key: cord-0796087-si1yyg5e authors: Noriega, Rodrigo; Samore, Matthew H. title: Increasing testing throughput and case detection with a pooled-sample Bayesian approach in the context of COVID-19 date: 2020-04-05 journal: bioRxiv DOI: 10.1101/2020.04.03.024216 sha: ada3869c1f1293c78f797cf763026e335ef77fda doc_id: 796087 cord_uid: si1yyg5e Rapid and widespread implementation of infectious disease surveillance is a critical component in the response to novel health threats. Molecular assays are the preferred method to detect a broad range of pathogens with high sensitivity and specificity. The implementation of molecular assay testing in a rapidly evolving public health emergency can be hindered by resource availability or technical constraints. In the context of the COVID-19 pandemic, the applicability of a pooled-sample testing protocol to screen large populations more rapidly and with limited resources is discussed. A Bayesian inference analysis in which hierarchical testing stages can have different sensitivities is implemented and benchmarked against early COVID-19 testing data. Optimal pool size and increases in throughput and case detection are calculated as a function of disease prevalence. Even for moderate losses in test sensitivity upon pooling, substantial increases in testing throughput and detection efficiency are predicted, suggesting that sample pooling is a viable avenue to circumvent current testing bottlenecks for COVID-19. detect a broad range of pathogens with high sensitivity and specificity. The implementation of 11 molecular assay testing in a rapidly evolving public health emergency can be hindered by resource 12 availability or technical constraints. In the context of the COVID-19 pandemic, the applicability of 13 a pooled-sample testing protocol to screen large populations more rapidly and with limited 14 resources is discussed. A Bayesian inference analysis in which hierarchical testing stages can have 15 different sensitivities is implemented and benchmarked against early COVID-19 testing data. 16 Optimal pool size and increases in throughput and case detection are calculated as a function of 17 disease prevalence. Even for moderate losses in test sensitivity upon pooling, substantial increases 18 in testing throughput and detection efficiency are predicted, suggesting that sample pooling is a 19 viable avenue to circumvent current testing bottlenecks for COVID-19. 20 2 Emerging infectious diseases pose a global hazard to public health, as exemplified by the COVID- 22 19 pandemic. Key epidemiologic strategies for control of community spread include contact tracing, 23 case isolation, ring containment, and social distancing (1-7). The use of microbiological testing to 24 identify disease cases is a crucially important element of these strategies. Some countries, including 25 the US, experienced a shortage of kits needed for COVID-19 diagnosis, which resulted in the 26 imposition of restrictive criteria to manage the selection of patients for testing. Constraints in the 27 supply of kits had a particularly significant impact on testing of mildly symptomatic individuals, as 28 well as asymptomatic contacts of confirmed cases. For some facilities that have been overwhelmed 29 by demand for testing as the pandemic progressed, test throughput continues to be a limiting factor 30 (8-10). Strategies for screening more individuals with a reduced burden on resources are highly 31 desirable. Using a Bayesian formalism, a hierarchical testing protocol based on sample pooling is 32 discussed. Anticipated benefits include easing the demand of constrained resources and enabling 33 more efficient detection of a larger number of cases. 34 Molecular assays are the predominant testing method for viral and bacterial pathogens (11-14). 35 Specifically, nucleic acid detection assays typically employ real-time polymerase chain reaction 36 (RT-PCR) for DNA targets and reverse-transcription real-time PCR (rRT-PCT) for RNA targets 37 (15, 16) . The popularity of such testing platforms is due to 1) their high sensitivity and specificity, 38 2) the widespread access to sequencing and synthesis technologies for the identification of nucleic 39 acid target sequences and probes, and 3) the development of fast, user-friendly, and cost-effective 40 equipment. While nucleic acid assays have powered a revolution in diagnostics and delivery of care 41 for individual patients, their application in large-scale infectious disease surveillance is hampered 42 partly by low throughput at a population level. 43 The information content of a diagnostic test can be evaluated with a Bayesian probability formalism 44 in the context of an individual sample or for repeated sampling from the same patient (17-19) by 45 3 taking into account the probability of detecting a positive case (assay sensitivity, or identification 46 rate ) and the probability of a positive result from healthy samples (false positive rate ). 47 Bayesian inference requires the assessment of a "prior" probability to the presence of disease in a 48 sample, ( ), which is updated to a "posterior" probability given a positive or negative test result Besides testing throughput, it is informative to assess the increased ability to detect cases using a 93 pooled-sample vs individual-sample scheme, which can be accomplished by comparing the ratio of 94 detected to missed cases in each protocol. Importantly, resource constraints are incorporated into 95 this comparison by accounting for unscreened cases in the standard 1:1 scheme (Eq. S6-8). The 96 relative increase in the detection-to-miss ratio between the pooled and standard 1:1 schemes 97 exhibits even more significant gains than those observed for testing throughput (Fig. 1) . is examined with a set of 186 positive rRT-PCR diagnostic test results for COVID-19 (Fig. 2) , Case Isolation and Ring Vaccination to Control Ebola Characterizing the Epidemiology of the 2009 Influenza A/H1N1 Pandemic in 174 Adaptive human behavior in epidemiological models The Effectiveness of Contact Tracing in Emerging 184 Epidemics Responding to global infectious disease outbreaks: Lessons from SARS on the 186 role of risk perception, communication and management Closure of schools during an influenza pandemic. The Lancet Infectious Diseases COVID-19-New Insights on a Rapidly Changing Epidemic What can early Canadian experience 195 screening for COVID-19 teach us about how to prepare for a pandemic? Carrier Transmission of COVID-19 Multiplex PCR System for Rapid Detection of Pathogens in Patients with Presumed 201 Utilization of Nucleic Acid Amplification Assays for the Detection of 203 Respiratory Viruses Detection of Respiratory Viruses by Molecular Methods Emerging Molecular Assays for Detection and Characterization of 207 Respiratory Viruses Real-Time PCR in Clinical 210 Microbiology: Applications for Routine Laboratory Testing Real-time PCR in virology The use of diagnostic tests: A probabilistic 215 approach Reasoning Foundations of Medical Diagnosis: Symbolic logic, 218 probability, and value theory aid our understanding of how physicians reason Diagnostic Decision: Probability Theory in the Use of Diagnostic Tests: An 221 Introduction to Critical Study of the Literature The detection of defective numbers of large populations Detection of 226 acute infections during HIV testing in North Carolina Utility of 229 pooled urine specimens for detection of Chlamydia trachomatis and Neisseria men attending public sexually transmitted infection clinics in Mumbai, India, by PCR Pooling 233 nasopharyngeal/throat swab specimens to increase testing capacity for influenza viruses by 234 PCR Estimating the prevalence of multiple 236 diseases from two-stage hierarchical pooling Group Testing Algorithms for Case Identification in the Presence of Test Error New coronavirus outbreak: 241 Framing questions for pandemic prevention Detection of 2019 novel coronavirus 247 (2019-nCoV) by real-time RT-PCR The exact solution to the two-stage group-testing problem ARUP COVID-19 rRT-PCR diagnostic testing results. Testing method is the EUA 251 approved Hologic Panther Fusion SARS-COV2 assay Detection of SARS-CoV-2 in 254 Different Types of Clinical Specimens Stratified Specimen Pooling Strategy for Chlamydia trachomatis and Neisseria gonorrhoeae Countries test tactics in 'war' against COVID-19