Date of Award

Spring 2022

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Public Health

First Advisor

Davis, J. Lucian (Luke)

Abstract

Background: Contact tracing is an evidence-based intervention that became a core component of many COVID-19 response plans throughout the globe. Despite its demonstrated success within other disease contexts, early studies of COVID-19 contact tracing have reported mixed results regarding the effectiveness and impact of contact tracing on ultimate pandemic objectives such as case incidence or mortality. Yet, there remains a knowledge gap regarding contextual factors that influence implementation outcomes of contact tracing that may in turn influence its ultimate impact. The primary objective of this dissertation is to address this knowledge deficit by evaluating an emergency contact tracing program developed at the onset of the pandemic and evaluating the performance of various Bluetooth-assisted contact tracing technologies. Methods: The emergency contact tracing program evaluated throughout this dissertation was established in partnership between the New Haven Health Department and Yale School of Public Health in March, 2020. In Aim 1 (Chapter 2), I evaluate implementation outcomes including reach (proportion of cases and contacts interviewed), timeliness (time from case testing to subsequent contact tracing steps), and sustainability of volunteer workforces. Within this Aim, I use quantitative methods including descriptive cascades to identify common reasons for case and contact drop-out, descriptive time measurements to evaluate timeliness of cumulative and individual contact tracing steps, and multilevel regression analyses to evaluate factors associated with successfully reaching cases and contacts. In Aim 2, I first use focus groups with contact tracers (Chapter 3) to qualitatively explore contextual elements associated with successful delivery of contact tracing. I use thematic analysis to analyze transcripts and the RE-AIM implementation science framework to organize the identified themes. In the second half of Aim 2 (Chapter 4), I draw on interviews with COVID-19 cases and contacts to qualitatively explore elements that influence behaviors upon which the uptake of contact tracing relies. These behaviors include testing, answering phone calls, participating in interviews, and isolating/quarantining. I thematically analyze transcripts and use the COM-B model of behavior change to organize the findings. In Aim 3 (Chapter 5), I evaluate two Bluetooth technologies used in a contact tracing pilot on a university campus. The first technology is an app-based approach, while the second uses a small, portable device to record Bluetooth data. I measure the sensitivity and specificity of each technology by comparing Bluetooth contact records to daily self-report records. I then use a post-participation survey to quantitatively and qualitatively evaluate the experiences of technology users and explore their perspectives. Results: Of the 1,705 cases reported in Aim 1, 545 (32%) were not reached due to missing key information, and another 334 (20%) were not reached due to their declining the calls or interview invitations. Ultimately, only 826 (48%) were interviewed. Of the 2,437 contacts reported by interviewed cases, 1,597 (66%) were not reached due to missing information, and another 153 (6%) were not reached due to their declining calls or interview invitations. Ultimately, only 687 (28%) were notified of their exposure. Median time to case interview from testing was five days and time to contact notification was 8 days. Various individual-, program-, and case-cluster factors were associated with successful outreach. Support from public health nurses was needed to stabilize the emergency contact tracing workforce due to surging caseloads and limited sustainability of volunteers (median time from sign-up to retirement from program was four weeks). In Aim 2, contact tracers identified many challenges and successes of the program’s implementation and made recommendations for improvement. Successfully engaging cases and contacts (the group hereafter referred to as “clients”) appeared dependent on outreach preferences, tracer communication skills, and sources of community mistrust. Effectiveness of contact tracing appeared threatened by time delays and the difficulties of isolation and quarantine. Adoption of a volunteer workforce appeared to rely on volunteer motivations, collaborative training, and supervision. Last, implementation efficiency was influenced by available tools and coordination with other agencies, and program maintenance was threatened by the low sustainability of volunteer workforces. Interviews with clients shed light on additional elements influencing specific behaviors required in successful contact tracing, and these findings were categorized within the Capability, Opportunity, Motivation, Behavior (COM-B) Model. Capability of clients to engage in tracing efforts was influenced by COVID-19 symptoms or baseline knowledge. Opportunities for engagement were influenced by structural and contextual resources and ties within social networks. Last, Motivation to engage in tracing efforts was influenced by symptoms, beliefs about deliverables and consequences of participating, trust in the health system, and emotional reactions of clients. In Aim 3, the portable device had higher sensitivity (94% vs 57%; p<.001) and specificity (95% vs 87%; p=.02) compared to the app-only technology. Participants largely considered Bluetooth contact tracing to be appropriate on a university campus but felt less comfortable with using GPS or Wi-Fi technologies. Most preferred technology that was developed and managed by the university compared to a third party, and privacy concerns were common. Conclusions: This dissertation presents some of the earliest efforts to better understand the contextual factors influencing success of COVID-19 contact tracing implementation, and its use of multiple and mixed methods to explore the implementation of contact tracing allows for the triangulation of findings from each individual Aim. Lacking information required for outreach posed a major barrier to reaching cases and contacts, although individual, case-cluster, and program-level factors associated with implementation success were also noted. While volunteers appeared to be an appropriate solution to emergency workforce needs, sustainability poses a significant threat to volunteer-driven programs as demonstrated quantitatively and qualitatively. Focus group and interview participants identified many individual-, program- and systems-level contextual elements influencing contact tracing delivery and uptake. Potential solutions to barriers as well as potential intervention activities to implement are discussed within these chapters. Last, Bluetooth technologies offer promising solutions to some contact tracing barriers, but the preferences of potential technology users and feasibility of managing such hardware/software approaches will be critical for uptake and adherence.

COinS