Automated vs. manual case investigation and contact tracing for pandemic surveillance: evidence from a stepped wedge cluster randomized trial. (January 2023)

• Record Type: Journal Article
• Authors: Raymond, Cameron; Ouyang, Derek; D'Agostino, Alexis; Rudman, Sarah L.; Ho, Daniel E.
• Is Part Of: EClinicalMedicine. Volume 55(2023)
• Publication Date: 2023-01
• Subjects: COVID-19 -- Contact tracing -- Health disparities; Medicine -- Research -- Periodicals; Medical policy -- Periodicals; Clinical Medicine; Health Policy; Public Health; Medical policy; Medicine -- Research; Periodical; Electronic journals; Periodicals; 613
• DOI: 10.1016/j.eclinm.2022.101726
• Languages: English
• ISSNs: 2589-5370

Abstract

Summary: Background: Case investigation and contact tracing (CICT) is an important tool for communicable disease control, both to proactively interrupt chains of transmission and to collect information for situational awareness. We run the first randomized trial of COVID-19 CICT to investigate the utility of manual ( i.e., call-based) vs. automated ( i.e., survey-based) CICT for pandemic surveillance. Methods: Between December 15, 2021 and February 5, 2022, a stepped wedge cluster randomized trial was run in which Santa Clara County ZIP Codes progressively transitioned from manual to automated CICT. Eleven individual-level data fields on demographics and disease dynamics were observed for non-response. The data contains 106, 522 positive cases across 29 ZIP Codes. Findings: Automated CICT reduced overall collected information by 29 percentage points (SE = 0.08, p < 0.01), as well as the response rate for individual fields. However, we find no evidence of differences in information loss by race or ethnicity. Interpretations: Automated CICT can serve as a useful alternative to manual CICT, with no substantial evidence of skewing data along racial or ethnic lines, but manual CICT improves completeness of key data for monitoring epidemiologic patterns. Funding: This research was supported in part by the Stanford Office of Community Engagement and the Stanford Institute for Human-Centered Artificial Intelligence.