Coronavirus Disease Contact Tracing Outcomes and Cost, Salt Lake County, Utah, USA, March–May 2020

Victoria L. Fields; Ian T. Kracalik; Christina Carthel; Adriana Lopez; Amy Schwartz; Nathaniel M. Lewis; Mackenzie Bray; Carlene Claflin; Kilee Jorgensen; Ha Khong, Walter Richards; Ilene Risk; Maureen Smithee; Madison Clawson; Lee Cherie Booth; Tara Scribellito; Jason Lowry; Jessica Huynh; Linda Davis; Holly Birch; Tiffany Tran; Joseph Walker; Alicia Fry; Aron Hall; Jodee Baker; Eric Pevzner; Angela C. Dunn; Jacqueline E. Tate; Hannah L. Kirking; Tair Kiphibane; Cuc H. Tran

Disclosures

Emerging Infectious Diseases. 2021;27(12):2999-3008. 

In This Article

Discussion

Our analysis of contact tracing of 184 index cases and 1,499 close contacts in Salt Lake County, Utah, highlights the substantial cost and time needed for these investigations. In addition, we found that, for successive generations of contacts traced, fewer cases were identified, and the time between symptom onset and SARS-CoV-2 testing decreased. However, changing quarantine or social distancing guidance during the investigation period also might have resulted in fewer cases in later generations. These findings highlight the effectiveness of contact tracing to guide control measures and reduce onward transmission of SARS-COV-2. Other jurisdictions can use these findings to examine their contact tracing yields, effort, and key COVID-19–associated time intervals to help guide programmatic changes.

Contact tracing is resource intensive.[8] Every index case investigated produced a transmission chain containing a median of 5 linked contacts. The median time to investigate these transmission chains was 4.16 (95% CI 4.06–4.72) hours at a cost of $107.22 (95% CI $92.60–$120.70). During the study period, 2,757 COVID-19 cases in Salt Lake County required investigation, which we estimate to have resulted in ≈$300,000 and ≈11,500 staff hours spent conducting these investigations. The time spent by contact tracers reflects resources needed to interview, educate, and enter data for cases and contacts and to write work excuse letters and conduct community notifications. The finding of lower yields in later generations highlights the need for further studies to examine the cost-benefit of tracing multiple generations of contacts.[20]

We found that 6% of contacts were unreachable or out of jurisdiction, which is lower than the 17% unreachable contacts identified through a text messaging–based system in a previous study.[21] However, consistent with another study,[22] we found a high proportion (83%) of index case-patients that did not have known contact with a laboratory-confirmed COVID-19 case-patient. The prevalence of cases without an identified epidemiologic link raises concerns over unrecognized transmission,[23] which suggests contact tracing efforts alone might not be sufficient to stop disease transmission.

Our contact tracing yields, laboratory confirmation of infection among 19% of contacts, were higher than those in South Korea (4%), and Shenzhen (15%) and Guangzhou (17%) in China.[2,7,24] Consistent with findings from recent studies,[1,2,24,25] we found household contacts were infected at a higher rate (32%) than nonhousehold contacts (16%). The finding of higher infection rates among household contacts reinforces the importance of evaluating prevention measures, such as using hotel services for contacts unable to separate themselves from other household members.[26] Compared with index cases (n = 184), confirmed secondary cases (n = 285) identified through contact tracing generated about one fourth of the contacts and less than one fifth of the secondary cases. During the study period, testing capacity was limited, delaying health department notifications and initiation of contact tracing investigations, which might have increased yields because case-patients spent more time not knowing their infection status.[8] In addition, because primarily symptomatic persons received testing, positive results might have resulted in higher rates and thus higher yields.

Modeling shows the probability of COVID-19 control decreases with long delays from symptom onset to case isolation, fewer cases ascertained by contact tracing, and increasing transmission before symptom onset.[8] Thus, time intervals between symptom onset, laboratory testing, and initial health department interview provide insight into the efficiency of contact tracing investigations.[27] One study found that contact tracing for COVID-19 reduced the time to test confirmation by 2.3 days and time to contact isolation by 1.9 days.[24] Similarly, we observed a 3-day decrease in the time from symptom onset to initial health department interview starting with first-generation contacts and noted to be the same or further decreasing in most subsequent generations. The time interval from symptom onset to initial health department interview was longer than that from symptom onset to first positive test or from symptom onset to isolation initiation. This time interval decreased between the first-generation and sixth-generation contacts; later generation contacts might have had more opportunity to follow health department recommendations and for the health department to promptly recommend testing when indicated. Although the usefulness of contact tracing in the setting of sustained SARS-CoV-2 transmission has been questioned,[28,29] consistent with other studies, our findings show that contact tracing reduced transmission; only one fourth of contacts traced and quarantined experienced COVID-19–like symptoms or tested SARS-CoV-2–positive.

New technologies, such as mobile telephone application–based symptom monitoring and electronic contact tracing platforms, might alleviate some of the burden needed to carry out investigations. In Utah, contacts could opt to receive daily phone calls or text message notifications. Text messaging might improve efficiency by decreasing time for contact follow-up, but it requires additional resources, a robust information technology infrastructure, and strong data protection safeguards.[21] Smartphone technology is another powerful tool for contact tracing; a widely accepted smartphone application that does not have major privacy concerns, including the collection of personal data such as location, might prove useful.[30] In addition, technology such as point-of-care SARS-CoV-2 testing, where results can be obtained within 48 hours, could reduce laboratory turnaround time. Rapid tests aid in quickly identifying index cases and contacts to implement isolation protocols (J. Joung et al., unpub. data, https://doi.org/10.1101/2020.05.04.20091231) and could improve contact tracing metrics. Online platforms that can identify how cases and contacts are linked, such as MicrobeTrace (https://microbetrace.cdc.gov/MicrobeTrace), also could aid in the management of investigations by reducing duplicative efforts, thereby improving efficiency.

The ongoing COVID-19 pandemic and emergence of the SARS-CoV-2 B.1.617.2 (Delta) variant have demonstrated the need for continuing layered prevention strategies, including contact tracing.[31] Our findings can help local and state jurisdictions determine the cost, effort, and yields associated with implementing a comprehensive contact tracing program, factors that are crucial for guiding policy decisions. Our data, coupled with further cost studies, can help inform resource allocation, including staffing needs and roles, technology requirements, and strategies to evaluate cost-effectiveness. In addition, our findings can be used to develop mathematical models to determine the need to scale up contact tracing to focus on all cases and contacts or to scale down and focus only on first-, second-, and third-generation contacts, as well as to decide who to interview, such as high-risk contacts or household contacts.

Our study's first limitation is that our approach might not be generalizable because Utah's surveillance system enables linkage between cases and contacts, which might not be available in other jurisdictions; differences could also exist in contact participation across jurisdictions. Second, during March 2020, testing was available only for persons meeting initial COVID-19 symptom criteria (Appendix), which might have reduced case identification and the ability to test contacts. Third, interventions such as social distancing guidance and stay-at-home-orders introduced during March–May 2020 might have decreased transmission. Fourth, information was derived from interviews, which have a potential for recall bias, including naming all contacts.[32] Fifth, costs of contact tracing are underestimated because we could not account for overtime benefits, such as time-and-a-half pay; overhead, such as staff health insurance and facility utility costs; staff training time; time spent providing services to the community, such as time to drop off masks; and other expenditures. Sixth, we could not track how many persons complied with recommendations to self-isolate or quarantine; the ability to determine whether cases and contacts complied with recommendations would aid in further quantifying contact tracing yield and effort. Finally, patients who do not seek care, potentially because of presymptomatic or asymptomatic infection, are a further challenge to preventing additional cases because SARS-CoV-2 shedding is highest early in illness.[8] We found that 2% of asymptomatic contacts tested SARS-CoV-2–positive and 76% of asymptomatic contacts were not tested. Therefore, the attack rate might have been underestimated given the large proportion of asymptomatic contacts who did not get tested.

In conclusion, our analysis highlights the importance of contact tracing to reduce transmission of SARS-CoV-2. However, the effectiveness of contact tracing is contingent upon availability of substantial resources and rapid testing capacity. Persons should seek testing as soon as they experience COVID-19–like symptoms and begin isolation while results are pending. Because of early viral shedding, health department messaging should strongly direct contacts to obtain testing when possible, especially contacts with a higher risk for exposure, such as caregivers within households, populations in congregate settings, and contacts with underlying conditions; or for contacts who have an occupation requiring them to be in contact with other vulnerable persons, such as long-term care facility workers, daycare workers, and those who work with unvaccinated persons.[33,34] Contact tracing metrics evaluated in this study can help other jurisdictions design, improve, and scale up contact tracing programs as needed for their specific epidemiologic contexts. Health departments should consider adjusting their approach to contact tracing as the situation evolves and adopting new technologies as these become available.

processing....