84

them warned that this would happen and that the pat-tern adopted was unsatisfactory; a more cautious,experimental and piecemeal approach was felt to bedesirable. Now we are faced with no development during1976/77 and a possible reduction in expenditure ofbetween 5 and 10% over 1977/79. Community medicineshould bear its share of these cuts, but we must be quitecertain that this share is a fair one in comparison withnon-medical and non-nursing administrative staff whosenumbers have increased by up to 30-40% in some cases.The risk of an excessive burden on community medicinefollows from two important facts. Firstly, the totalnumber of practitioners originally working in publichealth before 1974 was cut by a significant proportionduring reorganisation, and is only now beginning slowlyto recover as establishments expand. Secondly, thestandstill in 1976/77 is based on staff in post in March,1976, thereby excluding all vacant posts an importantpoint in a specialty with approximately 20% of its estab-lished posts vacant.

All manpower forecasts must be integrated with eco-nomic objectives, particularly at the present time, and itwould be unwise to attempt to expand community medi-cine at an unrealistic rate. However, the rate of increasemust take full account of the numbers already partiallytrained and who wish to return to community medicine,in addition to a core of highly trained new recruits.The D.H.S.S. target of 30 new recruits per year to the

registrar grade is equivalent to a rate of increase of 4%(taking Department of Health doctors into account),falling steadily to 3-5% per annum after 5 years andcontinuing to fall slightly thereafter. This rate of in-crease may be thought inadequate in view of the presentnumber of vacancies, but it is unlikely that any higherrate could be sustained in view of lack of finance and ofthe recruitment possibilities. The morality of the actionof Government in allowing administrative structures tobuild up and in recommending recruitment-rates com-parable with those in clinical specialties, whilst beingaware of the inevitable need eventually to make drasticcuts, must be questioned. But it may be appropriateunder these circumstances, taking into account thenature of community medicine, to look carefully at theneed for a permanent sub-specialist grade, possiblyemployed part-time. This would enable many people tocontribute to the specialty who might not otherwise bein a position to do so. Many specialists have takenadvantage of the opportunity to purchase "added years"and their early retirement could well make a consider-able difference to the picture within the next few years.The importance of manpower planning in such a

small specialty cannot be over-emphasised. There is, inaddition, a need for counselling and advice at all stagesin the careers of those who are in the training grades,which should be a specific responsibility of regionalfaculty advisers, but which cannot be ignored by anycommunity physician. The people whom we should aimto attract must have had an adequate training in thebasic subjects of their specialty and the type of courseoffered must be no less rigorous or satisfying, in bothcontent and presentation, than equivalent training pro-grammes in clinical specialties. It is essential that the

community physician of the future should be able to

command the respect and confidence of all those withwhom he or she will need to work; medical leadership inthe provision of health care to populations is no less vital

than it has ever been, despite the widely spread responsi-bility for management in the more complex patterns ofthe new organisation.We thank Dr R. A. Franklin and Dr A. G. Brown for advice and

assistance in the preparation of this article.Requests for reprints should be addressed to P.J.H., Sheffield Area

Health Authority (Teaching), Westbrook House, Sharrow Vale Road,Sheffield S 11 8EU.

REFERENCES

1. Heath, P. J. Community Hlth, 1974, 5, 178.2. Report of the Royal Commission on Medical Education. H.M. Stationery

Office, 1968.3. Report of the Working Party on Transferred Medical Officers. British Medi-

cal Association, 1974.4. Third Report of the Review Body on Doctors’ and Dentists’ Remuneration

(First Supplement), H.M. Stationery Office, 1973.

Public Health

CHOLERA TRANSMISSION NEAR A CHOLERAHOSPITAL

RICHARD J. LEVINE*STANISLAUS D’SOUZA

MOTIUR R. KHANDAVID R. NALIN

Center for Disease Control, United States Public HealthService, Atlanta, Georgia, U.S.A., Epidemiology Division,Cholera Research Laboratory, Dacca, Bangladesh; JohnsHopkins University School of Hygiene, Department of

Biostatistics, and Johns Hopkins University InternationalCenter for Medical Research, Dacca, Bangladesh, and

Baltimore, Maryland 21205, U.S.A.

Summary A review of the incidence of cholerafrom 1964 through 1974 in Matlab,

Bangladesh, revealed that among the villages severalhad very high incidence-rates. Investigation indicatedthat high cholera-rates in two of these villages wereprobably related to water contamination from a nearbycholera hospital established in 1963. The data implythat heavy contamination can overcome any immunityresulting from repeated exposure.

INTRODUCTION

FIELD surveillance for cholera and collection of demo-graphic data began in Matlab in late 1963 and a localcholera hospital, first on a canal-barge, then in a building,was established. Free ambulance boats were available

day and night 1968-71 to bring patients to the hospital.By 1968 the surveillance area included 225 000 personsin 234 villages. 1-5McCormack et al.’ reviewed data for 1964-1966 and

found that most cases occurred during major winter andminor spring epidemics and were geographically widelyscattered. He noted that the establishment of the vaccinetrial area and hospital might alter natural cholera epi-demiology. In the present study, the incidence of choleraover 11 years in Matlab was reviewed. While the overallpattern described earlier remained, several villages hadconsistently high incidence-rates. This report examinesthe hypothesis that high rates in two of the villages (8.6,10-9 per 1000 annually) were related to contaminationof canal water associated with the nearby cholera hospi-tal.

*Present address: Epidemiology Bureau, Center for Disease Control, Atlanta,Georgia 30333, U.S.A.

85

Fig. 1-Map of study area.

METHODS

Cholera was diagnosed by rectal swab or faecal cultures6from all patients who came to the hospital for treatment ofdiarrhaea (outpatients and cases admitted to hospital). The cri-terion for admission to the hospital was profuse diarrhoea lead-ing to severe dehydration, usually with shock. During1963-1969 inapparent infections and ambulatory cases werealso detected by field surveys.’ Each family was visited at leasttwice weekly and rectal-swab stool cultures were obtained frompersons with diarrhoea. During cholera vaccine trials daily rec-tal swabs were obtained for 4 days from’trial participants infamilies of persons with diarrhoea severe enough to limit nor-mal activity. Demographic data were available from periodiccensuses and from ongoing registration of vital events startedin 1966.The two high-incidence villages were divided into four areas

for the purpose of study (fig. 1): area A bordering a canalto the north; area B of similar size where homes did not adjoina canal; area C demarcated from B by a road and situatedalong a canal whose currents flowed by the cholera hospital;and area D located on the same canal, but along a sectionwhere water did not pass the hospital. Records of diarrhoea!deaths for the two villages from 1944-1963 were examined todetermine whether the high incidence of cholera antedated theestablishment of the hospital. Although these records main-tained by village headmen are undoubtedly incomplete, thereis no reason to suspect bias in reporting vital events from dif-ferent parts of the villages.

In December, 1974, vibrio and coliform counts were per-formed on canal water from 17 sites, a sanitation survey wasconducted, and water usage was determined by direct observa-tion on 6 days and by questionary.

RESULTS

High-incidence and average-incidence areas of thetwo villages studied were almost the same distance fromthe hospital, and the ratio of mild to severely affectedcases seeking medical attention was 1/3 in both areas.This suggests that admission-rates did not vary with dis-tance from hospital.The high-incidence area (C; fig. 1) consisted of adja-

cent parts of the two villages. During 5 epidemics sur-

veyed from November, 1968, throughFebruary, 1971, the cholera hospita-lisation rate in area C near the cho-lera hospital was 13-6 cases per 1000persons per year compared with 3-9for areas A, B, and D, and 2.0 for all234 villages. The incidence-rate ratioof area C compared to areas A, B,and D was 4.0. Area C had the high-est median hospital admission-rate ofall village areas of similar size for the5 epidemics surveyed. Equally strik-ing was the high cholera reinfection-rate. Among reinfections from 56 vil-lages between 1963 and 19699 thereinfection-rate for individuals fromarea C was more than 13 times as

high as the rate for the other 54 vil-lages (26.4 vs. 2.0 per 1000 personyears).

Comparison of annual diarrhoea

mortality-rates 1944-1963 based on1963 population indicated no excesspre-hospital deaths in area C (see

accompanying table), suggesting that the high rates de-veloped after the hospital was established. Had severecholera been as common before as after creation of thehospital, this should have been reflected in the rate ofdiarrhoea deaths for area C. During the period understudy migrations into and out of the villages occurredwith the net effect of Moslems replacing Hindus in somesections. Yet the population distribution among the fourareas probably did not undergo major changes.

In the first year of hospital operation cholera-rates inareas A, B, C, and D were similar and low. Afterwardsarea C’s rate increased in proportion to the overallnumber of cholera admissions; however, rates in areasA, B, and D, and in an adjacent village did not followthe graph of cholera admissions closely (fig. 2).

Sanitation findings also suggested a relation betweenincreased cholera and establishment of the hospital onthe nearby canal (fig. 1). Patients brought by speedboatambulance often defalcated near the landing, and fae-

cally soiled boats were washed there. Patients’ relativesoften infected themselves,10 used the canal bank fortoilet, bathing, and washing contaminated clothes. Thehospital sewage system was blocked; and leaking sewagedrains and surface runways emptied untreated effluentsfrom soakpits and sinks into the canal. The system wasnot sealed and could mix with monsoon flood waters.One of seventeen samples from seventeen canal sites

yielded cholera vibrios. This site adjoined the hospitaland had the highest coliform count (3000/ml; other sites30 to 1000/ml).

Most cholera patients from area C lived along thecanal where water containing hospital effluents couldreach (fig. 1). The area distant to the stagnant zone,DIARRHOEA DEATH-RATES (1944-1963) BEFORE ESTABLISHMENT OF THE

HOSPITAL

86

Fig. 2-Annual hospital cholera admissions and rates of cholera.

where tidal currents from the 2 river inlets cancelledeach other, was fed from a separate inlet and had astrikingly lower incidence-rate.

During the 11 years, 1964-1974, cholera incidence-rates in the village north of the canal were 13.6, 8.7,and 3-3/1000 per year, respectively, for families report-ing use of canal water, reporting no use but livingnearby (C, fig. 1), and reporting no use but livingfarther away (B, A, fig. 1). Differences between first andlast groups were highly significant (p 1-tailed=0.0000001 with rate ratio 4-1 and 90% confidencelimits for rate ratio of 2-6 and 6-3). Hospital cases ofnon-cholera diarrhoea (3’3/10000/year, diverse actiolo-

gies) did not cluster near the canal.From 1964 through 1973, 6927 severely ill cholera

patients were admitted to the hospital. According to pre-vious mortality reports," approximately 60% wouldhave died without treatment. Fluid and electrolyte re-placement reduced mortality of hospital patients to only0-6%. With a risk ratio 4-0 it can be estimated that

during the same period an excess of 71 cholera casesfrom area C (or about 1% of the total) were admitted tohospital.

DISCUSSION

The association of high incidence-rate with canal-water usage and the decline in rate with distance fromthe canal indicate that canal water was probably the ve-hicle of transmission of cholera. The combined data onincidence-rates, reinfection, mortality from diarrhoeaand sanitation, and water flow suggest that hospital-re-lated contamination of the canal accounts for the highcholera-rate in area C. The increased rate among those

living near the canal but reporting no canal-water usagemay be due to contamination of alternative water

sources by infected persons using the canal or throughmonsoon connections to the canal. Although the canal iscontaminated by boatmen,12 bazaar visitors, and others,the hospital is the most likely regular source of heavyvibrio contamination. Other canals within the surveil-lance area of 234 villages, including the canal borderingarea A, are contaminated by boatman and bazaarvisitors. But these have not resulted in unusual aggrega-tions of cholera cases. The high reinfection-rate in areaC indicates that heavy contamination of canal water canovercome any immunity resulting from repeated expo-sure.

The fact that cases of non-cholera diarrhoeas did notcluster around the canal suggests that water may not beimportant in their transmission. However, these non-cholera diarrhoeas have multiple aetiologies, and patternsof spread may differ, so no firm conclusion can bedrawn.

This is the first report of cholera ascribed to contami-nation from a hospital. The benefit of the hospital to thecommunity in terms of lives saved greatly outweighs thefew extra cases in the hospital vicinity. Nevertheless, thedata indicate the need for a simple inexpensive methodof sterilising vast quantities of diarrhoea fluid in an areawith a high water table and seasonal flooding.We thank the staff of the Cholera Research Laboratory for expert

technical assistance and P. Brachman, R. Cash, A. Choudhury, W.Greenough, A. Langmuir, K. Monsur, and K. Rothman for valuablesuggestions; and Dr Eugene Gangarosa for advice and encouragementin the preparation of the manuscript. This study was funded by N.I.H.research agreement R07AI10048-15 with the Johns Hopkins Centerfor Medical Research and by the Cholera Research Laboratory, anautonomous organisation supported by the Governments of the Peo-ple’s Republic of Bangladesh, U.S.A., U.K., and Australia.

Requests for reprints should be addressed to R. J. L.

REFERENCES

1. McCormack, W., Mosley, W. H., Fahimuddin, M., Benenson, A. S. Am. J.Epidem. 1969, 89, 393.

2. Oseasohn, R. O., Benenson, A. S., Fahimuddin, M. Lancet, 1965, i, 450.3. Benenson, A. S., Mosley, W. H., Fahimuddin, M., Oseasohn, R. O. Bull. Wld

Hlth Org. 1968, 38, 359.4. Mosley, W. H., McCormack, W. M., Fahimuddin, M., Aziz, K. M. A., Rah-

man, A. S. M. M., Chowdhury, A. K. M. A., Martin, A. R., Feeley, J. C.,Phillips, R. A. ibid. 1969, 40, 177.

5. Mosley, W. H., Woodward, W. E., Aziz, K. M. A., Rahman, A. S. M. M.,Chowdhury, A. K. M. A., Ahmed, A., Feeley, J. C. J. infect. Dis. 1970,121, suppl., 1.

6. Woodward, W. E., Mosley, W. H., McCormack, W. M. ibid. 1970, 121,suppl., 10.

7. District Commissioner’s Record Room, Comilla, Bangladesh.8. Miettinen, O. S. J. Am. stat. Ass. 1974, 69, 380.9. Woodward, W. E. J. infect. Dis. 1971, 123, 61.

10. Oseasohn, R. O., Ahmad, S., Islam, M. A., Rahman, A. S. M. M. Lancet,1966, i, 340.

11. Pollitzer, R. Monograph Ser. W.H.O. 1959, no. 43.12. Khan, M., Mosley, W. H. E. Pak. med. J. 1967, 11, 61.

FAILURE OF SANITARY WELLS TO PROTECTAGAINST CHOLERA AND OTHER DIARRHŒAS

IN BANGLADESH

RICHARD J. LEVINE*STANISLAUS D’SOUZA

MOTIUR R. KHANDAVID R. NALIN

Center for Disease Control, United States Public HealthService, Atlanta, Georgia, U.S.A., Epidemiology Division,Cholera Research Laboratory, Dacca, Bangladesh; JohnsHopkins University School of Hygiene, Department of

Biostatistics, and Johns Hopkins University InternationalCenter for Medical Research, Dacca, Bangladesh, and

Baltimore, Maryland 21205, U.S.A.

Summary Within an area of Bangladesh in whichthe incidence of cholera was high, use of

sanitary pipe wells did not protect against cholera orrelated non-cholera diarrhœas because well users alsoused contaminated water sources regularly enough tomaintain high infection-rates. Protection was found tocorrelate with education and wealth.

*Present address: Epidemiology Bureau, Center for Disease Control, Atlanta,Georgia 30333, U.S.A.