Introduction

Soil-transmitted helminths (STHs) are among the most common chronic infections worldwide, estimated to infect over a billion individuals in tropical and subtropical regions, mainly in low and middle income countries [1]. A number of intestinal worms fall under the STH umbrella: the roundworm Ascaris lumbricoides, whipworm Trichuris trichiura and two species of hookworm, Ancylostoma duodenale and Necator americanus [2]. STHs are predominantly transmitted when faeces containing eggs are deposited into the environment, develop to an infective stage and are transmitted via ingestion or across the skin boundary (hookworm). Thus prevalence is highest in areas where hygiene is poor, safe water and sanitation facilities are lacking and health services are insufficient [1]. STH infections rarely result in death, but they affect nutrition, resulting in anaemia, loss of appetite, intestinal damage and reduced absorption of vitamin A [2, 3], impacting on growth and cognition at critical stages of a child’s development [4]. School-age children are particularly vulnerable to these negative outcomes as they harbour the greatest worm burdens [2].

Control of STHs is therefore paramount and control initiatives centre around three themes: chemotherapy, sanitation, and health promotion. Regular school-based deworming is a proven, cost-effective strategy that can avert the health and educational consequences of STH infections [5, 6]. In 2001, World Health Assembly Resolution 54.19 set endemic countries the target of ensuring that at least 75% of school-age children at risk of morbidity were receiving regular treatment by 2010 [7]. However, the global coverage of treatment in at-risk areas at the end of 2009 was 31% [8]. Successfully implemented sanitation interventions and effective health promotion can reduce exposure to faecal material and thereby may impact substantially on helminth transmission [9, 10]. As deworming programmes are scaled up and sustainable methods of delivering treatment are established, there is renewed focus to determine how sanitation and behaviour change messages can best complement deworming activities.

In India, one of the most populous of the high burden endemic countries [11], an estimated 241 million children are in need of deworming [12]. Bihar State in eastern India is one of the most populated states with some of the lowest development indicators [13]. Prior to implementation of a school-based deworming programme between February and April 2011, the State Health Society Bihar conducted two sets of surveys to determine the geographical distribution of STH infections in this region. These surveys aimed to inform the state treatment strategy and to provide comprehensive epidemiological baseline data for future programme evaluation. Previous surveys conducted in Bihar have revealed large spatial heterogeneity in STH prevalence and it was hypothesised that caste could play a role in explaining this heterogeneity and may be a factor to consider epidemiologically when determining treatment strategies. In Hindu society the caste system divides individuals into social strata that determine privileges and limitations that are inherited from one generation to the next. Although now abolished, the caste system is an engrained part of society in some places, including Bihar. A person’s caste can define not only socio-economic status but also type of job [14], aspects of diet and possibly some behavioural characteristics too. The government released data collected during the large prevalence surveys to allow additional exploration of caste, socioeconomic and behaviour-related risk factors for helminth infections to try and understand influencing factors in heterogenous prevalence areas. The aim of the analysis here was to report the determined prevalence and intensity of STH in Bihar and to inform ongoing programmatic efforts to support deworming with behaviour change and sanitation interventions.

Methods

Statistical analysis

Egg counts were multiplied by 24 to adjust them to eggs per gram and classified as ‘light’, ‘medium’ or ‘heavy’ intensity infections according to WHO guidelines [16]. Prevalence statistics were used to produce maps in Arc Map 9.3 (ERSI, Redlands, CA, USA). Questionnaire data were double entered into MS Excel, and any missing or unclear data were clarified with the child during sample collection. The data were cleaned and transported to Stata 12 for analysis (StataCorp, 2011). Variables detailing the frequency (always, sometimes or never) of use of different defecation sites (open field, jungle, river or latrine) were recoded to create one variable for defecation practice. One variable for hand-hygiene practices was similarly created from questions on whether ash, soil, water only, or soap are ever used to wash hands after defecation. Due to data recording issues, analysis of defecation practices was restricted to Aurangabad and Gopalganj districts only. Castes are described in the categories outlined in Table 1 (Castes 1 to 6). As only twelve children from Caste 6 were included in the survey, they were combined with children from Caste 5 to allow for stratification by caste in further analysis. The relationship between STH infection and demographic and socio—economic variables (household assets), maternal education (proxy based on maternal literacy), caste, defecation practice, hand-hygiene practice, and school-level variables was explored in univariate analysis, taking into account the cluster effect of school. A log binomial model (family: binomial, link: log) was developed to allow for and to explore between-school variance in infection levels, assuming a fixed effect of individual and school level variables and using generalised estimating equations to account for school-level clustering. Explanatory variables significant at P < 0.2 level in univariate analysis were included in the model, which looked for associations significant at P < 0.05.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Local caste classification and adopted categories used to describe castes.

https://doi.org/10.1371/journal.pntd.0003790.t001

Results

Soil-transmitted helminth infection

The overall prevalence of any STH infection was 68%, ranging from 10% to 96% across schools (Fig 1). Prevalence varied by district: Aurangabad 49%; Muzaffarpur 62%; Araria 80% and Gopalganj 82% (Table 2). All surveyed districts exceeded the WHO-defined prevalence threshold for mass distribution of anthelmintic drugs. A. lumbricoides was the most prevalent infection (52%), followed by hookworm species (42%) and T. trichiura (5%). Between a third and a half of infected children in each district harboured more than one helminth species, but triple infection was rare (Table 2). T. trichiura and hookworm infections were almost all light intensity (52/54 and 475/483 infections for T. trichiura and hookworm respectively). Five hookworm infections and 17% (n = 102/600) of A. lumbricoides infections were of moderate intensity. Heavy intensity infections were only found in three children, all of whom were infected with hookworm. Medium and heavy intensity hookworm infection (n = 8) were found in Aurangabad and Gopalganj in the West of the state, while the majority (69.6%) of the 102 medium intensity A. Lumbricoides infections were found in one district: Araria, in the North East.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Prevalence of soil-transmitted helminth infections among primary school children in the four surveyed districts (N = 1157).

https://doi.org/10.1371/journal.pntd.0003790.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Prevalence of soil-transmitted helminths in the 20 schools in Bihar.

https://doi.org/10.1371/journal.pntd.0003790.g001

Risk factors for infection

Potential risk factors for STH infection were explored. T. trichuris was dropped due to low numbers of infections. The main findings are summarised in Table 3. There was no sex difference in the risk of either infection. Older children (over 12 years) were more likely to be infected with hookworm than children aged 4 to 7 (RR = 1.3, 95%CI 1.0–1.6), but there was no evidence that the prevalence of A. lumbricoides infection varied with age. Ownership of a phone or motorbike, both indicative of greater household wealth, were protective against hookworm infection when compared with children without a phone or any means of transport (phone ownership: RR = 0.9, 95%CI 0.8–0.9; motorbike: RR = 0.6, 95%CI 0.4–0.9). Children from households with motorbikes were also less frequently infected with A. lumbricoides (RR = 0.9, 95%CI 0.7–1.0). Children with literate mothers (proxy for maternal education) were at reduced risk of both infections, particularly hookworm (RR = 0.5, 95%CI 0.4–0.7) (Table 3). Household assets and maternal literacy were closely associated and only maternal literacy was retained during later analysis. Results also suggest that belonging to a caste other than Caste 1 or being Muslim might be associated with risk of hookworm infection (Table 3). Attending school barefoot did not increase the risk of either hookworm or A. lumbricoides infection (hookworm: 41% and 43%, P = 0.561; A. lumbricoides: 52% and 52%, P = 0.925 for children with and without shoes respectively). Children who solely use latrines were considerably less likely to be infected with hookworm than children who always open defecate (13% vs. 49% respectively, P <0.001); however, latrine usage was not found to be associated with A. lumbricoides infection (Table 3). Soap use appeared protective against hookworm and A. lumbricoides infection when compared with other methods of hand-cleansing after defecation (mainly use of soil), although these associations had only borderline significance (Table 3). Soap use at this juncture is indicative of soap use in general even though children do not acquire infection directly from hands contaminated at this time.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Risk factors for hookworm and A. lumbricoides infection, Bihar India (N = 1157).

https://doi.org/10.1371/journal.pntd.0003790.t003

After adjustment in the multivariable model (see Table 3 for variables included in the model), increasing age, lack of maternal literacy and Castes 3, 5 and 6 remained significantly associated with hookworm infection. Absence of a hand-washing station at the schools was independently associated with A. lumbricoides infection after adjusting for gender, age, caste, maternal literacy and hand-hygiene practice.

Discussion

The survey revealed high prevalence of STHs in all four districts under study. Hookworm and A. lumbricoides infections were widespread, but T. trichiura was rarely detected, probably due to the high temperatures in Bihar during certain times of the year that may not allow the full development of T. trichiura eggs. Together with survey data from two additional states, point-level environmental data (annual rainfall and temperature) and socio-economic data (literacy at district level), the state-wide prevalence of STHs was modelled. The modelled prevalence data resulted in the government decision to treat the whole state with anthelmintics twice a year, integrating school based deworming with the Lymphatic Filariasis programme which includes Albendazole. Between February and April 2011, a school-based deworming in Bihar’s 67,000 government schools reached 17 million school-age children, including one million un-enrolled children, of the targeted 21 million, one of the largest school-based deworming programmes globally to date (www.dewormtheworld.org). A second round of treatment is planned in 2012. The data collected in this survey will provide a useful baseline of prevalence and intensity of STH infection, as well as detailed contextual information. Both of which are valuable for future measurement of the impact of the deworming programme and will form part of the state’s monitoring and evaluation programme.

This paper further describes behavioural, socio-economic (based on a proxy of literacy in the model) and caste-related factors found to be associated with STH infection, useful for informing campaigns accompanying deworming activities in the Bihar context. Our findings on each of these topics are considered in turn. Behaviour is a key factor in determining exposure to infective stages and consequently affects the transmission and reinfection rate. As with other studies [22–24], we found that open defecation increased the risk of hookworm infection. Defecation fields are likely to be highly-contaminated environments with high rates of transmission, particularly as hookworm larvae can directly penetrate the skin and children often go around barefoot. We did not identify the absence of footwear as a risk factor for infection as some other studies have shown [25], possibly because all children spend some time barefoot and are similarly exposed, because the majority of shoes are open shoes such as flip flops and might afford little protection, or because transmission is not only via this route: N. americanus can penetrate other skin surfaces and A. duodenale can be transmitted orally. Shoe campaigns may not be the way forward. Unlike hookworm, A. lumbricoides is only transmitted via ingestion of infective stage eggs; and in concordance with this we found that open defecation was less important for transmission (being bare foot in an open defecation field would not contribute to A. Lumbricoides transmission). We could expect that behaviours which increase the likelihood of ingesting eggs play more of a role. We have some evidence at univariate level that soap use protects against infection and that soil use is a risk factor. The latter may be because the soil is infected with infective eggs if taken from an area close to the defecation site. Self-reported measures of hand-washing tend to overestimate soap use [26], so it is possible that hand-washing with soap actually takes place too infrequently to reduce risk in highly-contaminated environments. We are not able to firmly conclude on the role of hand-washing with soap or the risk posed by using soil for hand-cleansing. Given the “soil-mouth” transmission route, it is plausible that cleaning hands with soil in defecation fields contaminated with infective stage eggs (if soil is taken from the surrounding area) could be a risk factor for A. lumbricoides. A recent review found inconclusive evidence that hand-washing can reduce A. lumbricoides infection [27].

We now turn our attention to the findings pertaining to maternal literacy (our proxy for socio-economic status) and caste. It is unsurprising that defecation and hand-hygiene practices were closely related to each other; a latrine offers a more enabling environment for hand-washing with soap than an open field where soil was more commonly used for hand-cleansing. However, latrine use, and therefore relative frequency of soil or soap use, may be associated with maternal literacy. Maternal literacy was negatively associated with STH infection. This observation is concordant with other contexts [24], where persistent health differences are seen across different socio-economic groups; education is an important social determinant of health [28]. The complex, socially-stratified caste system in India, within which some castes are typically poorer and less educated than others—by our definition Caste 1 would be considered more educated and wealthy than Caste 6—plays an important role in determining health status [18]. As you are born into a caste and remain in it, some castes are trapped at a certain social level which makes it difficult to break the cycle of poverty and ill health. Defecation practice, hand-hygiene behaviour and maternal literacy—risk factors for helminth infection—all vary by caste, yet some caste differences in risk of infection remain in the adjusted model. This suggests that other intrinsic behavioural or cultural differences exist between the castes that were not captured by the indicators collected in the questionnaire. It is possible that children from different castes have different environmental exposure to helminth eggs if they live in geographically separate parts of the village and have differential exposure to faecal material. It was not possible to analyse this spatially. All castes have high prevalence of infection and there is no indication that after accounting for differences in socio-economic status (literacy) and hygiene behaviour that caste itself plays a significant role in determining prevalence in a particular area.

Water, sanitation and hygiene (WASH) behaviours are important in STH transmission and therefore WASH forms an important element of STH control alongside regular deworming. Should resources be available to support large-scale WASH and behaviour change campaigns, successfully implemented interventions could also complement school-based deworming. Such interventions could include a behaviour change campaign to accelerate the uptake and use of latrines, preceded by formative research to understand sanitation adoption behaviour and inform the intervention design [29, 30], or behaviour change campaigns to improve hand-washing practices and reduce the prevalence of cleansing hands with soil.

Increasing the coverage of (working) hand-washing stations in schools enables hand-washing to take place and increases general water use. It is interesting to see that children in schools with hand-washing stations had fewer STH infections. Although hand-washing stations can facilitate hand-washing, in this case it appears that their presence also allows use of water for other purposes which could reduce the risk of transmission. Indeed, survey teams observed that cooking utensils and plates were washed at hand pumps where available, but using soil—which could be contaminated with eggs—in their absence. Available water also makes the practice of washing fruit, vegetables and other food picked from the ground feasible. However, the relative importance of different transmission routes (e.g. unwashed food) is currently unknown and should be further quantified.

Our study has some limitations. Although Kato Katz is the recommended methodology for this type of survey, the technique is known to have less than perfect sensitivity as a diagnostic test especially at lower intensity infections and is limited by taking only one days sample rather than consecutive days when egg excretion is known to fluctuate. The actual prevalence of infection could therefore be higher than identified. One of the surveyed districts, Aurangabad, was a second choice district in this area because the first district selected could not be surveyed for security reasons. The risk factor analysis is restricted to the measured variables and may not present a complete picture of risk patterns and dynamics at play. Only three schools remembered the LF deworming programme despite extensive probing. Although we cannot discount the possibility that the reach of the LF programme was larger than this study indicated, we conclude that few children would have received Albendazole at this time, and even if they had, there was sufficient time for reinfection to take place before the survey and prevalence is unlikely to be underestimated for this reason. As previously mentioned, defecation and hygiene practices described are based on self-report; therefore it is possible that the frequency of “good” practice has been overestimated. Given that school-children answered the questionnaire, it was not possible to better define socio-economic status, but more information on household assets, parental occupation and income could have been useful. Only 4% of the survey population were from the Caste 1. Although current data is not available on the caste composition of Bihar state, the 1931 Census found that the “Upper Castes” represented 13.7% of the population, [31] and it is possible that Caste 1 has been under-represented in this survey.

STH prevalence in Bihar is high and justifies mass treatment. This survey provides a baseline for the monitoring and evaluation process for the Bihar State School Based Deworming Programme; a paper on lessons learnt from deworming over 17 million children is under consideration (Drake et al.). In Bihar, open defecation is common-place and hands are often cleansed using soil which may be contaminated with STH eggs in open defecation fields. Maternal literacy appears to have the biggest influence on risk of infection. Caste differences exist in the prevalence of different risk behaviours and caste appears to be associated with helminth infections. However, prevalence of infection is high in all castes and caste would not be a relevant targeting tool for treatment strategies or complementary campaign activities. The findings reported have been used to help direct the health and behaviour messaging being developed as part of ongoing school based deworming campaigns in Bihar. Additionally, the results hypothesise control of helminths as an important potential benefit of WASH programmes in this context. Community awareness campaigns and health and behaviour messaging around deworming programmes should consider the low levels of female literacy and develop pictorial or visual messages as appropriate.

Supporting Information

Acknowledgments

The authors would like to thank Shri C.K. Mishra, IAS, then Principal Secretary, Department of Health & Family Welfare, Government of Bihar and Shri Rajesh Bhushan, IAS, then State Project Director-Bihar Education Project Council and Secretary-Public Relations Department, Government of Bihar for their leadership and support in enabling this study. Several organisations collaborated on this work, including—the Bihar Education Project Council and the State Health Society Bihar, in Bihar, India; Deworm The World, Washington DC, USA and Partnership for Child Development, London, UK.

Particular thanks are extended to: Dr. Prabhat Kumar and all staff at Nalanda Medical College whose generosity with laboratory space and time was boundless; Alissa Fishbane, Deworm the World for her support and management; Dr. Yogita Kumar and Kriti Sharma for their technical and planning support in-country; Nicola Lloyd for her considerable contributions to survey planning and logistics; Rakesh Kumar and Stalin Chakrabarty for their in-state time and support; Babul Prasad and Gopal Sharma for management and logistical support in districts; and Rachel Pullan and Simon Brooker at the London School of Hygiene and Tropical Medicine for their technical input to the survey. Our sincere appreciation is further extended to the district hospitals in Bihar for their provision of space and laboratory facilities, all schools (teachers and children) who participated in the survey and to all the technicians and community volunteers who made the survey possible.

We would also like to thank Professor Nilanthi De Silva (University of Kelaniya, Sri Lanka) Dr. Joe Brown (London School of Hygiene and Tropical Medicine) and Professor Donald Bundy (Bill & Melinda Gates Foundation) for their comments on earlier drafts of this paper.