mHealth Series: Factors influencing sample size calculations for mHealth–based studies – A mixed methods study in rural China

BACKGROUND: An important issue for mHealth evaluation is the lack of information for sample size calculations. OBJECTIVE: To explore factors that influence sample size calculations for mHealth–based studies and to suggest strategies for increasing the participation rate. METHODS: We explored factors influencing recruitment and follow–up of participants (caregivers of children) in an mHealth text messaging data collection cross–over study. With help of village doctors, we recruited 1026 (25%) caregivers of children under five out of the 4170 registered. To explore factors influencing recruitment and provide recommendations for improving recruitment, we conducted semi–structured interviews with village doctors. Of the 1014 included participants, 662 (65%) responded to the first question about willingness to participate, 538 (53%) responded to the first survey question and 356 (35%) completed the text message survey. To explore factors influencing follow–up and provide recommendations for improving follow–up, we conducted interviews with participants. We added views from the researchers who were involved in the study to contextualize the findings. RESULTS: We found several factors influencing recruitment related to the following themes: experiences with recruitment, village doctors’ work, village doctors’ motivations, caregivers’ characteristics, caregivers’ motivations. Village doctors gave several recommendations for ways to recruit more caregivers and we added our views to these. We found the following factors influencing follow–up: mobile phone usage, ability to use mobile phone, problems with mobile phone, checking mobile phone, available time, paying back text message costs, study incentives, subjective norm, culture, trust, perceived usefulness of process, perceived usefulness of outcome, perceived ease of use, attitude, behavioural intention to use, and actual use. From our perspective, factors influencing follow–up were: different caregivers participating in face–to–face and text message survey, sending text messages manually, participants responding incorrectly, and technical issues. Participants provided several recommendations for improving follow–up and we added our views to these. CONCLUSIONS: This is the first study to evaluate factors influencing recruitment and follow–up of participants in an mHealth study in a middle–income setting. More work is needed to assess effectiveness of our suggested strategies. This work would improve evaluation of mHealth interventions.