Tracking health seeking behavior during an Ebola outbreak via mobile phones and SMS

Shuo Feng, Karen A. Grépin, Rumi ChunaraNpj Digit. Medicine 2018 · OpenReview

TL;DR

This paper demonstrates how SMS-based mobile surveys can be a powerful, low-cost tool for rapidly collecting population-level health-seeking behavior data during public health emergencies like the Ebola outbreak in Liberia. By using propensity score matching to compare SMS data with traditional household surveys, the study reveals significant shifts in facility-based delivery patterns during the outbreak, highlighting the potential of mobile data for real-time public health intelligence.

What Problem Does This Paper Address?

During public health emergencies, especially in regions with fragile health systems like West Africa during the Ebola outbreak, traditional data collection methods often fail to provide timely and comprehensive information on population health behaviors. Existing health information systems can be disrupted, and retrospective studies may not capture the dynamic changes occurring in real-time. This gap makes it difficult to understand the immediate impact of outbreaks on healthcare utilization and to inform mitigation strategies. Specifically, the authors highlight the challenge of accurately measuring maternal health service utilization, as traditional data sources often underrepresent private facilities and can be outdated.

Methods

The core methodological contribution of this paper lies in its innovative use of SMS-based mobile surveys and a robust statistical approach to analyze the data.

  • SMS-based Mobile Survey: The researchers deployed a national SMS poll in Liberia to collect data on health-seeking behaviors from 6694 individuals between March and June 2015. This method allowed for rapid data collection at a population level, overcoming the logistical challenges of traditional surveys during an emergency.
  • Propensity Score Matching: To address the potential bias of mobile survey participants not being representative of the general population, the authors employed propensity score matching. This technique was used to create balanced subsamples by matching SMS survey respondents with participants from the 2013 Liberian Demographic Health Survey (DHS). This allowed for a more rigorous comparison of health-seeking behaviors between the two data sources and across different time periods (during and after the Ebola outbreak).
  • Time-Series Analysis: By collecting data at multiple time points during and after the outbreak, the study could analyze temporal trends in health-seeking behaviors, specifically focusing on facility-based deliveries.

Datasets

  • GeoPoll (GP) SMS Survey Data: This is the primary dataset collected by the researchers via SMS from 6694 individuals in Liberia between March and June 2015. This dataset provided individual-level health and health-seeking behavior information.
  • 2013 Liberian Demographic and Health Survey (DHS): This is a nationally representative household survey used as a benchmark for comparison. The DHS data provided information on maternal health outcomes prior to and during the period of the SMS survey.

Key Insights and Ideas

  • Mobile Surveys as a Rapid Response Tool: The study powerfully demonstrates that mobile phone surveys, specifically via SMS, can be deployed rapidly and cost-effectively to gather crucial population-level health data during crises, filling a critical gap left by disrupted traditional systems.
  • Uncovering Subtle Shifts in Behavior: Propensity score matching was essential in revealing that while overall facility-based deliveries decreased during the Ebola outbreak, the shift was more pronounced in public facilities. This nuanced finding might have been obscured without rigorous statistical adjustment.
  • Differentiating Public vs. Private Healthcare Utilization: SMS surveys enabled the researchers to distinguish between the use of public and private healthcare facilities, a distinction often difficult to make with traditional data sources that primarily rely on government reporting.
  • The Importance of Methodological Rigor: The paper underscores that while mobile data offers immense potential, careful methodological approaches like propensity score matching are vital to account for potential biases and ensure valid inferences about population-level behaviors.

Key Results

  • Decreased Facility-Based Deliveries: The study found a significant decrease in facility-based deliveries during the Ebola outbreak compared to the post-outbreak period (p < 0.05).
  • Return to Baseline for Public Facilities: After the outbreak, the proportion of births in public facilities returned to baseline levels comparable to those observed in the DHS data.
  • Private Facility Use: The matched analysis indicated an increase in deliveries in private facilities after the outbreak.
  • Demographic Representation: While the SMS survey reached a generally younger and more educated population than the DHS, propensity score matching successfully created comparable groups for analysis.

Limitations and Future Work

  • Representativeness of Mobile Data: The authors acknowledge that mobile phone users may not be perfectly representative of the general population. While propensity score matching helps, unmeasured confounding variables could still exist.
  • Outdated DHS Data: The DHS data used for comparison was from 2013, several years before the 2015 outbreak. The authors note that this data may not fully reflect the pre-outbreak population dynamics.
  • Future Directions: The paper suggests that prospective SMS surveying and analysis could offer real-time insights into health disruptions and inform mitigation strategies. They also propose integrating mobile phone number collection into household surveys for longitudinal monitoring and exploring other mobile data collection methods like Interactive Voice Response (IVR) for populations with lower literacy. Increasing the number of covariates used in matching is also suggested for further reducing confounding.

Summary updated

Abstract

The recent Ebola outbreak in West Africa was an exemplar for the need to rapidly measure population-level health-seeking behaviors, in order to understand healthcare utilization during emergency situations. Taking advantage of the high prevalence of mobile phones, we deployed a national SMS-poll and collected data about individual-level health and health-seeking behavior throughout the outbreak from 6694 individuals from March to June 2015 in Liberia. Using propensity score matching to generate balanced subsamples, we compared outcomes in our survey to those from a recent household survey (the 2013 Liberian Demographic Health Survey). We found that the matched subgroups had similar patterns of delivery location in aggregate, and utilizing data on the date of birth, we were able to show that facility-based deliveries were significantly decreased during, compared to after the outbreak (p &lt; 0.05) consistent with findings from retrospective studies using healthcare-based data. Directly assessing behaviors from individuals via SMS also enabled the measurement of public and private sector facility utilization separately, which has been a challenge in other studies in countries including Liberia which rely mainly on government sources of data. In doing so, our data suggest that public facility-based deliveries returned to baseline values after the outbreak. Thus, we demonstrate that with the appropriate methodological approach to account for different population denominators, data sourced via mobile tools such as SMS polling could serve as an important low-cost complement to existing data collection strategies especially in situations where higher-frequency data than can be feasibly obtained through surveys is useful.
BibTeX
@inproceedings{feng2018tracking,
  author = {Shuo Feng and Karen A. Grépin and Rumi Chunara},
  title = {{Tracking health seeking behavior during an Ebola outbreak via mobile phones and SMS}},
  booktitle = {Npj Digit. Medicine},
  year = {2018},
  url = {https://openreview.net/forum?id=kn89kS6W6y},
}
Figure