Bacterial contamination of the lacteal contents of feeding bottles in metropolitan São Paulo, Brazil.

Bacterial contamination of the lacteal contents of feeding bottles in metropolitan Sao Paulo, Brazil* T.B. Morais,1 M.B. Morais,2 & D.M. Sigulem3 Reported are the results of a study in Sao Paulo, Brazil, to evaluate the bacterial contamination of the lacteal contents of feeding bottles prepared in urban households of low (LSE) and high (HSE) socioeconomic groups, involving 100 and 32 mothers of infants, respectively. Samples of the lacteal contents of the feeding bottles were cultured and the medians (25th and 75th percentiles) of the counts (bacteria per ml) were significantly higher in the LSE group: mesophilic bacteria, 555000 (17250-4 350 000) in the LSE group and 1615 (20-500000) in the HSE group; coliforms, 2400 (19-150 000) in the LSE group and 7 (0-7800) in the HSE group. Escherichia coli was isolated from 26% (26/100) of the samples from the LSE group and from 6% (2/32) of those from the HSE group ( P = 0.03). In the HSE group, higher coliform counts were associated with foodhandlers other than the mother, lower levels of maternal education, the use ofpasteurized milk, and the addition of ingredients other than milk. In the LSE group, feeding bottles prepared using tap water and those prepared for infants aged over 6 months had higher coliform counts. In general, the feeding bottles prepared in the households studied were heavily contaminated, especially in the LSE group. Introduction Diarrhoea remains one of the commonest illnesses among infants, and is also one of the major causes of mortality in developing countries (1). Infants who are not breast-fed exhibit a higher incidence and severity of diarrhoeal illnesses, demonstrating the connection between infant feeding practices and the risk of diarrhoea (2). The relative risk of diarrhoea- associated death among completely weaned infantsa is 14 times higher than of their exclusively breast-fed counterparts (3). The introduction of complemen- tary foods is concomitant with an increased risk of diarrhoea (4, 5). Food, particularly milk, may serve as a medium for the multiplication of enteric bacte- rial pathogens and/or the production of enterotoxins (6-8), accounting for a substantial proportion of diarrhoeal diseases and associated malnutrition among infants. (9). Nevertheless, the importance of safe food preparation practices in the prevention of diarrhoeal diseases has often been neglected (10), * From: Escola Paulista de Medicina, Universidade Federal de Sao Paulo, Vila Clementino, Sao Paulo, Brazil. I Food Bromatology and Microbiology Laboratory. Requests for reprints should be sent to Ms Morais at Rua Araioses 75, 05442- 010 Sao Paulo, Brazil. 2 Chief, Pediatric Gastroenterology, Department of Pediatrics. I Coordinator, Post-graduate course in nutrition. a No longer being breast-fed but still being fed milk of some sort. Reprint No. 5838 even though numerous studies have demonstrated that complementary foods prepared under unhygienic conditions are frequently and grossly contaminated (4, 6-8,11-16). Most of the studies on contamination of com- plementary foods have evaluated rural communities (4, 6, 8, 11), where living conditions are quite differ- ent from those in metropolitan areas of developing countries. In the latter, the low-income population can benefit from easier access to basic infrastructures, as well as to education and health care, which could influence household food preparation practices. The present study investigated bacterial con- tamination in the lacteal contents of feeding bottles of infants belonging to families of high (HSE) and low (LSE) socioeconomic levels in the metropolitan area of Sao Paulo, Brazil. The relationships between contamination, socioeconomic level and feeding bottle preparation practices were also evaluated. Materials and methods The study was carried out in the metropolitan area of Sao Paulo (population, 15 million) in south-eastern Brazil, the most industrialized city in Latin America. Sao Paulo has adequate water and electricity sup- plies, good solid waste collection but deficient sewer- age. It also experiences the usual range of urban problems, including pollution, overcrowding, traffic- jams, crime, and unemployment. Bulletin of the World Health Organization, 1998, 76 (2): 173-181 © Worid Health Organization 1998 173 T.B. Morais et al. Study population In Brazil, socioeconomic class is frequently associ- ated with a particular type of health care system. Individuals from higher socioeconomic classes generally attend private clinics, those from the mid- dle classes often make use of various health insur- ance companies, while those from the lower classes usually attend public hospitals and outpatient clinics. The study used this tendency to define the two socio- economic groups. A total of 100 mothers who brought their infants to the outpatient clinic of the Department of Pediatrics, Universidade Federal de Sao Paulo, which provides free regular clinical checks, routine immunizations and health education to the low- income population, were interviewed and allocated to the LSE group. The HSE group comprised 32 mothers attending private paediatric clinics. The criteria for selection in both groups were that the child should not exceed 2 years of age and that the mother should be bottle-feeding her infant and be able to provide a feeding bottle for sampling. There were 47 boys and 53 girls in the LSE group, and 18 boys and 14 girls in the HSE group; the median (25th and 75th percentiles) age was respec- tively 7.0 (4.0-10.0) months and 5.5 (4.0-9.0) months (P = 0.26, Mann-Whitney test). The mean (+SD) of the birth weight was 3023 ± 547 g in the HSE group and 2857 + 625 g in the LSE group (P = 0.18, Student's t-test). The median age at weaning was 1.0 month for both groups. The median number of children per family was 2.0 for both groups. The mothers were requested to give specific de- tails of how they prepared the feeding bottles: the type of milk powder used, whether they boiled the water used to dilute the powder, whether they added other ingredients, and who prepared the feeding bot- tle. They were also asked about the method they used to wash and disinfect the bottles and teats. Background socioeconomic data were obtained on educational level, the type of building material of the dwelling, ownership of a stove and refrigerator, the availability of a bathroom, kitchen, water tap and sink, and access to drinking-water, sewerage, an electricity supply and waste collection. Socioeconomic characterization. Differences in the mother's education level were found between the groups, the medians being 6 years and 15 years of education in the LSE and HSE groups, respectively (P < 0.0001, Mann-Whitney test). A previous study in Sao Paulo demonstrated that incomes are markedly lower among those with a lower level of education (17). We estimated that the income of approximately 70% of the LSE group was <US$ 100 per capita per month, the level considered the mini- mum for a family in Sao Paulo (18). In the HSE group, about 85% had incomes >US$ 100 per capita per month; of these, 70% had incomes >US$ 200 per capita per month. Despite their low incomes, the families in the LSE group had reasonable housing conditions. More than 90% lived in brick-built dwellings, had a kitchen and a bathroom of their own, and facilities such as a sink, tap, and running water. All of them owned stoves, while 88% possessed a refrigerator. All the households in the LSE group had an electricity supply and more than 90% benefited from drinking-water and waste collection services. Only 50%, however, were connected to a sewerage sys- tem. These data agree with the results obtained in the previous study (17). Households in the HSE group had all the facilities and services included in the survey. Methods Sampling and microbiological procedures Samples of the lacteal content of the feeding bottles were collected in order to perform the microbiologi- cal analyses. None of the mothers had previously been informed that the samples would be collected, thus ensuring that the bottles had been prepared under the usual conditions. Samples were collected immediately before the contents of the bottles were to be eaten and were kept at <100C until they reached the laboratory, which was within 2 hours of sampling. The samples were analysed for numbers of mesophilic and coliform bacteria and Escherichia coli. All samples were processed in conformity with standard microbiological methods (19). Statistical methods The data were analysed using Epi Info 6.0 software (20) and Sigma Stat for Windows (21). The statistical analysis of the relationships between contamination and feeding bottle preparation practices was based only on coliform counts using the Mann-Whitney test (level of significance P < 0.05). Results Microbiological results In terms of the three indicator organisms, the lacteal contents of the feeding bottles of the LSE group were significantly more contaminated than those of WHO Bulletin OMS. Vol 76 1998174 Bacterial contamination of feeding bottles in Sio Paulo, Brazil the HSE group (Table 1). In the LSE group, most bottles were heavily contaminated: 42% had aerobic mesophilic bacterial counts >106/ml and 27% had >105/ml coliform bacteria. E. coli was isolated from 26% of the samples. The results from the HSE group were also strik- ing, but the proportion of heavily contaminated feeding bottles was lower than in the LSE group. Of the samples analysed, 19% presented mesophilic bacterial counts >106/ml (HSE group, 6/32 versus LSE group, 42/100; P = 0.03, x2 test); 12% had coliform bacterial counts >105/ml (HSE group, 4/32 versus LSE group, 27/100; P = 0.14, x2 test) and E. coli was recovered from 6% of the samples (HSE group, 2/32 versus LSE group, 26/100; P = 0.03, X2 test). Factors related to feeding-bottle contamination Ingredients. Milk powder was the commonest ingre- dient used in the preparation of feeding bottles by both groups, but the addition of sugar and other ingredients was statistically higher in the LSE group (sugar, 82/100 versus 7/32, P = 0.000,X2 test, other ingredients, 70/100 versus 8/32, P = 0.000, x2 test) (Table 2). The most usual ingredients added by the LSE group were sugar (82%), corn starch (36%), commercial cereal formulae (9%) and rice starch (9%). Some feeding bottles were prepared with more than one of these ingredients. For the HSE group the contaminant load of feeding bottles prepared with pasteurized milk was statistically significantly higher than that of bottles prepared with milk powder, but statistically there was no difference between the HSE and LSE groups for bottles prepared with pasteurized milk. In con- trast, feeding bottles prepared with milk powder were less contaminated in the HSE group. The addition of other ingredients was associated statis- tically with a higher level of contamination of feeding bottles in the HSE group. For the LSE group no statistically significant associations were found between the type of milk, the addition of other ingredients, and coliform counts. Food handling and preparation practices. In both groups, most feeding bottles were prepared by the mothers (see Table 3). However, for the HSE group, Table 1: Counts of mesophilic bacteria, coliform bacteria and Escherichia coli in the lacteal contents of feeding bottles in the LSE and HSE groups Mesophilic bacteria Coliform bacteria Escherichia coli LSE HSE LSE HSE LSH HSE No. of bacteria/ml (n) (n) (n) (n) (n) (n) 0 - - 19 14 74 30 (19.0)a (43.8) (74.0) (93.8) 1-10 2 5 2 4 3 1 (2.0) (15.6) (2.0) (12.5) (3.0) (3.1) 10-102 6 6 10 1 6 - (6.0) (18.8) (10.0) (3.1) (6.0) 102-103 5 5 11 1 8 - (5.0) (15.6) (11.0) (3.1) (8.0) 103-104 9 2 16 4 5 1 (9.0) (6.3) (16.0) (12.5) (5.0) (3.1) 104-105 20 4 15 4 1 _ (20.0) (12.5) (15.0) (12.5) (1.0) 105-106 16 4 11 3 2 - (16.0) (12.5) (1 1.0) (9.4) (2.0) 106-107 24 4 16 1 1 - (24.0) (12.5) (16.0) (3.1) (1.0) ¢107 18 2 - - - - (18.0) (6.3) 25th percentile 17250 20 19 0 0 0 Median 555000 1615 2400 7 0 0 75th percentile 4350000 500000 150000 7800 4 0 P-valueb 0.0001 0.0044 0.0178 a Figures in parentheses are percentages. b Mann-Whitney test. WHO Bulletin OMS. Vol 76 1998 175 T.B. Morais et al. Table 2: Median coliform counts (bacteria/ml) in the lacteal contents of feeding bottles, according to the ingredients used in the LSE and HSE groups LSE group HSE group n Coliform count n Coliform count P-valuea Type of milk Powder 74 2 400; 21 0; 0.000 (74.0)b 15-240000c (65.6) 0-2400 Pasteurized 26 2400; 11 2 420; 0.887 (26.0) 23-46000 (34.4) 9-46000 P-valuea 0.573 0.035 Added sugar Yes 82 4 600; 7 240; 0.232 (82.0) 23-180 000 (21.9) 0-24000 No 18 460; 25 4; 0.146 (18.0) 9-24 000 (78.1) 0-4 600 P-valuea 0.279 0.460 Other added ingredients Yes 70 4 600; 8 24 000; 0.785 (70.0) 23-240000 (25.0) 2305-78000 No 30 425; 24 0; 0.020 (30.0) 0-240000 (75.0) 0-1320 P-valuea 0.112 0.008 a Mann-Whitney test. b Figures in parentheses are percentages. c Figures in italics are the 25th and 75th percentiles. Table 3: Median coliform counts (bacteria/ml) in the lacteal contents of feeding bottles, according to bottle handler, boiling of pasteurized milk and use of tap water in the LSE and HSE groups LSE group HSE group n Coliform count n Coliform count P-valuea Bottle handler Mother 95 2 400; 25 2; 0.001 (95.0)b 15-150000c (78.1) 0-4600 Other 5 240; 7 4 600; 0.713 (5.0) 23-46000 (21.9) 240-110000 P-valuea 0.855 0.086 Boiling of pasteurized milk Yes 22 2400; 10 12 120; 0.822 (84.6) 23-24 000 (90.9) 9-110000 No 4 25 300; 1 - (15.4) 2300-3523 000 (9.1) P-valuea 0.411 Use of tap water Yes 14 165000; 0 - (18.9) 15-2400000 (0.0) No 60 1 100; 21 0; 0.002 (81.1) 15-46000 (100.0) 0-2400 P-valuea 0.084 a Mann-Whitney test. b Figures in parentheses are percentages. c Figures in italics are the 25th and 75th percentiles. WHO Bulletin OMS. Vol 76 1998176 Bacterial contamination of feeding bottles in Sio Paulo, Brazil a statistically higher frequency of other people (grandmothers, childminders, and maids) were in- volved in the preparation (7/32 versus 5/100, P = 0.008, x2 test). Feeding bottles not prepared by the mothers had a higher median coliform count in the HSE group (P = 0.086). Bottles prepared by the mothers in the HSE group were significantly less contami- nated than those prepared by the mothers in the LSE group. Boiling the pasteurized milk was a predominant practice in both groups. The use of untreated tap water to dilute the milk powder was observed for 18.9% of the feeding bottles from the LSE group. This practice gave rise to a higher level of contami- nation (P = 0.084) than when some type of treated water had been used. In this group, 66.2% boiled the water, 13.5% used filtered water and 1.4% used bot- tled water. Untreated tap water was not used by the HSE group: 64.0% boiled the water, 18% used filtered water and 18% used bottled water. Washing and disinfection procedures. Washing of bottles and nipples, using brushes and detergents and some type of disinfection, was common, al- though 24% of the bottles in the LSE group were not disinfected (see Table 4). The following methods of disinfection were mentioned: boiling, scalding, and the use of a sterilizing agent. Scalding (60%) and boiling (75%) were the most usual practices in the LSE and HSE groups, respectively. In both groups, the coliform counts of washed and disinfected bot- tles did not differ significantly from those of bottles not so treated, though the washed and disinfected bottles had a lower median coliform count. In the LSE group, even washed and disinfected bottles had a statistically higher level of contamination than those of the HSE group. Maternal education and children's birth order and age. No statistically significant associations were found in the LSE group between bacterial counts, on the one hand, and maternal education or birth order, on the other (Table 5). In this group, the feeding bottles of infants aged >6 months had a higher level of contamination (P = 0.067). In the HSE group, no statistically significant correlations were found between birth order or age, on the one hand, and bacterial counts on the other. In this group, the feeding bottles of infants whose mothers had had fewer than 15 years of education were more contami- nated (P = 0.078). Table 4: Median coliform counts (bacteria/ml) in the lacteal contents of feeding bottles, according to washing and disinfection procedures in the LSE and HSE groups LSE group HSE group n Coliform count n Coliform count P-valuea Use of detergent Yes 94 2 400; 27 4; 0.004 (94.0)b 15-180000c (84.4) 0-11000 No 6 1 430; 5 4600; 0.712 (6.0) 23-75000 (15.6) 0-4600 P-valuea 0.641 0.724 Use of brush Yes 99 2 400; 31 4; 0.003 (99.0) 15-150000 (96.9) 0-11000 No 1 - 1 - - (1.0) (3.1) P-valuea Disinfection Yes 76 2 400; 28 4; 0.004 (76.0) 19-60500 (87.5) 0-7800 No 24 13 200; 4 2 420; 0.338 (24.0) 19-780000 (12.5) 120-14 300 P-valuea 0.403 0.512 a Mann-Whitney test. b Figures in parentheses are percentages. c Figures in italics are the 25th and 75th percentiles. WHO Bulletin OMS. Vol 76 1998 177 T.B. Morais et al. Table 5: Median coliform counts (bacteria/ml) in the lacteal contents of feeding bottles, according to maternal education level, birth order and age of child in the LSE and HSE groups Maternal education Matemal education <Mediana -Mediana P-valued Birth order First child Second or subsequent child P-valued Age of child s6 months >6 months P-valued LSE group n Coliform count 58 (58.0)b 42 (42.0) 7 800; 14-460000c 2 400; 75-24 00Cc 0.461 44 4 450; (44.0) 167-460000 56 1 750; (56.0) 11-84000 0.253 46 1 750; (46.0) 14-24 000 54 7 800; (54.0) 23-460000 0.067 HSE group n Coliform count 10 4600; (31.3) 4-24 000 22 0 (68.7) 0-2400 0.078 13 43; (40.6) 0-24000 19 0 (59.4) 0-4600 0.186 19 0; (59.4) 0-4 600 13 9; (40.6) 0-24000 0.326 a LSE, median = 6.0 years; HSE, median = 15.0 years. b Figures in parentheses are percentages. c Figures in italics are the 25th and 75th percentiles. d Mann-Whitney test. Discussion Introduction of complementary foods to infants' diets has been associated with increased diarrhoeal morbidity and mortality (2, 3). Among these foods, milk has been associated with a higher degree and frequency of bacterial contamination (6-8, 13, 14). Most studies on contamination of complemen- tary foods have evaluated poor populations living in rural communities (4, 6, 8, 12, 13, 15) or in urban slums (7,16), where unfavourable conditions for food preparation prevail. The present study, on the other hand, was performed in a large metropolitan area where the socioeconomic and environmental condi- tions of the LSE group were apparently satisfactory. Most of the households had an electricity supply, drinking-water, bathroom, kitchen, refrigerator and stove. Maternal education level was considerably better than that reported in other studies (7, 12, 13). In the present study, 26% and 6% of the feeding bottles from the LSE and HSE groups, respectively, were faecally contaminated, whereas other investi- gators have reported proportions ranging from 49% to 100% (6-8, 14). This may suggest that better housing conditions and maternal education were related to improved microbiological quality of the feeding bottles, especially in the LSE group. These results are in agreement with those of a study conducted in India (7), which demonstrated a higher level of contamination in the feeding bottles of the low-income group. This suggests that comple- mentary foods represent a higher risk factor among populations of low socioeconomic level. Contaminated ingredients and water, together with inadequate preparation, washing and disinfec- tion of the utensils used to prepare feeding bottles, are traditionally considered to be risk factors for contamination (14, 22). Other factors linked to so- cioeconomic conditions, such as maternal education, access to drinking-water and sewerage, as well as the availability of a sink, water tap, stove and refrigera- tor, may also be correlated with the microbiological quality of the feeding bottles. False beliefs and taboos, and traditional kitchen practices and food preferences, may contribute to a lack of safety in food preparation (10). In our study, we did not find false beliefs or taboos, but different kitchen practices and food pref- erences were observed in the two groups evaluated and these could be associated with the microbiologi- WHO Bulletin OMS. Vol 76 1998178 Bacterial contamination of feeding bottles in Sao Paulo, Brazil cal quality of the feeding bottles. In the HSE group, a significantly higher proportion of feeding bottles were prepared without the addition of other ingredients to the milk compared with the LSE group. Also, in the HSE group, people other than the mothers were more frequently involved in feeding- bottle preparation. The use of untreated tap water to dilute milk powder was exclusive to the LSE group. The variety of preparation practices and the different kinds of ingredients employed in the pre- paration of feeding bottles may have precluded identification of statistical associations between the probable risk factors of contamination and coliform counts. Nevertheless, the risk factors for contamina- tion differed according to socioeconomic level. In the HSE group, the use of pasteurized milk rather than milk powder was associated with a higher de- gree of coliform contamination, which may reflect the poor microbiological quality of the pasteurized milk sold in Sao Paulo. A previous study showed that 53% of 40 samples had coliform counts above the Brazilian standard of 10 coliforms/ml (23). This could be due to poor temperature control at the retail outlet rather than inadequate pasteurization. The addition of cereals and starches to the milk in the bottles was also correlated with a higher coliform count in the LSE group, but the addition of sugar was not associated with a higher level of contamination in either group. The use of untreated tap water was associated with higher coliform counts, despite the good bacterial quality of the water supplied to the city of Sao Paulo; this suggests that the water was inadequately stored or was not from the authorized public supply. In the HSE group, a higher coliform count was observed when the feeding bottles were prepared by people other than the mother; these people may have had a lower level of education. In the HSE group, there was a tendency for feeding bottles prepared by mothers who had less than 15 years of education to be more contaminated. In contrast, in the LSE group the mothers' educational level had no effect on the degree of contamination of feeding bottles. Birth order was not correlated with a higher level of contamination in either group. The feeding bottles of infants aged >6 months in the LSE group tended to be more contaminated, probably reflecting less care by the mother after the child's first 6 months of life. Even after washing with detergents and dis- infection, the feeding bottles of the LSE group showed a higher degree of contamination than those of the HSE group. The results also show that improvements in en- vironmental conditions and a better maternal educa- tion on their own were not associated with a lower risk of household food contamination. The existence of a basic social infrastructure in the LSE group was not enough to ensure that feeding bottles had the same microbiological quality as those prepared in the HSE group. Theoretically, the LSE group presented the minimum conditions and knowledge necessary for the safe preparation of feeding bottles. Nevertheless, contradictions could be detected between the state- ments made in the interviews and the high contami- nation levels of the feeding bottles. This could be due to one of the following hypotheses. First, unknown or undetected factors could have led to the poor microbiological quality of the feeding bottles in the LSE group. Second, although mothers may have reported using adequate preparation procedures, mainly those related to washing and disinfection, they did not use them effectively. In surveys, there is a tendency to overreport practices that are perceived to be "good" (24). In the LSE group, it may be necessary to study the preparation of feeding bottles in a systematic way to identify the factors that lead to contamina- tion. In this respect, use of the hazard analysis critical control point (HACCP) system (22) could be of value. This consists in identifying the hazards associ- ated with any stage of food production, processing or preparation, the assessing of the related risk, and the determination of the operations where control procedures will be effective (25). Once information on socioeconomic and cultural background and on food habits and preparation is known, appropriate educational programmes could be implemented to improve the safety of complementary foods prepared at home and to prevent feeding-bottle contamination (10, 26). A multidisciplinary approach would certainly broaden understanding about these problems and allow health education to be more effective. This may not be an easy task, however, since bringing about behavioural changes is frequently difficult (27). Acknowledgements Ms T.B. Morais thanks the Fundacao de Amparo a Pesquisa do Estado de Sao Paulo for the award of a scholarship. Resume Contamination bacterienne des complements lactes pour biberons dans la ville de Sio Paulo, Bresil L'introduction de compl6ments alimentaires pour nourrissons est associ6e a un risque augmente de WHO Bulletin OMS. Vol 76 1998 179 T.B. Morais et al. diarrhee. La pr6sente etude s'est attachee a 6tudier la contamination bact6rienne des produits lact6s utilis6s pour les biberons par des families appar- tenant aux couches socio-economiques favoris6es et defavorisees de la ville de Sao Paulo, Br6sil. La plupart des 6tudes ant6rieures ont ete effectu6es dans des communaut6s rurales. On a interrog6 au total 100 meres ayant amene leur nourrisson dans un service public de soins ambulatoires et appartenant a des groupes d6- favoris6s. Le groupe des meres appartenant a un milieu favoris6 comptait 32 personnes ayant con- sult6 dans des services de p6diatrie prives. La duree moyenne de scolarisation des meres 6tait respectivement de 6 ans et de 15 ans dans les deux groupes (P<0,0001). Malgre leur faible revenu, les families du groupe defavoris6 6taient correctement log6es. On a recueilli des 6chantillons du contenu des biberons pour faire une num6ration des micro- organismes indicateurs-bacteries mesophiles, coli- formes et Escherichia coli. La m6diane des numerations (25e et 75e per- centiles) (bact6ries/ml) a e sensiblement plus elev6e chez le groupe defavorise: 555000 (17250- 4350000) bacteries m6sophiles dans ce groupe, contre 1 615 (20-500000) dans le groupe favorise; 2400 (10-150000) coliformes dans le groupe de- favoris6, contre 7 (0-7 800) dans le groupe favorise. On a isol6 E. coli dans 26% (26/100) des 6chan- tillons du groupe d6favorise, contre 6% (2/32) de ceux du groupe favoris6 (P=0,03). Dans le groupe defavorise, 42% des 6chantillons pr6sentaient un nombre de colonies m6sophiles >106/ml et 27% plus de 105 bacteries coliformes/ml. La proportion de biberons fortement con- tamin6s a ete plus faible dans le groupe favorise que dans le groupe d6favorise: 19% des echantil- Ions du premier ont presente un nombre de bact6ries m6sophiles superieur a 106/ml (soit 6 femmes sur 32, contre 42 sur 100 dans le groupe defavoris6; P=0,03); et 12% plus de 105 coliformes/ml (soit 4 femmes sur 32, contre 27 sur 100: P=0,14). Dans le groupe favoris6, la charge de contami- nants des biberons prepares avec du lait pasteurise a ete significativement plus 6lev6e sur le plan statistique que celle des biberons prepar6s avec du lait en poudre. L'adjonction d'autres ingredients a 6te egalement statistiquement associ6e a un plus fort degr6 de contamination chez les femmes favorisees. Dans ce meme groupe, les biberons qui n'6taient pas pr6par6s par les meres montraient un nombre median de coliformes plus important (P=0,086). Les biberons prepares par les meres dans ce groupe 6taient sensiblement moins contamin6s que ceux pr6par6s par les meres du groupe d6favoris6. Dans ce dernier groupe, meme les biberons lav6s et d6sinfect6s ont montr6 un degre de contamination statistiquement plus 6lev6. Toujours dans le groupe d6favoris6, les biberons des nourrissons de plus de 6 mois etaient davantage contamines que ceux des enfants des femmes favorisees (P=0,067). Dans ce dernier groupe, les biberons appartenant a des nourrissons dont les meres avaient ete scolaris6es pendant une duree inferieure a 15 ans montraient une contami- nation plus importante (P=0,078). References 1. Grant JP. State of the world's children. New York, Oxford University Press, 1989. 2. Jason JM et al. Mortality and infectious disease asso- ciated with infant-feeding practices in developing countries. Pediatrics, 1984 (Suppl): 702-726. 3. Victora CG et al. Evidence for protection by breast- feeding against infant deaths from infectious diseases in Brazil. Lancet, 1987, 2: 319-322. 4. Rowland MGM et al. Bacterial contamination in tradi- tional Gambian weaning foods. Lancet, 1978, 1: 136- 138. 5. Guerrant RL et al. Prospective study of diarrheal illnesses in Northeastern Brazil: patterns of disease, nutritional impact, etiologies and risk factors. Journal of infectious diseases, 1983, 148: 986-997. 6. Black RE et al. Contamination of weaning foods and transmission of enterotoxigenic Escherichia coli diar- rhoea in children in rural Bangladesh. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1982, 76: 259-264. 7. Mathur R, Reddy V. Bacterial contamination of infant foods. Indian journal of medical research, 1983, 77, 342-346. 8. Vadivelu J et al. Enterotoxigenic Escherichia coli in the domestic environment of a Malaysian village. Epidemiology and infection, 1989, 103: 497-51 1. 9. Esrey SA. Food contamination and diarrhoea. World health, January-February 1990: 19-20. 10. Motarjemi Y et al. Contaminated weaning food: a major risk factor for diarrhoea and associated malnu- trition. Bulletin of the World Health Organization, 1993, 71: 79-92. 11. Echeverria P et al. Identification by DNA hybridisa- tion of enterotoxigenic Escherichia coli in homes of children with diarrhoea. Lancet, 1984, 1: 63-66. 12. Imong SM et al. The bacterial content of infant weaning foods and water in rural Northern Thailand. Journal of tropical pediatrics, 1989, 35: 14-18. 13. Imong SM et al. Maternal behaviour and socioeco- nomic influences on the bacterial content of infant weaning foods in rural Northern Thailand. Journal of tropical pediatrics, 1995, 41: 234-240. 14. Michanie S et al. Critical control points for foods prepared in households in which babies had 180 WHO Bulletin OMS. Vol 76 1998 Bacterial contamination of feeding bottles in Sio Paulo, Brazil salmonellosis. International joumal of food micro- biology, 1987, 5: 337-354. 15. Henry FJ et al. Bacterial contamination of weaning foods and drinking-water in rural Bangladesh. Epidemiology and infection, 1990, 104: 79-85. 16. Molbak K et al. Bacterial contamination of stored water and stored food: a potential source of diarrhoeal disease in West Africa. Epidemiology and infection, 1989, 102: 309-316. 17. Monteiro CA et al. [Study of the health of children in metropolitan Sao Paulo, SP (Brazil), 1984-1985. methodological aspects, socioeconomic aspects, and physical environment]. Revista de saude publica, 1986, 20: 435-445 (in Portuguese). 18. Departamento Intersindical de Estatisticas e Estudos Socio-Econ6micos. Custo de Vida. Boletim DIEESE, 1994, no. 164: 104 (in Portuguese). 19. Standard methods for the examination of dairy products, 15th ed. Washington, DC, American Public Health Association, 1985. 20. Dean AG et al. Epi-Info, version 6.0: a word process- ing, database, and statistics program for epidemiology on microcomputers. Atlanta, GA, Centers for Disease Control, 1994. 21. Fox E. User's manual Sigma Stat: statistical software for windows. Germany, Jandel, 1994. 22. International Commission on Microbiological Specifications for Foods. Microorganisms in foods. 4. Application of the Hazard Analysis Critical Control Point (HACCP) system to ensure microbiological safety and quality. Oxford, Blackwell Scientific Publication, 1988. 23. Baruffaldi R et al. [Hygiene-sanitary condition of type B pasteurized milk sold in the city of Sao Paulo, SP (Brazil) in the period February to August 1982]. Revista de saude publica, 1984, 18: 367-374 (in Portuguese). 24. Curtis V et al. Structured observations of hygeine behaviours in Burkina Faso: validity, variability, and utility. Bulletin of the World Health Organization, 1993, 71: 23-32. 25. Application of the hazard analysis critical control point (HACCP) system for the improvement of food safety. WHO-supported case studies on food prepared in homes, at street vending operations, and in cottage industries. Geneva, World Health Organization, 1993 (unpublished document WHO/FNU/FOS/93.1, available upon request from Division of Food and Nutrition, World Health Organization, 1211 Geneva 27, Switzerland). 26. Bryan FL et al. Hazards analyses of foods prepared by migrants living in a new settlement at the outskirts of Lima, Peru. Journal of food protection, 1988, 51: 314-323. 27. Rasmuson MR et al. Communication for child sur- vival. Washington, DC, U.S. Agency for International Development, 1988. WHO Bulletin OMS. Vol 76 1998 181