Application of Structural Equation Modelling to Predict Acceptance and Use of mHealth Interventions at the Design Stage

Abstract

The health status of women and children is considered as a reflection of the present society and predictor for the future generations. Unfortunately, challenges due to poverty, inadequate resources, illiteracy and socio-cultural barriers contribute to poor health and high maternal and newborn mortality in the developing countries. To address some of these challenges, there are several mHealth initiatives seeking to exploit opportunities provided by over 90% mobile penetration. However, most of these interventions have failed to justify their value proposition to inspire acceptance and use. It is our contention that the observed low up-take of mHealth innovations require holistic approach to align the solutions to consumer needs and expectations. In this paper, we demonstrate how to apply structural equation modelling to predict acceptance and use of mHealth interventions in low-resource settings. To identify factors that influence acceptance and use of mobile-based solutions, ninety five randomly selected antenatal and postnatal women were invited to participate in formal discussions on how mobile phones can be used to enhance access to maternal and newborn care. Seventy nine participants filled out self-assessment questionnaire as a prestudy to evaluate how their reaction would predict post-implementation reactions formed on the basis of significant hands-on experience. Based on the discussions and the results of the pre-study, we identified key factors that influence acceptance and use of mHealth interventions in low-resource settings. These factors were configured into a conceptual model comprising of nine variables used to predict post-deployment acceptance (fit) and use of mHealth prototype named Mamacare. After deploying mamacare in a rural hospital, a cohort of seventy nine subjects were recruited into a longitudinal experiment that involved sending of targeted SMS alerts on appointment reminders, safe delivery, danger signs, nutrition, preventive care, and adherence to medication. The experiment was designed to allow a within-subjects comparison in order to examine how structural equation modelling can be used to predict likelihood of acceptance and use of Mamacare based on reactions taken during the prestudy. After analyzing the prestudy dataset using SmartPLS, the results predicted 80.2% acceptance and 63.9% likelihood of use. However, results obtained from the first post-deployment user experience revealed lower rates of Mamacare acceptance and use at 69.1% and 50.5% respectively. The difference between prediction and actual outcome necessitated improvement of Mamacare using reactions obtained from the first post-test evaluation. Three months later, we conducted a follow-up post-test that recorded further drop in the acceptance from 69.1% to 60.3% but improved usage from 50.5% to 53.7%. Despite this variations, the study demonstrated that structural equation modelling is crucial to predicting acceptance and use of mHealth interventions in the early design stage.