key: cord-0900466-vvqh03p2 authors: Skittrall, J. P. title: Asymptomatic SARS-CoV-2 testing: predictors of effectiveness; risk of increasing transmission date: 2020-11-24 journal: nan DOI: 10.1101/2020.11.24.20236950 sha: 19b54bf6bfd9fc0ae454725b2da1519f32d660ce doc_id: 900466 cord_uid: vvqh03p2 Testing asymptomatic people for SARS-CoV-2 aims to reduce COVID-19 transmission. Screening programs' effectiveness depends upon testing strategy, sample handling logistics, test sensitivity, and individual behavior, in addition to dynamics of viral transmission. We investigated the interaction between these factors to determine how to optimize reduction of transmission. We show that under idealistic assumptions 70% of transmission may be averted, but under realistic assumptions only 7% may be averted. We show that programs that overwhelm laboratory capacity or reduce isolation of those with minor symptoms have increased transmission compared with those that do not: programs need to be designed to avoid these issues. Our model allows optimal selection of whom to test, quantifies the balance between accuracy and timeliness, and quantifies potential impacts of behavioral interventions. (with isolation following a positive screen) alone results in only a modest reduction in infections. When the presence of screening results in a relaxation of precautions taken by those with minor symptoms, we show that this combination can result in an overall increase in infections compared with no screening. We demonstrate that the success of screening depends upon a rapid turnaround of tests. As a result, we show that a screening program running comfortably under 5 capacity is more successful than one that pushes capacity and generates backlogs. Our derived expression can be used to compare the effectiveness of proposed testing strategies in complex scenarios where there are different infection rates and different test availabilitiessuch situations might occur in a small screening program, for example in a hospital with ward-to-ward variation in infection, or in a large program, for example in a national program with city-to-city 10 variation in infection and in laboratory locations. We began by considering two different screening scenarios. The first scenario is an ideal (maximum impact) scenario. In this scenario, screening is performed daily with a high-sensitivity test. Test turnaround is rapid. All those eligible for screening present on every occasion, and all those with positive screens immediately isolate. The second scenario is a realistic scenario for 15 mass population screening. In this scenario, screening is performed weekly with a highsensitivity test. Samples for testing must be transported by courier from the sampling site to the testing laboratory. Test turnaround follows that of a large laboratory operating within capacity. There is attrition reducing those eligible who present for screening and those who isolate following positive screens, similar to the attrition observed in other screening programs. The 20 difference between the two scenarios is marked. In the ideal scenario, we estimate screening alone can eliminate 70% of onward transmissions. In the realistic scenario, the proportion of transmissions eliminated reduces to 7%. We next considered what happens if the presence of screening reassures those with minor symptoms, so that instead of isolating they continue with their daily lives. We divided our 25 infected population into three categories: those who never display any symptoms of infection and always continue with their daily lives, those who display typical symptoms and isolate as soon as these symptoms manifest, and those who display minor symptoms and variably reduce their contact with others when such symptoms manifest. If, instead of isolating, those with minor symptoms behave as usual, the total number of transmissions increases, and because on average 30 an infected person remains infectious for longer (having not isolated), the proportion of transmissions eliminated by screening increases. However, because screening does not identify all the additional individuals in the population with minor symptoms but behaving as usual, the net result is a relative increase in the number of transmissions. If the behavior change is seen in a sufficiently high proportion of those with minor symptoms, the net result of the screening Model output where the proportion of those developing minor symptoms ("paucisymptomatic people") behaving as usual, rather than isolating, is varied from 0 to 1, with parameters otherwise 5 as in our realistic scenario (see Supplementary Methods). As fewer people with minor symptoms isolate, screening detects and leads to isolation of more infectious people (apparent change in transmissions line, showing estimated change in transmissions varying from -6.7% when all paucisymptomatic people isolate, to -7.3% when all paucisymptomatic people behave as usual). However, when the percentage change in transmissions is normalized to a fixed denominator (the 10 number of transmissions without screening when paucisymptomatic people all isolate), it can be seen that the combined effect of screening and behavior change is at best a reduced effectiveness in screening, and at worst an increase in the number of transmissions (normalized change in transmissions line, showing estimated change in transmissions varying from -6.7% when all paucisymptomatic people isolate, to +13.7% when all paucisymptomatic people behave as 15 usual). Following this, we considered the impact of testing turnaround times on the ability of screening to reduce viral transmission. In general, one would expect a greater number of screening tests to be able to detect a greater number of infections, and therefore to yield a greater reduction in transmissions. However, when the number of tests requested exceeds a laboratory's capacity, a 20 backlog develops with consequent increase in turnaround time. This effect was seen in English laboratories at the end of April 2020, when a policy of testing large numbers of asymptomatic people in residential facilities was implemented. We modeled the effect of exceeding laboratory capacity using turnaround time data from April-June 2020 in our regional clinical microbiology and public health laboratory in Cambridge, England ( Fig. 2 ) (see Supplementary Methods). Our 25 output shows that reliably keeping laboratory demand slightly below capacity results in a greater reduction in transmissions than when capacity is exceeded. We then proceeded to consider the general effect of laboratory turnaround time on the ability of screening to reduce viral transmission. Our output shows that the extent of transmission reduction depends strongly upon turnaround time. 30 All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this this version posted November 24, 2020. ; https://doi.org/10.1101/2020.11.24.20236950 doi: medRxiv preprint A full consideration of screening effectiveness takes into account individual engagement with screening. We considered this in our model, via the proportion of those offered screening who ever take it up, the proportion of those who attend each screening event, and the proportion who isolate when asked. The first and last of these have a linear effect on testing effectiveness 20 (Fig. S1) . For a weekly screening interval, the second is also approximately linear, reflecting that this screening interval only gives one opportunity to prevent most transmission. With more frequent screening intervals, the impact on transmissions of increased per-screen uptake becomes nonlinear, with diminishing returns as the uptake is higher (Fig. 3) . Because our model takes infectivity and testing distributions as input, limited only by the tractability of the resultant numerical integration, it can be used to predict the impact of different testing strategies in complex scenarios where rates of infection, access to testing or uptake vary within a population (Fig. 4) . This allows us to, for example, consider the impact of a program 10 using rapid near-patient tests daily, where such tests have lower sensitivity than laboratory-based nucleic acid amplification testing (9). Assuming total and per-test uptake remains the same (increased frequency balanced by increased convenience), our model predicts a reduction in transmissions from such a program between 25% and 32%, compared with the 7% reduction from a centralized mass testing program. (A) Scenario where testing capacity is sufficient to offer screening to the whole population every 20 days (taking non-attendance into account), so that a smaller proportion of the population is screened if testing is offered more frequently. The proportion/frequency combination does not 20 affect testing impact, unless testing is underutilized (right side) or redundant (left side of plot). (B) As (A), but with the population structured: half the population are healthcare workers in contact with vulnerable patients, so are always screened before others. Each half of the population further divides into two, with the infection rate in one quarter double that of the other. Those with the higher infection rate are screened first, subject to capacity and the vulnerable 25 patient contact-first rule, i.e. the screening priority is higher rate healthcare workers, lower rate healthcare workers, higher rate others, lower rate others. (C) Scenario in which there are two identical cities with two identical laboratories (realistic testing scenario), save that the first city has an infection prevalence greater than the second city. Laboratory capacity is sufficient to offer All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this this version posted November 24, 2020. ; https://doi.org/10.1101/2020.11.24.20236950 doi: medRxiv preprint screening to everybody every 10 days. Each city's laboratory can be used separately to offer screening to its local population (dashed line). Alternatively, both laboratories can be used to screen every 5 days those in the city with higher infection prevalence, but with an additional two-day turnaround delay for samples sent between cities (solid line). The more effective strategy depends upon relative infection rates in the two cities. 5 Our model is sufficiently robust to changes in modeling assumptions regarding viral transmission dynamics to make predictions that can be used in practice (Table S1 ). Ultimate control of the COVID-19 pandemic is unlikely prior to deployment of effective vaccines. Screening and isolation acts as a bridge to this goal, saving lives and permitting some resumption of economic and social activity. It must, however, be recognized that unintended 10 behavioral responses to screening may not only remove the opportunity to resume desired activity, but indeed may make disease transmission worse than if no screening had occurred. Pursuing centralized testing strategies may rapidly increase testing capacity. However, the critical impact of turnaround time on testing effectiveness means that such strategies must have highly streamlined logistics chains to retain effectiveness. If rapid centralized turnaround of tests 15 is not possible, then less accurate, localized testing may be more effective. As with the historical introduction of screening programs, in SARS-CoV-2 testing it is crucial we move from simply counting numbers of tests, to more sophisticated measures of their effectiveness. Engagement with screening substantially impacts success and must not be taken for granted (10) . A holistic approach considering the social, economic and political impacts, acknowledging the 20 incidence of false positive results and balancing their impact, and combined with good communication, is therefore a key part of a successful program (11, 12) . Testing strategies that employ confirmatory testing to reduce false positive rates need clear protocols for communicating with those awaiting confirmatory testing, to retain the benefit of earlier isolation. Our results can only be as accurate as the estimates on which our model is based (for example, of 25 test sensitivity). However, where such estimates limit the accuracy of our predictions, our model highlights the additional information required. (Our model is also limited by numerical accuracy in the calculations, but the underlying estimates represent a greater uncertainty.) In any case, our model demonstrates the impact of the different modifiable factors in a screening program, enabling policymakers to prioritize those that give most benefit. 30 Our study does not account for the potential impact on transmission rate of isolating contacts of those who screen positive. The impact of contact isolation on transmission reduces as the impact of screening increases (in the extreme case that screening stops all infection, no contact tracing is necessary). As screening identifies more infectious individuals, both the scale of contact tracing required and the impact from isolating non-infectious contacts increase. The relative efficacy of 35 screening is not impacted by contact tracing (although if contact tracing is successful, there should be fewer infectious individuals to screen), so can be considered separately. Our work allows policymakers to design SARS-CoV-2 screening programs to maximize impact in reducing transmission. Whilst the data used pertain to SARS-CoV-2, the underlying methodology is applicable to any transmissible disease, and can therefore be applied to other 40 epidemics and pandemics. 1. L. Rivett, S. Sridhar, D. Sparkes, M. Routledge, N. K. Jones, S. Forrest, J. Young, J. All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. Figures S1-S3 Table S1 Code S1 5 References (14-38) All rights reserved. No reuse allowed without permission. preprint (which was not certified by peer review) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this this version posted November 24, 2020. ; https://doi.org/10.1101/2020.11.24.20236950 doi: medRxiv preprint Screening of healthcare workers for SARS-CoV-2 highlights the role of asymptomatic carriage in COVID-19 transmission Universal and Serial Laboratory Testing for SARS-CoV-2 at a Long-Term Care Skilled Nursing Facility for Veterans Serial Laboratory Testing for SARS-CoV-2 Infection Among Incarcerated and Detained Persons in a Correctional and Detention Facility -Louisiana Viral dynamics of SARS-CoV-2 infection and the predictive value of repeat testing. medRxiv Suppression of a SARS-CoV-2 outbreak in the Italian municipality of Vo' Covid-19: Mass population testing is rolled out in Liverpool Covid-19 : Government plans to spend £100bn on expanding testing to 10 million a day Antibody responses to SARS-CoV-2 in patients of novel coronavirus disease 2019 Specificity and positive predictive value of SARS-CoV-2 nucleic acid amplification testing in a low prevalence setting Diagnostic tool or screening programme? Asymptomatic testing for SARS-CoV-2 needs clear goals and protocols BETS: The dangers of selection bias in early analyses of the coronavirus disease (COVID-19) pandemic The incubation period of coronavirus disease 2019 (CoVID-19) from publicly reported confirmed cases: estimation and application Clinical characteristics of 24 asymptomatic infections with COVID-19 screened among close contacts in Nanjing The CITIID-NIHR COVID BioResource Collaboration, Point of Care Nucleic Acid Testing for SARS-CoV-2 in Hospitalized Patients: A Clinical Validation Trial and Implementation Study Quantifying SARS-CoV-2 transmission suggests epidemic control with digital contact tracing 10 Serial interval of SARS-CoV-2 was shortened over time by nonpharmaceutical interventions Uptake in cancer screening programmes: A priority in cancer control Prevalence of Sars-Cov-2 infection in health workers (HWs) and diagnostic test performance: the experience of a teaching hospital in central Italy Differential occupational risks to healthcare workers from SARS-CoV-2 observed during a prospective observational study Sensitivity" and "specificity" reconsidered: The meaning of 30 these terms in analytical and diagnostic settings Evaluating the accuracy of different respiratory specimens in the laboratory diagnosis and monitoring the viral shedding of 2019-nCoV infections False-negative results of initial RT-PCR assays for COVID-19: a systematic review False negative tests for SARS-CoV-2 infection -challenges and implications Ke, Clinical, immunological and virological characterization of COVID-19 patients that test re-positive for SARS-CoV-2 by RT-PCR Assessing Viral Shedding and Infectivity of Asymptomatic or Mildly Symptomatic Patients with COVID-19 in a Later Phase Predicting infectious Severe Acute Respiratory Syndrome Coronavirus 2 from diagnostic samples SARS-CoV-2 virus culture and subgenomic RNA for respiratory specimens from patients 15 with mild Coronavirus disease Correlation between the interval of influenza virus infectivity and results of diagnostic assays in a ferret model If you ask them, will they come? Predictors of quarantine compliance during a hypothetical avian influenza pandemic: results from a statewide survey How to improve adherence with quarantine: rapid review of the evidence Self-isolation compliance in the COVID-19 era influenced by compensation: findings from a recent survey in Israel Improving adherence to screening follow-up among women with 30 abnormal pap smears: results from a large clinic-based trial of three intervention strategies Women with abnormal screening mammography lost to follow-up: an experience from Taiwan