IJE Advance Access originally published online on May 31, 2008
International Journal of Epidemiology 2008 37(4):805-815; doi:10.1093/ije/dyn093
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
A hierarchical model for studying risk factors for childhood diarrhoea: a case–control study in a middle-income country
1 Instituto de Saúde Coletiva, Federal University of Bahia, Salvador, Brazil.
2 Escola Bahiana de Medicina e Saúde Pública, Salvador, Brazil.
3 Department of Pediatrics, School of Medicine, Federal University of Bahia, Salvador, Brazil.
4 London School of Hygiene and Tropical Medicine, London, UK.
5 Infectious Diseases Epidemiology Unit, London School of Hygiene and Tropical Medicine, London, UK.
* Corresponding author. Federal University of Bahia, Instituto de Saúde Coletiva, 40.110-170, Salvador, Bahia, Brazil. E-mail: mauricio{at}ufba.br
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Objective To identify factors associated with diarrhoea occurrence in children in a city in a middle-income country, with high access to water and sanitation.
Methods A case–control study in the city of Salvador, north-eastern Brazil was conducted from November 2002 to August 2004. The study population consisted of children presenting at a health facility. A total of 1688 cases of diarrhoea and 1676 controls were selected. Data collection was by a questionnaire and structured observation during home visits. The explanatory variables were grouped according to a conceptual model defined previously. Analysis was done using a hierarchical approach, to provide a more dynamic view of the transmission characteristics of childhood diarrhoea. Non-conditional logistic regression was used, and odds ratio and population-attributable fractions were estimated.
Results Socioeconomic factors contributed most to determining diarrhoea occurrence, followed by interpersonal contact, while factors related to food preparation, the environment and water and sanitation made a smaller contribution.
Conclusion The findings indicate that the transmission of diarrhoea is influenced by factors from all hierarchical levels, with interpersonal transmission playing a relatively higher role than previously thought. This is compatible with a predominance of viruses and other agents spread by interpersonal routes including Shigella, Giardia and Cryptosporidium. Diarrhoea control strategies in similar settings (middle-income countries in which a large proportion of the population has access to water and sanitation) must give greater emphasis to policies geared towards reducing person-to-person transmission for the prevention of diarrhoea.
Keywords Diarrhoea, urban, transmission, person-to-person, risk factors, children, Brazil
Accepted 30 April 2008
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Despite significant reductions in mortality caused by diarrhoea over recent decades,1–4 diarrhoea continues to be a significant health problem in developing countries, particularly among children. Reductions in diarrhoea mortality have been attributed to general improvements in living conditions, including better nutritional situations, access to medical care, increased vaccine coverage, increased coverage of potable water and sewerage systems and the growing use of oral rehydration therapy.5,6 The evidence indicates that the reduction was more marked in mortality than morbidity7 but, despite this decrease in mortality, in developing countries, around 1.8 million children aged under 5 years continue to die because of diarrhoea every year.7
In Brazil, significant reductions in diarrhoea mortality took place over the last two decades.8 This decreasing pattern of mortality has also been observed in the city of Salvador.9 More recently, an evaluation of a large sanitation programme the city used to compare two longitudinal studies conducted among pre-school children conducted in the periods 1997–99 and 2003–04, (therefore, before and after the intervention), found a 26% reduction in the prevalence and 11% reduction in incidence of diarrhoea.10 In developed countries, mortality among children due to diarrhoea is rare, and the impact of the disease is very often measured in terms of its costs to the health system.11
Although the majority of diarrhoea in children is caused by an infectious agent, the web of determinants for diarrhoea in children is complex and the relative contribution of each factor varies as a function of complex interactions between the different aetiological agents and several other factors. These factors could be grouped as socioeconomic,12 environmental (those that facilitate faecal-oral transmission of infectious agents),13–15 related to contacts (those that increase the chance of contact and person-to-person transmission of infection, such as crowding or high density of housing)16 and associated with microbiological contamination of food (such as inadequate cooking or time temperature abuse). Factors can also be classified according to their place in a chain of causation, or a causal network as distal (for example, socioeconomic conditions that lead to lack of a refrigerator) or proximal (for example, consumption of food that was not kept in a refrigerator), and according to whether they are long lasting (for example, access to adequate sewage disposal) or transitory (for example, contact with someone with diarrhoea), although of course ‘long lasting’ factors may change in time, for example as a result of improvements in the water supply or sewerage network.
The objective of this study was to investigate the set of factors associated with the occurrence of diarrhoea (presented to the health services) among children aged up to 10 years living in a large city in a middle-income country. At the time of this study the city was undergoing a process of rapid expansion of its sewerage network, reaching >60% of its population at the end of the study (from 25% in 1997); piped treated water was available to 96.7% of homes from the start. Our investigation strategy aimed to explore the relative role of these factors.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Study design and population
This study was conducted among children of up to 10 years of age (120 months) who were living in the city of Salvador, Bahia, north-eastern Brazil. Salvador has a population of
2.5 million inhabitants and is characterized by great social inequality. This was a case–control study, with information collected using a questionnaire, stool examination and home visits. Cases and controls were selected at five large outpatient clinics for children, located in different areas of the city; four of them were state funded and free at the point of use (SUS, the Brazilian National Health System) and one was private (which also attended state funded patients). The study was performed between November 2002 and August 2004. Cases were selected consecutively during the working period of each clinic. Cases were not excluded if they had taken antibiotics prior to being seen in the clinic. They were defined as children whose main reason to attend the clinic, according to the guardian, was diarrhoea, and whose guardian consented to their participation in the study. Controls were children presenting at the same clinics for reasons other than diarrhoea and included: (i) healthy children presenting to the well baby clinic and immunization services; (ii) children attending for mild, non-chronic diseases: orthopaedic procedures, pre- or post-operative assessments relating to small surgical procedures and dermatological problems known not to be related with hygiene and sanitation. Cases and controls who reported any episode of diarrhoea during the preceding 3 weeks and 3 months, respectively, were excluded. Controls were frequency matched to cases on two variables: age and how they paid for the consultation: either through SUS or through a private health plan. For cases aged under 60 months, six age groups were defined (0–5, 6–11, 12–23, 24–35, 36–47 and 48–59 months). Controls were selected to reflect the distribution of cases within the same age groups. For cases aged over 60 months, an acceptable age match spanned 1 year below and 2 years above, provided that cases and controls were within the range of 60–119 months.
Data collection
The person responsible for the child answered a standard, pre-coded questionnaire. This elicited information on socioeconomic conditions including age, race as reported by the mother, mother's and father's schooling levels, family income, employment situation and activity sector of the head of the family and goods owned (number of household goods owned by the family, including radio, fridge, stove, television, air conditioning, car and computer). The questionnaire also covered environmental conditions of the home (type of construction material, floor material, number and types of rooms, presence of a separate place used as the kitchen, characteristics of the water supply and water use, garbage collection and disposal and presence of animals), and the area around the house (visible presence of wastewater, presence of faeces and garbage in surrounding areas, among others). The questionnaire also asked for information on the child's health (vaccinations, breast-feeding, birth weight, type of water offered, day care attendance and mother's age at childbirth, among others) and on transitory exposures during the 10 days before the beginning of the episode (for cases) or the interview (for controls), including time spent outside of the home, contacts with people with diarrhoea, consumption of foods or liquids different from those usually eaten or consumed outside of the home and absence from home of the person who usually looks after the child. For the full list of variables see Tables 1 and 2. For the cases, there were also questions about the diarrhoea episode (starting date, symptoms and treatment).
|
|
Approximately 1 week after the first contact, a home visit was made to each case or control. During this visit, complementary information on the characteristics of the home and surrounding environment were collected by direct observation.
Data analysis
The overall sample size was estimated with the aim of obtaining a large enough number of cases for separate analyses of diarrhoea cases associated with the most frequent aetiological agents. The present analysis addresses diarrhoea as a public health issue, without considering the aetiological diagnosis obtained in the stool examination, and included all 1688 cases and 1676 controls. Specific analyses for some aetiological agents will be presented in separate papers. At a precision of 95%, this study had an 80% power to detect odds ratio (OR) of 1.3 for risk factors with a frequency of 15% or more among the controls.
Statistical analysis was conducted in accordance with a predefined conceptual model (Figure 1). The model defined two hierarchical levels: the first (distal) level included only one block with all socioeconomic variables and the second (proximal) level included three blocks, one for each of the (proposed) most frequent routes for pathogen transmission in diarrhoea (environmental contamination, food preparation and person-to-person contact).
|
Univariate analysis and multivariate non-conditional logistic analysis were conducted, always keeping in the models the variables used in the frequency matching (age and how the consultation was paid for), as well as sex and the outpatient clinic where recruited. The variables that presented P
0.05 in the univariate analyses were included in the next stage, which was an intra-block multivariate analysis. Finally, the set of variables with P 0.05 from the multivariate analysis in each block was input to hierarchical analysis following the order defined in the conceptual framework. Level 1 variables were introduced into the hierarchical model first, and level 2 variables next since the effect of the variables at the distal level can be mediated by the variables at the proximal level. Variables in all blocks in level 2 were introduced together since we postulate that they operate at the same level. Because we were interested in the effect of the distal level variables (even if they are mediated by the proximal variables) our final estimate for the effect of distal variables is that before the introduction of proximal variables, whereas estimates of effect of the variables at the proximal level must be made after introducing the variables at distal level into the model, excluding variables with P > 0.05. The variables in the multivariate model within each level, and the variables at proximal level in the final model, were kept in the model when the corresponding P 0.05. Age was tested as a possible interaction variable by means of the likelihood ratio test (Wald test). The analyses were carried out using the STATA statistical package, version 9.0 (STATA Corporation, 2003). In addition to OR, the proportion of cases in each age group caused by each variable (population attributable fraction or PAF) was calculated, using the formula: [(% of exposed cases) x (OR –1)/OR]. For the distal variables, the PAF was calculated in two ways: using the OR for the level alone, to reflect the fraction attributed to the variable whether mediated by a proximal variable or not (this is presented in Table 3), and using the OR after including the proximal variables, to reflect the proportion of cases attributed to the effect of the variable, which is not mediated by the proximal variables included in the final model.
|
The protocol for this study was granted approval by the Research Ethics Committee of the Instituto de Saúde Coletiva, Federal University of Bahia. The children were only included in the study after the adults responsible for them had read and signed a statement of free and informed consent.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
The total population of this study was 3364 children (1688 cases and 1676 controls), of whom 1983 (59%) were <2–years-old, 944 (28.1%) were between 2- and 5-years-old and 435 (12.9%) were between 5- and 10-years-old. In 94% of the cases, three or more evacuations were reported on at least 1 day.
With the exception of race and cohabitation with the biological father, all the factors analysed were significantly associated with diarrhoea in the univariate analysis, (P 0.05) Living in a shack had the strongest association (OR = 3.5; 95% CI = 2.2–5.5). In the intra-block multivariate analysis, inadequate housing type remained with the highest OR (2.9; 95% CI = 1.7–4.7) (Table 1).
As shown in Table 2 the majority of environmental exposures were associated with diarrhoea in univariate analysis (with a P 0.05), with the exception of the variable relating to the presence of domestic animals in the house. In the intra-block multivariate analysis, the following variables remained with P 0.05: presence of faeces in the backyard, irregularity of water supply and two garbage-related variables: closeness to a garbage dump and irregular collection of garbage. Among the exposures related to food preparation, the absence of the person who routinely prepared the food, the lack of a room exclusively dedicated to be a kitchen (separate kitchen) in the home and the child having eaten outside the home were associated with diarrhoea in the univariate and intra-block multivariate analysis. Among the factors relating to person-to-person contact that were associated with diarrhoea, the following were significantly associated (P < 0.05) in the univariate and the multivariate analysis: contact with a person with diarrhoea, day care attendance and crowding at home.
The results from the final multivariate hierarchical analysis are presented in Table 3. In accordance with the sequence defined previously, the estimated values for socioeconomic variables are those in the final intra-block model, and for the proximal levels in the model including the distal level. The effects of housing type, sewage disposal, type of the toilet and regularity of the water supply were greater in children aged 3–8 years; the effect of day care attendance was greatest in children under 12-months-old (data not shown). The variables that increased the risk of diarrhoea by >50% were: shack-type housing (OR = 3.1), contact with people with diarrhoea (OR = 2.2), day care attendance (OR = 1.9) and absence of the person who routinely prepared the child's food (OR = 1.6). The variables with the strongest effect were not those responsible for the largest number of cases. The PAF for socioeconomic variables was 41%, and the large proportional contribution towards this was from low schooling level for the mother (17%) and ownership of less than four household items (17%). The changes after inclusion of variables in the next level were as follows (data not shown): the proportion of cases attributable to socioeconomic variables reduced from 41% to 35%; the PAF for the mother's schooling level reduced from 17% to 14% and the PAF for ownership of goods reduced from 17% to 13%. The block of variables relating to person-to-person contact presented a total PAF of 29%, predominantly made up of contact with people with diarrhoea in the previous 10 days, and day care attendance, with PAFs of 14 and 13%, respectively.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Our results show that a child was more likely to be a case of diarrhoea presenting to the health service (as measured by the ORs adjusted in an multivariate hierarchical model) if they (i) lived under more precarious socioeconomic conditions (in shack-type housing or in a family that did not own more than four items) and if their mother was young, had low schooling or worked outside the home; (ii) lived under environmental conditions that favoured faecal-oral transmission, either in public spaces, in the areas surrounding the home or within private spaces inside crowded homes (iii) in the previous 10 days had eaten outside of the home, or had their food prepared by someone other than the usual person (iv) in the previous 10 days had contact with someone with diarrhoea or were attending a day care centre. The estimates for ORs and PAFs made it possible to identify factors with a high risk (high OR) and those that contributed many cases to the overall diarrhoea burden (high PAF). Several of the exposures with a high OR were very rare, and so the PAF may be more useful for defining priorities for interventions that would have a greater impact in reducing the disease burden. Each of the following variables contributed >10% of the diarrhoea burden (as estimated by the PAF): low schooling of mother, ownership of less than four items, contact with someone with diarrhoea and day care attendance. Unfortunately, none of these is easy to change. In the present analysis, diarrhoea occurrence was influenced by factors from all hierarchical levels with interpersonal transmission playing a relatively higher role than previously thought.
The presence of variables from all the blocks defined in the conceptual model shows that, in the context studied, the occurrence of diarrhoea is a complex multifactorial process that is related to the precarious living conditions, but also related to transitory factors that occurred shortly before the episode. By building up a hierarchical model involving different causal levels and different blocks of factors that were potentially associated with diarrhoea, we present the relative importance of these various factors within a city in a developing country, with high coverage of treated water and sewer systems. Only multivariate techniques allow determination of the relative roles of these multiple factors. Hierarchical multivariate models go further, making it possible to include variables in the model at different levels of the causal chain, with the sequential introduction of variables from the distal to the proximal level, following a prior conceptual model.
The interpretation of conceptual models is a complex exercise and alternative explanations are often possible. In our results socioeconomic and environmental factors, factors relating to food handling and factors relating to contacts with the child were associated with diarrhoea occurrence. Various infectious agents (parasitic, bacterial or viral) are implicated in to the aetiology of diarrhoea. These have different transmission mechanisms and depend on complex social and environmental factors, particularly in large cities. The way in which these factors are presented and relate to each other in different contexts defines the relative importance of each factor in the causal chain, and has implications regarding the prevalent pattern of aetiological agents associated with diarrhoea.
Our results show that socioeconomic factors (level 1 of the conceptual model) were responsible for a large proportion of the diarrhoea burden, with PAF of 41% that reduced to 35% after controlling for the exposures studied (level 2) that covered the best known transmission routes for diarrhoea-related infectious agents (environmental contamination, food preparation and person-to-person contact). These results suggest that other factors must mediate the effect of socioeconomic variables. The proximal level explained a total of 49% of all cases. The PAFs for environmental and food exposures were not large (7 and 12%, respectively), while the PAF for factor relating to person-to-person contact was 29%: a third of all cases of diarrhoea in this population resulted for known exposures in the person-to-person contact block. This pattern, with a lower PAF for factors relating to environmental contamination, characterizes a profile of factors relating to diarrhoea occurrence that is close to what occurs in the context of developed countries,17 in which viral aetiological agents predominate over the bacterial agents; a context where the importance of direct contact between people has been consistently demonstrated. Thus, it has been shown that children attending day care centres are at greater risk of diarrhoea, especially those below the age of 3 years, as do children who, although looked after in their own homes, live together with other children in the house.18
This emphasizes the importance of direct contact between children for the transmission of pathogenic agents in these environments and accords with the growing importance of viral agents, and maybe other agents also spreading by interpersonal routes including Shigella, Giardia and Cryptosporidium, in the pathogenesis of diarrhoea in Salvador, particularly in its most severe forms. In a community study carried out in the years 2000 and 2001, entero-viruses were found in 17% of the stool samples from the diarrhoea cases examined, and half of these were rotavirus.19 On the other hand, pathogenic bacteria were found in 22% of the samples (<2% being Salmonella sp. or Shigella sp.) and diarrhoeagenic Escherichia coli in 11%, although its presence is not a guarantee that it is causing the diarrhoea episode as it is frequently found in healthy people.20 Also in Salvador, a hospital-based study among paediatric patients admitted to hospital with diarrhoea in the years 1999, 2000 and 2002 showed that, respectively, 22% (37/170), 28% (34/120) and 38% (137/358) presented rotavirus in stool.21 Giardiasis is also common in Salvador, where 13.7% of young children were found to be infected in a recent community-based study.22
As for food, we found an association between diarrhoea and the child having eaten outside of the home, and also an association with temporary absence from home of the person who usually prepared the child's food. These may be related to the quality of the care taken in preparing the food offered to the child, which was not measured in our study. There is evidence that food prepared at home carried a lower risk of diarrhoea, while the introduction of new foods to young children increases the risk of diarrhoea.16,23,24
The exposures traditionally associated with diarrhoea, such as deficient water supply and inadequate sewage disposal,25–27 did not appear in the final model. More than 95% of the homes in Salvador28 had access to potable piped water so even people who did not have access to good-quality piped water in their homes had ways of obtaining piped water in the neighbourhood. Irregularity in water supply (that may reduce the frequency of good hygiene practices like washing hands and objects, etc, or could even lead to direct contamination of the water)29,30 showed an association with diarrhoea in the intra-block multivariate model but did not enter in the final hierarchical regression model. The fact that excreta disposal did not appear in the final model is of great interest, given its history of strong association with diarrhoea. Moreover, in a longitudinal study carried out in Salvador,31 over a period immediately preceding the study now presented, access to a sewerage system was found to have a strong influence on the incidence of diarrhoea. Even though these two studies utilized different methodologies (longitudinal and case–control), one possible explanation for this difference may be the rapid expansion of the sanitation network in the city, which went from coverage of 25% of the homes in 1998 to >60% in 2003. This, together with other sanitation strategies such as septic tanks, covers the great majority of the city's population and 90% of our sample, taking both cases and controls into account, had adequate excreta disposal.10 Therefore, in our study, the environmental factors that were found to be associated with diarrhoea were the ones related to the presence of garbage dumps in the area surrounding the home, together with inadequate frequency of garbage collection and also the observation of the presence of faeces in the backyard of the home. The importance of garbage in transmitting diarrhoea in Salvador had been described previously.15
Even in underdeveloped urban contexts, social policies continually change a series of parameters affecting the daily lives of the populations. In epidemiological terms, this means that various risk factors have become less important as public health measures were implemented, with immediate consequences for the occurrence of different health problems. The findings presented here suggest that the epidemiology of diarrhoea is changing in the city of Salvador. For example, the risk factors identified do not include water and sanitation, traditional risk factors for diarrhoea, probably reflecting the investment in sanitation improvements that began at the end of the 1990s and has been consolidated in the 2000s. A previous study, also in the city of Salvador, has already indicated the importance of behavioural factors in the occurrence of diarrhoea.16 These points showed a need for the diarrhoea control strategies in contexts like the one studied here to consider the reduction of factors related with person-to-person contact. Finally, the relationship between diarrhoea and poverty in a context of such inequality as the city of Salvador needs to be further investigated, because despite the wide range of factors included in the model, the investigation still did not establish what mediates 35% of the contribution of poverty to the diarrhoea burden.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Financial support for this study was provided by Programa de Núcleos de Excelência (PRONEX, CNPq/MCT, Brazil), Contract no. 661086/1998-4.
Conflict of interest: None declared.
KEY MESSAGES
|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
1 Yach D, Strebel PM, Joubert G. The Impact of Diarrhoeal Disease on Childhood Deaths in the RSA, 1968–1985. S Afr Med J (1989) 76:472–75.[Web of Science][Medline]
2 El-Rafie M, Hassouna WA, Hirschhorn N, et al. Effect of diarrhoeal disease control on infant and childhood mortality in Egypt. Report from the National Control of Diarrheal Diseases Project. Lancet (1990) 335:334–38.[CrossRef][Web of Science][Medline]
3 Shaikh K, Wojtyniak B, Mostafa G, Khan MU. Pattern of diarrhoeal deaths during 1966–1987 in a demographic surveillance area in rural Bangladesh. J Diarrhoeal Dis Res (1990) 8:147–54.[Web of Science][Medline]
4 Moore SR, Lima AA, Schorling JB, Barboza MS Jr, Soares AM, Guerrant RL. Changes over time in the epidemiology of diarrhea and malnutrition among children in an urban Brazilian shantytown, 1989–1996. Int J Infect Dis (2000) 4:179–86.[CrossRef][Medline]
5 Bern C, Martines J, de Zoysa I, Glass RI. The magnitude of the global problem of diarrhoeal disease: a ten-year update. Bull World Health Organ (1992) 70:705–14.[Web of Science][Medline]
6 Victora CG, Bryce J, Fontaine O, Monasch R. Reducing deaths from diarrhoea through oral rehydration therapy. Bull World Health Organ (2000) 78:1246–55.[Web of Science][Medline]
7 World Health Organization. The World Health Report 2005, Make every mother and child count (2005) Geneva: WHO. Annex Table 3. Available at: http://www.who.int/whr/2005 (Accessed March 2008).
8 Carmo EH, Barreto ML, Silva JB Jr. Mudanças nos padrões de morbimortalidade da população brasileira: os desafios para um novo século. Epidemiologia e serviços de saúde (2003) 12:63–75.
9 Guimaraes ZA, Costa MC, Paim JS, Silva LM. [Decline and social inequalities of infant mortality caused by diarrhea]. Rev Soc Bras Med Trop (2001) 34:473–78.[Medline]
10 Barreto ML, Genser B, Strina A, et al. Effect of city-wide sanitation programme on reduction in rate of childhood diarrhoea in northeast Brazil: assessment by two cohort studies. Lancet, (2007) 370:1622–28.[CrossRef][Web of Science][Medline]
11 Thapar N, Sanderson IR. Diarrhoea in children: an interface between developing and developed countries. Lancet (2004) 363:641–53.[CrossRef][Web of Science][Medline]
12 Fuchs SC, Victora CG. Risk and prognostic factors for diarrheal disease in Brazilian infants: a special case-control design application. Cad Saude Publica (2002) 18:773–82.[Medline]
13 Yeager BA, Lanata CF, Lazo F, Verastegui H, Black RE. Transmission factors and socioeconomic status as determinants of diarrhoeal incidence in Lima, Peru. J Diarrhoeal Dis Res (1991) 9:186–93.[Web of Science][Medline]
14 Moraes LR, Cancio JA, Cairncross S, Huttly S. Impact of drainage and sewerage on diarrhoea in poor urban areas in Salvador, Brazil. Trans R Soc Trop Med Hyg (2003) 97:153–58.[CrossRef][Web of Science][Medline]
15 Rego RF, Moraes LR, Dourado I. Diarrhoea and garbage disposal in Salvador, Brazil. Trans R Soc Trop Med Hyg (2005) 99:48–54.[CrossRef][Web of Science][Medline]
16 Strina A, Cairncross S, Barreto ML, Larrea C, Prado MS. Childhood diarrhea and observed hygiene behavior in Salvador, Brazil. Am J Epidemiol (2003) 157:1032–38.
17 Ethelberg S, Olesen B, Neimann J, et al. Risk factors for diarrhea among children in an industrialized country. Epidemiology (2006) 17:24–30.[CrossRef][Web of Science][Medline]
18 Barros H, Lunet N. Association between child-care and acute diarrhea: a study in Portuguese children. Rev Saude Publica (2003) 37:603–8.[Web of Science][Medline]
19 Barreto ML, Milroy CA, Strina A, et al. Community-based monitoring of diarrhea in urban Brazilian children: incidence and associated pathogens. Trans R Soc Trop Med Hyg (2006) 100:234–42.[CrossRef][Web of Science][Medline]
20 Nataro JP, Kaper JB. Diarrheagenic Escherichia coli. Clin Microbiol Rev (1998) 11:142–201.
21 Santos N, Volotao EM, Soares CC, Campos GS, Sardi SI, Hoshino Y. Predominance of rotavirus genotype G9 during the 1999, 2000, and 2002 seasons among hospitalized children in the city of Salvador, Bahia, Brazil: implications for future vaccine strategies. J Clin Microbiol (2005) 43:4064–69.
22 Prado MS, Strina A, Barreto ML, Oliveira-Assis AM, Paz LM, Cairncross S. Risk factors for infection with Giardia duodenalis in pre-school children in the city of Salvador, Brazil. Epidemiol Infect (2003) 131:899–906.[CrossRef][Medline]
23 Rodrigues LC, Cowden JM, Wheeler JG, et al. The study of infectious intestinal disease in England: risk factors for cases of infectious intestinal disease with Campylobacter jejuni infection. Epidemiol Infect (2001) 127:185–93.[Medline]
24 Sobel J, Gomes TA, Ramos RT, et al. Pathogen-specific risk factors and protective factors for acute diarrheal illness in children aged 12–59 months in Sao Paulo, Brazil. Clin Infect Dis (2004) 38:1545–51.[CrossRef][Web of Science][Medline]
25 Checkley W, Gilman RH, Black RE, et al. Effect of water and sanitation on childhood health in a poor Peruvian peri-urban community. Lancet (2004) 363:112–18.[CrossRef][Web of Science][Medline]
26 Heller L, Colosimo EA, Antunes CM. Environmental sanitation conditions and health impact: a case-control study. Rev Soc Bras Med Trop (2003) 36:41–50.[Medline]
27 Fewtrell L, Kaufmann RB, Kay D, Enanoria W, Haller L, Colford JM Jr. Water, sanitation, and hygiene interventions to reduce diarrhoea in less developed countries: a systematic review and meta-analysis. Lancet Infect Dis (2005) 5:42–52.[CrossRef][Web of Science][Medline]
28 Instituto Brasileiro de Geografia e Estatística (IBGE). Pesquisa Nacional de Saneamento Básico. Ministério do Planejamento Orçamento e Gestão. (Accessed September 2006). Available at: http://www.ibge.gov.br.
29 Curtis V, Cairncross S. Effect of washing hands with soap on diarrhoea risk in the community: a systematic review. Lancet Infect Dis (2003) 3:275–81.[CrossRef][Web of Science][Medline]
30 Cairncross S. Water supply and sanitation: some misconceptions. Trop Med Int Health (2003) 8:193–95.[CrossRef][Web of Science][Medline]
31 Genser B, Strina A, Teles CA, Prado MS, Barreto ML. Risk factors for childhood diarrhea incidence: dynamic analysis of a longitudinal study. Epidemiology (2006) 17:658–67.[CrossRef][Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||