Summary of findings
An estimated 2.6 billion people or 39% of the world's population lack access to improved facilities for the disposal of human excreta, such as a basic pit latrine, a toilet connected to a septic tank or piped sewer system, or a composting toilet according to the World Health Organization (WHO) and the United Nations Children's Fund (UNICEF) (WHO/UNICEF 2010). In low-income regions, where people are most vulnerable to infection and disease, only one in two people is covered by improved sanitation. More than one billion people still practice open defecation. In sub-Saharan Africa and southern Asia coverage is just 31% and 33%, respectively. While the global population in 2006 is about equally divided between urban and rural dwellers, more than seven out of 10 people living without improved sanitation are rural inhabitants (WHO/UNICEF 2010).
The shortfall in sanitation coverage is not the result of a failure to recognize the need for it or declare goals to meet this need at the highest international level. The 1977 Mar del Plata Declaration by the United Nations expressed the goal of providing safe water and sanitation for all by 1990, launching the Water and Sanitation Decade (1981 to 1990). In 1990 the United Nations renewed the call and extended the deadline to the end of the 20
Definitions of sanitation
In the broadest sense, sanitation deals not only with the collection, storage, treatment, disposal, reuse or recycling of human excreta (faeces and urine), but also the drainage, disposal, recycling, and re-use of wastewater and storm water (sullage), and household, industrial, and hazardous solid waste. The MDG target, which is expressed in terms of 'basic sanitation', followed this broader approach and also included concepts of affordability, cultural acceptability, and environmental sustainability (United Nations 2002). The United Nations Millennium Task Force on Water and Sanitation attempted to consolidate these notions, defining basic sanitation as "the lowest-cost option for securing sustainable access to safe, hygienic, and convenient facilities and services for excreta and sullage disposal that provide privacy and dignity, while at the same time ensuring a clean and healthful living environment both at home and in the neighbourhood of users" (Millennium Project 2005). The MDG definition is context specific. In dispersed, low-income rural areas it may include a simple pit latrine, while in congested urban slums with a reliable water service, household-based solutions would be deemed inadequate and low-cost sewerage systems would be necessary to ensure the proper collection, treatment, and disposal or reuse of excreta and household wastewater (Millennium Project 2005).
The Joint Monitoring Programme for Water Supply and Sanitation (JMP) defines improved sanitation and unimproved sanitation in terms of the facilities for the disposal of human excreta (WHO 2002). Improved sanitation includes a private flush or pour-flush toilet or latrine connected to a piped sewer system or septic system, a simple pit latrine with a slab, a ventilated improved pit (VIP) latrine or a composting toilet. Unimproved sanitation includes any other flush or pour-flush latrine, an open pit latrine, bucket latrines, a hanging latrine, any public or shared facility or open defecation (WHO/UNICEF 2002).
Neither set of definitions is strictly health-based. The MDG definition addresses not only safety and hygiene but also convenience, cost, privacy, and dignity. It emphasizes household and community impact, sustainability, and actual use by the population so protected. While the JMP classification is intended to reflect the health risk associated with safe excreta disposal, its distinction between improved and unimproved facilities is based mainly on observable criteria developed by the JMP to facilitate surveys of provision and promote the upgrading of facilities. The differences between these definitions are not merely academic; funding and other resources are largely directed at increasing levels of provision reported by the JMP and meeting MDG targets (UNDP 2007). At a minimum, however, both definitions agree that sanitation must include the safe disposal of human excreta, a criterion that is founded in health.
For this reason, this review focuses on sanitation interventions to introduce or expand the provision or use of facilities for excreta disposal. This includes steps to reduce open defecation by constructing basic sanitation in accordance with the MDG target. It also includes interventions to improve the disposal of child faeces, such as by promoting potties, when accompanied by the safe disposal of their contents (Traore 1994). This review does not extend, however, to interventions that are not principally aimed at the sanitary disposal of human faeces. Thus, it does not include efforts to use human waste in agricultural applications; an activity that may actually increase risks to health.
Diarrhoeal disease, disease agents, and pathways
Diarrhoeal diseases kill an estimated 1.8 million people each year (WHO 2005). Among infectious diseases, diarrhoea ranks as the third leading cause of both mortality and morbidity (after respiratory infections and HIV/AIDS). Young children are especially vulnerable, bearing 68% of the total burden of diarrhoeal disease (Bartram 2003). Among children younger than five years of age, diarrhoea accounts for 17% of all deaths (United Nations 2005).
The immediate threat from diarrhoea is dehydration, and a loss of fluids and electrolytes. Thus, the widespread promotion of oral rehydration therapy has significantly reduced the case-fatality rate associated with the disease. Such improvements in case management, however, have not reduced morbidity, which is estimated at four billion cases annually (Kosek 2003). And since diarrhoeal diseases inhibit normal ingestion of foods and adsorption of nutrients, continued high morbidity is an important cause of malnutrition, leading to impaired physical growth and cognitive function (Guerrant 1999; Petri 20089), reduced resistance to infection (Baqui 1993), and potentially long-term gastrointestinal disorders (Schneider 1978). With continued high attack rates, diarrhoeal disease is also an enormous economic burden, resulting in significant direct costs to the health sector and patients for treatment as well as in lost time at school, work, and in other productive activities (Mulligan 2005).
The infectious agents associated with diarrhoeal disease are transmitted chiefly through the faecal-oral route (Byers 2001). A wide variety of bacterial, viral, and protozoan pathogens excreted in the faeces of humans and animals are known to cause diarrhoea. Among the most important of these are Escherichia coli, Salmonella spp., Shigella spp., Campylobacter jejuni, Vibrio cholerae, rotavirus, norovirus, Giardia lamblia, Cryptosporidium spp., and Entamoeba histolytica (Leclerc 2002). The importance of individual pathogens varies between settings, seasons, and conditions.
These pathogens may be transmitted through the ingestion of contaminated food, water or other beverages, by person-to-person contact, and by direct or indirect contact with infected faeces. Because of this variety of pathways, environmental interventions for the prevention of diarrhoeal disease typically include steps to improve the proper disposal of human faeces (sanitation), as well as improving water quality (Clasen 2006), water quantity and access, and promoting hand washing and other hygiene practices (Curtis 2003; Ejemot 2008).
In addition to diarrhoea, there are other important risks to health associated with poor sanitation. These include schistosomiasis, soil-transmitted helminth infection (including ascariasis, trichuriasis, and hookworm infection), trachoma (Emerson 2004), and tropical enteropathy. Tropical enteropathy, a subclinical disorder of the small intestine caused by faecal bacteria ingested in large quantities by young children living in conditions of poor sanitation and hygiene, may be a substantial cause of under-nutrition in young children that is entirely separate from diarrhoea (Humphrey 2009).
Excreta disposal and diarrhoea
While the biological association between diarrhoea and exposure to human faeces is well established, there is little rigorous epidemiological evidence of the effectiveness of sanitation interventions to prevent disease. Much of the evidence of the effectiveness and mechanisms of improved sanitation to prevent diarrhoea derives from observational studies (Barreto 2007; Genser 2008; Green 2009). There is little evidence of this from intervention studies. A previous Cochrane Review examined environmental sanitary interventions but it was limited to interventions to prevent active trachoma (Rabiu 2005).
There are three previous reviews of excreta disposal interventions (Esrey 1985; Esrey 1991; Fewtrell 2005; Waddington 2009 ). Esrey and colleagues identified 10 studies of improvements in excreta disposal with a median reduction in diarrhoea of 22% (ranging from 0% to 48%) (Esrey 1985). A subsequent review of 'sanitation' interventions reported a median reduction of 30% from 11 studies (36% from the five studies the investigators deemed to be rigorous) (Esrey 1991). Esrey and colleagues based their conclusions chiefly on observational studies. In addition to the confounding and bias inherent in such studies, Esrey and others have pointed out significant and widespread methodological problems in these studies (Blum 1983; Esrey 1986). Although these previous reviews were helpful in identifying the broad questions and suggesting answers, they did not employ the more rigorous methodologies and statistical methods of a systematic review (Egger 2001). In terms of coverage, for example, neither study (Blum 1983; Esrey 1986) involved a comprehensive search strategy. The reviews were also limited to studies in the English language. With respect to statistical methods, the simple use of the median fails to take into account the size of the study and the variance observed in the results (Deeks 2001). Moreover, they did not distinguish between the various case definitions (Moy 1991) and measures of diarrhoea morbidity (Morris 1996; Pickering 1987). Also, while Esrey attempted to incorporate qualitative criteria in the reviews, there was no independent assessment of the study quality or, for that matter, whether the studies identified met the inclusion criteria.
Fewtrell and colleagues conducted a more formal systematic review and meta-analysis of environmental interventions against diarrhoea (Fewtrell 2005). They identified just four such studies of improved sanitation, only two of which provided data that they could use in a meta-analysis. Fewtrell and colleagues reported that the interventions weree protective, with a pooled risk ratio (RR) of 0.68 (95% confidence interval (CI) 0.53 to 0.87) − a 32% reduction in diarrhoea that would appear to be consistent with Esrey's findings. In addition to being based on just two studies, however, there are other reasons to question the evidential weight of this estimate. Firstly, the review was limited to published studies and did not include a search of Chinese-language databases in which a number of articles on sanitation interventions are believed to be indexed. Secondly, one of the two studies identified in the review (Azurin 1974) had cholera rather than general diarrhoea as its outcome. Cholera is usually a source of epidemic diarrhoea against which environmental interventions tend to be more effective than can be expected for general diarrhoea, leading to results that cannot be generalized to endemic diarrhoea (Gundry 2004). The other study (Daniels 1990) followed an observational design, and thus did not meet the eligibility criteria. Emerson and colleagues (Emerson 2004) have demonstrated that a health impact (trachoma) from latrines can be investigated using the more rigorous randomized, controlled trial design.
Waddington and colleagues undertook an update of the Fewtrell review (Waddington 2009). They identified six sanitation studies that met their inclusion criteria, yielding a pooled RR of 0.63 (95% CI 0.43 to 0.93). However, the pooled estimate of the three studies they regarded as being of 'high quality' was not statistically significant (0.64, 95% CI 0.37 to 1.10), and none of the six sanitation studies included in their review followed an intervention design.
Beyond the paucity of rigorous epidemiological evidence in support of sanitation interventions, there is relatively little evidence of the acceptability, scalability, and sustainability of steps to improve excreta disposal, especially in rural settings where provision is lowest (Jenkins 2005).
This review employs the rigorous methodology and other benefits of the Cochrane Collaboration to identify and summarize evidence on the effectiveness of sanitation interventions to prevent diarrhoeal diarrhoea. By highlighting such evidence or the paucity thereof, it seeks to remove a knowledge barrier that may be contributing to a comparatively slow progress in achieving the sanitation target of the MDGs.
To assess the effectiveness of interventions to improve human excreta disposal for preventing diarrhoeal disease.
Criteria for considering studies for this review
Types of studies
Randomized, quasi-randomized, and non-randomized controlled trials (RCTs). The unit of randomization may include individuals, families, households, villages, communities or other clusters.
Types of participants
Children and adults in any country or population.
Types of interventions
Interventions aimed at introducing or expanding the coverage and use of facilities designed to reduce direct or indirect contact with human faeces. Such facilities including simple pit latrines, VIP latrines, bucket latrines, hanging toilets, water sealed pour-flush toilets (whether or not connected to a vault, septic tank or sewer), and composting toilets. It also includes the promotion of apparatus to improve the safe disposal of child faeces, such as potties and scoops, when accompanied by improved disposal of their contents.
We included interventions that combine improvements in excreta disposal with other environmental interventions such as improvements in water quantity or access, and in water quality or in hygiene practices.
Study participants who practice open defecation or who continue to follow their current practices with respect to excreta disposal rather than the prescribed intervention.
Types of outcome measures
Diarrhoea episodes among individuals, whether or not confirmed by microbiological examination.
The WHO's definition of diarrhoea is three or more loose or fluid stools (that take the shape of the container) in a 24-hour period (WHO 1993). We defined diarrhoea and an episode in accordance with the case definitions used in each trial. We excluded trials that had no clinical outcomes; for example, trials that report only on microbiological pathogens in the stool. Where data were provided, we extracted and analysed data from the studies describing the method of diarrhoea surveillance and reporting, the severity of diarrhoea, hospital admission, and measures taken by individuals in response to diarrhoea.
- adverse events
Search methods for identification of studies
We attempted to identify all relevant studies regardless of language or publication status (whether published, unpublished, in press or ongoing).
We searched the following databases using the search terms detailed in Appendix 1: Cochrane Infectious Disease Group Specialized Register; Cochrane Central Register of Controlled Trials (CENTRAL), published in The Cochrane Library; MEDLINE; EMBASE; and LILACS. We also searched Chinese-language databases (Fung 2008) available under the China National Knowledge Infrastructure (CNKI-CAJ) using comparable Chinese-language search terms. We also searched the metaRegister of Controlled Trials (mRCT) using 'diarrhoea' and 'sanitation or latrine or toilet or privy or disposal' as search terms.
We searched the conference proceedings of the following organizations for relevant abstracts: International Water Association and the Water, Engineering and Development Centre, Loughborough University, UK.
Organizations and pharmaceutical companies
We contacted researchers and organizations including the Water, Sanitation and Health Programme of the WHO; World Bank Water and Sanitation Program; UNICEF Water, Environment and Sanitation; Environmental Health Project; IRC International Water and Sanitation Centre; Foodborne and Diarrheal Diseases Branch, Division of Bacterial and Mycotics Diseases, Centers for Disease Control and Prevention (CDC); US Agency for International Development (USAID); and the UK Department for International Development (DFID) for unpublished and ongoing trials.
We checked the reference lists of all studies identified by the above methods.
Data collection and analysis
Selection of studies
Except for Chinese-language search results, Thomas Clasen (TC) and Kristof Boeston (KB) independently reviewed the titles and abstracts resulting from the searches and selected all studies that potentially met the inclusion criteria for the review. After obtaining full copies of all such studies, they independently determined if the trial met the inclusion criteria by completing an eligibility form. For Chinese-language search results, Isaac Fung (IF) undertook the same process individually and summarized the article in English, and TC and Wolf-Peter Schmidt (WS) reviewed the summaries to independently determine the eligibility of the study. Potentially relevant studies that we ultimately deemed did not meet the eligibility criteria for the study are nevertheless identified together with the reason for exclusion in the Characteristics of excluded studies section.
Data extraction and management
One author used a pre-piloted form to extract and record the data described in Appendix 2, and attempted to contact authors to supply missing data. We recorded morbidity based on the measure used in the trial. We recorded whether the effect of the intervention on diarrhoea was expressed as a prevalence ratio (binary outcome), a cumulative RR (binary outcome) or an incidence rate ratio (count variable). We extracted the number of participants and events to calculate risk : prevalence ratios. We extracted the number of events and person-time at risk for the calculation of incidence rate ratio. TC entered the extracted data into Review Manager 5.
Assessment of risk of bias in included studies
Two authors independently assessed the risk of bias of each trial using an assessment form. While the protocol for the review contemplated assessing RCTs using the generation of allocation sequence, allocation concealment, blinding, and loss to follow-up, no RCTs were actually identified that met the review's eligibility criteria. We assessed quasi-randomized and non-RCTs using the following criteria.
- Comparability of characteristics between the intervention and control groups with respect to relevant baseline characteristics such as water quality, diarrhoeal morbidity, age, socioeconomic status, access to water, hygiene practices, and sanitation facilities. We classify this as adequate if no substantial differences are present; unclear if they are not reported or it is not known if substantial differences exist; or inadequate if one or more substantial difference exists.
- Data collection for intervention and control groups at the same time. We classify this as adequate if the data were collected at similar points in time; unclear if this is not reported or not clear from the trial; or inadequate if data were not collected at similar points in time.
- Loss to follow-up. We classed the inclusion of randomized participants in the analysis as adequate if 90% or more of all participants enrolled at the outset of the trial were included in the analysis; unclear if it is not clear what portion of the participants who were enrolled at the outset of the trial were included in the analysis; or inadequate if less than 90% of all participants enrolled at the outset of the trial were included in the analysis.
Assessment of reporting biases
We plotted the RR against number of clusters in each study to explore the existence of publication bias. While this is not a classic funnel plot where the effect size is plotted against the standard error of the effect, this approach was the only option since most studies included too few clusters to calculate standard errors (Donner 2000).
We compiled the data using Review Manager 5. Although trials of environmental intervention such as sanitation assess outcomes on an individual level, the unit of randomization is often not the individual but a household, a group of households, a school, a neighbourhood or a village. Some trials correct for this design effect by adjusting for the intra-cluster correlation.
Where applicable, we calculated overall point estimates and 95% CI as the mean of the log cluster-level data, following methods for paired and unpaired cluster randomized trials suggested by Bennett and colleagues (Bennett 2002). If cluster-level disease rates were not reported, we relied on the point estimate and 95% CI given by the trial authors, provided that clustering had been accounted for in the analysis. In trials meeting the inclusion criteria but failing to adequately adjust for clustering (or to provide the data to allow adjustment), we extracted the unadjusted point estimates but rejected the 95% CI. We did not calculate the CI for trials with fewer than four clusters per study arm (Donner 2000). We pre-specified that cluster randomized trials that did not adjust for clustering would not be combined with individual randomized trials in the meta-analysis or tables. Due to substantial methodological heterogeneity among the studies included in the review and the absence of unreliable CI for most estimates of effect, we determined that a meta-analysis was inappropriate.
Subgroup analysis and investigation of heterogeneity
Due to substantial clinical and methodological heterogeneity among the studies included in the review, we determined that subgroup analysis to investigate heterogeneity in outcomes was inappropriate. For the same reason, no sensitivity analysis was undertaken.
Description of studies
Execution of the search strategy yielded 2028 titles and abstracts. These titles and abstracts were screened, and the full text articles of 38 studies were obtained for further assessment. Of these 38 studies, 13 met the review's inclusion criteria (see Characteristics of included studies), while 25 were excluded for other reasons (see Characteristics of excluded studies). Of the 13 included trials, all were published in journals. Seven of the studies were published in Chinese (Hu 1988, Wei 1998; Xu 1990; Xu 1994; Yan 1986; Zhang 2000; Zhu 1997), five in English (Aziz 1990; Garrett 2008; Huttly 1990; McCabe 1957; Rubenstein 1965), and one in French (Messou 1997). We worked with the original language version of each study.
In all the trials included in the review, the intervention was allocated at a cluster (village/community, school, household) level and not at the individual level. Further details on the number and types of clusters is included in Characteristics of included studies.
The trials used a variety of methods for defining, assessing, and reporting outcomes. Eight trials used the WHO definition of diarrhoea (three or more loose stools in 24 hours) as the case definition of the disease (Aziz 1990; Garrett 2008; Hu 1988; Huttly 1990; Wei 1998; Xu 1994; Zhang 2000; Zhu 1997), two used hospital records (Rubenstein 1965; Xu 1990), and three reported no case definition (McCabe 1957; Messou 1997; Yan 1986). The episodes were either reported to the investigator without clinical confirmation (Aziz 1990;Garrett 2008; McCabe 1957; Messou 1997; Wei 1998; Xu 1994; Yan 1986), recorded by the participant or head of household (Huttly 1990; Zhu 1997) or based on hospital admission records (Rubenstein 1965; Xu 1990). Recall for diarrhoeal disease episode reporting was over periods of seven days (Aziz 1990; Garrett 2008), eight days (Huttly 1990), 15 days (Messou 1997), one month (Hu 1988; McCabe 1957; Xu 1994; Yan 1986) or a calendar season (Wei 1998). In one study, surveillance was through individuals reporting to clinics, with a four-week follow-up of non-reporters (Zhang 2000). For each of the studies included in the review, the measure of diarrhoeal disease frequency was the cumulative incidence risk and the estimate of the effect of the intervention was a RR.
Only Garrett 2008 adjusted the data for clustering at the household level and for repeated observations of the same patients. As data collected in such cases are not independent observations, the failure to make such adjustments may result in overstating the precision of the measures of effect reported by such studies.
In addition to diarrhoea, other health outcomes included anthropometrics (Huttly 1990), outpatient visits and admissions for diarrhoea and all causes (Rubenstein 1965), the prevalence of positive shigella cultures from stool samples (McCabe 1957), and deaths related to diarrhoea (Messou 1997),
Study participants and settings
Details of the participants and setting for each trial appear in the Characteristics of included studies. Collectively, the 13 trials included in this review covered at least 33,417 participants, excluding two studies (Hu 1988; Wei 1998) that did not report the number of participants. The study populations ranged from 124 to 14,787 participants, with a mean of 3245 and a median of 1732.
Five studies reported outcomes for all ages (McCabe 1957; Xu 1990; Xu 1994; Yan 1986; Zhang 2000). The other studies limited participants to children under one year of age (Rubenstein 1965), under five years of age (Aziz 1990; Garrett 2008; Messou 1997), under six years of age (Hu 1988; Huttly 1990), and primary or secondary school children (Wei 1998; Zhu 1997). As the effect of sanitation on diarrhoea and certain infections may vary considerably with age, the comparability of results must be viewed in light of these significant differences in ages of the study populations.
All the included studies were conducted in low-income or middle-income settings. Seven were conducted in China (Hu 1988; Wei 1998; Xu 1990; Xu 1994; Yan 1986; Zhang 2000; Zhu 1997), two in rural USA (McCabe 1957; Rubenstein 1965), and one each in Bangladesh (Aziz 1990), the Ivory Coast (Messou 1997), Kenya (Garrett 2008), and Nigeria (Huttly 1990). Most studies were done in rural settings, though others were done in urban, suburban or school settings (Characteristics of included studies).
There was uncertainty, and considerable heterogeneity, in the pre-intervention (control) settings in which the interventions were implemented ( Table 1). Six of the included studies did not provide clear information on the pre-intervention excreta disposal facilities of the study population even though this served as the control. Among the other seven studies that did provide this information, one reported open defecation (Aziz 1990) or a combination of open defecation or open pits (Garrett 2008; Huttly 1990). Others described the sanitation facilities of the controls as open pits (Zhang 2000), 'pit latrines' (Yan 1986), 'unsatisfactory facilities, including surface privies' (McCabe 1957) or 'shallow pit or single urns' (Hu 1988). Other pre-intervention (control) conditions potentially relevant to the outcome of interest also varied or were not described in the studies. With respect to water supplies, five studies reported systems that would qualify as improved under WHO/UNICEF JMP definitions, two were unimproved and six studies provided insufficient information from which water supplies could be assessed ( Table 1). None of the studies reported previous hygiene instruction in the settings prior to undertaking the intervention.
The studies included in this review involved a wide variety of interventions to improve excreta disposal facilities ( Table 2). These consisted of the promotion or construction of VIP latrines (Huttly 1990), VIP latrines or sanitary platforms over pit latrines (Garrett 2008); borehole latrines with wooden superstructures and concrete slabs (McCabe 1957); household flush toilets connected to septic tanks or biogas reactors (Hu 1988); private latrines connected to a piped water system (Rubenstein 1965; Zhang 2000); private, multi-compartment water-sealed toilets (Aziz 1990; Yan 1986; Zhang 2000); school-based latrines (Wei 1998; Zhu 1997), and public latrines (Messou 1997; Xu 1990).
In two studies the intervention appeared to be mainly concerned with the capture and use of human faeces in biogas reactors (Hu 1988; Xu 1994 ). While this intervention meets the eligibility criteria for this review, it may have been driven by energy rather than health objectives.
As can be seen in Table 2, only five studies (Hu 1988; McCabe 1957;Xu 1990; Xu 1994; Yan 1986) included interventions that consisted solely of improvements to excreta disposal. In all but one of the other studies (Zhu 1997), the improvement in excreta disposal was, at the least, accompanied by improvements in water supplies, and in all but two studies (Rubenstein 1965; Zhang 2000), there was also a hygiene promotion component. In the other studies the excreta disposal and water supply activities were combined with other activities that could potentially impact on diarrhoea. Garrett 2008 evaluated a project involving not only improved latrines, water supplies (shallow wells and rainwater harvesting) and hygiene promotion but also point-of-use water treatment with chlorine. Wei 1998 also included the promotion of point-of-use water treatment (boiling water at school). The intervention reported in Messou 1997 included the provision of oral rehydration salts (ORS). Several studies included steps to improve the management of excreta disposal in addition to infrastructure enhancements (Wei 1998; Yan 1986; Zhu 1997). For these multiple-component interventions it is not possible to isolate the effect of the improvement in excreta disposal or ascribe the difference in outcome solely to thesanitation component.
Despite the diversity in technologies and types of improvement in excreta disposal, it appears that all the interventions would result in 'improved' sanitation, as the term is used by the JMP (WHO/UNICEF 2002), except for those implemented in schools (Wei 1998; Zhu 1997 ) or other public settings (Messou 1997; Xu 1990), where the improvements would fail to meet the definition since they were shared facilities.
Coverage and use of the intervention
In most cases, details on the baseline coverage - and in some cases, the end-point coverage - were not reported ( Table 1; Table 2). Garrett 2008 reported that at baseline, 39% of intervention households had latrines compared with 28% of the control households; at the follow-up point eight weeks after the launch of the intervention, coverage had increased among the intervention group, but to just 49%, compared to 27% for the controls (P < 0.001). McCabe 1957 observed that 52% of the intervention community had "unsatisfactory facilities" at the outset of the study, while the others were served by a community sewer treatment system; following implementation of the intervention, all households had satisfactory facilities. Two other studies reported that at the end of the intervention period, latrine coverage was 80% (Xu 1994) or 85% (Yan 1986): in neither of these cases, however, did the investigators report latrine coverage in the control group or at baseline. Studies evaluating public latrines reported coverage of one toilet for 10 inhabitants (Messou 1997). Other studies did not actually report on on the increase in coverage that was achieved as a result of the intervention.
Some studies reported on latrine use rather than intervention coverage. Aziz 1990, who reported that use of the latrine was constantly monitored, recorded that 88% of households used the latrines in 1987 against 83% in 1993. In 1993 a total of 64% of the latrines available in 1987 were still functioning properly. A subsequent evaluation reported that that most children aged 36-59 months were said to use latrines, but few below this age did so (Hoque 1996). Huttly 1990 reported that 46% of households at the intervention were using a VIP latrine.They also observed that latrine use was high among adults but low (19%) among children. McCabe 1957 observed that "almost everyone" used the installed privies.
Risk of bias in included studies
In none of the studies was the allocation of the intervention among the participating clusters random. Rather, in each case, the intervention was allocated either by the researchers (Wei 1998,Zhu 1997) or the reasons were not specified in the reports or were provided subsequently (see Characteristics of included studies ). The allocation was not concealed and the intervention was not blinded.
The studies included in this review were assessed for risk of bias based only on the comparability of characteristics between the intervention and control groups and the contemporaneitys of the data collection. All studies met the definition of 'adequate' for the contemporaneity of the data collection, and 11 studies met the definitions of 'adequate' under both criteria ( Table 3). Information on loss to follow-up is presented in Table 3. Only three studies reported loss to follow-up or provided information from which it could be calculated. Among these, only Messou 1997 and Xu 1990 had a less than 10% loss to follow-up and were thus characterized as adequate on this criterion.
Because of the lack of random allocation and the subjective reporting of diarrhoea as outcome measure, the risk of selection bias as well as of observer and responder bias in all trials must be considered high, even where the aforementioned criteria are satisfied.
Effects of interventions
The measure of effect for each of the 13 studies included in the review appears in Summary of findings for the main comparison.
Only two of the 13 studies included in this review reported no protective effect from the intervention (Huttly 1990; Xu 1994). Huttly and colleagues suggested that the intervention may not have been effective in preventing diarrhoea in the particular setting due to problems in implementation of the sanitation intervention that led to low levels of coverage and to low utilization by children.
Except for two studies (Garrett 2008; Xu 1994), CI could not be provided due to the insufficient number of clusters for which data was reported (Donner 2000). For the Garrett study, we relied on theCI provided by the authors. It is noted that the authors reported adjusting for clustering. For the Xu 1994 study we calculated the CI according to the methods proposed by Hayes and Moulton (Hayes 2008).
Due to the heterogeneity of studies and unavailability of reliable CI, no pooled effect was calculated. In general, however, the consistency of a protective effect against diarrhoea seen across the included studies indicates that these findings may be unlikely to be due to chance alone if there were no other factors that could explain the effect (ie baseline differences between communities, observer / responder bias).
Only one trial reported on mortality as a study outcome. Messou 1997, which involved a combination of improved latrines with source water improvement, an oral rehydration intervention and hygiene instruction, reported an 85% reduction (from 27% to 4%) in the proportion of deaths related to diarrhoea in the villages with the intervention compared with no reduction in control villages. That trial also reported an 85% reduction (from 5.3% to 0.8%) in the mortality associated with diarrhoea among intervention villages with no correspondingly decline in control villages. We emphasize that the trial was primarily designed to investigate the impact of the intervention on death using a before and after study design, which is not optimal, especially since only two intervention and two control villages was included. Direct comparison between intervention and control villages resulted in a RR of 0.16 (mortality in intervention arm 2%, in control arm 11%). CIs could not be calculated due to an insufficient number of clusters.
None of the trials reported on adverse events from the intervention.
We were unable to create classic funnel plots that plot the effect size against the standard error of the effects, a method suitable to explore publication bias. However, the number of clusters can be regarded as a proxy for study power. Figure 1 shows the graph plotting the effect size against the number of clusters. The figure shows that studies with fewer clusters reported markedly larger effect sizes compared to studies that included more clusters. It is noted, however, that such asymmetry may also be due to clinical and methodological heterogeneity. Since we found substantial evidence of such heterogeneity, we cannot conclude that the funnel plot convincinlgy demonstrates evidence of publication bias in this case.
|Figure 1. Effect size according to number of clusters per arm to explore publication bias|
Results suggest that interventions to improve excreta disposal are effective in preventing diarrhoeal disease. While the studies reported a range of effects, most found the intervention to be protective against diarrhoea. This is true not only for studies that combined the sanitation intervention with improvements in water or hygiene, but also for those that consisted solely of improvements in excreta disposal.
Nevertheless, this review provides only limited and preliminary evidence, and does not allow the quantification of such an effect. We were able to identify a significant number of studies that met the review's inclusion criteria, including five studies where the improvement in excreta disposal was not accompanied by other environmental interventions and thus could be investigated independently of activities that are also believed to prevent diarrhoea. Few if any of these studies were included in previous reviews of sanitation interventions (Esrey 1985; Esrey 1991; Fewtrell 2005; Waddington 2009). However, substantial heterogeneity among interventions, settings, and methodologies, and the absence of reliable CIs for most studies made it impossible to use meta-analysis to calculate a pooled estimate of effect from the intervention. Accordingly, while we extracted and present information on pre-intervention settings, type of intervention, coverage, compliance, and other components of the intervention, the analysis is only descriptive as no pooling of effects was possible. Despite most studies reporting a protective effect, it is not possible to actually derive a reliable estimate of the size of the effect.
Moreover, the strength of evidence must be qualified by certain methodlological issues presented by the studies included in the review. Firstly, while we were able to identify 13 studies that met the review's inclusion criteria, none randomized the intervention among the clusters comprising the study population. Most of these compared one or only a few intervention sites with a similar number of control sites. Many of the reasons for choosing one community for the intervention over the other may well be associated with disease risk (eg willingness to co-operate, the presence of infrastructure, level of education, wealth), potentially introducing a systematic bias. Secondly, none of the studies included in the review assessed the effect of the sanitation intevention on intermediate outcomes, such as the quality of drinking water, microbial contamination of foods or presences of flies, which could suggest an objective impact of the intervention on common transmission pathways. Thus, there is no independent way to confirm that the intervention reduced exposure, much less disease.
A further shortcoming of most studies is the potential for observer and responder bias in assessing the disease outcomes. None of the studies was blinded, though this may be unavoidable with sanitation interventions. Most also relied on reported diarrhoea. The assessment of diarrhoeal diseases by active surveillance conducted by field workers is prone to cause bias both on the side of the study participants (responder bias) and the field worker (observer bias). These factors are known to potentially shift the effect measure towards a greater effect (Wood 2008). It may be possible to minimize such bias by including a passive surveillance component, where households are encouraged to seek treatment at a specified health centre, as was done in two of the included studies. However, the extent to which this outcome has increased validity has yet to be investigated.
In addition to the methodological limitations of all studies, differences in study populations, and settings, baseline sanitation levels, water, and hygiene practices, methodologies, case definitions, and outcome surveillance, and types of interventions limit the comparability of results from the studies included in this review. Failure to record or adjust for differences in the coverage and use of the excreta disposal intervention in many studies also raises issues about the conclusions that can be drawn from the findings. It may be expected, for example, that increasing coverage or use from 10% to 70% would yield different outcomes from those obtained by increasing coverage from 60% to 70%; however, this information is missing from most studies. Population density may also play an important role in the effectiveness of sanitation interventions but details on on density are absent from most studies.
Moreover, only five of the 13 included studies consisted solely of such improvements in excreta disposal. In all the other studies included in this review the sanitation intervention was accompanied at least by improvements in drinking water supply. And in some studies, there were hygiene promotion or other components to the intervention. Of the five sanitation-only studies, four were concentrated geographically in a single country (China) and the other was conducted in a developed country (USA) more than 50 years ago.
The studies included in this review are suggestive of the wide variety of interventions being undertaken to improve excreta disposal in low-income settings, both at the household and community level, and the extent to which they may be effective in minimizing human contact and pathogen exposure. They also suggest the considerable variations in quality, coverage, use, and sustainability of the interventions. Even uniform interventions implemented in a manner that is equally effective in containing excreta are nevertheless likely to yield different levels of effectiveness in reducing diarrhoea, depending on other exposure pathways. Thus, single, pooled estimates of the contribution that sanitation can make to preventing diarrhoeal disease are not only methodologically unsound but also misleading.
This review highlights some of the challenges in estimating the contribution of sanitation interventions to prevent diarrhoea and some of the shortcomings in the studies conducted to date. However, the more salient conclusion is the paucity of rigorous studies on the effectiveness of these sanitation interventions given the substantial burden of disease that is associated with direct or indirect contact with faeces. Research could help identify not only innovative solutions but also the key factors associated with the effectiveness of sanitation interventions and how they can best be targeted and delivered. This lack of research on the health impact of sanitation parallels, and may be partially responsible for, the lag in progress in extending sanitation coverage as reflected in the vast shortfall in achieving the MDG sanitation target.
Implications for practice
This review provides some evidence that excreta disposal interventions are effective in preventing diarrhoeal diseases. However, this conclusion is based largely on the consistency of some protective effect (11/13 studies). The quality of the evidence is generally poor and does not permit quantification of any such effect. The wide range of estimates of effect from the intervention may be due to clinical and methodological heterogeneity among the studies, including differences in study design, case definitions, outcomes, method, and length of follow-up, and methods of estimating measures of effect. These and other differences could impact on levels of observer and responder bias. There were also important differences in the study populations, the levels of ambient exposure, the types of interventions, and the levels of coverage and use. The range in effect may also suggest that the contribution that excreta disposal interventions can make in preventing diarrhoea may depend on the local context and the exposure scenarios, and transmission dynamics that research to date cannot fully explain.
Implications for research
Perhaps the major finding of this review is the paucity of rigorous evidence demonstrating the effect of basic sanitation in preventing diarrhoea, a leading killer of young children. This is clearly not from a lack of attention to this sector by the public health community at the highest international level. Sanitation has attracted support (if not funding) at that level over the past 30 years. Neither is it because sanitation interventions cannot be assessed using experimental designs: Emerson 2004 demonstrated how a cluster-based approach can be used to randomize excreta disposal interventions in a rigorous cluster RCT study design. While the MDG target for sanitation is intended to inspire the political will to advance the implementation of basic sanitation, it is possible that the pace of implementation is being retarded by this dearth of reliable evidence of the health outcomes that may be achieved thereby, and how they vary with exposure setting, type of improvement, and coverage achieved. Future research should address this need.
Rigorously conducted RCTs to assess interventions to improve basic sanitation will help clarify the potential contribution of such interventions. While conventional blinding may not be possible, it is important that such trials take steps to minimize an exaggerated effect from bias associated with open trials of non-objective outcomes. If possible, such studies should assess the impact of sanitation on objective intermediate outcomes associated with exposure to excreta, such as contaminated food and water. Limiting exposure to child faeces, using potties or scoops, and properly disposing of their contents may be a particular priority, but no intervention studies have been conducted to date to determine or optimize the effect of such measures. Multiple trials in different settings will also help identify the circumstances in which improvements in excreta disposal should be targeted and given priority.
The effect of sanitation in the context of other environmental interventions to prevent diarrhoea should also be explored. Because sanitation is a primary barrier to faecal-oral transmission it seems plausible under some circumstances to prevent most diarrhoea by implementing sanitation even without improvements in water supply or quality. It is also important to evaluate such interventions in rural versus urban settings where the challenges of implementation, transmission pathways, and exposure levels may vary. In high density settings it would also be useful to investigate whether and to what extent any benefit from increasing coverage and the use of latrines is conferred on non-adopters, as with insecticide-treated bednets, rendering the health impact of the intervention a "public good" in economic terms.
There is also a need for longer-term effectiveness studies in programmatic (not-research driven) settings. Rigorous observational studies and project evaluations can also contribute valuable evidence on the scalability and sustainability of sanitation interventions. Differences in programmatic approaches to optimize the adoption and long-term utilisation of sanitation should also be investigated.
Finally, we note that this review does not address the potential contribution of improved excreta disposal to preventing other important health threats associated with inadequate sanitation, including schistosomiasis, other helminth infections, malnutrition, stunting and tropical enteropathy.
The editorial base for the Cochrane Infectious Diseases Group is funded by the UK Department for International Development (DFID) for the benefit of low- and middle-income countries.
Data and analyses
This review has no analyses.
Appendix 1. Search methods: detailed search strategies
Appendix 2. Data extracted from included studies
Protocol first published: Issue 2, 2008
Review first published: Issue 6, 2010
Contributions of authors
Sandy Cairncross and Thomas Clasen conceived of the review. All the authors assisted in the design and development of the protocol, including the literature review used in the Background section. Thomas Clasen and Kristof Bostoen developed the search strategy. Wolf-Peter Schmidt and Sophie Boisson provided additional epidemiological and statistical support. Isaac C-H Fung assisted in developing the search strategy for the Chinese-languagee databases. Marion Jenkins, Beth Scott, and Steven Sugden provided technical advice on sanitation interventions. All authors provided general advice for the review, commented on the protocol, and gave final approval on the version to be published.
Declarations of interest
Sources of support
- No sources of support supplied
- WaterAid, UK.
- UNICEF, USA.
Medical Subject Headings (MeSH)
MeSH check words
Adult; Child; Humans
* Indicates the major publication for the study