Household Global Positioning System (GPS) coordinates duplication in health surveys – a case study of the Vaccination Coverage Surveys in the Democratic Republic of Congo (DRC)

Abstract Background Household-based surveys are fundamental for estimating immunization coverage in low- and middle-income countries. Increasingly, household GPS coordinates are collected in surveys to improve spatial resolution for targeted intervention. Geolocation data quality impacts coverage estimates, program targeting, and policy decisions. However, little attention has been paid to the accuracy and uniqueness of these data. This study investigated the extent of GPS coordinate duplication in vaccination coverage survey (VCS) in the Democratic Republic of the Congo (DRC), and explored potential causes. Methods Three rounds of VCS were conducted in January 2022 (VCS1), February 2023 (VCS2), and March 2024 (VCS3). In each round, duplicated household GPS coordinates were identified. Duplication patterns were analysed by health region, interviewer ID, tablet device ID, and interview date. Data quality in other fields was compared between interviews recorded by devices with > 90% duplicated GPS coordinates and those recorded by devices with only unique coordinates. Results In VCS1, 17% of GPS points had substantial duplications, with 5 provinces (Mongala, Nord-Ubangi, Sud-Kivu, Kwilu, and Haut-Lomami) showing over 40% duplications. Of all 501 devices IDs retrieved, 116 recorded exclusively unique household GPS coordinates, while the interviews recorded on 10 devices all showed household GPS duplication (7 were deployed to the 5 provinces with the highest duplication rates). In Kwilu Province and Haut Lomami, in particular, tablets deployed to the same health zones and days showed similar duplication patterns. The quality of other data elements in interviews with duplicated GPS coordinates was comparable to those with unique coordinates. Notably, duplication rates in VCS2 and VCS3 were substantially reduced to within 1%. Conclusions Significant GPS coordinate duplication was observed in VCS1 in the DRC, but this issue was largely mitigated in subsequent survey rounds. The duplication identified highlights the need for improved field protocols, including enhanced interviewers training, robust validation steps, and clear hardware and software guidelines. GPS errors in household survey threaten the integrity of spatial data, downstream analyses, and policy decisions based on geospatial metrics. Our findings underscore the need for standardized quality assurance in the geolocation data collection practices of global health survey implementations.