A cohort study of enteric campylobacter infection in children from birth to two years in Bangui (Central African Republic)

122 Trw-is~crm~s

OF THE ROYAL SOCIETY OFTROPICALMEDICINE AND HYGIENE (1990) 84, 122-125

A cohort study of enteric campylobacter infection two years in Bangui (Central African Republic) M. C. George+Courbot*, Pasteur de Ban@,

Institut

Republic

Abstract A cohort of 111 children from Bangui, Central African Republic, was followed for enteric campylobatter infection from birth until the aae of 2 vears. Stools were examined at each episode-of diarrhoea, and bi-weekly up to the age of 6-months irrespective of the nresence of diarrhoea. 349 enisodes of diarrhoeai illness were recorded (1.6 per child-year). Campylobacters were isolated from 41 (11.7%) of the 349 episodes, but in half of them another enteric pathogen was also isolated. Campylobacters were statistically associatedwith diarrhoea only before the age of 6 months. Bi-weekly sampling up to this age detected 75 infections (1.3 ner child-vear). vet onlv 12 (16%) were associated‘with diarrhoea. d&&ylob&ter coli was isolated slightly more often (51%) than C. jejuai (49%); biotyping and serogrouping showed that no strain was especially associatedwith disease.Fewer children who had campylobacter infection before the age of 6 months suffered campylobacter diarrhoea between 6 and 24 months of age than those who did not, but the difference did not reach statistical significance. A significantly higher rate of isolation was found in the homes of infected children (human and animal contacts) than of non-infected children. Campylobacter infections were statistically associated with the presenceof live poultry and the lack of piped water in homes. Introduction Enteric Campylobacter speciesare more prevalent in developing than the developed countries (BLASER& RELLER, 1981; BLASERet al., 1983; PAL & NAIR, 1984). Several authors have reported the existence of many healthy excretors in developing countries such as Bangladesh (BLASERet al., 1980; GLASS et al., 1983), South Africa (BOKKENHEUSER et al., 1979), Mexico (CALVA et al., 1988; CRAVIOTO et al., 1988) and Liberia (MBLBAK et al., 1988). We obtained similar results in Bangui in 1981-1982 in a study of infantile diarrhoea (GEORGESet al., 1984). In a previous work, we reported the results of a prospective study of campylobacter infections in a cohort of 127 children followed from birth to 6 months of age in Bangui (GEORGES-COURE~OT et al., 1987). Of these, 111 were followed up until 2 years of agefor campylobacter infections. We present here the final results of this study, which lasted from October 1983 to June 1987. Materials and Methods A cohort of 111 children born in a maternity ward in the city of Bangui, Central African Republic ‘Author for correspondence.

from birth to

A. M. Cassel-Beraud, I. Gouandjika, J. Monges and A. J. Georges

Central African

Population studied

in children

(Maternid des Castors) was followed until 2 years of age. All the children were from parents living in the neighbourhood of Castors. Most of the people in that district live in small closed houses, at street level, with narrow passageways,and animals roam free 111the streets and homes. Few people have piped municipal water supplies in the home; most of them take water from either wells or public fountains. At the beginning of the study, 223 children were enrolled, but onlv 111 could be followed during the 2 first vears of lifer 112were removed from the &hort either because they moved or because of lack of parental cooperation. All the children were monitored by a nurse visiting them at home twice a week until the age of 6 months and then once a week until they were 2 years old. Data collection and samgling

Episodes of diarrhoea (defined as the passageof at least 3 liquid stools per day) were recorded for each child, and faecal sampleswere examined for bacterial enteropathogens, rotavirus and eukaryotic parasites. In addition, up to the age of 6 months, faecal samples were taken twice a week, irrespective of the presence of diarrhoea, and examined for campylobacters and rotavirus. After that age, for each child with diarrhoea, a specimen was taken from a child in the same age group within the cohort who had not had diarrhoea within the preceding 14 d. The case definition of an episode of campylobacter diarrhoea was one in which the first campylobacter isolation coincided with the onset of diarrhoea, even if the child showed prolonged excretion after recovery from the diarrhoea. Household studies

Once or twice weekly, we looked for campylobacters in the faeces of household contacts (adults and children) of children infected with campylobacter and in animals and water samples from their homes. As a control, the samewas done in the homes of children with negative stool cultures and without diarrhoea recorded during the preceding 3 d. The mean age of the household diarrhoea contacts and non-diarrhoea contacts was similar. Laboratory methods Faecal sbecimens.For the isolation of camnvlobac-

ters faecai specimens were plated on sheep blood Columbia axar (Oxoid. Basinnstoke, UK) SUDplemented w?th Butxler ‘antibio& supplement (virion, Cham? Switzerland) (GOOSENSet al., 1983). Plates were mcubated at 42°C in a candle jar for 48 h. Identification was based on growth at 42°C on the selective medium, colony morphology, bacterial morphology in Gram-stained smears, motility in dark-

123 infection in children with and without diarrhoea is shown in Table 2. Rotaviruses are frequently detected in diarrhoeal stools until the age of 12 months, but much less frequently thereafter. The incidence of rotaviruses in control children was low even before the ageof 12 months and none was detected after that age.

field preparations, catalase and oxidase production. Speciesidentification and biotyping were carried out according to the schemeof LIOR (1984). Other enteric pathogens were detected by standard methods. Enterotoxigenic Eschtichia coli (ETEC) LT-producing strains were detected by enzyme-linked immunosorbent assay(ELISA), and ST-producing strains by the infant mouse assay. Rotaviruses were detected by ELISA. Water samples. For the detection of campylobacters one-litre samples of water were filtered through membranes of 0.45 pm pore size which were then placed in Oosterom’s enrichment medium (OOSTEROM et al., 1981). After 48 h incubation at 42°C (with caps loosened in a candle jar), one drop was plated in the same way as faeces. Campylobacter serogrouping.Serogrouping was done by slide agglutination according to the scheme of LIOR et al. (1982). The antisera were prepared and provided by H. Lior (Laboratory Centers for Disease Control, Ottawa, Canada); 3 of the serawere prepared against strains isolated in Bangui.

Incidence of cam&obacter

Results Incidence of diawhoea and of pathogens other than campyiobacters

Sequential campylobacter infections

Up to the age of 6 months, 75 episodesof infection (with and without diarrhoea) were detected in 48 children of the cohort of 111 (1.3 episodes per child-year): 28 had one episode, 14 had two episodes, 5 had three, and 1 child had 4 episodesof infection. In 37 of these episodesexcretion of campylobacters was prolonged, but in 9 of 26 cases studied sequential isolates were of different serogroups-up to 4 in one child (an episodewas defined as a period during which campylobacters were isolated in consecutive stools, even if different strains were found). Episodes of

Over the 2 years of the survey 349 episodes of diarrhoea were recorded in 104 children, giving a mean of 1*57f 1.13 diarrhoea episodesand 12-l+ 10.6 days of diarrhoea per child-year. The diarrhoea rate increased slighly up to the age of 18 months but then declined (Table 1). Only 7 children did not have diarrhoea during the first 2 years of their lives. The isolation rates of enteropathogens other than campylobacters and rotavirus from diarrhoeal stools are shown in Table 1. The incidence of rotavirus Table 1. Episodes of diarrhoea to 2 years of age

and eat&c

diawhoea

Campylobacter isolation rates are shown in Table 2. Campylobacters were isolated from 41 (11.7%) of the 349 diarrhoeal stools tested; 21 (51%) of them had another enteric pathogen in addition to the campylobatter. These 41 episodesof diarrhoea associatedwith campylobacter were distributed in 34 (3 1%) of the 111 children in the cohort, 7 having 2 episodes before 24 months of age. In children below the ageof 6 months, the campylobacter isolation rate from diarrhoeal stools was significantly higher than from non-diarrhoeal stools, but after that age there was no significant association with diarrhoea. No consistent seasonaltrends were seen during the 4 years of the study.

pathogens other than campylobacter

and rotavirus

isolated fmm cohort of 111 children from birth

Diarrhoea

of episodes of diarrhoea Number

Age (months) o-6

7-12 13-18 19-24

rate (episodes/ child-year) EPEC’ 1.6

90 94 98 67

1.7 1.8 1.2

8 (8.9)b 11 (11.7) 47 (7.1) (6.0)

ETEC’

Number of episodes from which pathogen isolated Slll~llil Shigella E. histolytica Gimdia

4 (4.4) 4 (4.3) 2l(1.5) (2.0)

l(1.1)

0 (0)

0 0l(1.0)

2 (2.1) 3 (3.1) 2 (3.0)

0 1 (1.1) 10 (10.2) 11 (16.4)

A. lumbrimides

1 (1.1) 8 (8.5) 14 (14.3) 12 (17.9)

0 3 (3.2)

10 (10.2) 12 (17.9)

“EPEC=enteropathogenic E. coli; ETEC=enteroroxigenic E. coli. bNnmbers in parentheses are percentages.

Table 2. Campylobacters and rotaviruses in children with and without diarrhoea studied from birth

Campylobacter Children with diarrhoea Age (months) E2 13-18 1%24 O-24 “Numbers ‘Numbers ‘NS=not

n=lO4 No. of episodes No. positive z z; 349

:; g;j” 12 (12.2) 4 (6.) 41 (11.7)

Control No. of tests

children No. positive

5073 152 73 147 5445

127 (2.5) 17 (11.2) 3 (4.1) 15 (10.2) 162 (3.0)

of episodes and tests are the same as those for campylobacter. in parentheses are percentages.

significant.

P

Children with diarrhoea No. positive

to the age

Rotavirus” Control children No. positive

-co.05 %’ NS

of

:i I::::{

‘Y I3

: I:::; 30 (8.6)

fl 31 (0.6)

2 years

P
124 Table

3. Presence of campylobacters

in the homes of infected

Homes of infected children No. tested No. positive Human contacts Animals Water Homes

988 16

68 (6.9)

1;:

: ‘X 45’(33.6)

and non-infected

chidren

Homes of non-infected children No. tested No. positive 361 ‘8 (4.4) 2: 57 l&17.5)

P b

ii NS -co*05

aNumbers in parentheses are percentages. bNS=not significant. ‘Any sample (human, animal, water) positive. Table 4. Incidence of campylobacter infection in relation to the presence of domestic animals and municipal piped water supply io the home

Exposurein home Poultry

Exposed Unexposed Casein home” No casein home Casein home” No casein home 20 :t 28 :; 11 4:: 4; :: 62

Dogs Goats No piped water 1 14 “Home with a child having at least one enteric campylobacterinfection before 6 months of age. bRelativerisk (95% confidenceinterval). ‘Signi6cant at 0.05 level. infection occurring more than 1 month apart were always with different strains. Only 12 (16%) of the 75 infections were associated with diarrhoea. Fewer of the children who had a campylobacter infection before 6 months of age (8 of 48: 16%) suffered campylobacter diarrhoea between 6 and 24 months of age than children who had no such early infection (16 of 63: 25%). However, this difference was not statistically significant: the relative risk (RR) of getting a symptomatic infection after the age of 6 months in the two groups was 1.70 (95% confidence interval [CI] 0*65-4*18). Similarly, only 1 (8%) of 12 children who had symptomatic infection before the age of 6 months had subsequent campylobacter diarrhoea, compared with 23 of 99 (23%) who had asymptomatic or no infection, but again the difference did not reach statistical significance (RR 2.98; 95% CI 0.36-12.48). Bacteriology of campylobacter isolates

The species, biotype and Lior serogroup of 88 isolates obtained from 35 diarrhoeal and 53 control stools were defined. Just over half of them (51%) were C. coli, representing both biotypes (I and II) and 16 serogroups, the most frequent of which were LIO 28, 29, 21 and 2. The remaining 49% of isolates were C. jejwti, mainly of biotypes I and II; 19 serogroups were represented, LIO 4 accounting for 19% of C. jejuni isolates. No biotype or serogroup was more frequently associated with diarrhoea than the others. Clinical features of campylobacter infection

Of the 41 children with diarrhoea associatedwith campylobacter, 14 had fever, 25 had vomiting and 14 had light dehydration. Only 2 children presented with severe &hydration needing intravenous rehydration; both were also infected by rotavirus. The children with campylobacter diarrhoeas were treated with oral erythromycin (40 mg/kg body weight) daily and by

RR(95% c1y 2.5 (1.08b5.54) 0.64 (0.28- 1.51) 5.63 (0.63-27.8) 13.42 (1.65-52.1)

oral rehydration when dehydration was not severe.All the children recoveredcompletely without evidence of malabsorption or malnutrition. Household surveys

Cultures were prepared in 134 homes of children infected with campylobacter and in 57 homes of non-infected children (Table 3). Campylobacters were isolated from at least one sample in the homes of 33.6% of the infected children and 17.5% of noninfected children. This difference is statistically significant (P
The incidence of 1.6 episodes of diarrhoea per child-year in Bangui was lower than in many developing countries, where rates of at least 3 per child-year are usual. Yet enteric campylobacter infections are frequent: 48 (43.2%) of a cohort of 111 children presented at least one infection before six months of age and 34 (30.6%) had diarrhoea associated with campylobacter infection before 2 years of age. This rate is probably underestimated as isolation was done in a candle jar, which is less effective than gas mixtures containing less oxygen (GOO~SENS et al., 1983). As we showed previously (GEORGES-COURBOT et al., 1987), campylobacter infections were clearly associated with diarrhoea before the age of six

125 months. The present study shows that after that age,

the association was lost; moreover in more than half of the caseswhere campylobacters were associatedwith diarrhoea, another enteric pathogen was present. We also found that the incidence of camtwlobacter diarrhoea was lower in children who had had an infection (symptomatic or not) before the age of 6 months than in children who did not have one, although the numbers tested were too low to reach statistical significance. CALVA et al. (1988) found in Mexico that an early symptomatic infection protected against subsequent symptomatic infection. The early isolation of campylobacter in the stools of children and the high frequency of symptomless excretion indicate high rates of transmission from birth. It is notable that in our previous survey (GEORGESCOURBOT ei al., 1987) all 9 infections detected under the ageof 1 month were symptomless, suggesting that maternal antibodies protected from invasive infection. In Bangui half of the campylobacter isolates were C. coli (GEORGES-COIJBBOT et al.. 1986). This unusually high proportion of C. coli is paralleled only in Yugoslavia (POPOVI&UROI~,1989); in most parts of the world only 5-10% of isolates are C. coli. Contrary to a previous study in 1982-1984, in which serogrouping by haemagglutination of campylobacters isolated from children less than 15 years old showed one serotype (PEN 37/56) to be highly associated with diarrhoea (GEORGES-COURBOTet al., 1986), in the present study no strain showed such an association. We found that the homes of children infected with campylobacter were more likely to contain another person or an animal with campylobacter infection than homes of uninfected children. Moreover, we found a significant association between the presence of live po&ry in the home and the rate of campylobatter isolation. This was also found bv GBADOS et al. (1988) in Peru. Few chickens could be tested for campylobacters, yet one-third of them were positive. Similarly campylobacter isolation rates were significantly higher in families lacking a municipal water supply in the home. The fact that only one campylobatter isolation was made from water does not negate this tinding! as the organism may be present intermittently and m numbers that are difficult to detect by culture. The main conclusion from this study is that although campylobacters do cause diarrhoea in early childhood in a developing community, most infections are inapparent. They are also complex, as several strains were commonly found witbin a brief period in a child both with and without diarrhoea. Acknowledgements

We thanks Dr Lior for providing sera for campylobacter serotyping. This study was funded by a grant from the World Health Organization (Diarrhoeal Diseases Control Programme). We also thank Fred La Sor (US Embassy, Bangui) and Dr M. B. Skirrow for helping in the preparation of the manuscript. References

Blaser, M. J., Taylor, D. N. & Feldman, R. A. (1983). Epidemiology of Campylobacter jejuni infections. Epidemiologic Reviews, 5, 157-176. Bokkenheuser, V. D., Richardson, N. J., Bryner, J. H., Roux, D. J., Schutte,A. B., Koornhof, H. J,.,Freitnan, I. & Hartman, E. (1979). Detection of entertc campylobacteriosis in children. Journal of Clinical Microbiology, 9, 227-232.

Calva, J. J., Ruiz-Palacios, G. M., Lopez-Vidal, A. B., Ramos. A. & Boialil. R. (1988). Cohort studv of intestinal infection’ with campylobacter in Mexican children. Lancer, i, 503-505. Cravioto, A., Reyes, R. E., Ortega, R., Femandez, G., Hemandez, R. & Lopez, D. (1988). Prospective study of diarrhoeal diseasein a cohortof rural Mexican children: incidence and isolated pathogens during the first two years of life. Epidemiology and Infection, 101, 123-134. Georges, M. C., Wachsmuth, I. K,., Meunier, D. M. Y., Nebout, N., Didier, F., Siopathis, M. R. & Georges, A. J. (1984). Parasitic, bacteriai, and viral enteric pathogens associatedwith diarrhoea in the Central African Republic. Joumal of Clinical Microbiology, 19, 571-575. Georges-Courbot, M. C., Baya, C., Beraud, A. M., Meunier, D. M. Y. & Georges, A. J. (1986). Distribution and serotypes of Campylobacter jejuni and Campylobatter coli in enteric Campylobacter strains isolated from children in the Central African Republic. 3ournul of Clinical Microbiology,

23, 592-594.

Georges-Courbot, M. C., Beraud-Cassel, A. M., Gouandjika, I. & Georges, A. J. (1987). Prospective study of enteric Campylobacter infections in children from birth to 6 months in the Central African Republic. Jownul of Clinical Microbiology,

25, 836-839.

Glass, R. I., Stall, B. J., Huq, M. I., Struelens, M.J., Blaser, M. & Kibriya, A. K. M. G. (1983). Epidemiologic and clinical features of endemic Campylobacter jejuni infection in Bangladesh. Joumal of Znfectious Diseases, 148, 292-296.

Goossens!H., De Boeck, M. & Butxler, J. P. (1983). A new selective medium for the isolation of Campylobacter jejuni from human faeces. European Journal of Clinical Microbiology, 2, 389-394.

-

Grados. 0.. Bravo. N.. Black. R. E. & Butzler. I. P. (1988). Paediatric campylobacter &arrhoeafrom household exposure to live chickens in Lima, Peru. Bulletin of the World Health Organization,

66, 369-374.

Lior, H. (1984)., New exended biotyping scheme for Campylobacferrlqunr, Campylobqcter coli, ,an< ,Yampylok&,,“ridu 3ownal of Clamcal Mzc~ooaology, 20,

Lior, H., Woodward, D. L.,. Edgar, J. A., Laroche? L. J. & Gill, P. (1982). Serotypmg of Campylobacter lejuni by slide agglutination based on heat-labile antigenic factors. @~rnal

of Clinical Microbiology,

15, 761-768.

Mlbak, K., Hojlyng, N. & Gaarslev, K. (1988). High prevalence of campylobacter excretors among Liberian children related to environmental conditions. Epidemiology and Znfection, 100, 227-237.

Oosterom, J., Vereijken, M. J. G. M. & Engels, G. B. (1981). Campylobacter isolation. Veterinary Quarterly Review, 3, 104. Pal, S. C. & Nair, G. B. (1984). Epidemiology of campylcbacteriosis in developing countries. Indian Council of Medical Research Bulletin, 14, (unnumbered pages). Propovic-UroiC, T. (1989). Campylobacw jejuni and Campylobacter coli diarrhoea in rural and urban populations in Yugoslavia. Epidemiology and Infection, 102, 59-67.

Blaser, M. J. & Reller, L. B. (1981). Campylobucrerenteritis. New England Journal

of Medicine, 305, 1444-1452.

Blaser, M. J., Glass, R. I., Huq, M. I., Stoll, B., Kibriya, G. M. & Alim, A. R. M. A. (1980). Isolation of Campylobacter fetus subsp. jejuni from Bangladeshi children. 3oumal of Clinical Microbiology, 12, 744-747.

Received 26 October 1988; revised 2June 1989; accepted 4 July 1989

fm publication