated and stored at -20 OC until they were analysed for HBsAg and anti-HAV IgM using commercially available micro-enzyme- linked immunosorbent assay (micro-ELISA) kits. Other data Information on the fatal cases admitted to various government hospitals was obtained from the district health authorities. Statistical methods Data were analysed using X2 tests for contingency tables and for trend, as appropriate (7). Odds ratios (OR) and their 95% confidence intervals (CI) for disease risk for various groups were also calculated (8). An alpha (a) level of 0.05 was considered signi- ficant. Results Epidemiological data The sample survey covered 657 families, consisting of 3666 members, of whom 138 persons had icteric hepatitis during the study period (overall incidence, 3.76% (Table 1)) while one individual (an 80-year- old male) died from hepatitis. Application of the overall incidence of hepatitis found for the sampled population to the entire city produced an estimate that 79 091 cases of hepatitis had occurred in Kanpur from the beginning of the epidemic until 4 May 1991. Of the 138 persons who had hepatitis in the sur- veyed population, 89 (64.5%) were males and 49 (35.5%) were females. The age distribution of these individuals is shown in Table 2. A total of 81.1% of the affected persons were aged 10-39 years, and only 8 (5.8%) were under 10 years. The latter had a much lower risk (8/6184 (1.29%)) of contracting jaundice than those in older age groups (130/3048 (4.26%)): X2 = 11.7; degrees of freedom (df) = 1; P = 0.0006; OR = 0.29, 95% CI = 0.13-0.62). The inci- dence of hepatitis among the males and females sur- veyed was 5.30% and 3.30%, respectively (X2 = 6.18; P = 0.013), and the odds ratio for males rather than females being affected was 1.73 (95% CI = 1.19-2.50). The distribution of onset of the disease had two maxima: a minor peak at the end of Febru- Table 1: Incidence of icteric hepatitis in the seven municipal wards of Kanpur city covered in the sample survey No. of Population Ward No. families surveyed No. of cases Mixed water 16 95 556 58 (10.4)a. b area (MW) 13 96 515 36 (7.0)b 18 110 622 18 (2.9)b 34 72 442 7 (1.6)b Total 373 2135 119 (5.6)C Tubewell 47 40 243 6 (2.5)d water 49 182 986 10 (1.)d area (TW) 1 62 302 3 (1.0)d Total 284 1531 19 (1.2)c Overall total 657 3666 138 (3.76) a Figures in parentheses are percentages. b Difference within various MW areas: X2 = 48.75; degrees of freedom (df) = 3; P < 0.000001. c Difference between the incidences in MW and TW areas: X2 45.01; df = 1; P < 0.000001. d Difference within various TW areas: X2 = 3.56; df = 2; P>0.05. Table 2: Age and sex distributions of hepatitis cases in the sample survey Age group (years) 0-9 10-19 20-29 30-39 40-49 50-59 260 No. of persons Male Female Total 6 2 8 29 19 48 31 15 46 10 8 18 7 2 9 5 2 7 1 1 2 (5.8) a (34.8) (33.3) (13.0) (6.5) (5.1) (1.5) Total 89 49 138 a Figures in parentheses are percentages. ary and a major and sustained peak in the latter half of March and in April 1991 (Fig. 2). Incidence of hepatitis and the water supply The incidence of hepatitis in the MW areas varied from 1.6% to 10.4% (mean, 5.6%), while that in the TW areas varied from 1.0% to 2.5% (mean, 1.2%). The incidence in the MW areas was significantly higher (X2 = 45.01, P <10-6). The odds ratio of having hepatitis for a person who lived in an MW area rather than in a TW area was 4.70 (95% CI = 2.82-7.91). The incidence of icteric hepatitis in different MW areas varied significantly (X2 = 48.75, df = 3; P <10-6). In contrast, the incidences in the three WHO Bulletin OMS. Vol 70 1992 599 S.R. Naik et al. Fig. 2. Distribution of the frequency of onset of hepati- tis in Kanpur city, December 1990 to April 1991. No. of cases 10 5 0 TW areas were not significantly different from each other (X2 = 3.56, df = 2; P = 0.17). Within the MW areas the following sources of drinking-water were available for families: munici- pal taps; municipal taps and a handpump; only a handpump; and wells (Table 3). Only four families (24 individuals) used well-water for drinking pur- poses and were therefore excluded from further analysis. The other three groups showed a signifi- cant increase in the incidence of hepatitis as the water source changed from a handpump (4.3%) to both a handpump and municipal tap-water (5.8%) to purely municipal tap-water (6.8%) (X2 for trend = 5.19; df = 1; P = 0.023). This implicates the muni- cipal tap-water, derived from the river Ganges, in causing the hepatitis. Water analysis data The records of the water analysis laboratory indica- ted that the concentration of chloride and nitrite ions Table 3: Relationship between the drinking-water source and the incidence of hepatitis in mixed water areas, Kanpur, 1991 Water source Municipal tapa Tap + handpumpa Handpumpa Wellc and the chemical oxygen demand of the water at the intake point from the river increased considerably beginning in June 1990 (Fig. 3) compared with the levels over the previous 5-year period. These data are consistent with recent faecal contamination of the water prior to the point of intake from the river. The records for water chlorination and coliform counts (tested every 7-19 days) at the city's water treatment plant (Fig. 4) correlated with the incidence of hepatitis in various areas. For example, water from reservoir No. 4 had high coliform counts (35 per dl) and traces of free chlorine on 16 February 1991; this reservoir supplied water to wards 16 and 13, which had higher incidences of hepatitis (10.4% and 7% respectively), while wards 18 and 34, with incidences of 1.6% and 2.9%, respectively, received water from the other reservoirs, whose coliform counts were in the permissible range but whose free chlorine content was at times below the normal Fig. 3. Plots showing the monthly river raw water chemical analysis data for chloride, nitrite, and chemi- cal oxygen demand from January 1988 to March 1991. 2 o.X o.a o.a o.a o.a No. of No. of persons No. of persons families at risk affected 162 50 157 976 295 840 66 (6.8) b 17 (5.8) 36 (4.3) 4 24 0 Jan Apr Jul Oct Jan Apr Jul Oct Jan Apr Jul Oct Jan 1988 1989 1990 1991 x 0m0, WHO Bulletin OMS. Vol 70 1992 0 0 Nitrite levels (mg/I) 20_------------------- 10-----------_ 12 -------_---------- -1l H--------__-_---- 4 - Chemical oxygen demand (mg/l)A n~~~~~2 1------------------------_ a %2 test for trend using 0, 1, and 2 as the respective weights for the three water supplies; x2 = 5.19; degrees of freedom = 1; P = 0.023. b Figures in parentheses are percentages. c Data were excluded from the analysis because the numbers were too small. 2.5 sn . _1,1,11 ........................---------------------..................................... A I3 1 600 Viral hepatitis E epidemic in Kanpur, India Fig. 4. Plots showing the residual free chlorine content and coliform counts in four treated water reservoirs from January to March 1991. 8 40 4 Reseoir 1 Reservoir 3 30.---- 3 10- -- -- -2---1 2040 -. 0 Reservoir 4 /! \ I 30 20 10 Io Jan Feb Feb Mar Mar Jan Feb Feb Mar Mar 17 8 20 9 20 28 18 27 11 30 Coliform count Chlorine content WHO 92375 range. The incidence of hepatitis in areas that received water from reservoir No. 4 was signifi- cantly higher than those supplied from other reser- voirs (x2 = 40.68; P <10-). Intrafamilial spread Having established that the hepatitis was waterbome, we examined whether there was any evidence of intrafamilial (secondary) spread. The number of cases in each family is shown in Table 4. Of 138 hepatitis cases in the sample survey, 86 were the only cases in their family, while of the remaining 52 cases, 21 were "multiple" first cases in the same family. Thus a total of 107 of 138 (77.5%) cases were first cases in their families. Of the 31 "fresh" cases, 23 (74%) occurred within 2 weeks of the onset of clinical disease in the index case, three at 2-4 weeks later, and five at 4-6 weeks later; these eight cases that occurred after 2 weeks could be con- sidered to be indeterminate, since they could have arisen because of either primary or secondary intra- familial spread. No family had any case that pre- sented later than 6 weeks after the occurrence of the index case. Table 4: Distribution of the 138 hepatitis cases in fami- lies with single or multiple cases No. of cases No. of No. of cases per family families Index Fresh 1 86 86 0 2 13 13 13 3 6 6 12 4 2 2 6 Total - 107 31 Reservoir 2 *- ~~----.... Serological data HBsAg or IgM anti-HAV was not detected in any of the 41 sera tested, indicating that none of these patients had evidence of recent viral hepatitis A or B. Other data Hospital records revealed that 48 persons, including 13 pregnant women, died of viral hepatitis from January to April 1991 in various govemment hospi- tals in Kanpur. Discussion We estimate that over 79 000 cases of viral hepatitis occurred in the Kanpur epidemic-the largest num- ber yet recorded. The largest previous epidemic of viral hepatitis, which occurred in Delhi in 1955-56, affected approximately 29 300 persons (4). Such large-scale hepatitis epidemics are invariably water- bome and caused by an enterically transmitted hepa- titis non-A, non-B agent (1-3) (now called hepatitis E virus (HEV)) (2). The serological tests showed that hepatitis A or B virus was not responsible for the epidemic in Kanpur. Using a reverse transcription- polymerase chain reaction based on the sequence of a previously described HEV clone (ETI.1) (9), we demonstrated the presence of the HEV genome in 6 of 10 stool samples obtained from persons affected in the epidemic (10). The epidemiological features of the Kanpur epi- demic resemble those reported for previous HEV epidemics (1-3), i.e., a high attack rate among adult males but relatively few cases involving children. Thus, only 6% of the cases that we observed in the sample survey were aged 0-9 years, 81% were aged 10-39 years, and 65% were males. The Kanpur epidemic affected a significantly larger proportion of persons who lived in the MW areas, whose water supply was derived from the river Ganges, than those who lived in areas that used exclusively tubewell water. Extremely high inci- dences of 10.4% and 7.0% were observed in the MW areas that received their water from reservoir No. 4, where the chlorination had been unsatisfactory. In addition, the incidence of hepatitis among persons who consumed handpump-water, even though they resided in MW areas, was significantly lower. This evidence clearly indicates that the epidemic occurred because of contamination of the city's water supply system. To understand the origin of this contamination, it may be important to trace the history of the Kan- pur water supply system. The water intake station at Bhaironghat was constructed in 1892 on the banks of WHO Bulletin OMS. Vol 70 1992 -1U . .... -1 601 S.R. Naik et al. the river Ganges. Subsequently, the river has gradu- ally changed its course, and as a result this intake point is now about 7 km from the main stream of the river. A looped diversion channel was therefore dredged to carry water from the main river to the Bhaironghat pumping station and to a thermal power- house located approximately 1-km downstream of the pumping station (Fig. 5). An open drain carrying domestic sewage discharges into the diversion chan- nel, about 100-m upstream of the pumping station's intake point. Sewage from this drain had previously caused no major health problems, possibly because it was diluted by the continuous flow of river water in the channel. In 1990, however, the powerhouse closed down and regular dredging of the water chan- nel downstream of the water intake point stopped. The channel thus became a virtual cul-de-sac, ending at the water pumping station, and undiluted sewage from the drain began to contaminate the raw water supply to the city. From June 1990 the monthly raw water chemical analyses data confirm gross faecal contamination. It is notable that the water in reser- voir No. 4 had gross coliform contamination and a very low free chlorine content as early as the middle of February 1991 and that the city areas supplied by this reservoir had the highest incidence of hepatitis. Inadequate chlorination was therefore an important additional factor in causing the epidemic, and this was also considered to be an important factor in the epidemic in Delhi (4). A free residual chlorine concentration of at least 0.5 mg/l for a minimum of 30 minutes is considered adequate to ensure the "viral quality" of drinking-water (12). Routine coli- form counts have been used as a surrogate marker for Fig. 5. Location of the water pumping station, thermal powerhouse, the channel bringing water to these, and a sewage drain opening Into the channel. The speckled area shows the site of the blockage in the channel caused by the stoppage of dredging operations. the viral quality of public water supplies, since chlo- rination reduces the load of coliforms and of enteric viruses (13). No data are yet available on the efficacy of chlorination in reducing the load of HEV in water. The present epidemic had two peaks. The first, in late February 1991, was probably related to the continuing faecal contamination of water at the river intake point from June 1990. The delay of nearly 8 months between these events is, however, difficult to explain, although it is possible that the concentra- tion of the virus in the water attained a critical level only around December 1990. The second peak, in March-April 1991, was more sustained and was most probably related to chlorination failure. Our analysis of the primary (waterbome) and secondary (intrafamilial) modes of spread of infec- tion revealed that 77.5% of the cases in the sample survey were first cases in their families, and thus primary cases. Of the 31 cases that followed the index cases, 23 occurred within 2 weeks of onset of disease in the index cases. Since this is shorter than the lower limit for the incubation period of viral hepatitis E (2-8 weeks), we consider these cases also to be primary, i.e., 130 of 138 cases (94.2%) were primary. For the remaining eight cases (5.8%) hepatitis developed within 2-6 weeks of that in the index cases. No case occurred more than 6 weeks later, indicating that secondary spread was not a major feature in this epidemic, at least when the survey was carried out. Similar findings have been reported previously by Khuroo for a hepatitis epidemic in Kashmir (11). Our analysis of the data on the spread of infec- tion in the epidemic has several limitations. Firstly, individuals can be exposed to a waterbome infection not only at home but also at their place of work or elsewhere. However, it was virtually impossible to obtain a detailed account of the places visited by any individual in the 2-8 weeks prior to the survey. Our study therefore examined only the water source used at home. Furthermore, there were no data on the extent to which river water was supplemented with tubewell water at different points in the supply sys- tem. Variations in the composition of mixed water may also partly explain variations in the incidences in different MW areas. The relatively higher attack rate (2.5%) in ward 47 (a TW area) perhaps requires explanation, although it was not significantly differ- ent from the rate in the two other TW areas (1.0%). One reason may have been that in ward 47 people consumed raw vegetables grown on farms that were irrigated with untreated sewage water; however, in the absence of a convenient method for detecting HEV in food, we are unable to confirm or refute this. Our findings highlight the continuing problems posed by contamination of water supply systems in WHO Bulletin OMS. Vol 70 1992 RIVER GANGES WATER CHANNER P S WATER PUMPING STATIONi POWERHOUSE 602 Viral hepatitis E epidemic in Kanpur, India urban areas of India and the pressing need for a com- prehensive policy to provide safe potable water and proper sewage disposal facilities. Acknowledgements We thank Lupin Laboratories Ltd., Bombay, for financial help; UNICEF, Ranbaxy Laboratories Ltd., Bombay, and Gyan Scientific Traders, Lucknow, for providing con- sumables; Dr L. Prasad, Dr A.R. Pandey, Dr R. Babu, Dr D.N. Tewari, Dr R.C. Joshi, Dr A.K. Mallik, Dr O.P. Misra, Dr P.K. Ray, Dr P.K. Seth, Mr Rehman, Vaidya U.C. Sharma, Dr M. Das, and Dr A. Kumar, for help in conducting the field survey; and Mr K.K. Shukla for provid- ing the water analysis data. Mr S.N. Semwal and Mr T.S. Negi provided technical help and Ms S. Yadav prepared the figures. R.A. was supported by the Scientist's Pool Scheme of the Council of Scientific and Industrial Research, New Delhi. Resume Vaste 6pidemie d'h6patite virale E transmise par l'eau b Kanpur, Inde Le sous-continent indien est regulierement frap- p6 par des 6pid6mies d'h6patite. Le present article d6crit l'tude 6pid6miologique de la plus vaste 6pid6mie d'h6patite E (h6patite non-A non-B transmise par voie ent6rique) jamais rap- port6e, survenue a Kanpur, en Inde, en 1991. Une enqu6te pilote r6alis6e dans la ville a montr6 que l'incidence de I'hepatite icterique dans une zone fortement atteinte 6tait de 10,6%. Comme les epid6mies d'h6patite sont g6n6ralement transmises par l'eau, ce que confirment les r6sultats de l'enqu6te pilote, nous avons r6alis6 une enqu6te d6taill6e afin d'6tu- dier la relation entre l'incidence de l'h6patite et la source d'approvisionnement en eau de la ville. Nous avons calcul6 que, pour d6celer une diff6rence d'incidence de I'h6patite dans deux r6gions ayant un approvisionnement en eau dif- f6rent (en supposant des taux d'incidence de 5% et 2,5%), il fallait enqu6ter au moins 152 foyers dans chaque r6gion. Au moyen d'une technique d'6chantillonnage en plusieurs stades et de tables de nombres au hasard, nous avons proc6d6 a un sondage dans 420 foyers r6partis dans sept des 50 quartiers de la ville. Dans quatre des sept quartiers choisis, 1'eau du reseau 6tait un m6lange d'eau du Gange et d'eau de puits instantan6s, tandis que dans les trois autres quartiers, il ne s'agissait que d'eau de puits instantan6s. Entre d6cembre 1990 et avril 1991, l'incidence de l'h6patite ict6rique dans 1'6chantillon 6tudi6 a ete de 3,76% (138 sujets sur 3666) ce qui, extrapol6 a 1'ensemble de la ville, conduit a une estimation de 79 091 personnes atteintes. Le taux d'atteinte 6tait plus eleve chez les sujets de sexe masculin (5,3% contre 3,3%; P = 0,013) et chez les adultes que chez les enfants de moins de 10 ans (4,26% contre 1,29%; P = 0,0006). Parmi les sujets affectes, 82% appartenaient au groupe d'age 10-39 ans. L'incidence de l'h6patite 6tait plus elev6e dans les r6gions recevant de l'eau mixte que dans celles recevant de l'eau de puits (5,6% contre 1,2%; P = 106). Dans ces pre- mieres r6gions, I'incidence diminuait lorsque l'eau de boisson 6tait prise non au robinet seu- lement mais a la fois au robinet et a la pompe a main, ou uniquement A la pompe a main (7,8%, 6,8% et 4,3% respectivement; P = 0,023, test du x2). Ces donn6es font apparaitre une contami- nation de l'eau du robinet dans les r6gions ali- ment6es en eaux mixtes, et permettent d'incrimi- ner l'eau du Gange comme source de l'6pid6mie. Rien ne montre qu'il y ait propaga- tion secondaire intrafamiliale des cas. Des pr6le- vements de s6rum r6alis6s chez 41 sujets ont revel6 'absence de virus de l'h6patite A ou B, et nous avions d6ja rapport6 la pr6sence du geno- me du virus de l'hepatite E dans des 6chan- tillons de selles recueillis chez des malades lors de cette 6pid6mie. L'incidence de l'h6patite pr6sente deux pics, le premier en fWrier 1991 et le deuxieme, a la fois plus haut et plus large, en mars-avril 1991. Les teneurs en chlorure et en nitrate ainsi que la demande chimique d'oxygene dans l'eau du fleuve montrent qu'il y a eu contamination f6cale depuis juin 1990, phenomene responsable du premier pic d'incidence. La contamination a ete mise en relation avec le rejet d'un 6gout dans le canal alimentant la station de pompage. De plus, I'analyse de l'eau dans l'un des reservoirs de l'usine de traitement a montre l'absence de chlore libre et une num6ration des coliformes de 35 par d6cilitre le 16 fWvrier 1991, ce qui permet de penser qu'une chloration insuffisante de l'eau a t a l'origine du deuxieme pic. Notre 6tude montre 6galement de fagon indirecte que la chloration de 1'eau joue un r6le important dans l'inactivation du virus de l'hepatite E. References 1. Ramalingaswami, V. & Purcell, R.H. Waterborne non-A, non-B hepatitis. Lancet, 2: 571-573 (1988). 2. Bradley, D.W. Enterically transmitted non-A, non-B WHO Bulletin OMS. Vol 70 1992 603 S.R. Nalk et al. hepatitis. British medical bulletin, 46: 442-461 (1 990). 3. Khuroo, M.S. Hepatitis E: enterically transmitted non-A, non-B hepatitis. Indian journal of gastroen- terology, 10: 96-100 (1991). 4. Vishwanathan, R. Infectious hepatitis in Delhi (1955-56): a critical study; epidemiology. Indian journal of medical research (suppl.), 45: 1-30 (1957). 5. Census of India 1991, Uttar Pradesh: provisional population tables. Lucknow, Director of Census Operations, Uttar Pradesh, 1991, pp. 21-22. 6. Armitage, P. & Berry, G. Statistical methods in medical research, 2nd ed. Oxford, Blackwell, 1987, pp. 160-185. 7. Armitage, P. & Berry, G. Statistical methods in medical research, 2nd ed. Oxford, Blackwell, 1987, pp. 371-407. 8. Armitage, P. & Berry, G. Statistical methods in medical research, 2nd ed. Oxford, Blackwell, 1987, pp. 455-483. 9. Reyes, G.R. et al. Isolation of a cDNA from the virus responsible for enterically transmitted non-A, non-B hepatitis. Science, 247: 1335-1339 (1990). 10. Ray, R. et al. Hepatitis E virus genome in stools of hepatitis patients during a large epidemic in north India. Lancet, 338: 783-784 (1991). 11. Khuroo, M.S. Study of an epidemic of non-A, non-B hepatitis: possibility of another human hepatitis virus distinct from post-transfusion non-A, non-B type. American journal of medicine, 68: 818-824 (1980). 12. Guidelines for drinking water quality, vol. 1: recom- mendations. Geneva, World Health Organization, 1984. 13. Hurst, C.J. Presence of enteric viruses in fresh- water and their removal by the conventional drink- ing water treatment process. Bulletin of the World Health Organization, 69: 113-119 (1991). 604 WHO Bulletin OMS. Vol 70 1992