Digital Targeted Client Communication for Effective Coverage of Antenatal Care : Assessment, Development, and Evaluation

Abstract

The effective provision of essential interventions in antenatal care (ANC) is crucial for a positive pregnancy experience. However, the effective coverage of essential interventions in ANC is often low, even in settings where there is high coverage of ANC visits. Untimely use, and low quality, of services are both important factors that contribute to low effective coverage. Digital health interventions are often moderately effective in improving the timeliness of attendance to ANC. The interventions with high effectiveness are often those that are individually tailored, theory based, and co-designed with the end users. Most research reports have focused on evaluating effectiveness, while the design, development, and implementation of such digital health interventions have been a “black box”. This limits the possibility to replicate interventions that proved to be effective, which ultimately results in a waste of research resources. Digital health interventions that aim to change the behavior of end users, such as targeted client communication (TCC), need to be carefully designed to maximize effectiveness and minimize any unintended consequences. This can be achieved through underpinning the design on relevant behavior change theories; careful assessment of the context; co-creating the content, frequency, and mode of the intervention; and iteratively testing and improving the final intervention. To ensure that the finalized intervention results in the intended effects, both effectiveness and any potential unintended consequences should be estimated using robust study designs such as randomized controlled trials. In this thesis, we present the approaches we took to develop an acceptable and safe automated digital TCC intervention and to test its potential unintended consequences in a cluster randomized controlled trial. The effectiveness of the digital TCC intervention is not included in this thesis. This thesis is part of a larger project that combines a national implementation of a digital maternal and child health registry in Palestine (called MCH eRegistry) with several effectiveness trials of digital health interventions.

We have used both quantitative and qualitative research methods. We started by identifying the gaps in effective coverage and worked with stakeholders to identify and target contextually relevant essential interventions. A national expert panel identified anemia, hypertensive disorders of pregnancy, diabetes mellitus, fetal growth restriction, and untimeliness of attendance to the routine visits as a priority – in this thesis simply referred to as “the target conditions”. Quantifying the coverage and effective coverage of the essential interventions informed the nature and goal of the digital health intervention – the digital TCC. Understanding the perceptions of women and the healthcare providers was essential towards developing the digital intervention. This was conducted using in-depth interviews guided by a theory. The content of the TCC was developed in an iterative process based on the results of the in-depth interviews, recommended message framing approaches, and feedback from users and other stakeholders. We used concepts and models proven to be helpful in guiding the process of choosing the right words, which are simple to understand, culturally acceptable, and effectively convey the intended messages without increasing worries in pregnancy. We developed algorithms in the digital MCH eRegistry to tailor messages automatically to individual women based on selected variables. Finally, we evaluated any unintended effects of the TCC intervention, such as an increase in pregnancy-related worries (measured by the Cambridge Worry Scale) or reduced satisfaction with ANC services, in a cluster randomized controlled trial. In public primary healthcare clinics in Palestine, we found that 60% of women attended four or more ANC visits and the majority of women started their first visit early. Coverage of at least one screening test of the target conditions was high, but the effective coverage was low. The main contributing factor for the low effective coverage was that the pregnant women did not attend care in a timely manner. We identified three constructs of the Health Belief Model (HBM) that were important in understanding the patterns of ANC attendance. The contents of the intervention, therefore, focused on influencing women’s perceptions of the risks and seriousness (perceived susceptibility and severity) of the target conditions and the benefits of timely ANC attendance. Through the iterative process of co-designing, testing, and refining using tools and methodologies for composing behavior change communication messages, we designed a digital TCC intervention considered acceptable by pregnant women, healthcare providers, and health education experts in Palestine. The TCC, which contained information about the risks of exposure and complications, did not increase pregnancy-related worries among the recipients of the intervention compared to the controls. The women were equally and highly satisfied with the ANC they received in both intervention and control arms. This work contributes to the rapidly growing field of design and implementation of digital health interventions for maternal and child health and beyond. It adds to the limited literature on the scientific development process of TCC, and estimations of potential side-effects using robust study designs. Demonstrating the application of automated digital TCC to women using routine data from the MCH eRegistry, without adding burden to the healthcare providers, can further improve the investment returns of such digital health registries.