BMJ No 7123 Volume 315
Paper Saturday 20/27 December Christmas 1997 issue
Risk factors for winter outbreak of acute diarrhoea in France: case-control study
Laurent Letrilliart, Jean-Claude Desenclos, Antoine FlahaultAbstract
Objectives: To assess the potential role of consumption of shellfish (particularly raw oysters) and tap water in the winter epidemic of acute diarrhoea in France.
Design: Population based, case-control study during the 1995-6 winter epidemic of acute diarrhoea in France.
Setting: A national network comprising 1% of general practitioners in France.
Subjects: 568 pairs of cases and controls consulting in general practice and interviewed by 209 doctors from 26 December 1995 to 31 January 1996. Cases and controls were matched for age, doctor, and time of consultation.
Main outcome measures: Adjusted relative risk of diarrhoea estimated from conditional logistic regression.
Results: The risk of acute diarrhoea was not increased in people who had recently eaten raw oysters (odds ratio 1.1; 95% confidence interval 0.9% to 1.4%) or other shellfish such as clams, cockles, and mussels, or in those people who usually consumed tap rather than bottled water (0.8; 0.6% to 1.1%). The risk was, however, increased in people who had had recent contact with a person with diarrhoea, either within the household (adjusted odds ratio 5.0) or in the workplace (3.1), and in people who lived with a child of 2 years of age or under (1.6). Recent treatment with either oral penicillin or cephalosporin was also independently associated with acute diarrhoea in winter.
Conclusions: The winter epidemic of acute diarrhoea in France is probably not caused by consumption of either shellfish or tap water. A viral aetiology, however, is suggested by the speed with which the acute diarrhoea is transmitted.
Introduction
| In developed countries acute diarrhoea is a major cause of
morbidity and medical expense. Each year in the United States about 99
million people experience acute diarrhoea or vomiting, of whom 8.2
million seek medical help (0.036 per person year).(1)
According to data from France's electronic surveillance of
communicable disease, around 3.3 million people in France consulted a
doctor in 1995 for acute diarrhoea, totalling 0.057 consultations per
person year.(2,3) In France large outbreaks of acute diarrhoea begin regularly at the end of December or early in January (fig 1). During the epidemic of January 1996 we estimated that about 670,000 people had consulted a doctor for acute diarrhoea throughout continental France.(4) |  | |
| Fig 1: Incidence of acute diarrhoea as reported weekly by doctors in France, December 1990 to January 1997 (epidemic threshold is upper limit of 95% confidence interval of regression model applied to non-epidemic series). |
The consumption of shellfish has been implicated in the winter outbreaks of acute diarrhoea in France because of the increased consumption of shellfish, particularly raw oysters, between Christmas and the New Year(5); the annual peak of diarrhoea occurs just after 1 January, when shellfish is consumed on a large scale. Another theory is that drinking water becomes contaminated by pathogens and toxic agents that have been washed into the public water supply during flooding.(6,7) Although one adverse effect of antibiotics is acute diarrhoea, the role of antibiotics in morbidity has yet to be ascertained.(8)
Little is known about the risk factors for acute diarrhoea in Europe, particularly during winter outbreaks. We therefore conducted a case-control study among doctors in France during the winter of 1995-6 to identify the predominant modes of transmission and risk factors for the winter outbreak of acute diarrhoea.
Patients and methods
About 1% of general practitioners participate in the French communicable disease surveillance scheme called Sentinelle(2); our population based, case-control study was of the patients of these doctors. Our doctors took part in the study on a voluntary and unpaid basis, as they do for the surveillance scheme.
Cases
Cases were patients who had consulted one of the doctors in the
Sentinelle scheme for acute diarrhoea - that is, had had loose stools
for <15 days(3) - during the winter outbreak of acute
diarrhoea from 26 December 1995 to 31 January 1996. Each doctor was
asked to include the first three of their patients who had either
presented to the surgery or required a home visit for acute diarrhoea.
Controls
Controls were matched to cases of acute diarrhoea for age (0-4
years, 5-14, 15-59, and 60 and over), doctor, and time of consultation
(within two days). The first patient to meet these criteria was
included as a control by the doctor. Patients who reported
gastrointestinal symptoms over the past three months were not eligible
for inclusion in the study.
Data collection
The doctors were asked to collect data on cases and controls
during the consultation; the parent or guardian was consulted if the
patient was <12 years of age. Data included demographic
characteristics (age, sex, and whether the patient resided in a rural
or urban area); attendance at a nursery, school, retirement home, or
other; the presence of a child 2 years of age or under at home; consumption
of raw shellfish within 10 days before consultation (oysters, clams,
cockles, and mussels; if patients had consumed cockles and mussels they
were asked whether they had eaten them raw or cooked); consumption of
tap water (always or often compared with sometimes or never); and type
of oral antibiotic treatment, including trade name of drug, during the
month before consultation. Antibiotics were subsequently classed into
four groups: penicillins, cephalosporins, macrolids, and combined other
families. Transmission from person to person was ascertained by asking
the patient whether they had had any contact with a person with
diarrhoea within the previous 10 days and, if so, whether this had
occurred in the household, workplace, or a confined setting such as a
nursery, school, retirement home, or other. Information was also
obtained on the date of onset of diarrhoea in the contact; only this
date was recorded because repeat contacts may have occurred. From the
date of the onset of diarrhoea in the contact we estimated a surrogate
variable for the incubation time as the time lag between the onset of
diarrhoea in the case minus that in the contact - from these data we
could approximate the median incubation time for the diarrhoea.
At the time of the study 209 of the 438 doctors who participated in the Sentinelle scheme took part in the case-control study; data included 588 cases and 568 controls, constituting 568 matched pairs.
Statistical analysis
Data were stored in a database (version 7, Oracle, Redwood City,
CA). Statistical analysis was performed on sas software
(version 6, sas Institute, Cary, NC). Matched odds ratios
and their 95% confidence intervals were used to estimate relative
risks in univariate analyses. Conditional logistic
regression(9) was then used to assess the independent
contribution of risk factors for acute diarrhoea identified in
univariate analyses: all variables with P= 0.20 and under in univariate
analyses were introduced in the initial multivariate model and deleted
through a backward procedure; variables with a significance level of
0.05 (two tailed formulation) were kept in the final model.
Results
| Demography - Cases and controls did not differ by sex and
place of residence but controls were on average 2.1 years older than
their matched case (SE 0.41 years) and were on average included into
the study 1.1 days after the matched case. Controls were more likely
than the cases to attend their general practitioner's surgery: 454 out
of 562 (81%) v 331 out of 565 (59%).
Consumption of shellfish and tap water - The risk of acute diarrhoea was not increased among people who had consumed shellfish (raw or cooked) in the 10 days before the onset of acute diarrhoea (table 1); even when there was an increase in consumption of shellfish we did not find a higher risk of acute diarrhoea. Usual consumption of tap water was not associated with acute diarrhoea (table 1). |  | |
| Do shellfish cause the winter outbreaks of diarrhoea in France? |
| Table 1 - Recent consumption of shellfish and tapwater, and exposure to contacts and antibiotics (conditional logistic regression models) in occurrence of acute diarrhoea | |||
| Risk factor | Cases (n=568) |
Controls (n=568) |
Odds ratio (95% CI) |
|---|---|---|---|
| Consumption | |||
| Any raw shellfish | 210/564 (37.2%) |
191/557 (34.3%) |
1.1 (0.8 to 1.5) |
| Raw oysters | 206/566 (36.4%) |
191/564 (33.9%) |
1.1 (0.9 to 1.4) |
| Raw clams | 20/566 (3.5%) |
12/564 (2.1%) |
1.7 (0.8 to 3.6) |
| Cockles: | |||
| None | 550/565 (97.3%) |
553/564 (98.0%) |
1.0 |
| Raw | 8/565 (1.4%) |
3/564 (0.5%) |
2.6 (0.7 to 10.0) |
| Cooked | 7/565 (1.2%) |
8/564 (1.4%) |
0.9 (0.3 to 2.8) |
| Mussels: | |||
| None | 511/564 (90.6%) |
501/564 (90.0%) |
1.0 |
| Raw | 9/564 (1.6%) |
8/564 (1.4%) |
1.1 (0.4 to 2.8) |
| Cooked | 44/564 (7.8%) |
48/564 (8.6%) |
0.9 (0.6 to 1.3) |
| Tapwater | 359/564 (63.7%) |
376/564 (66.4%) |
0.8 (0.6 to 1.1) |
| Contacts | |||
| In nursery | 5/568 (0.9%) |
7/568 (1.2%) |
0.5 (0.1 to 2.0) |
| In school | 89/568 (15.7%) |
94/568 (16.5%) |
0.2 (0.0 to 1.4) |
| In retirement home | 6/568 (1.1%) |
1/568 (0.2%) |
5.0 (0.6 to 42.8) |
| Child 2 years of age or under at home | 93/550 (16.9%) |
58/560 (10.4%) |
1.9 (1.3 to 2.7) |
| Contact with person with diarrhoea: | |||
| No contact | 334/564 (59.2%) |
475/561 (84.7%) |
1.0 |
| In household | 181/564 (32.1%) |
58/561 (10.3%) |
5.0 (3.4 to 7.3) |
| In workplace | 30/564 (5.3%) |
14/561 (2.5%) |
3.4 (1.7 to 6.8) |
| In retirement home | 3/564 (0.5%) |
1/561 (0.2%) |
5.0 (0.5 to 55.4) |
| In school | 11/564 (2.0%) |
10/561 (1.8%) |
1.7 (0.7 to 4.3) |
| Other* | 5/564 (0.9%) |
3/561 (0.5%) |
6.0 (1.0 to 36.6) |
| Oral antibiotics | |||
| Penicillins | 43/568 (7.6%) |
28/568 (4.9%) |
1.6 (1.0 to 2.6) |
| Cephalosporins | 20/568 (3.5%) |
11/568 (1.9%) |
1.8 (0.9 to 3.8) |
| Macrolids | 16/568 (2.8%) |
13/568 (2.3%) |
1.2 (0.6 to 2.6) |
| Other | 9/568 (1.6%) |
7/568 (1.2%) |
1.3 (0.5 to 3.5) |
| *Includes contacts in nursery (one case), neighbourhood (one case, three controls), nurse's house (one case), friend's house (one case), or hostel (one case). | |||
| Confined setting - Acute diarrhoea was not associated with being in a confined setting such as a nursery, school, or retirement home (table 1). The presence of a child 2 years of age or underin the household was, however, associated with an almost twofold increased risk of acute diarrhoea (1.9; 1.3 to 2.7). Patients who had recently been in contact with a person with diarrhoea were 4.3 times more likely to develop acute diarrhoea (3.1 to 6.0). The risk of acute diarrhoea was greatest after exposure to a person with diarrhoea in the household (5.0; 3.4 to 7.3). The distribution of the time interval between the date of onset of diarrhoea in exposed cases and the date of onset of diarrhoea in a contact (reported by 181 of the 241 cases exposed to a person with diarrhoea) ranged from 0 to 18 days, with a median time of 2.0 days (fig 2). (4) |  | |
| Fig 2: Distribution of surrogate incubation time (onset of diarrhoea in case minus that in contact) among 181 cases who reported recent contact with person with diarrhoea and date of onset of diarrhoea for contact. Median time = 2 days. |
Antibiotics - Treatment with antibiotics was associated with acute diarrhoea in the month before consultation (1.7; 1.1 to 2.4). An increased risk of acute diarrhoea was associated with penicillins (1.6; 1.0 to 2.6) and cephalosporins (1.8; 0.9 to 3.8) (table 1).
Multivariate analysis - The consumption of clams and tap water, attendance at a retirement home, the presence of a child of 2 years of age or under at home, contact with a person with diarrhoea (dummy variables for household contact, contact in the workplace, and contact elsewhere), and treatment with penicillins, cephalosporins, and macrolids were introduced in the initial multivariate model. In the final model, contact with a person with diarrhoea either in the household or in the workplace, the presence of a child 2 years of age or under at home, or treatment with penicillins and cephalosporins were independently associated with acute diarrhoea (table 2).
| Table 2 - Association of risk factors with acute diarrhoea in conditional logistic regression model | |
| Risk factor | Adjusted odds ratio (95% CI) |
|---|---|
| Contact with person with diarrhoea: | |
| No contact | 1.0 |
| In household | 5.0 (3.4 to 7.3) |
| In workplace | 3.1 (1.6 to 6.3) |
| Other* | 2.7 (1.2 to 5.8) |
| Child 2 years of age or under at home | 1.6 (1.1 to 2.4) |
| Oral penicillins | 1.9 (1.1 to 3.3) |
| Oral cephalosporins | 2.5 (1.1 to 5.9) |
| *Includes contacts in retirement home, school, and other setting. | |
Discussion
The large size of the samples in our study (568 pairs of cases and controls) implies that the consumption of shellfish was not important in the winter epidemics of acute diarrhoea in France from 26 December 1995 to 31 January 1996. Although an epidemic of acute diarrhoea due to the consumption of raw oysters was described in a coastal area in France during the last week of 1992,(10) the statistical power of our study in detecting any effect of eating raw oysters was estimated after the study as 90% (for a minimal relative risk of 1.5). This rules out the consumption of raw oysters as a cause of the national epidemic in 1996. Even after stratification for coastal areas, our results remained quite similar (data not shown). Neither consumption of shellfish nor consumption of tap water was associated with acute diarrhoea. Exposure to a contact with diarrhoea within the past 10 days, particularly within a household, was identified as the main risk factor for acute diarrhoea. The median incubation time for acute diarrhoea was estimated to be two days. Living with a child of 2 years of age or under without knowing that he or she had diarrhoea (adjusted odds ratio 1.6) and exposure to antibiotics within the month before consultation (odds ratio 1.7) were associated with the occurrence of acute diarrhoea; penicillins and cephalosporins were identified as independent risk factors, with adjusted odds ratios of 1.9 and 2.5. In both univariate and multivariate analyses the recent intake of macrolids was not associated with the occurrence of acute diarrhoea.
Although reports have shown a risk of acute diarrhoea within confined environments such as nurseries, schools, and retirement homes,(11, 12) this was not the case in this study. There may be a selection bias with attendance at nursery and school as children with diarrhoea are more likely to stay at home.
Validity of study
The difference observed between cases and controls in the place of
consultation may indicate a selection bias for the controls. This
should not, however, have had much of an impact on the other results as
the place of consultation is related more to the state of the patient
than to the cause of his or her disease. In fact, cases are more likely
to request a home visit because of their acute condition compared with
controls who may have less impairment. The presence of a child of 2
years of age or under in the household made no difference as to whether the
controls were seen by their doctor at home or if they went to the
surgery.
Of the cases included in the study, 92% met the criteria of the World Health Organisation for acute diarrhoea - that is, three or more stools passed for at least 24 hours.(13) For controls the observed rate of consumption of tap water (66.4%) was consistent with that estimated from a representative sample of the population in early 1995 (64.4%).(14)
Most cases of acute diarrhoea are likely to be caused by infection.(15) Given that stool cultures were not requested in this study, we could not identify the causative organism.
A French phenomenon?
Although, both in the United Kingdom and in the United
States,(16,17) more people are admitted to hospital for
diarrhoea during the winter than at other times of the year, no
definite seasonal pattern is reported from these countries for
morbidity in the general population.
Antibiotic intake
Diarrhoea is a common side effect of treatment with antibiotics
and has been reported in up to 20% of people taking
antibiotics.(18) The main antibiotics implicated in
diarrhoea are ampicillins, amoxicillins, cephalosporins, and
lincosamids.(8)(19) Our results confirm the risk of acute
diarrhoea related to penicillins (including ampicillins and
amoxicillins) and to cephalosporins but not to macrolids. We could not
evaluate the role of lincosamids in the occurrence of acute
diarrhoea - for example, clindamycin is well recognised in diarrhoea
caused by Clostridium difficile.(8) In
addition, lincosamids are rarely prescribed by general practitioners in
France, so this antibiotic family was not reported for any case or
control.
Conclusions
Our results indicate that the winter outbreaks of acute diarrhoea
in France are caused primarily by a person coming into contact with
someone with diarrhoea, implying a viral aetiology, and not by the
consumption of shellfish or tap water.
| Key Messages | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
We thank the doctors who participate in the Sentinelle scheme
who provided data for the survey, and Dr Laurent Beaugerie, Pierre-Yves
Boëlle, Dr Sylvie Chevret, Marguerite Guiguet, Professor Henry
Tuckwell, and Professor Alain-Jacques Valleron for their ongoing
support and critical review of the manuscript.
Funding: This study was supported by a scholarship to LL from
the Fondation de France.
Conflict of interest: None.
INSERM Unit 444,
Institut fédératif
Saint-Antoine de Recherches sur la Santé,
Paris,
France
Laurent Letrilliart, primary care physician
Antoine Flahault, head of Sentinelle system
Réseau National de Santé Publique,
Saint-Maurice,
France
Jean-Claude Desenclos, head of
infectious diseases unit
Correspondence to: Dr Letrilliart
email: letrilli@b3e.jussieu.fr
References 1 Garthright W E, Archer D L, Kvenberg J E. Estimates
of incidence and costs of intestinal infectious diseases in the United
States. Public Health Rep 1988;103:107-15.
2 Fourquet F, Drucker J. Communicable disease surveillance: the
Sentinelle network. Lancet 1997;349:794-5.
3 Flahault A, Chauvin P, Massari V, Carrat F, Farran N, Retel O,
et al. Epidémiologie des maladies transmissibles en médecine
libérale: Bilan du réseau 'Sentinelles' en 1995. Bull
Epidemiol Hebd 1996;33:143-5.
4 Flahault A, Garnerin P, Chauvin P, Farran N, Saidi Y, Diaz, et
al. Sentinelle traces of an epidemic of acute gastroenteritis in
France. Lancet 1995;346:162-3.
5 Morse D J, Guzewich J J, Hanrahan J P, Stricof R, Shayegani M,
Deibel R, et al. Widespread outbreaks of clam- and oyster-associated
gastroenteritis. Role of Norwalk virus. N Engl J Med
1986;314:678-81.
6 Météo France. Bulletin climatique annuel
1995. Toulouse, France: Météo France, 1996.
7 Laursen E, Mygind O, Rasmussen B, Ronne T. Gastroenteritis: a
waterborne outbreak affecting 1600 people in a small Danish town.
J Epidemiol Community Health 1994;48:453-8.
8 Bartlett J C. Antibiotic-associated diarrhea. Clin
Infect Dis 1992;15:573-81.
9 Hosmer D W, Lemeshow S. Applied logistic regression. New York:
Wiley, 1989.
10 Daurat G. Une épidémie de gastroentérite aiguë á
virus Norwalk-like liée á la consommation d'huîtres dans
l'Hérault, décembre 1992. Bull Epidemiol Hebd
1994;37:170-1.
11 Thompson S C. Infectious diarrhoea in children: controlling
transmission in the child care setting. J Paediatr Child
Health 1994;30:210-9.
12 LeBaron W, Furutan N P, Lew J F, Allen J R, Gouvea V, Monroe S S.
Viral agents of gastroenteritis. MMWR Morb Mort Wkly Rep
1990;39:1-24.
13 World Health Organisation. The management and prevention
of diarrhoea. Practical guidelines. 3rd ed. Geneva,
Switzerland: WHO, 1993.
14 Dufour A. Opinions des FranÍcais sur l'environnement
et appréciations sur l'eau du robinet. Paris: Centre de
Recherche pour l'Etude et l'Observation des Conditions de vie, 1995.
15 Watson A J. Diarrhoea. BMJ 1992;304:1302-4.
16 Ryan M J, Ramsay M, Brown D, Gay N J, Farrington C P, Wall P G.
Hospital admissions attribuable to rotavirus infection in England and
Wales. J Infect Dis 1996;174(suppl 1):S12-8.
17 Glass R I, Kilgore P E, Holman R C, Jin S, Smith J C, Woods P A, et
al. The epidemiology of rotavirus diarrhea in the United States:
surveillance and estimates of disease burden. J Infect
Dis 1996;174(suppl 1):S5-11.
18 Surawicz C M, Elmer G W, Speelman P, McFarland L V, Chinn J, van
Belle G. Prevention of antibiotic-associated diarrhea by Saccharomyces
boulardii: a prospective study. Gastroenterology
1989;96:981-8.
19 George W L. Antimicrobial agent-associated colitis and diarrhea:
historical background and clinical aspects. Rev Infect
Dis 1984;6(S1):S208-13.
Home | Current issue | Past issues | Classified ads | Career Focus | Feedback
Collections | About this site | About the BMJ | BMA | Medline