Study design and setting
The present pre–post intervention study with a two-stage (schools and classrooms) cluster sampling was conducted on 1,781 students studying in 8th to 12th grades between July 2018 and December 2018. The study did not use external controls. The study was conducted in Ujjain district, Madhya Pradesh, Central India. Ujjain district has 5 sub-districts, covering 6,091 square km, and is a plateau [17]. The district has a population of approximately 2 million (1,986,864);approximately 61% (616,353) of the population resides in rural areas, with mostly an agrarian economy [17].
Sample size calculation
To calculate the sample size, a pilot study was conducted on 65 students, in which the students answered 54% of the questions correctly. These 65 students were selected from five schools by convenient sampling. Sample size calculation was performed with one sample comparison of proportion 54%, two-sided alpha of 0.05, and power of 90%. The minimum sample size obtained after calculation was 1,613, to which 10% was added to account for attrition or refusal rate. Thus, the estimated sample size was 1,774. The students who participated in the pilot study were not included in the main study.
Sampling strategy and data collection
A list of public and private higher secondary schools (grades: 8th–12th; age: 14–19 years) along with the number of students in each class was obtained from the district education officer. In the first sampling stage, public and private schools were randomly selected from two separate lists of all public and private schools. Figure 1 illustrates the sampling procedure and the inclusion and exclusion criteria for school selection. Of the 514 public and private higher secondary schools in the Ujjain district, 72 schools with at least 40–50 students in each class of 8th–12th grade were selected to reduce the number of visits required to obtain the estimated sample size. Of the 72 selected schools, 12 schools from rural areas and 12 schools from urban areas were selected randomly using computer-generated random numbers.
A structured WASH-knowledge questionnaire was developed in English, which was then translated to Hindi according to WHO recommendations for questionnaire translation [18]. The WASH-knowledge questionnaire comprised 15 questions, which were divided into the following four sections: water (1 question), sanitation (5 questions), hygiene (3 questions), and knowledge about diarrhea (6 questions on definition, causes, signs and symptoms, and community treatment) (Additional file 1: questionnaire in English).
The questionnaire also included limited demographic information such as name, age, grade, and gender of the participating students. Each questionnaire required approximately 20 min to complete, with some questions possessing multiple correct options. Each correct answer was given a score of 1. The minimum and maximum possible scores were 0 and 44, respectively. The same questionnaire was used after educational intervention to assess the effect of the intervention. The questionnaires were distributed and collected by 4–6 trained research assistants, who were present in class during the session. Although they assisted the students in understanding the questions in case of any difficulty, they did not assist the students in answering the questions.
Educational intervention
A visit was scheduled for each school before starting the intervention. Informed written consent was obtained from both the school principal and parents of the students, and written assent was obtained from the students. The trained research assistants ensured that the structured WASH-knowledge questionnaires were completely filled by the students. In case of any missing information, they interviewed the students and filled the missing details. The principal researcher visited the schools to supervise the survey activities. No efforts were made to contact the students who were absent on the day of data collection.
The educational intervention comprised an approximately 60-min practical training session, which included a form of a flip chart and appropriate illustrations and pictures in a Microsoft PowerPointTM presentation to convey the WASH-related messages to the students. The training module was based on the “Save The Children” community intervention module for childhood diarrhea [19]. Although the training was provided in Hindi in all schools, the medium of instruction in the included schools was either Hindi (regional language) or English. The medium of instruction refers to the language that is used to teach the contents of the educational curriculum.
Two class periods (90-min duration) were required to complete the 20-min pre-intervention questionnaire, whereas the 60-min intervention was provided on the same day. The students were not informed about the post-intervention questionnaire. After a minimum gap of one month following the intervention, the students were asked to complete the same WASH-knowledge questionnaire in 20 min. Overall, 144 sessions were conducted in 6 months, which included 72 pre-intervention and 72 post-intervention sessions.
Fidelity of intervention
To maintain fidelity in implementation of the intervention, the research assistants received training by the principal investigator. A 4-h training session was conducted to explain the intervention module. All slides in the power point presentation, pictures, and videos were discussed with regards to content and the method of delivering the content. The concepts were reinforced by providing the research assistants an opportunity to engage in role-playing. The training session was repeated once every fortnight during the study period. A training manual was used to articulate the content and delivery of the educational practical session. During the intervention at least two research assistants delivered the intervention, and one senior researcher was present to ensure consistency and fidelity in delivery of the intervention.
Ethical considerations
The Institutional Ethics Committee of R D Gardi Medical College, Ujjain, India approved the research protocol (IEC-RDGMC-493). Permission was obtained from the District Magistrate, Ujjain to approach the schools and perform the intervention.
Data management and analysis
Data was collected in schools through paper-based questionnaires which were later entered in Epi InfoTM (Version 7.2). Data analysis was done using Stata (Version 16.1, Stata Corp, College Station, Texas, USA). Descriptive statistics were used to determine the proportion of correctly answered questions by the students in the pre- and post-intervention. Pearson χ2 was used to test the significance of the difference in proportions at pre- and post-intervention timepoints. For continuous variables, range, mean, and standard deviation was presented. Means of pre- and post-test scores were compared using repeated measures analysis of variance (ANOVA). The effect size of intervention was determined by calculating Cohen’s d. Conventionally effect size is considered large if Cohen d is greater than or equal to 0.8. Multivariate quantile regression models were used to test the association between difference in pre- and post-intervention score (outcome) and independent variables. The study used quantile regression modeling to capture the full distribution of the outcome-difference in pre- and post-intervention scores, which is superior to an arbitrary binary cut-off for pass or fail. Such a cutoff would be required for binomial regression modeling. The coefficient (b), and 95% confidence interval were estimated for 10th, 25th, 50th (median), 75th and 90th quantiles of the difference in pre- and post-intervention score based on 500 bootstrap samples. The multivariate quantile regression analysis was performed using the simultaneous quantile regression command in Stata (Version 16.1, Stata Corp, College Station, Texas, USA). A P value <0.05 was considered significant.