An mHealth intervention to improve vaccination uptake and other health outcomes among rural Indian children: pilot study for a cluster-randomized trial

The primary aim of this pilot study is to assess the feasibility of the planned main study. We will view the study as feasible if the following ex-ante criteria are met:
1. Recruitment and randomisation are feasible if: (i) 70% of villages approached agree to participate and accept randomisation; (ii) in participating villages, 70% of households with children in the target age range agree to participate and accept randomisation. Criteria related to recruitment and randomisation are assessed using data collected from project administrative records at study baseline (Week 0).
2. Interventions are feasible if uptake is sufficient to demonstrate acceptability and potential for impact, quantified as: 50% of households recruited to the study participate in interventions in some form (by listening to one or more outbound calls at home, and/ or by placing one or more calls, and/ or by attending community meetings). Participation in interventions was measured through two sources: (1) Call data were measured continuously throughout the study using the Tika Vaani interactive voice-response (IVR) system; (2) participation in community meetings was measured at community meetings taking place at Weeks 0, 4, 8, and 12.
3. Retention is feasible if 50% of households participating in the baseline survey agree to participate in the end line survey. This is assessed using data collected from project administrative records at study baseline (Week 0) and end line (after Week 13).
4. The design is feasible if the contamination proportion is below 15%. Contamination is assessed continuously throughout the study using the Tika Vaani interactive voice-response (IVR) system to identify the proportion of calls from callers outside the intervention group (i.e. control group villages, or non-study villages). The IVR system is used to identify unknown numbers and a moderator follows up on calls from non-registered callers to identify call origin. A contamination proportion in excess of 15% is a threat to the feasibility of using a cluster-randomised design with village as the unit of randomisation and geographical distances between villages similar to those in the pilot study.

1. Implementation fidelity is assessed by evaluating: (i) whether the activities are implemented as planned (content), (ii) whether the number of planned activities and the selected territory are respected (coverage) and (iii) if the planned frequency and duration of the activities are respected (frequency and duration). We will also explore whether there are specific moderating factors that can explain the degree of fidelity obtained. Implementation fidelity is assessed using 5 data sources: (1) structured observation with a checklist used by each field worker to verify if planned activities are implemented as specified in terms of content, coverage, duration and frequency, continuously throughout the study; (2) semi-structured interviews with field workers involved in intervention delivery at study end line (after week 13); (3) surveys administered via mobile phone (open for completion continuously from Week 7 to Week 13); (4) semi-structured interviews with health workers and local government officials at study end line (after week 13); (5) Analysis of project documents and IVR metadata collected continuously throughout the study.
2. Acceptability, accessibility, and satisfaction of the interventions to end users is assessed using 3 data sources: baseline and end line surveys (administered at Week 0 and following Week 13), mobile phone surveys administered through the IVR (open for completion continuously from Week 7 to Week 13), and qualitative interviews (community meetings taking place at Weeks 0, 4, 8, and 12, and semi-structured exit interviews and group discussions administered after Week 13 following the end line survey).
3. Knowledge, attitudes, practices, and behaviours of end users related to child health themes addressed by the intervention are assessed using 2 data sources: baseline and end line surveys (administered at Week 0 and following Week 13), and qualitative interviews (exit interviews, group discussions) administered after Week 13 following the end line survey.
4. Community-level indicators (numbers and characteristics of participants in community events in intervention villages, numbers and characteristics of callers outside of the target group who use the IVR system in intervention villages, and numbers and characteristics of additional participants who join the end line survey in intervention and control villages) are captured continuously from the project administrative records and IVR platform. We also access administrative records of the Assistant Nurse Midwife (ANM) to identify the numbers of children vaccinated per immunisation centre in intervention and control villages over the study period, using data collected retrospectively after Week 13 following the endline survey.