Study area and population

The study was carried out in the Taabo HDSS, located in south-central Côte d’Ivoire [26]. The study area and the baseline situation with regard to hygiene and sanitation conditions have been described elsewhere [27]. In brief, open defecation is common. In addition to agricultural activities, local populations raise animals such as pigs, goats, sheep and chicken in their immediate living environment that may also influence disease transmission (e.g. pigs consume human faeces). Preventive chemotherapy was implemented in 2008, 2009 and 2010 using albendazole, praziquantel and ivermectin for the control of soil-transmitted helminthiasis, schistosomiasis, onchocerciasis and lymphatic filariasis.

The study was implemented in two villages (Katchénou and Sahoua), and seven hamlets belonging to different villages, Yobouékro (village Sahoua), Ouattafouékro and Kouadio Kouamékro (Ahondo), Boussoukro (Tokohiri), Amani Kouadiokro (Sokrogbo) and Bêh N’Guessankro and Allah Thérèsekro (Léléblé). These villages and hamlets were selected because of their characteristics that are in favour of meaningful and successful implementation of CLTS, namely (i) small population sizes; (ii) clear signs of practiced open defecation; (iii) inhabitants that have the potential to become natural leaders; and (iv) relatively homogeneous population structure in terms of culture and socioeconomic status. Communities were randomly assigned to the intervention and control group taking into account for matching characteristics such as population size, hygiene status, village affiliation and geographic position. All residents of the villages and hamlets were invited to participate. During the study period a non-governmental organisation (NGO) visited Amani Kouadiokro and carried out CLTS. Given that for our study this hamlet was a control site, we considered this location retrospectively to be within the intervention group for analysis. Although this community had not received the additional health education package it qualified as intervention community in light of rapidly achieved ODF status that is the ultimate goal of the intervention and to be related with all outcomes assessed at follow-up.

Study design

Figure 1 shows the study design. In July 2011, shortly before the annual round of preventive chemotherapy, a cross-sectional survey was conducted to determine peoples’ infection status with helminths and intestinal protozoa and to assess their KAPB regarding helminthiases and hygiene [27]. Interventions (i.e. CLTS and health education) were implemented subsequently, including a second round of preventive chemotherapy. Approximately 1 year after baseline, a follow-up survey was conducted, using the same field and laboratory procedures.

Parasitological survey

Our research team visited all villages and hamlets to obtain local approval and to explain the parasitological survey and questionnaire interviews. The day before the survey, field enumerators from the Taabo HDSS and community health workers distributed plastic containers (125 ml) to all households for stool and urine collection. Participants were invited to submit early morning stool and urine samples on the following day to a communal central place in the village/hamlet. For parasitological examinations, faecal and urine samples were transferred to a mobile field laboratory in Léléblé or Sokrogbo, or the laboratory of the hospital in Taabo-Cité depending on the sampled localities’ proximity to the respective laboratories.

Stool and urine samples were processed and examined by experienced laboratory technicians on the day of collection. In brief, stool samples were subjected to duplicate Kato-Katz thick smears, using 41.7 mg plastic templates [28]. The number of eggs of Schistosoma mansoni, Ascaris lumbricoides, Trichuris trichiura and hookworm were counted and recorded separately. Egg counts were multiplied by a factor of 24 to estimate the number of eggs per gram of stool (epg) [29]. Urine samples were subjected to a filtration method. Ten ml of vigorously shaken urine was passed through a Nytrel filter (Sefar AG; Heiden, Switzerland; diameter: 13 mm, mesh size: 20 μm). Individual filters were placed on a microscope slide, a drop of Lugol’s iodine added and then examined under a microscope. The number of S. haematobium eggs were counted and expressed as eggs per 10 ml of urine [30]. For quality control, 10% of the samples were re-examined by a senior laboratory technician and discrepancies discussed until accordance was reached.

Additionally, from each stool sample 1–2 g was processed using an ether-concentration method. The sodium acetate-acetic acid-formalin (SAF)-fixed stool samples were forwarded to a laboratory at the Centre Suisse de Recherches Scientifiques en Côte d’Ivoire (CSRS; Abidjan, Côte d’Ivoire) for examination of intestinal protozoa infections. Standard protocols were followed and intestinal protozoa (Blastocystis sp., Chilomastix mesnili, Entamoeba coli, Entamoeba hartmanni, Entamoeba histolytica/E. dispar, Endolimax nana, Giardia intestinalis and Iodamoeba bütschlii) recorded semi-quantitatively [31].

Questionnaire survey

For the interviews with KAPB questionnaires, households were visited by a researcher accompanied by a trained field enumerator from the Taabo HDSS who speaks the local language. The questionnaire was addressed to the household head or a present adult household member in case of absence of the earlier. The questionnaire has been presented in detail elsewhere [27]. In brief, the questionnaires consisted of basic questions pertaining to demographic factors, socioeconomic indicators, sanitation, hygiene and defecation behaviour, opinions, taboos and beliefs, and concepts on intestinal parasitic infections.

CLTS and health education

The CLTS intervention was carried out according to Kar & Chambers [21]. Pre-triggering was integrated in the information visit to communities about the upcoming surveys and further enhanced during the KAPB survey, while information was collected and defecation sites located. Subsequently, a workshop was organised to train personnel from the Taabo HDSS, local health authorities, members of the mayor’s office and other stakeholders from the sub-prefecture of Taabo with regard to the CLTS approach. At this event, experienced CLTS facilitators from different institutions (CSRS, UNICEF and the Ministry of Health and Public Hygiene) presented the CLTS approach and the different tactics of implementing CLTS and leading the community to the ignition moment. Teams were then built, each including a lead-facilitator, co-facilitators, content and process recorders and one person responsible for organisational issues and communities that were chosen for the intervention visited. The visit (triggering) lasted half a day and went as follows. First, a meeting with the community leaders was held, encouraging all community members to participate. Secondly, several activities were done with the community members, including mapping of community and defecation areas, identifying dirtiest neighbourhoods on the map, calculation of “caca” and medical expenses due to diarrhoea, transect walk through the open defecation areas (“walk of shame”), triggering disgust of faecal contamination of food, water or hands [showing, for example, how flies fly from faeces to food (“caca-to-food”)]. The latter activity is where the ignition moment took place in general, namely when people realise that they are eating each other’s faeces. In a next step, the community was motivated to plan their decisions, actions and time schedule until when the community will be ODF. Third, a closing ceremony was organised by the team where active and motivated community members identified during the intervention (natural leaders) were invited to present in front of all invited local authorities their plan of action (Fig. 2).

After triggering, the action of the community was supported and followed-up by regular visits every 2 weeks by the facilitators. During these follow-ups, the facilitators visited the communities to assist and monitor the construction of latrines. People were constantly encouraged to help each other to build latrines or share latrines until every household had its own latrine [21, 22, 32]. ODF status was evaluated 10 and 28 weeks after triggering through active search for remaining open defecation sites during unannounced visits in the community and its surroundings.

The whole process of CLTS was supported by specific health education sessions using participatory rural appraisal tools. This approach included three major aspects, namely (i) evaluating existing knowledge; (ii) health education provided to the whole community and to separate groups (e.g. men, women, children and health committees) through focus group discussions led by a social scientist; and (iii) set up of an action plan for continued provision of health education. The facilitators discussed the health risks associated to inappropriate hygiene, open defecation and other risk behaviour with groups of community members. Risk behaviour for helminth and intestinal protozoa infections, transmission cycles and preventive measures were described using visual tools such as booklets, posters and pictures. These tools were handed over to community health committees in charge of promoting good hygiene behaviour in the community that were built as part of the intervention.

Statistical analysis

Data were double-entered and cross-checked using EpiInfo version 3.5.1 (Centers for Disease Control and Prevention, Atlanta, GA, USA) and analysed using STATA version 10.0 (Stata Corp., College Station, TX, USA). Only participants with complete datasets (i.e. those with duplicate Kato-Katz thick smears, one SAF-fixed faecal sample and one urine filtration) and confirmed administration of albendazole treatment were included in the final analyses. Participants were stratified into five age groups (i.e. < 6; 6–15; 16–29; 30–45; and > 45 years). Logistic regression models for infection status, behaviour and awareness with regard to defecation and hygiene were used to assess the association with the intervention at baseline and the 1-year follow-up. Prevalence reduction was defined as (prevalence at baseline – prevalence at follow-up). Negative binomial regression models were used to assess the association between faecal egg counts of single helminth species infections and the intervention. Egg count reduction, expressed as egg reduction rate (ERR), was calculated as [1 – (geometric mean epg at follow-up/geometric mean epg at baseline)] multiplied by a factor of 100. Significant changes between baseline and follow-up frequencies for infection status, behaviour and awareness with regard to defecation and hygiene within the respective group were analysed using the McNemar’s test. To assess differences over time in infection status between control and intervention communities, taking into account for different baseline prevalence levels, we compared (i) total proportion of change; (ii) proportion of decrease; and (iii) proportion of increase [33, 34]. Differences in proportion of change were considered significant in case of non-overlapping 95% confidence intervals (CIs) based on the following formula:

where SE is the standard error, p_g1 is the proportion of change in group 1, p_g2 is the proportion of change in group 2, N_g1 is the number of subjects in group 1, and N_g2 is the number of subjects in group 2.

Univariate and multivariable relationship analysis to assess guiding risk factors for the follow-up infection status was done using logistic regression modelling, including a random effect coefficient on community level. Results of this analysis were presented as crude and adjusted odds ratios (ORs). The latter were adjusted for sociodemographic indicators, namely sex, age group, socioeconomic status and ethnic origin.

Participants

There were 3420 people living in the nine selected communities according to the Taabo HDSS census database and 3289 were present and participated during the baseline survey in 2011. A total of 2413 provided stool samples, 2294 submitted urine samples and questionnaire data were available from 3152 people. Overall, 1894 people had complete parasitological and questionnaire data. Results from the baseline survey have been presented elsewhere [27]. For the 1-year follow-up survey done in 2012, a total of 3290 people were present, of which 2064 provided stool and 1902 urine samples, while questionnaire data were obtained from 3182 people. Complete parasitological and questionnaire data were available from 1481 people.

A total of 810 individuals had complete parasitological and questionnaire data from both the baseline and 1-year follow-up surveys and had received anthelminthic treatment after the second ODF evaluation in March 2012. Age and sex characteristics of this final cohort are presented in Additional file 1: Table S1.

Community response following the interventions

Table 1 summarises information on latrines owned by households before and after the intervention. At baseline, coverage of latrine ownership at the unit of the community varied from 0% (i.e. Katchéneou) to 50% (i.e. Allah Thérèsekro). Following two rounds of evaluations, four out of five intervention communities reached a 100% household coverage with latrines and were verified ODF. The remaining intervention community reached a latrine coverage of more than 80% but continued open defecation was still recorded. The proportion of latrine ownership at follow-up in control communities remained comparable with the baseline proportion. Dynamics of latrine construction among intervention communities are shown in Fig. 3. Kouadio Kouamékro and Amani Kouadiokro (not shown in figure) already reached 100% latrine coverage and ODF status after the first environmental evaluation 10 weeks after triggering. The other three intervention communities only started to extensively construct latrines following a specific health education package.

Type of intervention	Community	Registered inhabitants (n = 3420)	Registered households (n = 487)	Households with latrines at baseline (%)	Households with latrines at follow-up (%)	ODF statusc
Preventive chemotherapy	Allah Thérèsekro	320	56	28 (50.0)	34 (60.7)
Preventive chemotherapya	Boussoukro	294	48	12 (25.0)	15 (31.3)
Preventive chemotherapya	Ouattafouékro	157	29	1 (3.4)	1 (3.4)
Preventive chemotherapy	Sahoua	872	118	25 (21.2)	31 (26.3)
Preventive chemotherapy + CLTSb	Amani Kouadiokro	253	24	0 (0)	24 (100)	Yes
Preventive chemotherapy + CLTS + health education	Bêh N’Guessankro	393	55	15 (27.3)	45 (81.82)	No
Preventive chemotherapy + CLTS + health education	Katchénou	713	91	0 (0)	91 (100)	Yes
Preventive chemotherapy + CLTS + health education	Kouadio Kouamékro	171	29	1 (3.4)	29 (100)	Yes
Preventive chemotherapy + CLTS + health education	Yobouékro	247	37	3 (8.1)	37 (100)	Yes

Abbreviations: CLTS community-led total sanitation, ODF open defecation free

^aBoussoukro and Ouattafouékro are two hamlets situated 1 km apart from each other. For randomisation these two hamlets were considered as one unit

^bAmani Kouadiokro, a hamlet at the border of the sub-prefecture, was initially attributed to the control group during randomisation, however, an NGO intervening at the adjacent sub-prefecture erroneously visited the hamlet and carried out a CLTS intervention. Thus, this community was attributed to the intervention group for analysis of parasitological data

^cTwo evaluations took place. The first evaluation was implemented on 26–27 November 2011, the second took place on 19–25 March 2012. Kouadio Kouamékro and Amani Kouadiokro were certified ODF following the first evaluation, whereas Katchénou and Yobouékro were certified ODF following the second evaluation

Impact of the intervention on prevalence and intensity of infection

Helminth infection intensities at baseline and follow-up are shown in Additional file 2: Table S2. ERRs for hookworm were considerably higher for intervention than control communities (64.9 vs 15.2%) due to higher baseline egg counts in the intervention communities. There was no significant difference in hookworm egg counts between intervention and control group after treatment.

Risk factors guiding infection status at follow-up

Figure 4 shows reinfection patterns with STHs in intervention and control communities, stratified by age group and sex. While soil-transmitted helminth reinfection patterns by age group in control communities followed baseline patterns with highest infection rates among adolescent and young adults (16–29 years), a peak shift towards school-aged children (6–15 years) was observed in intervention communities. In intervention communities 19.8% of school-aged children were found soil-transmitted helminth-positive compared to 4.4% in control communities (P <  0.001). In the former, almost half of all new infections (23/50, 46%), were diagnosed in school-aged children. Similar to age, reinfection patterns in control communities showed continued sex differences, as already observed at baseline with significantly higher rates of males infected in 2012 than females (11.2 vs 2.7%, P = 0.001), while no significant sex differences in infection status in 2012 was observed among intervention communities (14.4 vs 8.9%, P = 0.076).

In contrast to soil-transmitted helminth infection, WASH indicators and intervention group (OR = 0.68, P = 0.015) and ODF community (OR = 0.69, P = 0.017) had lower odds for intestinal protozoa infection at the 1-year follow-up. Among the WASH indicators, households with newly constructed latrines (OR = 0.63, P = 0.007) showed the highest effect on reduction of intestinal protozoa infection.

Impact of the intervention on KAPB

Community response and latrine construction

Our package of interventions resulted in a strong response in the five selected communities. Indeed, four out of the five communities were declared ODF within six months, whilst in the fifth community more than 80% of the households had latrines (Fig. 3). Particularly small communities (e.g. Amani Kouadiokro and Kouadio Kouamékro) rapidly reached 100% latrine coverage, and hence, were certified ODF. This corroborates recent findings from a CLTS programme conducted in Ghana that found the largest impact in small, remote villages with low exposure to prior water and sanitation projects [41]. Hamlets and small villages have the advantage to be usually socially cohesive with a more clear and respected hierarchy that may result in a faster and higher degree of social mobilisation of the associated community members. In larger villages, such as Katchénou where 17 weeks after triggering still fewer than 10% of the households possessed a latrine, conducting a health education campaign showed an impressive boosting effect. This underlines the need for integrated approaches combining CLTS or other sanitation programmes with health education. Future interventions in large communities, however, may represent a challenge for community mobilisation and need to address the role of communication and leadership (e.g. by repeated triggering, choice of “good” religious and/or secular leaders and community facilitators) [21, 42]. Compared to other CLTS programmes conducted in Eastern Africa, we reached ODF status within a relatively short period of time. Crocker et al. [43] explain low changes in open-defection reduction and increase in latrine ownership with generally low open-defecation in Ethiopia.

Impact on helminth and intestinal protozoa infection

Interestingly, we did not observe significant differences in soil-transmitted helminth reinfection rates when comparing intervention and control communities. This may be partly explained by still relatively low reinfection rates in both groups (control: 7.0%, intervention: 11.8%), perhaps indicating a too short follow-up period (in our case five months). Our assumption was that infection levels in the control communities after a single round of preventive chemotherapy would have approached baseline levels (here > 25%) within about a year [3]. Impact assessment for helminth infection was further hampered by significant discrepancies in baseline infection levels between groups. While soil-transmitted helminth infections were more prevalent among intervention communities, schistosomiasis showed a highly focally distribution, as it mainly occurred in a single locality that was part of the control group. It should be noted, however, that a negative relationship between hookworm infections and toilet ownership and use was evident in the baseline data of the same study cohort [27]. Further, the finding on a significant difference in the extent of proportion of decrease in hookworm infection indicates that baseline positives that became negative were also more likely to remain negative five months after treatment within the intervention group (Table 3).

For intestinal protozoa, although not tackled through the administration of anthelminthic drugs as part of the preventive chemotherapy component, we found significantly lower infection rates in the intervention communities at follow-up, while infection levels in control communities remained unchanged (Table 2). Baseline uninfected individuals were further found to be less likely to become newly infected with E. nana or Blastocystis sp. if they belonged to the intervention group as highlighted by a statistically significantly lower proportion of increase (Table 3). In the multivariable logistic regression analysis, the following variables were associated with significantly lower odds for intestinal protozoa infections: intervention community, ODF status, ownership and use of toilet. Lower intestinal protozoa infection levels and lower hookworm egg counts in intervention communities may serve as biological indicators for decreased environmental faecal pollution [44] and reduced abundance of flies acting as potential carriers [45]. Hence, our findings strengthen the assumption of a direct impact from the intervention package.

Other studies on community level incorporating a sanitation package to fight enteric parasite infections showed mixed results. Studies that reached high latrine coverage and with long-term follow-up [10] resulted in higher impact on prevalence and incidence, compared to studies achieving low latrine coverage [16] and high levels of continued open defecation [15].

Potential factors guiding infection status at follow-up

Reinfection patterns for soil-transmitted helminths highlight potential sources for continued transmission despite ODF status in intervention communities. School-aged children were at the highest risk of reinfection (23 out of 50 new infections in the intervention communities), which might, at least partly, be explained by children attending the nearest school that might be located in another community not yet declared ODF. Such reinfections might be considered as imported infections, dampening the impact for soil-transmitted helminthiasis control in intervention communities. Another potential source for reinfection could be seen in plantation sites, where open defecation is continued, as latrines were only built in close proximity to the households. In intervention communities, reported open defecation while people pursue agricultural activities remained relatively high (41.1%). Logistic regression modelling indicated potential cultural difference in hygiene practices, as expressed by a higher risk for intestinal protozoa (re)infections among the autochthonous population and reduced risk among Muslim community members. In previous research, males have been found to be more prone for reinfection, as they are less adhering to good defecation practices and hygiene behaviour [9, 46, 47]. In our study, however, this gender-related reinfection pattern was no longer observed in the intervention communities compared to control communities where it persisted. How much of this can be related to positive change in defecation and hygiene behaviour among males and to what extent the interventions triggered it remains to be clarified.

Impact on behaviour and attitudes

Apart from access to sanitation facilities, attitudes towards defecation and hygiene practices, acceptance and (appropriate) use of these infrastructures are key determinants for discontinuation of open defecation and thus successful interruption of faecal contamination of the environment [15, 16]. Our findings showed high proportion of behaviour change within the intervention communities with considerably reduced reported open defecation in general (before: 95.8%, after: 44.6%), going hand-in-hand with an increase in stated toilet use (before: 15.5%, after: 94.6%). Especially open-defecation in close proximity to the households, which probably serves as the main source for infections and transmission via flies, plummeted in both adults (before: 75.0%, after: 16.7%) and children (before: 78.3%, after: 4.5%). Similarly to findings from a CLTS trial in rural Mali [13], household heads from intervention communities significantly more often reported washing hands after defecating. The intervention significantly increased perceptions of open defecation as polluting the environment among beneficiary communities. This may be explained by the emphasis given to activities focusing on projection of the extent of environmental contamination through faeces during CLTS triggering (e.g. mapping, calculating “caca”, transect walk and identifying the dirtiest neighbourhoods) [22].

Nevertheless, it is difficult to say to what extent CLTS alone or the combination of health education and preventive chemotherapy improved people’s behaviour of defecation and general hygiene and cleanliness. Lawrence et al. [48] identified shame from the CLTS triggering as one of the most important factors for behavioural change in rural Zambia. Yet, CLTS may also influence adherence through other social and emotive factors that include disgust, pressure from hierarchical powers and community groups, and competition among villages to achieve ODF status. Health education complements by addressing lacking knowledge on disease transmission and prevention. Raising awareness is particularly important where local concepts of helminthiases and diarrhoeal diseases may foster risk-related behaviour or reduce adherence to best practices [49]. Further research on factors improving adherence to CLTS and health education interventions is warranted and may contribute to enhancing community effectiveness of such interventions.

Strengths and limitations

A strength of our study is the rigorous sanitation approach, namely CLTS, which implies a very high private latrine ownership coverage and, ideally, the certification of ODF status in an intervention community. Furthermore, adherence may be enhanced due to non-subsidized latrine provision [16, 50, 51]. However, our assessment on behaviour change and adherence to hygiene practices is based on reported frequencies from cross-sectional questionnaire surveys administered to household heads and is not supported by direct observations. Thus, the assumption of the link between access to a private latrine and its use might not be entirely valid.

Certain considerations with regard to study design should be taken into account in future studies to generate stronger evidence. Specifically, inclusion of more localities is pivotal, so that bias from local heterogeneity of helminth and intestinal protozoa infection can be mitigated. Furthermore, a cluster randomised controlled trial design generates stronger evidence, since potential baseline differences in disease prevalence, population size, level of existing water and sanitation infrastructure are addressed through randomisation [7]. Sustainability of CLTS is a critical issue [36] and remains to be assessed in different social-ecological settings. Our study had a relatively short follow-up period between the last round of anthelminthic treatment and the follow-up parasitological assessments and thus might not have allowed measuring the full impact on helminth infections. A prolonged survey period would thus improve surveillance data with regard to reinfection patterns as well as create new evidence on long-term effects of CLTS and health education as a means for controlling and eliminating neglected tropical diseases [10].