key: cord-1052372-1ro1p0l9
authors: Love, J.; Wimmer, M. T.; Toth, D. J. A.; Chandran, A.; Makhija, D.; Cooper, C. K.; Samore, M.; Keegan, L.
title: Comparison of antigen- and RT-PCR-based testing strategies for detection of Sars-Cov-2 in two high-exposure settings
date: 2021-06-06
journal: nan
DOI: 10.1101/2021.06.03.21258248
sha: a95cb4a9447b80530ea33cce236d66e10abcf1bf
doc_id: 1052372
cord_uid: 1ro1p0l9

Surveillance testing for infectious disease is an important tool to combat disease transmission at the population level. During the SARS-CoV-2 pandemic, RT-PCR tests have been considered the gold standard due to their high sensitivity and specificity. However, RT-PCR tests for SARS-CoV-2 have been shown to return positive results when administered to individuals who are past the infectious stage of the disease. Meanwhile, antigen-based tests are often treated as a less accurate substitute for RT-PCR, however, new evidence suggests they may better reflect infectiousness. Consequently, the two test types may each be most optimally deployed in different settings. Here, we present an epidemiological model with surveillance testing and coordinated isolation in two congregate living settings (a nursing home and a university dormitory system) that considers test metrics with respect to viral culture, a proxy for infectiousness. Simulations show that antigen-based surveillance testing coupled with isolation greatly reduces disease burden and carries a lower economic cost than RT-PCR-based strategies. Antigen and RT-PCR tests perform different functions toward the goal of reducing infectious disease burden and should be used accordingly.

Since its emergence in late 2019, SARS-CoV-2, the virus responsible for COVID-19, has spread 22 rapidly, causing significant global morbidity and mortality. Although early outbreaks were 23 concentrated in China and Italy, the United States (US) was the global epicenter for most of 24 2020, accounting for approximately one fourth of all cases globally by March 2021 (Dong, Du, 25 & Gardner, 2020). Despite the approval of multiple SARS-CoV-2 vaccines, production, 26 distribution, and uptake hurdles combined with the emergence of novel viral variants indicates 27 that achieving herd immunity remains a distant prospect (Moore & Offit, 2021; Sallam, 2021) . 28 Therefore, comprehensive testing, contact tracing, and infectious case isolation remain important 29 interventions to continue slowing the spread of SARS-CoV-2 and maintaining health system 30 integrity (Love et al., 2021) . 31

As testing availability has increased, epidemiological questions have arisen regarding the optimal 32 deployment of different test strategies. Diagnostic tests for SARS-CoV-2, none of which 33

perfectly reflect viral carriage (Woloshin et al. 2020) , fall into two broad categories: antigen tests 34 and real-time reverse transcription polymerase chain reaction (RT-PCR) tests. While both tests 35 diagnose active SARS-CoV-2 infection, antigen tests detect the presence of a specific viral 36

antigen and are capable of returning results within 15 minutes, while RT-PCR amplify genomic 37 sequences and therefore require longer turn-around times (CDC 2021a). Substantial attention has 38 been paid to the lower sensitivity of antigen testing compared with that of RT-PCR testing 39 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice. (Scohy et al., 2020) . Early studies on SARS-CoV-2 antigen testing sensitivity reported relatively 40 low sensitivity with respect to RT-PCR, leading some public health officials to place lower 41 confidence in antigen testing than RT-PCR testing in ending the COVID-19 pandemic. However, 42

it has been suggested that when comparing antigen-and RT-PCR tests to viral culture, a proxy 43 for transmissibility, one rapid antigen test ( individual-based accounting of infectious state (Figure 1 ). There are two simultaneous processes 71 in the model: the disease process and the testing process. In the disease process (Figure 1 ), 72 individuals start as fully susceptible (S) and become exposed (E) according to a density-73 dependent probability of exposure. This probability is defined as the product of a rate and the 74 infectious proportion of the population, accounting for quarantine and isolation, as described 75

below (see Equation S1

). The value of is derived from the product of R0, which we assume to 76 be uniformly distributed between 1.2 and 1.5 (Zhang, Keegan, Qiu, & Samore, 2020), and the 77 recovery rate, , which we assume to be uniformly distributed between 1/2.6 and 1/6 per day 78 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 6, 2021. In the testing process (Figure 1 ), we assume a fixed proportion of the population, t, is tested each 87 day (surveillance testing). The assumptions and parameter values for our baseline simulation 88 settings are as follow (Table S1 ). We assume that susceptible individuals who test positive (TS; 89 false positive) do so at a rate of ! = 0.0001 to account for imperfect test specificity. For tested 90

exposed (TE) and tested recovered (TR) individuals, we use the negative percent agreement of 91 each test compared to culture reported in published analyses for RT-PCR ( " = 95.5%) and for 92 antigen tests ( " = 98.7%) (Pekosz et al., 2021) . For tested infectious individuals (TI), we use the 93 antigen test positive percent agreement # = 96.4% and RT-PCR positive percent agreement 94 # = 100% reported in the same analysis (Pekosz et al., 2021) . We assume that individuals who 95 test positive are isolated (Q) and are returned from isolation after an average of 14 days ( = 96 0.07 in figure 1, though return from isolation is determined as 1/w days after the start of 97

isolation; see below); we also explore a shorter 10-day isolation (see supplement). We assume 98 that patients will reduce their mixing while awaiting test results and therefore are both less 99 susceptible and less infectious. As a baseline assumption, we use a 50% reduction in mixing due 100 to this partial quarantine (q). We also explore two additional possible reductions of 25% and 101 75% ( Figures S6, S7 ). Other than this partial quarantine, tested individuals move through the 102 disease course as normal. Test results for antigen testing are returned on the same day as test 103 administration (q = 1), while results are returned in 48 hours for RT-PCR (q = 1/3), an 104 approximation for the US average time for RT-PCR test result turnaround (as in Larremore et al., 105

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint 2021). We explore two additional possible RT-PCR test turnaround times of 1 and 4 days in the 106 supplement. 107 start as susceptible (S) and become exposed (E) at a rate . Exposed individuals become 109 infectious (I) at a rate , and recover (R) at a rate . Simultaneously, surveillance testing occurs 110 at a rate proportional to the population make up, and individuals awaiting test results are in a 111

"leaky" quarantine. Tested susceptibles (TS) can become exposed (TE), then infectious (TI), and 112

can infect susceptibles (both S and TS). Since tested infectious individuals (TI) are quarantining, 113 their infectiousness is reduced by a factor q. Likewise, since tested susceptibles (TS), are also 114 quarantining, their susceptibility is reduced by a factor q. Individuals who test positive are 115 isolated (Q). We assume isolation is perfect, and thus individuals can only progress through their 116 disease process (QE -> QI -> QR); susceptibles who are isolated cannot become infected and 117

infected individuals who are isolated cannot cause infections. After the isolation period is over 118 (14 days in the standard condition), individuals are returned to the general population, retaining 119 their current disease state. Antigen and RT-PCR tests have a specified positive ( # ) and negative 120 percent agreement ( " ) with viral culture (see text). ! is 0.0001, representing imperfect test 121

specificity. Note that we use positive percent agreement for tests of infectious individuals (I) and 122 negative percent agreement for both exposed (E) and recovered (R) individuals, as it has been 123

shown that RT-PCR may detect infection among individuals who are no longer infectious 124 (Kohmer et al., 2021 

We explore five testing strategies for each of these two settings and assumed asymptomatic 144 screening of 1%, 2%, 5%, and 10% of the population each day. In each setting and for each 145 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint of the infected and quarantined proportion, the mixing rate reduction due to quarantine (q), 187

and . is determined by the product of random draws from uniformly-distributed values of R0 188 and g. The probability of becoming infectious is determined by the rate s. Recovery, however, is 189

modeled not with a population-level rate (g), but with an individual-based approach using a fixed 190 duration of infectiousness; for each infection, a recovery day is designated that is 1/g time steps 191

from the day of infection. Similarly, tests are returned by defining a day of test return for each 192 test that is 1/q time steps from the day of testing, and return from isolation is determined by 193

defining an end day to the isolation period as 1/w days after isolation begins. 194

We conducted 10,000 stochastic simulations for each testing strategy in each setting. 195

To improve the decision support aspect of our model, we layered a cost effectiveness analysis on 197

top of the epidemiological model to estimate the costs of each testing strategy relative to the 198 relevant outcomes. We used parameter values estimated from literature sources and expert input 199 ( . 205

We considered the direct cost of testing to be the product of the number of tests per day, price 207 per test per day, personal protective equipment (PPE) costs per day, and labor costs per day. For 208 each type of test, the total cost is the sum of these daily costs multiplied by the duration of the 209 simulation. We assume that RT-PCR is sent to an external laboratory and therefore has no direct 210 capital costs to the decision maker, and we assume PPE is the same for either type of testing. The 211 no test strategy by definition has no direct testing costs. 212

We then consider the direct costs to the decision makers who purchase and conduct testing in 214 each of our two settings: universities and nursing homes. For nursing homes, staff absenteeism 215 due to quarantine incurs costs measured in labor productivity loss, a cost limited to staff, and a 216 conservative proxy for true costs which could include temporary staff and other costs. Positive 217 test results in either residents or staff incur labor costs for the administrative burden of reporting 218 positives and initiating cleaning protocols. For positive residents, additional PPE and labor costs 219 are incurred due to assumed need for isolation and additional staff care for those residents. Total 220 outcomes cost is therefore calculated as: 221

CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. Compared to simulations without testing, all testing strategies reduced the peak and total 233 infections in simulated epidemics (Figure 1 ). Greater reduction in infections was achieved with 234 higher rates of daily screening. The relative differences between testing strategies' performance 235

in reducing infections were largely maintained across both nursing home and dormitory settings. 236

In the nursing home setting, no statistically significant differences were found in % infections 237 averted across testing strategies at low levels of surveillance testing (1%, 2%). At high levels of 238 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint surveillance testing (5%, 10%), the reflex to antigen strategy (34% and 53% averted) 239 outperformed standalone PCR (26% and 44% averted), which outperformed standalone antigen 240 (20% and 35% averted) (Figures 2,3) . The reflex symptomatic negatives to PCR strategy did not 241 statistically significantly improve performance versus standalone antigen in any case (Figure 2 ). 242

These same broad trends were also observed in the university residence hall setting. For 243

university dormitories under screening rates of 1%, the reflex to antigen strategy averted the 244 highest percentage of infections (~6%), followed by standalone PCR (~5%), then standalone 245 antigen (~3%). Standalone antigen testing and reflexing symptomatic to PCR strategies were 246 again statistically equivalent in infection reduction (~3%) (Figure 2 ). At higher daily screening 247 rates, this overall pattern was largely conserved, yet exaggerated. For 5% and 10% daily testing, 248

overall reduction of infection rates increased to 16-32%, and 34-64%, respectively (Figure 2) . 249

Since hospitalizations and deaths are calculated as proportions of infections, hospitalizations 250 averted and deaths averted follow patterns similar to that found in infections averted ( Figure S1 ). 251

Very few deaths occurred in the residence hall setting. 252

On a per-test basis, different testing strategies resulted in similar percent infections averted 253

( Figure S8 ). At higher surveillance levels (i.e., 5% and 10% daily surveillance) and with larger 254 populations (i.e., in the residence hall setting), standalone PCR testing showed a notably higher 255 mean per-test percent infection averted ( Figure S8 ). 256

Overall, surveillance testing reduces the disease burden in populations, and the effect of testing 257 on measures of disease burden was greatest at the highest rates of testing. We found similar 258 patterns in simulations that used different parameter values for days to PCR test return, 259 quarantine mixing reduction, and days of isolation (See Figures S2-S7 ). 260 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint When evaluating antigen testing by modeling test sensitivity and specificity with respect to PCR, 261 we find that the performance difference between standalone antigen and PCR testing is greater 262 than when these test features are modeled with respect to viral culture ( Figure S2 ). For example, 263 the difference between standalone PCR and antigen testing strategies in mean % infections 264 averted in the residence hall setting at 10% daily surveillance testing was roughly 20% when 265 evaluating test performance with respect to PCR ( Figure S2b ) but only roughly 15% when 266 evaluating test performance with respect to viral culture (Figure 3) . 267

The economic analysis revealed that antigen-based strategies carried an overall lower economic 268 cost than did RT-PCR-based strategies. At 2% and 10% screening rates, respectively, over the 6-269 month period in nursing homes, the testing strategy of standalone Notably, the degree to which the standalone PCR testing strategy prevents infections is highly 296 contingent on the effectiveness of quarantine during the test-result waiting period. At low 297

waiting-period quarantine effectiveness, standalone PCR testing outperforms standalone antigen 298 testing to a lesser degree than at higher waiting-period quarantine effectiveness (Figures 2, 3, S6 , 299 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint S7). If individuals awaiting surveillance test results do not undergo quarantine, as may occur in 300 some settings, the difference between these two testing strategies is likely to less pronounced. 301

While the reflex to antigen strategy was the most effective at reducing infections, standalone 302 antigen testing was the least expensive and most cost-effective testing strategy in both settings at 303 all screening rates, due primarily to differences in test prices. Standalone PCR testing performed 304

well, but it was outperformed by the reflex to antigen strategy, both in terms of economic cost 305 and of reducing disease burden. 306

During the COVID-19 pandemic, RT-PCR testing has been established as the standard by which 307

to measure other tests. Our modeling analysis demonstrates that using viral culture, which may 308

better reflect viral transmissibility (Pekosz et al., 2021) , as the test standard dramatically alters 309 the relative performance of different surveillance testing strategies. Under this paradigm, 310

antigen-based surveillance testing strategies coupled with infectious case isolation are shown to 311 strongly reduce disease burden at a level close to RT-PCR-based strategies, but at a much lower 312 economic cost (Figures 2,3) , somewhat in contrast to model results under a more typical RT-313 PCR test-standard paradigm ( Figure S2 ). This lower cost has the potential to make additional 314 resources available for other management, containment, or recovery efforts, and therefore has the 315 potential to substantially reduce disease burden during the COVID-19 pandemic or others in the 316 future. Understanding test results with the priority endpoint in mind (e.g., the test's ability to 317 identify currently infectious infections during surveillance testing programs) should be of 318 primary importance, and our modelling study should prompt further research on the relationship 319 between viral culture, diagnostic test results, and transmissibility for SARS-CoV-2 and other 320 infectious diseases. 321

The cost perspective of this model may be of use to public health decision makers in determining 322

whether or not to invest in surveillance testing, but it does not account for the broader costs to 323 society and the healthcare system. An expanded or alternative perspective to this model that 324 could estimate the indirect societal costs of infection, disease, and quarantine would likely yield 325 more robust cost-effectiveness values and ICERs compared to those we find here. 326

Our results support the work of other studies that demonstrate that frequency of testing can 327 overcome differences in sensitivity (e.g., Larremore et al. 2021 ). In an important addition, we 328 provide a pragmatic example of an affordable and effective strategy that is implementable in two 329 group-living settings. As vaccine uptake remains low in many group living settings (e.g., 330

Cavanaugh et al. 2021), and as a greater understanding of the potential for immune escape 331 mutants is developed (Garcia-Beltran et al., 2021), surveillance testing strategies that use antigen 332 tests can be considered as highly effective, cost-reducing alternatives to PCR testing strategies. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint

CDC (2021b) FLUVIEW: U.S. Influenza Surveillance System

CDC (2020a) Interim Guidance for Antigen Testing for Sars-Cov-2

CDC (2020b) COVID-19 Pandemic Planning Scenarios

An interactive web-based dashboard to track COVID-19 350 in real time

Circulating SARS-CoV-2 variants escape neutralization by vaccine-354 induced humoral immunity

The Comparative Clinical Performance of Four SARS-CoV-2 Rapid Antigen Tests 358 and Their Correlation to Infectivity In Vitro

Test sensitivity is secondary to frequency and turnaround time for COVID-19 362 screening

Prolonged SARS-CoV-2 RNA shedding: Not a rare 364 phenomenon

A conceptual model 367 for the coronavirus disease 2019 (COVID-19) outbreak in Wuhan, China with individual 368 reaction and governmental action

Continued need for non-pharmaceutical interventions after COVID-19 372 vaccination in long-term-care facilities

Rethinking Covid-19 Test Sensitivity -A 375 Strategy for Containment

SARS-CoV-2 Vaccines and the Growing Threat of Viral 378 Variants

Assessment of SARS-CoV-2 Screening 381 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted

Outbreak dynamics of COVID-384 19 in China and the United States

Antigen-Based Testing but Not Real-Time Polymerase Chain Reaction Correlates 388

With Severe Acute Respiratory Syndrome Coronavirus 2 Viral Culture

R: A language and environment for statistical computing

COVID-19 Vaccine Hesitancy Worldwide: A Concise Systematic Review of 393 Vaccine Acceptance Rates

Low performance of rapid antigen detection test as frontline testing 396 for COVID-19 diagnosis

Seasonal Incidence of Symptomatic Influenza in 399 the United States

Combination of RT-qPCR testing and clinical 402 features for diagnosis of COVID-19 facilitates management of SARS-CoV-2 outbreak

False Negative Tests for SARS

Infection -Challenges and Implications

Clinical evaluation of BD Veritor SARS-CoV-2 point-410 of-care test performance compared to PCR-based testing and versus the Sofia 2 SARS 411 antigen point-of-care test

The real time effective reproductive 414 number for COVID-19 in the United States

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 6, 2021. Table 1 . Results of the economic cost analysis. In both settings and at all levels of surveillance testing, the antigen-based testing strategy was least expensive. The "retest all negatives with antigen" strategy, which averted the most infections (Figure 3) , was less expensive than PCR-based testing. The "retest negative symptomatic with PCR" strategy was similar in costs to the antigen-based testing strategy.. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)The copyright holder for this preprint this version posted June 6, 2021. ; https://doi.org/10.1101/2021.06.03.21258248 doi: medRxiv preprint