‘MATRI-SUMAN’ a capacity building and text messaging intervention to enhance maternal and child health service utilization among pregnant women from rural Nepal: study protocol for a cluster randomised controlled trial

The study is conducted across 6 VDCs (52 wards) of Dhanusha district of Nepal. Dhanusha is one of the 75 districts of Nepal, which is situated in the southern part (also known as ‘Terai’ in local language). VDCs are the basic political unit of the district, which is further divided into nine smaller units called wards. Each ward functions as one cluster in our study. The study area is predominantly inhabited by rural communities with agriculture as a main source of occupation (particularly vegetable farming) and the study population are relatively stable.

The district borders with Siraha, Mahottari and Sindhuli districts in the east, west and north, respectively and India in the south. The main residents of the district are from the Maithili heritage and the adult literacy rate is 69%. Administratively, the district comprises one municipality and 101 VDCs with an estimated population of 754,777 in 2011 [22]. Each VDC consists of one government health facility. According to the skilled birth attendant (SBA) policy of Nepal, the health post serves as the first contact point (birthing centre) for institutional delivery [4, 23], with the deployment of at least one staff nurse. Azonal hospital (200 beds) and a private medical college act as referral hospitals for Dhanusha and adjoining districts [24, 25].

The study population consists of pregnant women aged between 15 and 49 years in their second trimester (gestation period between 13 and28 weeks). Pregnancy status was confirmed by a validated pregnancy test report. The gestational age will be determined based on first date of last menstrual period by asking with the participant. The expected date of delivery (EDD) based on gestational age will be confirmed through physical examination of pregnant women by nursing staff. Women were excluded if they had the following characteristics: (a) those not planning to stay at the study site during pregnancy and postpartum, (b) any physical and mental disability (which can adversely influence the study results), (c) women who have miscarriage or still birth during recruitment process, and (d) woman who are not willing (to give consent) to participate in the study.

Baseline information is collected after allocation of the clusters in two groups. At the same time, one-day extensive reinforcement training to the FCHVs in the intervention arm is conducted. Intervention group will receive text messages in their mobile phones. The principal investigator is responsible for sending text messages to the participants once they are recruited in the study and this process will continue until the fifth-month after childbirth. There will be three assessments during follow-up period and the data will be collected during these assessment periods: between 38 and 39 weeks of gestation, during delivery, and at 7 months after child birth. During the follow-up period, monitoring and supervision of FCHVs in the intervention arm will also be done to assess their performance. Additionally, assessment of child will be carried out at birth and 7 months after birth [Fig. 1].

The estimation of sample size was done considering the study design as two-parallel arm cluster RCT [26]. We assumed that 12% of the child births have been attended by a skilled birth attendant (SBA) at baseline in both arms and this rate will increase to 32% post- intervention in the intervention arm. 20% as the minimum difference between the two groups we would like to see after intervention with 0.80 power and 0.05 level of significance, and to account for intra-cluster correlation by taking the design effect of 2.29 for delivery attended by SBA, an adjusted sample size of 147 pregnant women in each arm is required for the study [4, 24]. With allowable 20% potential attrition, the final sample size is 177 for each arm. Thus, 354 pregnant women are estimated to be included in a trial. This number would be sufficient to detect a change in the effect sizes over a follow-up period in the intervention group as compared to control group. The study area consisted of 67,625populations (male/female ratio = 100:94) and the reported birth rate are 25.5 per 1000 per year [4], which gives an estimated 1724 expected pregnant women. Since, this study will consider the pregnant women in their second trimester, around 431 (=1724 × 0. 25) pregnant woman would be available at any point of time during a year, which is sufficient number for our study.

We employed multi-stage cluster sampling. First, two health facilities (consisting of one primary health care centre and one health post) catchment area of Dhanusha district has been selected purposefully. We selected these health facilities because there have been no MCH-related services from other organizations before, but the governmental. Furthermore, no research work has been conducted yet relating to this field.

Second, six VDCs have been selected in the catchment area of these two health care facilities by stratified random sampling technique. The population of each VDC in the catchment area range from 3500 to 19,000. Therefore, VDCs in both health care facilities catchment areas have been stratified into three strata based on population size (< 5000, 5000–10,000, and > 10,000), forming a total of six strata. One VDC has been randomly selected from each stratum [24, 25]. Third, those selected six VDCs comprised of 54 wards (here we called cluster for our study) where two of them are excluded as these wards are situated in the town area. The reminder 52 clusters have been selected for a trial. Finally, the complete enumeration of all households with pregnant women in their 2nd trimester from these clusters has been carried out.

All households with pregnant women in 52 clusters have been identified and screened applying inclusion and exclusion criteria. The identification of pregnant women has been made with the help of FCHVs’ antenatal registers. Pregnant women from each household in have been registered by FCHVs in their respective ward. All pregnant women with gestational period between 13 to 28 weeks are included in the study. Each household is coded and marked in the outside wall with a visible unique serial number to track participants during follow-up period.

The baseline information of these pregnant women was collected between July and September 2015. Selection of participants has been done by trained female research assistants (FRA), who are fluent in both Nepali and Maithili languages. Written informed consent have been taken from each participant who are willing to participate in the study. Participants are informed about the purpose of the study, advantages and any potential harms arising from a study. Head of the family, usually husband or close relatives of the relevant participants, were also informed about the study. For any adverse event or inquiry by the participants will be entertained or clarified.

Before baseline data collection, pre-trial inventory on antenatal, delivery and postnatal care, and number of pregnancies in the study area have been obtained from respective health facilities and FCHVs’ records. We have formed matched pair of clusters consulting with village secretary, chairman of the ward and FCHV of the respective ward. There are three criteria for matching: approximately equal numbers of pregnant women in the cluster; similar or predominant ethnic group distribution in each cluster; and proportionate SBA utilization rate across the cluster. The matched pair of cluster is made within each VDC. As there are nine wards in each VDC, an odd number of clusters in the VDC have been matched with odd number of cluster with other VDC. We have selected 26 matched pairs from52 clusters of the study area. Finally, 26 pairs of clusters were randomized into intervention and control arms, each arm holding 26 clusters.

Twenty-six pairs of matched clusters have been ranked from the top best matched to the bottom matched based on matching criteria. Within each pair, the random allocation rule has been used to assign either to the intervention group or control group (Additional file 1). There is no blinding of the researchers and participants in the study due to the nature of the intervention and ethical reasons. Data collectors during follow-up will be blinded to allocation.

Most of the clusters are geographically separated and that the distance between clusters are at least 500 m from each other which may prevent the possible treatment contamination between intervention and control arms. There may be potential contamination due to sharing of training knowledge but knowledge is not equal to implementation, moreover, the control group FCHVs will not be followed-up by our FRAs. The likely contamination of text messaging through mobile phone to the pregnant women will be confirmed by asking whether they received MCH information from the intervention area during follow-up period.

The intervention has been implemented by two approaches; i) Capacity building of FCHVs (reinforcement training on maternal and new-born health followed by regular supervision) for the promotion of health seeking behaviour among pregnant women, and ii) periodic text messaging to the pregnant women through a mobile phone for demand generation of MCH services.

Structured and semi-structured questionnaires were adapted from NDHS 2011 to collect quantitative information [4]. There are three sections in a questionnaire: i) baseline questionnaire ii) antenatal follow-up questionnaire, and iii) postnatal follow-up questionnaire. The baseline questionnaire includes three components: i) socioeconomic information, ii) obstetric/gynaecologic history, anthropometric measurement and existing service utilization of pregnant women, and iii) Food Frequency Questionnaire (FFQ) and dietary diversity questionnaire recommended by the UN Food and Agriculture Organisation [29]. The antenatal follow-up questionnaire also consists of three components: i) information on services utilization, ii) anthropometric measurement and iii) FFQ and dietary diversity questionnaire. Postnatal follow-up questionnaire consists of two components i) maternal and neonatal services utilization along with morbidity and mortality and ii) new-born anthropometric measurement and child care practices. A focus group discussion (FGD) guide is prepared to collect information on attitudes and beliefs of FCHVs and pregnant women on service utilization and nutrition practices to give an in-depth understanding on the status of dietary diversity and complement some of the quantitative findings.

Anthropometric variables: All the anthropometric measurements will be taken on each pregnant women and their subsequent child using standard anthropometric measurement techniques [30].

A total of 2 ml venous blood will be collected and kept in EethyleneDiamine Tetra-acetic Acid (EDTA) vial. Collected EDTA blood will be transferred to the clinical pathology laboratory of Janaki Health Care and Research Centre, Janakpur and Haemoglobin concentration will be estimated by Cyanmethemoglobin method.

The structured questionnaire has been piloted using responses from 20 households in the adjacent district. Necessary feedback is received and the amendment has been made in the final questionnaire before baseline data collection. Data will be collected by research assistants who are trained in the tools and techniques of data collection. Anthropometric assessment will be taken by six Female Research Assistants (FRAs) (one FRA per VDC).All measurements will be made three times and the median value will be recorded for analysis. The collection of information on pregnancy, outcomes and its determinants will be made longitudinally at different stages. The methods will employ structured interviews, FGD and anthropometric assessment.

Data will be collected at four chronological points. Baseline information on demographic characteristics, obstetric history, anthropometry, biochemical test, dietary survey and current practices on services utilization will be taken on 4–6 month prior to intervention. The reminders will be the follow-up period as described below;

Data entry will be performed in EpiData software version 3.1 and then it will be transferred to STATA software (Version 13.0) where all statistical analyses will be performed. The logistic regression model will be used to assess primary and secondary outcomes of the intervention. The independent t-test will be used to compare mean of different variables between the intervention and control groups. Chi-square test will be used to investigate the association between categorical variables in intervention and control groups. Within group analysis will also be done for outcomes whether it is attributed to either text message or capacity building or both.

For the repeated anthropometric measurement, biochemical test and dietary measures, we will use generalized estimating equation (GEE) and mixed model analysis to evaluate the effects of the intervention. The analysis will adjust for potential confounders for changes in weight gain and haemoglobin including baseline socio-demographic characteristics. We will also evaluate the possible interactions between the intervention indicators and participant attributes to assess the effect of intervention. Data will be analysed by ‘intention-to-treat’ principle.

All the FGDs will be recorded in a voice recorder. A coding framework will be prepared based on FGDs guideline. Transcription of data will be done in Nepali (National language), and then will be translated to English. Two researchers will be involved separately for coding based on pre-identified theme in the evaluation framework. Coding will be compared for consistency. The analysis will be done manually and no other software will be used for qualitative data analysis.

To ensure the quality, this study will use standard operating procedures for major processes. The field study team will receive extensive training on the objectives, methods and procedures as well as ethics coupled with frequent supervision and monitoring.

This trial MATRI-SUMAN is designed for the PhD research Project in 2014–17 and is currently completed.