High Community Transmission of SARS-CoV-2 Associated with Decreased Contact Tracing Effectiveness for Identifying Persons at Elevated Risk of Infection – Vermont

Abstract Vermont contact tracing (CT) consistently identified people at risk for COVID-19. However, the prevalence ratio (PR) of COVID-19 among contacts compared with noncontacts when viral transmission was high (PR = 13.5; 95% CI: 13.2–13.9) was significantly less than when transmission was low (PR = 49.3; 95% CI: 43.2–56.3).

Case investigation and contact tracing (CICT) is a core public health activity that seeks to break 2 chains of disease transmission and has been used for decades to control the spread of numerous 3 communicable agents [1]. CICT starts when an infected person (index case) is interviewed and 4 asked to identify people exposed to infection during the index case's infectious period. Trained 5 public health professionals then notify exposed persons to offer postexposure treatment (if 6 available) and guidance about how to protect themselves and prevent further spread of disease. 7 CICT programs seek to decrease disease burden across a population by focusing interventions 8 among exposed persons they can identify; CICT is therefore effective only when the risk for 9 infection among identified exposed persons is greater than that among the general population. 10 Since early 2020, CICT has been employed extensively worldwide to mitigate the spread of 11 SARS-CoV-2, the virus that causes COVID-19. Although models [2,3] indicate CICT can slow 12 or stop the spread of SARS-CoV-2, data concerning effectiveness are limited [1]. Moreover, 13 CICT is time-consuming, resource-intensive, and challenging to implement [4,5]

5
We used data from the National Electronic Disease Surveillance System Base System, a CDC-6 developed information system used by VDH to manage all reported electronic and non-electronic 7 COVID-19 laboratory results from hospitalized and non-hospitalized Vermont residents and all 8 index case interview records (including questionnaires to solicit close contacts), to identify 9 people who sought SARS-CoV-2 reverse transcription polymerase chain reaction (RT-PCR) 10 testing during May 3, 2020-October 30, 2021. Residents of long-term-care and correctional 11 facilities were excluded from the analysis. Recipients of antigen tests were also excluded because 12 these results were inconsistently reported to VDH. We stratified the analysis period into 39, two- 13 week periods based on specimen collection date and categorized people as index cases if any of 14 their RT-PCR test results within a two-week period were positive. People who were tested were 15 counted only once during any two-week period, but could be included again in subsequent 16 periods.

17
From index case interview records, we determined who (among those who sought RT-PCR 18 testing) had been reported to VDH as a close contact during the 14 days before their test. VDH  Test-seeking people from the general population who had not been reported to VDH as close contacts in the preceding 14 days were categorized as noncontacts. Attack rates (ARs) and 1 prevalence ratios (PRs) were calculated for the entire period and for each two-week period. We 2 calculated AR as the number of persons tested positive for COVID-19 (index cases) per total 3 people who sought testing, and PR as AR among close contacts divided by AR among 4 noncontacts. Two-week periods were also grouped into four SARS-CoV-2 community 5 transmission levels based on mean weekly COVID-19 incidence, as defined by CDC: low, 6 moderate, substantial, or high transmission levels corresponded to mean weekly incidence of 7 <10, 10-49.99, 50-99.99, and ≥100 new cases per 100,000 persons, respectively [10]. We 8 calculated COVID-19 incidences for each two-week period using official Vermont case counts 9 as reported by CDC [11]. PRs and 95% confidence intervals for COVID-19 among close 10 contacts to COVID-19 among noncontacts were calculated for each transmission level. This 11 activity was reviewed by CDC and was conducted consistent with applicable federal law and 12 CDC policy [12]. 13

15
Two peaks of SARS-CoV-2 transmission occurred in Vermont during the analysis period ( Figure   16 1a  Median number of index cases per two-week period was 724 (range: 57-2,784 index cases). ARs 21 among close contacts (range: 9.33%-34.6%) remained above ARs among noncontacts (range: 22 0.13%-2.64%) for every two-week period (Figure 1a).   who were ~50 times more likely to have COVID-19 than noncontacts (Figure 1c). However, the 1 effectiveness of CICT at identifying people at risk for COVID-19 was significantly less when 2 community viral transmission increased beyond moderate levels; above this threshold, 3 prevalence of COVID-19 among close contacts approached prevalence of COVID-19 among 4 noncontacts. 5 These findings indicate that the public health benefit of CICT diminished during periods of 6 increased transmission. One possible explanation for these findings is that, during transmission 7 surges, CICT programs were unable to identify an adequate number of close contacts at risk for 8 COVID-19; CICT staff might have become overwhelmed by increasing caseloads and unable to 9 fully interview all index cases to elicit close contacts. However, the possibility exists that 10 worsened CICT performance (e.g. timeliness) might have contributed to a rise in transmission.

11
CICT effectiveness might also have been influenced by viral variants (with different 12 transmissibility and incubation periods) and changes in public guidance, statewide mitigation 13 strategies, and public behaviors over time.
14 Index cases previously identified by VDH as close contacts represent instances in which CICT 15 had opportunities to intervene and break chains of viral transmission. Notably, only one-third of 16 index cases had been previously identified as close contacts. This indicates that, despite 17 aggressive efforts, VDH was unable to identify the majority of close contacts, and that CICT was 18 insufficient as a sole strategy for disease control [6].

19
Findings in this report are subject to at least six limitations. First, our exclusion criteria might 20 have resulted in an underestimation of index cases and close contacts. Second, test-seeking 21 behaviors might have differed between close contacts and noncontacts, and changed over time.

22
Third, this analysis only considers one measure of CICT; timeliness of close contact notification, adherence of close contacts to recommended guidelines, and other measures that determine 1 CICT's effectiveness to mitigate SARS-CoV-2 spread were not analyzed. Fourth, index cases 2 might have notified their own close contacts without reporting them to VDH. Fifth, people might 3 have been counted more than once if they had persistent positive PCR tests ≥14 days or 4 reinfections. Sixth, we only attempted to compare the relative effectiveness of Vermont's CICT 5 program across time and did not attempt to measure its absolute effectiveness on disease 6 prevalence, hospitalizations, or deaths. 7 CICT is an important public health tool to mitigate spread of communicable diseases, but it 8 requires substantial time and resources. In Vermont, COVID-19 CICT was most effective at 9 identifying people at higher risk for COVID-19 during periods of low and moderate viral 10 transmission, but appears insufficient as a sole mitigation strategy.