• View in gallery

    Histogram with diarrhea longitudinal prevalence as number of diseased days per 1,000 child-days of follow up shows data for children who had at least one episode of diarrhea. A diarrhea-positive day is defined as a day with two or more loose stools and/or blood and mucus in the stool. Disease days are defined based on caretaker response in weekly logs. Disease numbers are normalized to the number of days of follow-up per child (range = 117–133 days).

  • View in gallery

    Structural equation model of diarrhea describing the relationship between drinking water quality (DWQ), environmental quality EQ), household hygiene (HH), and food hygiene (FQ) indicators (risk categories) and longitudinal diarrhea prevalence (outcome variable). Each risk category is measured by four observed risk factors (refer to Table 2 for a detailed list of the indicators from which these were selected). All error terms and correlations are omitted to simplify the representation. The beta values of relationships estimated using a dataset collected from children less than five years of age in Jakarta, Indonesia, during January–July 2011 is given next to the arrows depicting these relationships. Covariates, such as poverty, location within the city, education of the head of household, and child age are also incorporated into the model. The estimated beta values are shown only for significant relationships. Level of significance is shown by using an asterisk next to the beta values as defined below. * Significant to 0.10; ** Significant to 0.05.

  • 1.

    Liu L, Johnshon HL, Cousens S, Perin J, Scott S, Lawn JE, Rudan I, Campbell H, Cibulskis R, Li M, 2012. Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000. Lancet 379: 21512161.

    • Search Google Scholar
    • Export Citation
  • 2.

    Lerman SJ, Shepard DS, Cash RA, 1985. Treatment of diarrhea in Indonesian children: what it costs and who pays for it. Lancet 2: 651654.

  • 3.

    Agtin MD, Soeharno R, Lesmana M, Punjab NH, Simanjuntak C, Wangsasaputra F, Nurdin D, Pulungsih SP, Rofiq A, Santoso H, Pujarwoto H, Sjahrurachman A, Sudarmono P, von Seidlein L, Deen JL, Ali M, Lee H, Kim DR, Han O, Park JK, Suwandono A, Ingerani, Oyofo BA, Campbell JR, Beecham HJ, Corwin AL, Clemens JD, 2005. The burden of diarrhoea, shigellosis, and cholera in north Jakarta, Indonesia: findings from 24 months surveillance. BMC Infect Dis 5: 89.

    • Search Google Scholar
    • Export Citation
  • 4.

    Lima AA, Guerrant RL, 1992. Persistent diarrhea in children: epidemiology, risk factors, pathophysiology, nutritional impact, and management. Epidemiol Rev 14: 222243.

    • Search Google Scholar
    • Export Citation
  • 5.

    WHO/UNICEF, 1985. The Management of Diarrhoea and Use of Oral Rehydration Therapy. Geneva: World Health Organization.

  • 6.

    Grantham-McGregor S, 1995. A review of studies of the effect of severe malnutrition on mental development. J Nutr 125: 2238S.

  • 7.

    Brown KH, 2003. Diarrhea and malnutrition. J Nutr 133: 32853325.

  • 8.

    Moore SR, Lima NL, Soares AM, Oria RB, Pinkerton RC, Barrett LJ, Guerrant RL, Lima AA, 2010. Prolonged episodes of acute diarrhea reduce growth and increase risk of persistent diarrhea in children. Gastroenterology 139: 11561164.

    • Search Google Scholar
    • Export Citation
  • 9.

    Niehaus MD, Moore SR, Patrick PD, Derr LI, Lorntz B, Lima AA, Guerrant RL, 2002. Early childhood diarrhea is associated with diminished cognitive function 4 to 7 years later in children in a northeast Brazilian shantytown. Am J Trop Med Hyg 66: 590593.

    • Search Google Scholar
    • Export Citation
  • 10.

    Morriss S, Cousens SN, Kirkwood BR, Arthur P, Ross DA, 1996. Is prevalence of diarrhea a better predictor of subsequent mortality and weight gain than diarrhea incidence? Am J Epidemiol 144: 582588.

    • Search Google Scholar
    • Export Citation
  • 11.

    Bernd G, Strina A, Teles CA, Prado MS, Barreto ML, 2006. Risk factors for childhood diarrhea incidence. Dynamic Analysis of a Longitudinal Study Epidemiology 17: 658667.

    • Search Google Scholar
    • Export Citation
  • 12.

    Fewtrell L, 2005. Water, sanitation, and hygiene interventions to reduce diarrhoea in less developed countries: a systematic review and meta-analysis. Lancet Infect Dis 5: 4252.

    • Search Google Scholar
    • Export Citation
  • 13.

    Aulia H, Surapaty SC, Bahar E, Susanto TA, Roisuddin, Hamzah TM, 1994. Personal and domestic hygiene and its relationship to the incidence of diarrhoea in South Sumatra. J Diarrhoeal Dis Res 12: 4248.

    • Search Google Scholar
    • Export Citation
  • 14.

    Blum D, Feachem RG, 1983. Measuring the impact of water supply and sanitation investments on diarrhoeal diseases: problems of methodology. Int J Epidemiol 12: 357366.

    • Search Google Scholar
    • Export Citation
  • 15.

    Esrey SA, Habicht J-P, 1986. Epidemiologic evidence for health benefits from improved water and sanitation in developing countries. Epidemiol Rev 8: 117129.

    • Search Google Scholar
    • Export Citation
  • 16.

    Esrey SA, Potash JB, Shiff RC, 1991. Effects of improved water supply and sanitation on ascariasis, diarrhoea, dracunculiasis, hookworm infection, schistosomiasis, and trachoma. Bull World Health Organ 69: 609621.

    • Search Google Scholar
    • Export Citation
  • 17.

    Semenza JC, Roberts L, Henderson A, Bogan J, Rubin CH, 1998. Water distribution and diarrheal disease transmission: a case study in Uzbekistan. Am J Trop Med Hyg 59: 941946.

    • Search Google Scholar
    • Export Citation
  • 18.

    Arnold BF, Colford JM, 2007. Treating water with chlorine at point-of-use to improve water quality and reduce child diarrhea in developing countries: a systematic review and meta-analysis. Am J Trop Med Hyg 76: 354364.

    • Search Google Scholar
    • Export Citation
  • 19.

    Sima L, Desai MM, McCarty KM, Elimelech M, 2012. Relationship between use of water from community-scale water treatment refill kiosks and childhood diarrhea in Jakarta. Am J Trop Med Hyg 87: 979984.

    • Search Google Scholar
    • Export Citation
  • 20.

    Semba RD, 2009. Purchase of drinking water is associated with increased child morbidity and mortality among urban slum-dwelling families in Indonesia. Int J Hyg Environ Health 212: 387397.

    • Search Google Scholar
    • Export Citation
  • 21.

    Pickering H, Hayes RJ, Tomkins AM, Carson D, Dunn DT, 1986. Alternative measures of diarrhoeal morbidity and their association with social and environmental factors in urban children in The Gambia. Trans R Soc Trop Med Hyg 81: 853859.

    • Search Google Scholar
    • Export Citation
  • 22.

    Manun’ebo MN, Haggerty PA, Kalengaie M, Ashworth A, Kirkwood BR, 1994. Influences of demographic, socioeconomic and environmental variables on childhood diarrhoea in a rural area of Zaire. J Trop Med Hyg 97: 3138.

    • Search Google Scholar
    • Export Citation
  • 23.

    Genser B, Strina A, Teles CA, Pardo MS, Barreto ML, 2006. Risk factors for childhood diarrhea incidence: dynamic analysis of a longitudinal study. Epidemiology 17: 658667.

    • Search Google Scholar
    • Export Citation
  • 24.

    Pruss A, Kay D, Fewtrell L, Bartram J, 2002. Estimating the burden of disease from water, sanitation, and hygiene at a global level. Environ Health Perspect 110: 537542.

    • Search Google Scholar
    • Export Citation
  • 25.

    Eisenberg JN, Scott JC, Porco T, 2007. Integrating disease control strategies: balancing water sanitation and hygiene interventions to reduce diarrheal disease burden. Am J Public Health 97: 846853.

    • Search Google Scholar
    • Export Citation
  • 26.

    Esrey SA, Feachem RG, 1989. Interventions for the Control of Diarrhoeal Disease among Young Children: Promotion of Food Hygiene. Geneva: World Health Organization.

    • Search Google Scholar
    • Export Citation
  • 27.

    Esrey SA, Habicht J-P, Casella G, 1992. The complementary effect of latrines and increased water usage on the growth of infants in rural Lesotho. Am J Epidemiol 135: 659667.

    • Search Google Scholar
    • Export Citation
  • 28.

    Barreto ML, Genser B, Strina A, Teixeira MG, Asssis AM, Rego RF, Teles CA, Prado MS, Matos SM, Santos DN, dos Santos LA, Cairncross S, 2007. Effect of city-wide sanitation programme on reduction in rate of childhood diarrhoea in northeast Brazil: assessment by two cohort studies. Lancet 370: 16221628.

    • Search Google Scholar
    • Export Citation
  • 29.

    Santos CA, Strina A, Amorim LD, Genser B, Assis AM, Prado MS, Barreto ML, 2012. Individual and contextual determinants of the duration of diarrhoel episodes in preschool children: a longitudinal study in an urban setting. Epidemiol Infect 140: 689696.

    • Search Google Scholar
    • Export Citation
  • 30.

    Fahrmeir L, Raach A, 2007. A Bayesian semiparametric latent variable model for mixed responses. Psychometrika 72: 327346.

  • 31.

    MacDonald RP, Ho RM, 2002. Principles and practice in reporting structural equation analyses. Psychol Methods 7: 6482.

  • 32.

    Badan Pusat Statistik, 2004. National Socio-Economic Survey (NSES). Jakarta, Indonesia: SUSENAS.

  • 33.

    Wright JA, Gaundry SW, Conroy R, Wood D, Du Perez M, Ferro-Luzzi A, Genthe B, Kirimi M, Moyo S, Mutisi C, Ndamba J, Potgieter N, 2006. Defining episodes of diarrohoea: results from a three-country study in sub-Saharan Africa. J Health Popul Nutr 24: 816.

    • Search Google Scholar
    • Export Citation
  • 34.

    Du Perez M, Conroy RM, Wright JA, Moyo S, Potgieter N, Gundry SW, 2008. Use of ceramic water filtration in the prevention of diarrheal diseases: a randomized controlled trial in rural South Africa and Zimbabwe. Am J Trop Med Hyg 70: 696701.

    • Search Google Scholar
    • Export Citation
  • 35.

    Levine MM, Caplan ES, Waterman D, Cash RA, Hornick RB, Snyder MJ, 1977. Diarrhea caused by Escherichia coli that produce only heat-stable enterotoxin. Infect Immun 17: 7882.

    • Search Google Scholar
    • Export Citation
  • 36.

    Kaplan JE, Schonberger LB, Varano G, Jackman N, Bied J, Gary GW, 1982. An outbreak of acute nonbacterial gastroenteritis in a nursing home. Am J Epidemiol 116: 940949.

    • Search Google Scholar
    • Export Citation
  • 37.

    Laffan AM, McKenzie R, Forti J, Conklin D, Marcinko R, Shrestha R, Bellantoni M, Greenough WBI, 2011. Lactoferrin for the prevention of post-antibiotic diarrhoea. J Health Popul Nutr 29: 547551.

    • Search Google Scholar
    • Export Citation
  • 38.

    Banks MR, Farthing MJ, 2012. Diarrhea. Hawkey CJ, Bosch J, Richter J, Garcia-Tsao G, Chan F, eds. Textbook of Clinical Gastroenterology and Hematology. New York: John Wiley and Sons, 3242.

    • Search Google Scholar
    • Export Citation
  • 39.

    Simanjuntak CH, Punjabi NH, Wangsasaputra F, Nurdin D, Pulungsih SP, Rofiq A, Santoso H, Pujarwoto H, Sjahrurachman A, Sudarmono P, Seidlein LV, Acosta C, Robertson SE, Ali M, Lee H, Park J, Deen JL, Agtini MD, Clemens JD, 2004. Diarrhoeal episodes and treatment-seeking behavior in a slum area of north Jakarta, Indonesia. J Health Popul Nutr 22: 119129.

    • Search Google Scholar
    • Export Citation
  • 40.

    Gupta SK, Suanto A, Gray A, Astuti BW, Jain N, Rolos R, Hoekstra RM, Quick R, 2007. Factors associated with E. coli contamination of household drinking water among tsunami and earthquake survivors, Indonesia. Am J Trop Med Hyg 76: 11581162.

    • Search Google Scholar
    • Export Citation
  • 41.

    Dibley MJ, Sadjmin T, Kjolhede CL, Moulton LH, 1996. Vitamin A supplementation fails to reduce incidence of acute respiratory illness and diarrhea in preschool-age Indonesian children. J Nutr 126: 434442.

    • Search Google Scholar
    • Export Citation
  • 42.

    Loehlin JC, 1998. Latent Variables Models: An Introduction to Factor, Path, and Structural Analysis. Third edition. Mahwah, NJ: Lawrence Erbaum Associates.

    • Search Google Scholar
    • Export Citation
  • 43.

    Arbuckle JL, 2010. AMOS: A Structural Equation Modeling Software. Chicago, IL: Small Waters Corporation.

  • 44.

    Byrne BM, 1989. Multigroup comparisons and the assumption of equivalent construct validity across groups: methodological and substantive issues. Multivariate Behav Res 24: 503523.

    • Search Google Scholar
    • Export Citation
  • 45.

    Steiger JH, Lind JC, 1980. Statistically Based Tests for the Number of Factors. Iowa City, IA: Psychometric Society.

  • 46.

    Browne MW, Cudeck R, 1989. Single sample cross-validation indices for covariance structures. Multivariate Behav Res 24: 445455.

  • 47.

    Pickering AJ, Julian TR, Marks SJ, Mattoili MC, Boehm AB, Schwab K, Davis J, 2012. Fecal contamination and diarrheal pathogens on surfaces and soils among Tanzanian households with and without improved sanitation. Environ Sci Technol 46: 57365743.

    • Search Google Scholar
    • Export Citation
  • 48.

    Julian TR, MacDonald L, Guo Y, Marks SJ, Kosek M, Yori PP, Pinedo SR, Schwab K, 2013. Fecal indicator bacteria contamination of fomites and household demand for surface disinfection products. Am J Trop Med Hyg 89: 869872.

    • Search Google Scholar
    • Export Citation
  • 49.

    Lin A, Arnold BF, Afreen S, Goto R, Huda T, Haque R, Raqib R, Unicomb L, Ahmed T, Colford JM Jr, Luby SP, 2013. Household environmental conditions are associated with enteropathy and impaired growth in rural Bangladesh. Am J Trop Med Hyg 89: 130137.

    • Search Google Scholar
    • Export Citation
  • 50.

    Stevens J, 1996. Applied Multivariate Statistics for the Social Sciences. Third edition. Mahwah, NJ: Erlbaum.

  • 51.

    Clasen T, Schmidt W, Rabie T, Roberts I, Cairncross S, 2007. Interventions to improve water quality for preventing diarrhoea: systematic review and meta-analysis. BMJ 334: 782.

    • Search Google Scholar
    • Export Citation
  • 52.

    Sobsey M, 2002. Managing Water in the Home: Accelerated Health Gains from Improved Water Supply. World Health Organization Sustainable Development and Healthy Environments. Geneva: World Health Organization.

    • Search Google Scholar
    • Export Citation
  • 53.

    Esrey SA, Feachem RG, Hughes JM, 1985. Interventions for the control of diarrhoeal diseases among young children: improving water supplies and excreta disposal facilities. Bull World Health Organ 63: 757772.

    • Search Google Scholar
    • Export Citation
  • 54.

    Barreto ML, Milroy CA, Strina A, Prado MS, Leite JP, Ramos EA, Ribeiro H, Alcantara-Neves NM, Teixeira M d G, Rodrigues LC, Guerreiro H, Trabulsi LR, 2006. Community-based monitoring of diarrhea in urban Brazilian children: incidence and associated pathogens. Trans Rl Soc Trop Med Hyg 100: 234242.

    • Search Google Scholar
    • Export Citation
  • 55.

    Black RE, Brown KH, Becker S, 1984. Effects of diarrhea associated with specific enteropathogens on the growth of children in rural Bangladesh. Pediatrics 73: 799.

    • Search Google Scholar
    • Export Citation

 

 

 

 

Modeling Risk Categories to Predict the Longitudinal Prevalence of Childhood Diarrhea in Indonesia

View More View Less
  • Department of Chemical and Environmental Engineering, Yale University, New Haven, Connecticut; School of Public Health, Yale University, New Haven, Connecticut

We present an innovative approach for analyzing diarrheal prevalence data that uses latent variables to model the relationships between multiple, interdependent environmental risk factors, and socioeconomic mediators. This strategy was applied to elucidate diarrheal longitudinal prevalence risk factors in children 1–4 years of age in low-income areas of Jakarta, Indonesia. Through a prospective cohort study, we identified 257 children who had at least one episode of diarrhea. At the onset of the study, we collected responses to 44 indicators, covering a wide range of previously identified diarrhea risk factors, including demographic and socioeconomic factors. We used exploratory factor analysis to uncover four latent categories of risk factors and their respective indicators from the initial pool of 44 indicators. Thereafter, we used structural equation modeling to model the relationships between the four risk categories and diarrheal longitudinal prevalence, controlling for socioeconomic and demographic covariates. Our final model elucidated several pathways to longitudinal diarrheal prevalence. Most notably, poverty exerts its effect on increased diarrheal prevalence via dual pathways: poor household hygiene and food quality, controlling for covariates. Implications of this and other findings for disease control in Jakarta are discussed.

Introduction

Diarrhea continues to be a leading cause of early childhood mortality and morbidity in lesser developed nations and is responsible for nearly one million deaths annually.1 In Indonesia, diarrhea is a leading cause of healthcare expenditures2 and mortality3 among infants. Most acute diarrheal episodes can be effectively treated with oral rehydration therapy. Conversely, persistent and prolonged diarrhea is more difficult to manage4 because of a vicious cycle of malnutrition, reduced immunity, and prolonged or repeated disease.58 Persistent and prolonged diarrhea causes numerous developmental impacts, including long-term cognitive deficits.9 Numerous studies have identified individual risk-factors for diarrheal disease incidence among children in developing countries, but few have investigated factors associated with the longitudinal prevalence,10 rather than the incidence of infections.4

Gastrointestinal disease can be transmitted by a number of interdependent environmental factors, which are in turn influenced by socioeconomic factors.4,11 Water and sanitation risk factors, including inconsistent hand washing,12,13 access to sanitation,14 and access to clean drinking water,1520 have been characterized, in addition to socioeconomic and age-related risk factors.4,2123 Infectious agents for diarrhea differ by transmission route, with different associated disease characteristics likely associated with differing risk-factors and transmission routes.

Previous analyses of environmental risk factors for diarrhea are most often based on statistical inference using a variety of regression models. Most studies identify individual risk factors for diarrhea.12,14,15,24,25 However, there is heterogeneity across and within studies with regard to the relative risk associated with various exposure routes, within different environments, possibly reflecting the relative contribution of environmental exposure routes for different pathogens that may cause diarrhea. Practitioners have promoted interventions such as water, sanitation and hygiene improvements to reduce risk of exposure via all routes,15,26,27 but there is little epidemiologic evidence of their benefits. A hierarchical-effect decomposition strategy that groups risk factors into discrete blocks23,28,29 has been introduced to improve analyses by enabling investigators to understand direct and indirect effects. However, because the methods used for this strategy rely only on measurable risk factors, it is not able to measure direct relationships among abstract variables that are difficult to measure or quantify, nor is it able to analyze complex relationships among covariates and these variables.

Structural equation modeling (SEM) is a more general version of generalized linear modeling that is able to incorporate various types of data.30,31 Regression methods are limited because they can model individual risk factors but not categories of risk. For example, to understand the relative importance of a large number of hypothesized risk factors of diarrheal prevalence, regression models would need to individually incorporate all hypothesized risk factors. Backward/forward selection strategies are thus typically used to identify individual risk factors that are significant. Often, several risk factors are concurrently significant, making it difficult to quantitatively compare various risk factors within the same study area or design.

In this report, we present an innovation in data analysis for water, sanitation, hygiene, and diarrhea studies. We propose a structural equation model to interpret environmental, socioeconomic, and health data collected form a child-cohort. This method is used to analyze childhood diarrhea prevalence in low-income areas of Jakarta, Indonesia.

Methods

Study design and population.

A prospective cohort study was conducted in two low-income areas of Jakarta, Indonesia (Tanjung Priok and Koja in north Jakarta and Pejaten Timur in south Jakarta). One thousand children 1–4 years of age were enrolled in the study. Participant selection and data collection are described elsewhere.19 In brief, study participants were selected from among all age-eligible children registered at public child health centers (Posyandu). Data were collected from children via two methods. A baseline interview was conducted during participant enrollment (February 2011) and recorded directly into a Google Chrome™ survey program onto netbooks to minimize data entry errors. Questions were modeled on those previously validated for use in an economic survey in Indonesia.32 Additional questions were incorporated to account for commonly discussed diarrheal risk factors.

A daily record of diarrhea was collected by using a Smiley diagram,33,34 which primary caregivers were trained to complete during February–July 2011. Daily water use and boiling were monitored by using weekly logs in a similar way. Information contained within the survey sheets was verbally confirmed during collection. All surveys and diagrams were translated from English to Indonesian and back-translated by different certified translators. Any inconsistencies in translation were resolved by consensus. Methods were validated during a pilot survey (n = 100) conducted in Padang, Indonesia during December 2010. Data from log sheets were recorded daily to laptop computers and validated by field managers. A randomly selected 10% subset of the log sheets was further validated by separate data entry specialists.

The dependent variable was the longitudinal prevalence of diarrhea. Consistent with previous studies,3538 diarrhea was defined as having two or more loose stools or a loose stool containing blood or mucus in a given day. Longitudinal prevalence of diarrhea was calculated by dividing the total number of diarrhea-days (numerator) by the total number of days of follow-up per child (denominator). All rates are reported per 1,000 child-days.28 To study longitudinal prevalence rather than incidence, we selected and analyzed data only for children who had at least one day of diarrhea.

We collected data on 44 risk factors that are deemed to be most important in Indonesia (Table 1).3,20,3941 Specifically, we analyzed household hygiene, environmental quality, food hygiene, and water quality.11,17,25,28 Covariates collected for analysis included whether the household, based on reported monthly income and household member number, did or did not fall below the official 2011 Indonesian poverty line (referred to as poverty status in the analytical section), child age, head of household education, and location within the city.

Table 1

Diarrhea indicator variables grouped into latent variable blocks, Indonesia*

Household hygieneEnvironmental qualityFood qualityDrinking water quality
Teeth brushing frequencyordNumber of family members sharing a living space (continuous)Eating frequencyordLocation of drinking watercat
Hand washing before eatingy/nElectricity within the homey/nPurchased meat in previous monthy/nDistance to drinking watercont
Hand washing after playingy/nHome ownership statuscatMeat qualityordWater typecat
Hand washing after toilet usey/nBathing water is purchased from street vendorcatMeat pricecontBoiling water prior to drinkingy/n
Bathing frequencyordLevel of sanitation sharingcatMilk type consumedcatWater use for tea, majoritycat
House cleaning frequencyordWastewater disposalcatInstant food preparation methodcatFamily member responsible for watercat
Milk bottle washing methodcatGarbage disposal methodcatConsumption of vegetable frequencyordHappiness with water qualityord
Dish washing methodcatPrivate access to water for washingcatStreet food eating frequencyordFrequency of use of improved water sourcescont
Use of soap frequencyordPopulation density in immediate block areacontRice preparation methodcatEscherichia coli in drinking waterord
Washing water container frequencyordHome typecatGovernment rice subsidy allowancey/nColiform in drinking waterord
  Cooking fuel typey/nWater storage time from collectionord
  Level of kitchen sharingy/nWater storage containerord
  Own a refrigeratory/n 

ord = ordinal response; cat = categorical response; y/n = yes or no response; cont = continuous response.

Analytic strategy.

Our analytic strategy involves two stages: first, selection of risk categories and second, modeling the risk categories, covariates and longitudinal prevalence of diarrhea. The first uses exploratory factor analysis to select risk categories from 44 observed risk factors. Each risk category is represented by four respective risk factors with the highest factor loadings.42 SPSS version 16.0.2 (IBM SPSS, Armonk, NY) was used to manipulate variables and perform the explanatory risk factor analysis.

The second stage elucidates the significant predictors of diarrhea longitudinal prevalence through SEM using AMOS version 18 (IBM, Armonk, NY).43 Parameters in the SEM were estimated by the maximum-likelihood method, which compares the fit of a hypothesized structural model to the observed variance-covariance matrix. Traditionally, SEM researchers used Chi-square to test for model fitness. However, the exact fit hypothesis tested by chi-square is over-restrictive and sensitive to sample size and violations of normality.42 A heuristic rule suggests that a model with a ratio of Chi-square to its degree of freedom (df) less than two is an acceptable fit.44 Steiger and Lind45 recommended root mean square error approximation (RMSEA) to evaluate model fit. The RMSEA measures discrepancy per degree of freedom and imposes a penalty for adding complexity to a model without substantially improving model fit. Smaller RMSEA values indicate a better model fit, with values less than 0.05 indicating a close fit. Values between 0.05 and 0.08 correspond to an acceptable fit, and RMSEA values greater than 0.10 suggest a poor fit.46 The Comparative Fit Index (CFI) and the Tucker-Lewis Index (TLI) measure the relative reduction in model misfit when comparing the target model relative to a baseline (independence) model. CFI and TLI values ≥ 0.90 have been considered an indication of an acceptable fit of the model to the observed data. Our paper uses RMSEA, CFI, and TLI to evaluate the fit of the proposed model. Several models were investigated and the best fitting was selected.

Unlike conventional approaches, SEM groups individual risk factors into risk categories, then models these risk categories against the outcome. In addition, the SEM approach is also preferable because 1) correlated risk categories can be incorporated into the same model (multi-co-linearity is a non-issue), and 2) models elucidate complex pathways involving multiple transmission and mediation routes. For example, regression analysis can show that poverty is a risk factor of diarrhea but SEM is able to show that the poverty–diarrhea link is mediated by poor household hygiene. Because SEM is able to model the complex relationships among risk factors, it is a more realistic interpretation.

Essentially, SEM offers a level of sophistication much needed to understand the complexities of diarrhea risks, and advantageously translates into more informed decision making for designing intervention programs. Against this background, we propose a structural equation model to interpret environmental, socioeconomic, and health data collected form a child cohort. We use this innovative method to analyze childhood diarrhea prevalence in low-income areas of Jakarta, Indonesia.

Ethical clearance.

This study was approved by the Secretary of Foreign Research Permits from the Indonesian Ministry of Research and Technology, the Jakarta Governor's Office Department of Population Research Ethics Committee, and the Yale University Human Investigation Committee. Collaboration and verbal consent was provided by community leaders and staff at public child health centers. Oral informed consent was obtained from the legal guardians of all children.

Results

Of the 1,000 children monitored, 257 children had at least one diarrhea episode by our definition. A description of this subset (the group of children who had at least one episode of diarrhea) is shown in Table 2. A slight majority of the children lived in households located in north Jakarta (66%) where the head of household had a high school education (51%), and nearly half (47%) had per-capita incomes below the national poverty line. The mean ± SD age of these children was 31 ± 13 months. The mean longitudinal prevalence of diarrhea for the children who had diarrhea was 25 diarrhea-days per 1,000 child-days, with a wide range (8–158 diarrhea-days per 1,000 child-days). As shown in Figure 1, data are right-tailed, with a Pearson's skewness coefficient of 2.5. Thus, of the children who were sick with diarrhea, most had an acute episode; a minority had persistent or reoccurring diarrhea. The conclusions from factor analysis and the model are community specific.

Table 2

Factor loading for measured variables from latent variables in children 1–4 years of age in Jakarta, Indonesia who had diarrhea during February–July 2011

Variable groupFactor loadingVariablePopulation valuesVariable shorthand
Covariates Head of household education 
 Elementary, no. (%)60 (23) 
 High school, no. (%)130 (51) 
 Middle school, no. (%)67 (26) 
 Age in months, mean (±SD)31 (±13) 
 Located in north Jakarta, no. (%)170 (66) 
 Per capita earnings below poverty, no. (%)121 (47) 
Household hygiene indicators0.642Teeth brushing frequencyHH1
 Never before, no. (%)87 (34) 
 Once a day, no. (%)53 (21) 
 Every time after eating, no. (%)117 (46) 
0.556Washes hands before eating, no. (%)238 (93)HH2
0.491Washes hands after playing, no. (%)240 (93)HH3
0.481Milk bottle washing methodHH4
 Sterilizes in boiled water, no. (%)71 (28) 
 Uses soap and untreated water, no. (%)68 (27) 
 Rinses with treated (boiled) water, no. (%)118 (46) 
Environmental quality indicators0.703Level of sanitation sharingEQ1
 Private toilet, no. (%)166 (65) 
 Shared with neighbors, no. (%)56 (22) 
 Use of public facility, no. (%)35 (14) 
0.619Wastewater disposalEQ2
 Public sewer, no. (%)170 (66) 
 Pit latrine or other underground, no. (%)27 (9) 
 Directly to surface water, no. (%)60 (24) 
0.710Bathing water is purchased from vendor, no. (%)67 (26)EQ3
0.689Garbage disposalEQ4
 Collected by sanitation services, no. (%)126 (49) 
 Thrown into the river or creek, no. (%)103 (40) 
 Disposed of in an open trash pit, no. (%)28 (11) 
Food hygiene indicators0.628Street food eating frequencyFQ1
 Daily, no. (%)183 (71) 
 Often (4–6 times/week), no. (%)23 (9) 
 Sometimes (1–3 times/week), no. (%)37 (14) 
 Never, no. (%)14 (5) 
0.690Type of milk consumedFQ2
 Mother's breast milk, no. (%)58 (23) 
 Formula, no. (%)123 (50) 
 Condensed milk, no. (%)76 (30) 
0.595Purchased meat within previous month, no. (%)59 (23)FQ3
0.294Owns a refrigerator, no. (%)94 (37)FQ4
Drinking water quality indicators0.431Water Distance, n (±SD)0.68 (±0.74)DWQ1
0.566Boils water prior to drinking, no. (%)189 (74)DWQ2
0.800Water type*DWQ3
 Branded bottled water, no. (%)30 (11) 
 Refills water from water kiosk, no. (%)51 (20) 
 Tap water, no. (%)59 (23) 
 Well water, no. (%)37 (14) 
 Combination, no. (%)80 (31) 
0.768Water for tea, majorityDWQ4
 Branded bottled water, no. (%)38 (15) 
 Refills water from water kiosk, no. (%)63 (25) 
 Tap water, no. (%)85 (33) 
 Well water, no. (%)71 (28) 

Water type is water used for at least 90% of the days monitored.

Combination describes children who did not use any one type of water more than 90% of the days monitored.

Water for tea is based on respondent selection during baseline survey.

Figure 1.
Figure 1.

Histogram with diarrhea longitudinal prevalence as number of diseased days per 1,000 child-days of follow up shows data for children who had at least one episode of diarrhea. A diarrhea-positive day is defined as a day with two or more loose stools and/or blood and mucus in the stool. Disease days are defined based on caretaker response in weekly logs. Disease numbers are normalized to the number of days of follow-up per child (range = 117–133 days).

Citation: The American Society of Tropical Medicine and Hygiene 89, 5; 10.4269/ajtmh.12-0540

Exploratory factor analysis.

Exploratory factor analysis was used to select the top risk factors for each of the following risk categories: household hygiene practices, environmental quality, food hygiene, and water quality. Principal components with varimax rotation were performed on indicator variables for each risk category. Household hygiene had an eigenvalue of 1.47, and explained 21.02% of the variance in the risk category; environmental quality, with an eigenvalue of 1.98, explained 33.07% of the variance; food hygiene, with an eigenvalue of 1.31, explained 32.78% of the variance; and water quality, with an eigenvalue of 1.89, explained 20.95% of the variance. The four risk categories and their indicator variables, including their factor loading are shown in Table 2. The bivariate correlation among indicators is shown in Table 3.

Table 3

Bivariate correlations among indicators used to predict values for latent variables, Indonesia

VariableVariable shorthandHousehold hygieneEnvironmental qualityFood hygieneDrinking water quality
HH1HH2HH3HH4EQ1EQ2EQ3EQ4FQ1FQ2FQ3FQ4DWQ1DWQ2DWQ3DWQ4
Household hygieneHH11               
HH20.206*1              
HH30.1240.344*1             
HH40.050−0.004−0.0041            
Environmental qualityEQ10.0430.002−0.178*0.0271           
EQ20.0430.027−0.0680.0770.284*1          
EQ30.0120.066−0.056−0.0990.316*0.361*1         
EQ40.009−0.0720.082−0.072−0.273*−0.219*−0.295*1        
Food hygieneFQ10.1390.1080.0640.1220.002−0.068−0.1600.00011       
FQ20.1590.189*0.0270.297*−0.039−0.211*−0.255*−0.0600.2351      
FQ30.0950.0840.071−0.0160.0400.1350.266*−0.062−0.081−0.1101     
FQ40.000−0.0630.105−0.065−0.290*−0.262*−0.1570.334*0.051−0.0060.200*1    
Drinking water qualityDWQ1−0.0440.0390.012−0.0920.382*0.294*0.410*−0.282*−0.199−0.217*0.048−0.261*1   
DWQ20.159−0.136−0.1240.0020.0830.070−0.026−0.0120.1450.018−0.113−0.076−0.0801  
DWQ3−0.011−0.053−0.054−0.0330.0140.0040.005−0.045−0.038−0.030−0.078−0.084−0.0090.1551 
DWQ40.019−0.191*−0.0150.080−0.091−0.053−0.1210.2020.110−0.076−0.1410.112−0.1230.327*0.184*1

Correlation is significant at the 0.01 level (two-tailed).

Correlation is significant at the 0.05 level (two-tailed).

Values for each risk factor, grouped by the risk category, are shown in Table 2. For continuous response variables, we present the mean and SD, and for categorical variables, we list the frequency and percentage of each value. Covariates, including socioeconomic status and child age, were also considered. The mean and frequency of these covariates also found in Table 2. Approximately half of all children came from families whose household head had not completed high school (49%) and from families that were below the poverty line (47%). Although an equal number of children from north and south Jakarta were surveyed, more than half (66%) of the diarrhea-positive cases (n = 257) were located in north Jakarta. Finally, considering that children who were monitored were all 12–62 months of age at baseline, it should be noted that the average age of those who had diarrhea was 31 months, and the SD was 13 months.

Structural equation modeling.

The structural equation model with four risk categories and diarrhea longitudinal prevalence as the outcome was selected to provide a good fit for the data (χ2 = 214.67, df = 148, χ2:df ratio = 1.45, CFI = 0.931, RSMEA = 0.042, TLI = 0.90). The model with respective standardized regression weights from each risk category to diarrheal rate is shown in Figure 2. Although errors in highly correlated variables were correlated, these are not shown in Figure 2 for simplification.

Figure 2.
Figure 2.

Structural equation model of diarrhea describing the relationship between drinking water quality (DWQ), environmental quality EQ), household hygiene (HH), and food hygiene (FQ) indicators (risk categories) and longitudinal diarrhea prevalence (outcome variable). Each risk category is measured by four observed risk factors (refer to Table 2 for a detailed list of the indicators from which these were selected). All error terms and correlations are omitted to simplify the representation. The beta values of relationships estimated using a dataset collected from children less than five years of age in Jakarta, Indonesia, during January–July 2011 is given next to the arrows depicting these relationships. Covariates, such as poverty, location within the city, education of the head of household, and child age are also incorporated into the model. The estimated beta values are shown only for significant relationships. Level of significance is shown by using an asterisk next to the beta values as defined below. * Significant to 0.10; ** Significant to 0.05.

Citation: The American Society of Tropical Medicine and Hygiene 89, 5; 10.4269/ajtmh.12-0540

In this model, environmental quality (β = 0.097, P = 0.21) and water quality (β = 0.011, P = 0.86) were not found to be significant predictors of diarrhea rate. Instead, food hygiene (β = 0.24, P = 0.008) and household hygiene (β = 0.15, P = 0.094) emerged as significant predictors of longitudinal diarrhea prevalence. The negative regression weights can be interpreted to mean that lower food hygiene and household hygiene were associated with longitudinal diarrhea prevalence. Furthermore, per capita income below poverty (β = 0.44, P < 0.0001) and location (β = 0.032, P = 0.039) were significant predictors of food hygiene, controlling for education and age. With respect to household hygiene, per capita income below poverty (β = 0.16, P = 0038), and age (β = 0.54, P < 0.0001) were significant, controlling for education and location. Per capita income below poverty was the common predictor for decreased food hygiene and household hygiene.

With respect to mediation and pathway analyses, the poverty–diarrhea prevalence link is mediated jointly by household hygiene and food quality. Specifically, households with more severe poverty tend to have poorer household hygiene (β = 0.16, P = 0038) that increased diarrhea prevalence (β = 0.15, P = 0.094); households with more severe poverty also tend to evidence lower food hygiene (β = 0.44, P < 0.0001) that increased diarrhea prevalence (β = 0.24, P = 0.008). With regards to the location–diarrhea link, children living in the north Jakarta (an urban slum) tended to have worst food hygiene (β = 0.032, P = 0.039) compared with those living in south Jakarta (a peri-urban development), and had a higher prevalence of diarrhea (β = 0.24, P = 0.008).

Discussion

In this report, we present an SEM approach to predicting the relative importance of risk categories, including household hygiene, environmental quality, drinking water quality, and food safety. Household hygiene and food hygiene risk categories were the most significantly linked to diarrhea prevalence. A number of recent studies have begun to investigate the relationship between surface contamination, hygiene, disease transmission, and other factors, such as clean water access and sanitation. More specifically, surfaces were found to harbor significant concentrations of fecal indicators, fomites, and diarrheal pathogens in Tanzanian47 and Peruvian households.48 Recently, published evidence has shown that environmental conditions in the household are linked to impaired childhood growth and other indicators of gut enteropathy in rural Bangladesh.49 Unlike water quality studies, the collection of pathogens from surfaces is not standard practice in the water, hygiene, and sanitation research field, and the importance of household hygiene may have been underestimated in previous research as a result. The survey-based indicators and methods outlined in this report demonstrate the importance of household hygiene without the expensive and cumbersome need to test for pathogens in every home.

The model was able to uncover links between poverty status, geographic location within the city, and hygiene status. The relative importance of disease transmission routes will likely differ in other areas of the world, but the methods outlined in this report are applicable to understand the relative importance of different disease transmission factors with locally relevant data wherever such data may exist.

This report presents an innovative modeling approach that offers distinct advantages. First, unlike multivariable regression, which becomes saturated with many individual risk factors and covariates,50 SEM is able to group individual risk factors into higher order categories and model these categories with outcome variables. Second, this method accounts for complex interdependent relationships among predictor variables, rather than considering each independent-dependent variable relationship separately. Third, SEM enables mediation and pathway analyses. Fourth, the grouping of individual risk factors into categories lay the groundwork for integrated interventions. For example, SEM enables us to analyze the association between overall household hygiene practices (a risk category) and diarrhea, as opposed to, for instance, focusing only on the association between hand washing to diarrhea. This model can be used to interpret data collected in other areas to improve the global understanding of disease transmission and recommended disease transmission interventions. Similar analyses can be used to understand the infection risks associated with other diseases in crowded slums.

An analytical strategy that is statistically rigorous and able to incorporate complex interactions among various factors can inform holistic intervention strategies, rather than piece-meal solutions. Hygiene education,13 food safety promotion,26 community-wide sanitation,28 and household water treatment,51,52 have all been recommended to prevent diarrheal infections. Combinations of several interventions are more effective than single approaches.12,27,53 Resources are often limited, especially for urban slum areas, requiring effective intervention approaches to address the most salient risk-factors.25 By elucidating the relationship among risk categories of diarrhea, this report identifies the most promising intervention types, including those that enhance household hygiene. These findings are similar to those of other groups, whose findings are showing the importance of surface decontamination. Data from a prospective study of childhood diarrhea rates in low-income areas of Jakarta, Indonesia were analyzed by using SEM and analysis showed that food hygiene and household hygiene practices are significantly associated with diarrhea length, controlling for environmental quality and drinking water quality. A possible explanation for this variation may be the length of infections associated with pathogens that are preferentially transmitted by different environmental routes. Findings from other areas confirm that diarrhea length and episode intensity differed by infectious agent in other areas of the world.54,55 Protozoa, viruses, and bacteria have unique chemical and physical characteristics that may lead them to be preferentially transmitted via different environmental routes. For example, a greater proportion of food-borne pathogens may lead to long diarrhea episodes, and a greater proportion of water-borne pathogens may lead to short-term episodes. The SEM model also enables us to estimate the relationship between covariates and latent predictors, showing that poverty status and location are significant predictors of food hygiene, and poverty status and age in months are significant predictors for household hygiene practices. The model also shows that, as anticipated, household poverty status and head of household education are significantly related.

There are several limitations of this study. First, this is a non-causal model in which all results reflect associations among variables, and no causal relationships among variables can be interpreted from these results. Second, data was collected from a cohort of children that is geographically and socioeconomically specific and results cannot be extrapolated to other population groups. These findings are only pertinent to middle or lower-class children in two areas of Jakarta because children are selected from free health care center records and wealthy children opt out of the public health care system in Jakarta.39 The data and conclusions are population specific and model specific.

In conclusion, this study presents a useful method for analyzing the complex relationship between various environmental factors, socioeconomic variables, and childhood diarrhea. Using a dataset from Jakarta, we showed that food hygiene and household hygiene are the most significant latent variables associated with longitudinal prevalence of diarrhea among children who had at least one episode of diarrhea. This method is promising and should have wide applicability in the field.

  • 1.

    Liu L, Johnshon HL, Cousens S, Perin J, Scott S, Lawn JE, Rudan I, Campbell H, Cibulskis R, Li M, 2012. Global, regional, and national causes of child mortality: an updated systematic analysis for 2010 with time trends since 2000. Lancet 379: 21512161.

    • Search Google Scholar
    • Export Citation
  • 2.

    Lerman SJ, Shepard DS, Cash RA, 1985. Treatment of diarrhea in Indonesian children: what it costs and who pays for it. Lancet 2: 651654.

  • 3.

    Agtin MD, Soeharno R, Lesmana M, Punjab NH, Simanjuntak C, Wangsasaputra F, Nurdin D, Pulungsih SP, Rofiq A, Santoso H, Pujarwoto H, Sjahrurachman A, Sudarmono P, von Seidlein L, Deen JL, Ali M, Lee H, Kim DR, Han O, Park JK, Suwandono A, Ingerani, Oyofo BA, Campbell JR, Beecham HJ, Corwin AL, Clemens JD, 2005. The burden of diarrhoea, shigellosis, and cholera in north Jakarta, Indonesia: findings from 24 months surveillance. BMC Infect Dis 5: 89.

    • Search Google Scholar
    • Export Citation
  • 4.

    Lima AA, Guerrant RL, 1992. Persistent diarrhea in children: epidemiology, risk factors, pathophysiology, nutritional impact, and management. Epidemiol Rev 14: 222243.

    • Search Google Scholar
    • Export Citation
  • 5.

    WHO/UNICEF, 1985. The Management of Diarrhoea and Use of Oral Rehydration Therapy. Geneva: World Health Organization.

  • 6.

    Grantham-McGregor S, 1995. A review of studies of the effect of severe malnutrition on mental development. J Nutr 125: 2238S.

  • 7.

    Brown KH, 2003. Diarrhea and malnutrition. J Nutr 133: 32853325.

  • 8.

    Moore SR, Lima NL, Soares AM, Oria RB, Pinkerton RC, Barrett LJ, Guerrant RL, Lima AA, 2010. Prolonged episodes of acute diarrhea reduce growth and increase risk of persistent diarrhea in children. Gastroenterology 139: 11561164.

    • Search Google Scholar
    • Export Citation
  • 9.

    Niehaus MD, Moore SR, Patrick PD, Derr LI, Lorntz B, Lima AA, Guerrant RL, 2002. Early childhood diarrhea is associated with diminished cognitive function 4 to 7 years later in children in a northeast Brazilian shantytown. Am J Trop Med Hyg 66: 590593.

    • Search Google Scholar
    • Export Citation
  • 10.

    Morriss S, Cousens SN, Kirkwood BR, Arthur P, Ross DA, 1996. Is prevalence of diarrhea a better predictor of subsequent mortality and weight gain than diarrhea incidence? Am J Epidemiol 144: 582588.

    • Search Google Scholar
    • Export Citation
  • 11.

    Bernd G, Strina A, Teles CA, Prado MS, Barreto ML, 2006. Risk factors for childhood diarrhea incidence. Dynamic Analysis of a Longitudinal Study Epidemiology 17: 658667.

    • Search Google Scholar
    • Export Citation
  • 12.

    Fewtrell L, 2005. Water, sanitation, and hygiene interventions to reduce diarrhoea in less developed countries: a systematic review and meta-analysis. Lancet Infect Dis 5: 4252.

    • Search Google Scholar
    • Export Citation
  • 13.

    Aulia H, Surapaty SC, Bahar E, Susanto TA, Roisuddin, Hamzah TM, 1994. Personal and domestic hygiene and its relationship to the incidence of diarrhoea in South Sumatra. J Diarrhoeal Dis Res 12: 4248.

    • Search Google Scholar
    • Export Citation
  • 14.

    Blum D, Feachem RG, 1983. Measuring the impact of water supply and sanitation investments on diarrhoeal diseases: problems of methodology. Int J Epidemiol 12: 357366.

    • Search Google Scholar
    • Export Citation
  • 15.

    Esrey SA, Habicht J-P, 1986. Epidemiologic evidence for health benefits from improved water and sanitation in developing countries. Epidemiol Rev 8: 117129.

    • Search Google Scholar
    • Export Citation
  • 16.

    Esrey SA, Potash JB, Shiff RC, 1991. Effects of improved water supply and sanitation on ascariasis, diarrhoea, dracunculiasis, hookworm infection, schistosomiasis, and trachoma. Bull World Health Organ 69: 609621.

    • Search Google Scholar
    • Export Citation
  • 17.

    Semenza JC, Roberts L, Henderson A, Bogan J, Rubin CH, 1998. Water distribution and diarrheal disease transmission: a case study in Uzbekistan. Am J Trop Med Hyg 59: 941946.

    • Search Google Scholar
    • Export Citation
  • 18.

    Arnold BF, Colford JM, 2007. Treating water with chlorine at point-of-use to improve water quality and reduce child diarrhea in developing countries: a systematic review and meta-analysis. Am J Trop Med Hyg 76: 354364.

    • Search Google Scholar
    • Export Citation
  • 19.

    Sima L, Desai MM, McCarty KM, Elimelech M, 2012. Relationship between use of water from community-scale water treatment refill kiosks and childhood diarrhea in Jakarta. Am J Trop Med Hyg 87: 979984.

    • Search Google Scholar
    • Export Citation
  • 20.

    Semba RD, 2009. Purchase of drinking water is associated with increased child morbidity and mortality among urban slum-dwelling families in Indonesia. Int J Hyg Environ Health 212: 387397.

    • Search Google Scholar
    • Export Citation
  • 21.

    Pickering H, Hayes RJ, Tomkins AM, Carson D, Dunn DT, 1986. Alternative measures of diarrhoeal morbidity and their association with social and environmental factors in urban children in The Gambia. Trans R Soc Trop Med Hyg 81: 853859.

    • Search Google Scholar
    • Export Citation
  • 22.

    Manun’ebo MN, Haggerty PA, Kalengaie M, Ashworth A, Kirkwood BR, 1994. Influences of demographic, socioeconomic and environmental variables on childhood diarrhoea in a rural area of Zaire. J Trop Med Hyg 97: 3138.

    • Search Google Scholar
    • Export Citation
  • 23.

    Genser B, Strina A, Teles CA, Pardo MS, Barreto ML, 2006. Risk factors for childhood diarrhea incidence: dynamic analysis of a longitudinal study. Epidemiology 17: 658667.

    • Search Google Scholar
    • Export Citation
  • 24.

    Pruss A, Kay D, Fewtrell L, Bartram J, 2002. Estimating the burden of disease from water, sanitation, and hygiene at a global level. Environ Health Perspect 110: 537542.

    • Search Google Scholar
    • Export Citation
  • 25.

    Eisenberg JN, Scott JC, Porco T, 2007. Integrating disease control strategies: balancing water sanitation and hygiene interventions to reduce diarrheal disease burden. Am J Public Health 97: 846853.

    • Search Google Scholar
    • Export Citation
  • 26.

    Esrey SA, Feachem RG, 1989. Interventions for the Control of Diarrhoeal Disease among Young Children: Promotion of Food Hygiene. Geneva: World Health Organization.

    • Search Google Scholar
    • Export Citation
  • 27.

    Esrey SA, Habicht J-P, Casella G, 1992. The complementary effect of latrines and increased water usage on the growth of infants in rural Lesotho. Am J Epidemiol 135: 659667.

    • Search Google Scholar
    • Export Citation
  • 28.

    Barreto ML, Genser B, Strina A, Teixeira MG, Asssis AM, Rego RF, Teles CA, Prado MS, Matos SM, Santos DN, dos Santos LA, Cairncross S, 2007. Effect of city-wide sanitation programme on reduction in rate of childhood diarrhoea in northeast Brazil: assessment by two cohort studies. Lancet 370: 16221628.

    • Search Google Scholar
    • Export Citation
  • 29.

    Santos CA, Strina A, Amorim LD, Genser B, Assis AM, Prado MS, Barreto ML, 2012. Individual and contextual determinants of the duration of diarrhoel episodes in preschool children: a longitudinal study in an urban setting. Epidemiol Infect 140: 689696.

    • Search Google Scholar
    • Export Citation
  • 30.

    Fahrmeir L, Raach A, 2007. A Bayesian semiparametric latent variable model for mixed responses. Psychometrika 72: 327346.

  • 31.

    MacDonald RP, Ho RM, 2002. Principles and practice in reporting structural equation analyses. Psychol Methods 7: 6482.

  • 32.

    Badan Pusat Statistik, 2004. National Socio-Economic Survey (NSES). Jakarta, Indonesia: SUSENAS.

  • 33.

    Wright JA, Gaundry SW, Conroy R, Wood D, Du Perez M, Ferro-Luzzi A, Genthe B, Kirimi M, Moyo S, Mutisi C, Ndamba J, Potgieter N, 2006. Defining episodes of diarrohoea: results from a three-country study in sub-Saharan Africa. J Health Popul Nutr 24: 816.

    • Search Google Scholar
    • Export Citation
  • 34.

    Du Perez M, Conroy RM, Wright JA, Moyo S, Potgieter N, Gundry SW, 2008. Use of ceramic water filtration in the prevention of diarrheal diseases: a randomized controlled trial in rural South Africa and Zimbabwe. Am J Trop Med Hyg 70: 696701.

    • Search Google Scholar
    • Export Citation
  • 35.

    Levine MM, Caplan ES, Waterman D, Cash RA, Hornick RB, Snyder MJ, 1977. Diarrhea caused by Escherichia coli that produce only heat-stable enterotoxin. Infect Immun 17: 7882.

    • Search Google Scholar
    • Export Citation
  • 36.

    Kaplan JE, Schonberger LB, Varano G, Jackman N, Bied J, Gary GW, 1982. An outbreak of acute nonbacterial gastroenteritis in a nursing home. Am J Epidemiol 116: 940949.

    • Search Google Scholar
    • Export Citation
  • 37.

    Laffan AM, McKenzie R, Forti J, Conklin D, Marcinko R, Shrestha R, Bellantoni M, Greenough WBI, 2011. Lactoferrin for the prevention of post-antibiotic diarrhoea. J Health Popul Nutr 29: 547551.

    • Search Google Scholar
    • Export Citation
  • 38.

    Banks MR, Farthing MJ, 2012. Diarrhea. Hawkey CJ, Bosch J, Richter J, Garcia-Tsao G, Chan F, eds. Textbook of Clinical Gastroenterology and Hematology. New York: John Wiley and Sons, 3242.

    • Search Google Scholar
    • Export Citation
  • 39.

    Simanjuntak CH, Punjabi NH, Wangsasaputra F, Nurdin D, Pulungsih SP, Rofiq A, Santoso H, Pujarwoto H, Sjahrurachman A, Sudarmono P, Seidlein LV, Acosta C, Robertson SE, Ali M, Lee H, Park J, Deen JL, Agtini MD, Clemens JD, 2004. Diarrhoeal episodes and treatment-seeking behavior in a slum area of north Jakarta, Indonesia. J Health Popul Nutr 22: 119129.

    • Search Google Scholar
    • Export Citation
  • 40.

    Gupta SK, Suanto A, Gray A, Astuti BW, Jain N, Rolos R, Hoekstra RM, Quick R, 2007. Factors associated with E. coli contamination of household drinking water among tsunami and earthquake survivors, Indonesia. Am J Trop Med Hyg 76: 11581162.

    • Search Google Scholar
    • Export Citation
  • 41.

    Dibley MJ, Sadjmin T, Kjolhede CL, Moulton LH, 1996. Vitamin A supplementation fails to reduce incidence of acute respiratory illness and diarrhea in preschool-age Indonesian children. J Nutr 126: 434442.

    • Search Google Scholar
    • Export Citation
  • 42.

    Loehlin JC, 1998. Latent Variables Models: An Introduction to Factor, Path, and Structural Analysis. Third edition. Mahwah, NJ: Lawrence Erbaum Associates.

    • Search Google Scholar
    • Export Citation
  • 43.

    Arbuckle JL, 2010. AMOS: A Structural Equation Modeling Software. Chicago, IL: Small Waters Corporation.

  • 44.

    Byrne BM, 1989. Multigroup comparisons and the assumption of equivalent construct validity across groups: methodological and substantive issues. Multivariate Behav Res 24: 503523.

    • Search Google Scholar
    • Export Citation
  • 45.

    Steiger JH, Lind JC, 1980. Statistically Based Tests for the Number of Factors. Iowa City, IA: Psychometric Society.

  • 46.

    Browne MW, Cudeck R, 1989. Single sample cross-validation indices for covariance structures. Multivariate Behav Res 24: 445455.

  • 47.

    Pickering AJ, Julian TR, Marks SJ, Mattoili MC, Boehm AB, Schwab K, Davis J, 2012. Fecal contamination and diarrheal pathogens on surfaces and soils among Tanzanian households with and without improved sanitation. Environ Sci Technol 46: 57365743.

    • Search Google Scholar
    • Export Citation
  • 48.

    Julian TR, MacDonald L, Guo Y, Marks SJ, Kosek M, Yori PP, Pinedo SR, Schwab K, 2013. Fecal indicator bacteria contamination of fomites and household demand for surface disinfection products. Am J Trop Med Hyg 89: 869872.

    • Search Google Scholar
    • Export Citation
  • 49.

    Lin A, Arnold BF, Afreen S, Goto R, Huda T, Haque R, Raqib R, Unicomb L, Ahmed T, Colford JM Jr, Luby SP, 2013. Household environmental conditions are associated with enteropathy and impaired growth in rural Bangladesh. Am J Trop Med Hyg 89: 130137.

    • Search Google Scholar
    • Export Citation
  • 50.

    Stevens J, 1996. Applied Multivariate Statistics for the Social Sciences. Third edition. Mahwah, NJ: Erlbaum.

  • 51.

    Clasen T, Schmidt W, Rabie T, Roberts I, Cairncross S, 2007. Interventions to improve water quality for preventing diarrhoea: systematic review and meta-analysis. BMJ 334: 782.

    • Search Google Scholar
    • Export Citation
  • 52.

    Sobsey M, 2002. Managing Water in the Home: Accelerated Health Gains from Improved Water Supply. World Health Organization Sustainable Development and Healthy Environments. Geneva: World Health Organization.

    • Search Google Scholar
    • Export Citation
  • 53.

    Esrey SA, Feachem RG, Hughes JM, 1985. Interventions for the control of diarrhoeal diseases among young children: improving water supplies and excreta disposal facilities. Bull World Health Organ 63: 757772.

    • Search Google Scholar
    • Export Citation
  • 54.

    Barreto ML, Milroy CA, Strina A, Prado MS, Leite JP, Ramos EA, Ribeiro H, Alcantara-Neves NM, Teixeira M d G, Rodrigues LC, Guerreiro H, Trabulsi LR, 2006. Community-based monitoring of diarrhea in urban Brazilian children: incidence and associated pathogens. Trans Rl Soc Trop Med Hyg 100: 234242.

    • Search Google Scholar
    • Export Citation
  • 55.

    Black RE, Brown KH, Becker S, 1984. Effects of diarrhea associated with specific enteropathogens on the growth of children in rural Bangladesh. Pediatrics 73: 799.

    • Search Google Scholar
    • Export Citation

Author Notes

* Address correspondence to Menachem Elimelech, Department of Chemical and Environmental Engineering, Yale University, 9 Hillhouse Avenue, New Haven, CT 06520. E-mail: menachem.elimelech@yale.edu

Authors' addresses: Laura C. Sima and Menachem Elimelech, Department of Chemical and Environmental Engineering, Yale University, New Haven, CT, E-mails: laura.sima@yale.edu and menachem.elimelech@yale.edu. Reuben Ng, School of Public Health, Yale University, New Haven, CT, E-mail: reuben.ng@yale.edu.

Save