Optimal COVID-19 quarantine and testing strategies

The study addresses whether the duration of COVID-19 quarantine can be safely shortened by optimally timed testing without increasing the probability of post-quarantine transmission (PQT). In the context of substantial socioeconomic and mental health burdens of 14-day quarantines and the need to safely reopen workplaces, schools, and other close-quarter environments, the authors evaluate how integrating RT-PCR testing with quarantine can reduce transmission risk. They note limitations of symptom-based screening due to presymptomatic and asymptomatic transmission and the risk of false negatives when testing too early. The purpose is to quantify PQT under various quarantine durations and testing strategies (entry, exit, both) for travelers (unknown exposure time) and contact-traced individuals (unknown or known exposure time), and to identify optimal testing times that can maintain or improve upon the effectiveness of a 14-day quarantine.

The paper situates its work within evidence that quarantine and isolation reduce transmission, acknowledges widespread use of 14-day quarantines per WHO/CDC guidance, and highlights the shortcomings of symptom-based screening given significant presymptomatic/asymptomatic spread. Prior modeling and empirical studies have examined impacts of testing, contact tracing, and quarantine on epidemic dynamics, but there is no consensus on optimal quarantine duration or test timing that minimizes PQT, especially considering time-varying RT-PCR sensitivity and infectivity profiles. Some jurisdictions implemented testing at both entry and exit of 14-day quarantines, but without leveraging testing to shorten quarantine burden. This study fills the gap by integrating infectivity profiles and temporal diagnostic sensitivity to optimize testing within quarantine frameworks.

- Framework: Mathematical modeling of PQT (probability of at least one post-quarantine transmission) combining time-varying infectivity and temporal RT-PCR diagnostic sensitivity. - Epidemiological parameters: Basic reproduction number R0=2.5; incubation period 8.29 days; baseline latent period 2.9 days (with sensitivity at 1.9 and 3.9 days); 30.8% asymptomatic infections; perfect self-isolation upon symptom onset; RT-PCR specificity assumed 100%. - Infectivity profile: Derived from transmission-pair data and constrained by incubation and latent periods. Infectivity during latent period modeled exponentially; infectivity set to zero by 20 days post-symptom onset. - Diagnostic sensitivity: Fitted logistic regression to time-varying RT-PCR sensitivity from 0–25 days post-symptom onset (digitized literature data), with extrapolation to pre-symptomatic period aligned to infectivity to obtain full time course of sensitivity. - Transmission stochasticity: Expected post-quarantine secondary cases modeled with a negative binomial distribution (dispersion k=0.257) to compute PQT; Poisson sensitivity analysis also conducted. - Testing logistics: Assumed a 1-day delay from sample collection to result; thus, an “exit” test occurs one day before quarantine ends. Individuals testing positive or developing symptoms are isolated until recovery. - Scenarios analyzed: 1) Travel quarantine: individuals enter quarantine at random times over their infectious course (unknown exposure time). 2) Contact-traced quarantine: entry to quarantine linked to index case symptom onset; analyzed with no delay and with a 1-day delay in tracing and quarantine initiation (generally earlier in infection than travelers). 3) Known exposure date: optimization of testing day given a known time since exposure at quarantine start. For each, tested strategies: no testing; entry-only testing; exit-only testing; both entry and exit testing; quarantine durations 1–14 days. - Prevalence/cohort analysis: PQT quantified for cohorts (e.g., n=40) under varying community prevalence to assess the impact of multiple simultaneously quarantined individuals. - Real-world application: Evaluated an offshore oil and gas company’s protocols. Early phase: 3-day quarantine with entry testing. Revised protocols: Region A adopted 7-day quarantine with entry and exit testing; Region B adopted 5-day quarantine with entry and exit testing. A total of ~4,010–4,040 RT-PCR tests between April 11 and August 26, 2020 were analyzed with positivity stratified by entry vs exit tests. Modeled impact on expected offshore transmission events was estimated based on detected positives and counterfactuals without exit testing.

- General optimization: - Testing on exit (or testing on both entry and exit) enables reducing a 14-day quarantine by approximately 50% while maintaining comparable PQT; testing on entry alone shortens quarantine by at most one day. - Without testing and assuming self-isolation at symptom onset, PQT falls below 0.25 with quarantine ≥3 days and below 0.05 with quarantine ≥8 days. - With a 1-day result delay, optimal single-test timing: on exit for quarantines ≤7 days; around day 6 for quarantines ≥8 days. - Travel (unknown exposure time): Exit testing provides the largest reduction in PQT across quarantine durations, outperforming entry-only testing. - Contact tracing: - Earlier entry into quarantine shifts optimal testing to later in quarantine; for quarantines ≤7 days, exit testing remains optimal. - A 7-day quarantine with exit testing, or a 6-day quarantine with both entry and exit testing, achieves PQT equal to or lower than a 14-day quarantine without testing. - Entry-only testing confers minimal benefit when exposure time is unknown. - Known exposure date: With a 14-day quarantine starting 1 day after infection, optimal testing day is approximately day 6; as quarantine starts later post-infection, the optimal test day decreases roughly linearly. For shorter quarantines, testing on exit is optimal when entry occurs early in infection. - Community prevalence effects (example cohort n=40, prevalence 1%): - 5-day quarantine: PQT ≈ 0.06 (entry-only) vs ≈ 0.005 (entry+exit). - 7-day quarantine: PQT ≈ 0.02 (entry-only) vs ≈ 0.001 (entry+exit). - Sensitivity analyses: - Varying latent period by ±1 day modestly affects infectivity profiles and optimal testing day (≤1 day shift) and tends to lower PQT with shorter latent periods for longer quarantines. - Modeling PQT with Poisson vs negative binomial distributions does not alter qualitative conclusions about exit testing superiority. - Real-world offshore implementation: - Total positives: 69 (~1.7%). With entry+exit testing, 47 positives were detected overall; 16 were negative on entry but positive on exit, preventing an estimated nine offshore transmission events that could have led to outbreaks. - Implementing exit testing reduced modeled PQT by ~98% with a 7-day quarantine (Region A) and ~93% with a 5-day quarantine (Region B) compared to a 3-day entry-only regime. - No workers who tested negative on both entry and exit subsequently developed COVID-19 while offshore.

The findings show that strategically timed testing substantially enhances quarantine effectiveness and allows shorter quarantines without increasing post-quarantine risk. Exit testing is consistently more impactful than entry testing when exposure time is unknown, because early testing suffers from low diagnostic sensitivity during the latent/early incubation period. In contact-tracing contexts, earlier entry into quarantine necessitates longer quarantines or later test timing, but combinations such as 7 days with exit testing or 6 days with tests on entry and exit can match or outperform the traditional 14-day quarantine without testing. Results are robust across reasonable variations in latent period and transmission dispersion and hold across different prevalence levels, though higher prevalence increases the value of additional tests to mitigate rare false negatives. In confined or high-risk environments (e.g., offshore rigs), exit testing prevented expected transmissions and potential outbreaks, illustrating real-world utility. Although absolute PQT depends on setting-specific R and mitigation measures (masking, distancing), the relative advantage of exit over entry testing remains.

Optimizing the timing of limited RT-PCR testing within quarantine can halve the standard 14-day quarantine while maintaining or lowering the probability of post-quarantine transmission. Exit testing—alone or combined with entry testing—provides the greatest benefit, whereas entry-only testing yields limited additional protection when the time of exposure is unknown. These insights inform policies for travel-related quarantine and contact tracing, enabling safer, less burdensome protocols across workplaces, universities, military, and other close-quarter environments. Future work should assess alternate diagnostics (e.g., saliva RT-PCR, rapid antigen tests), multiple-test schedules under varying prevalence, and setting-specific transmission risks to refine optimal strategies.

- Assumes constant baseline R0=2.5 and perfect self-isolation upon symptom onset, which may not hold universally. - RT-PCR temporal sensitivity was modeled from published datasets and extrapolated pre-symptomatically; real-world performance varies by assay, sampling site, and collection quality. - Specificity assumed to be 100%, potentially underestimating false positives and operational impacts. - One-day test result delay was assumed; longer or variable delays would shift optimal timing and PQT. - Asymptomatic infectiousness was assumed similar to symptomatic cases; deviations would affect estimated PQT. - Negative binomial dispersion parameter fixed (k=0.257); other dispersion values alter absolute PQT but not qualitative conclusions. - Real-world implementation data were from a specific industry and logistics (offshore rigs); generalizability to other settings may vary. - Limited exploration of extensive multi-test strategies due to test availability and uncertain correlations among repeated false negatives.