Performance of a Mobile Phone App-Based Participatory Syndromic Surveillance System for Acute Febrile Illness and Acute Gastroenteritis in Rural Guatemala

Abstract

Background: With their increasing availability in resource-limited settings, mobile phones may provide an important tool for participatory syndromic surveillance, in which users provide symptom data directly into a centralized database.

Objective: We studied the performance of a mobile phone app-based participatory syndromic surveillance system for collecting syndromic data (acute febrile illness and acute gastroenteritis) to detect dengue virus and norovirus on a cohort of children living in a low-resource and rural area of Guatemala.

Methods: Randomized households were provided with a mobile phone and asked to submit weekly reports using a symptom diary app (Vigilant-e). Participants reporting acute febrile illness or acute gastroenteritis answered additional questions using a decision-tree algorithm and were subsequently visited at home by a study nurse who performed a second interview and collected samples for dengue virus if confirmed acute febrile illness and norovirus if acute gastroenteritis. We analyzed risk factors associated with decreased self-reporting of syndromic data using the Vigilant-e app and evaluated strategies to improve self-reporting. We also assessed agreement between self-report and nurse-collected data obtained during home visits.

Results: From April 2015 to June 2016, 469 children in 207 households provided 471 person-years of observation. Mean weekly symptom reporting rate was 78% (range 58%-89%). Households with a poor (<70%) weekly reporting rate using the Vigilant-e app during the first 25 weeks of observation (n=57) had a greater number of children (mean 2.8, SD 1.5 vs mean 2.5, SD 1.3; risk ratio [RR] 1.2, 95% CI 1.1-1.4), were less likely to have used mobile phones for text messaging at study enrollment (61%, 35/57 vs 76.7%, 115/150; RR 0.6, 95% CI 0.4-0.9), and were less likely to access care at the local public clinic (35%, 20/57 vs 67.3%, 101/150; RR 0.4, 95% CI 0.2-0.6). Parents of female enrolled participants were more likely to have low response rate (57.1%, 84/147 vs 43.8%, 141/322; RR 1.4, 95% CI 1.1-1.9). Several external factors (cellular tower collapse, contentious elections) were associated with periods of decreased reporting. Poor response rate (<70%) was associated with lower case reporting of acute gastroenteritis, norovirus-associated acute gastroenteritis, acute febrile illness, and dengue virus-associated acute febrile illness (P<.001). Parent-reported syndromic data on the Vigilant-e app demonstrated agreement with nurse-collected data for fever (kappa=.57, P<.001), vomiting (kappa=.63, P<.001), and diarrhea (kappa=.61, P<.001), with decreased agreement as the time interval between parental report and nurse home visit increased (<1 day: kappa=.65-.70; ≥2 days: kappa=.08-.29).

Conclusions: In a resource-limited area of rural Guatemala, a mobile phone app-based participatory syndromic surveillance system demonstrated a high reporting rate and good agreement between parental reported data and nurse-reported data during home visits. Several household-level and external factors were associated with decreased syndromic reporting. Poor reporting rate was associated with decreased syndromic and pathogen-specific case ascertainment.

Keywords: Guatemala; acute febrile illness; app; dengue; diarrhea; mobile phone; norovirus; participatory; syndromic surveillance.

Figure 1

Study design and CONSORT diagram of study recruitment, enrollment, and completion. The participatory syndromic surveillance (PSS) cohort enrolled children from April to September 2015, followed by prospective observation with the weekly symptom diary for acute gastroenteritis and acute febrile illness episodes (dotted box) through June 2016.

Figure 2

Map showing clusters of participants with high (>70%; yellow circle) symptom diary app response rate versus moderate (40%-70%; orange circle) and low (<40%; red circle) response rates, in the Southwest Trifinio Region of Guatemala during the first 25 weeks of surveillance prior to allowing study nurses to manually enter syndromic data (April-October 2015).

Figure 3

Weekly syndromic reporting rate for acute febrile illness and acute gastroenteritis, April 2015-June 2016. Weekly syndromic reporting rate of participants using the Vigilant-e symptom diary mobile phone app (orange), manually entered data from nurse phone call (green), and combined mobile phone and manual data entry (blue). Several factors were associated with periods of decreased reporting, including a time period of high staff turnover (June-July 2015), a cellular tower collapse (August 2015), and primary and run-off presidential elections (October 2015). On October 2, 2015, study nurses were allowed to manually enter participant data if there was no response. In April 2016, study nurses performed an in-home visit to participating households to repair or replace malfunctioning phones and remind participants to use the Vigilant-e app for reporting when possible.