Our study demonstrated that in critically ill patients with COVID-19, almost 10% of patients continued to shed replication competent virus after 10 days. The median duration of shedding of live virus among the these patients was 13 [11–19] days post-initial testing. Almost three quarters of patients were still RT-PCR positive at a median of 14 [12–16] days after an initial positive test. The remaining 17.6% patients were RT-PCR negative for SARS-CoV-2 at 14 [12–16] days post-initial testing.
There was no difference in baseline medical comorbidities between the culture positive and culture negative groups. There was also no difference in the median number of days from a patient’s initial positive test, and their ability to grow replication competent virus upon repeat testing.
The TCID50/mL for culture positive samples was 562. Of note, this was not significantly lower than the median value that we had previously reported in a study of outpatients with COVID-19 (1780 [282–8511] vs. 562 [178–3160], p = 0.99) despite the ICU patients being sampled further into their illness (3 [2–4] vs. 13 [11–19] days, respectively, p < 0.001]. While the exact infectious dose of SARS-CoV-2 is not known, it is estimated to be between 36 and 179 virions, meaning that our value of 562 would theoretically be enough to cause infection .
Other groups have attempted to describe the relationship between RT-PCR positive and culture positive patients. Singanayagam et. al looked at a subset of 20 patients with critical illness in their larger study of 253 patients with COVID-19 and found that the median Ct of culture positive samples in this group was higher than ours (32.6 [28.4–33.4]) . The authors did not report their culture positive rate of this subset of patients, and they used SARS-CoV-2 RNA-dependent RNA polymerase as their target, making comparisons to our results difficult.
Cui et al. took samples from 21 critically ill patients and found that they were unable to grow live virus if the Ct value was greater than 28.4, and they had no positive viral cultures after 12 days in hospital . This group utilized the N gene of SARS-CoV-2, and only one of the patients was mechanically ventilated, and the median SOFA score was 0, suggesting that their patients were less ill than our patients.
The largest study to date looking at the duration of viral shedding was by van Kampen et al. who looked at the ability to recover replication competent virus in a group of 129 hospitalized patients . Eighty-nine of these patients were in ICU, and 91% of them were receiving mechanical ventilation. The authors did not report SOFA or other severity of illness scores, but presumably, with the high rates of mechanical ventilation, their patients were similar to ours. These authors found that the ability to recover live virus was statistically unlikely when the viral load was below 6.63 Log10 RNA copies/mL. As these authors used the same target gene as we did (E gene), this value corresponds to a Ct of ~ 22 which is consistent with our data. The slight variation is likely due to differences in the standard curve creation of Ct and Log10 RNA copies/mL.
Multivariable logistic regression, using culture positivity as a dependent variable and SOFA score, need for mechanical ventilation and RT-PCR cycle threshold demonstrated that only cycle threshold was predictive of a positive viral culture on repeat testing (OR 0.71, 95% CI 0.59–0.87, p = 0.001). This implies that for every one unit increase in Ct, there was a 29% decrease in the odds of being culture positive. Further, ROC analysis demonstrated that a Ct > 25 was highly predictive of not being able to recover live virus with a specificity of 100% (95% CI 70–100%). This also suggests that patient factors are not helpful in determining culture positivity, but this result is limited by the small number of positive culture results in our study.
Our results have implications for the de-escalation in isolation precautions for patients with COVID-19 in ICUs. There are a variety of policies that have been advocated for de-escalation of precautions in patients with COVID-19 [1, 2]. These include a time based or symptom-based strategy, or a strategy based on repeat testing. Our results suggest that duration between tests nor symptoms nor severity of illness can predict the presence of replication competent virus. In fact, there were two patients who remained culture positive greater than 20 days after their initial test. Patients can remain RT-PCR positive for a significant length of time after their initial test, and not grow live virus. The longest length of time a patient was RT-PCR positive was 29 days. This results in prolonged isolation requirements, and the need for more personal protective equipment use by heath care providers in patients who are not infectious. The CDC currently recommends a longer isolation period for critically ill patients with COVID-19 . Our results suggest that in some cases, this time may be excessive, and in others, insufficient to ensure the absence of live virus.
PCR and other nucleic amplification (NA) strategies have surpassed viral culture as the gold standard viral diagnostic, because of their wider application, higher sensitivity, rapid performance, and ability for field deployment. A major drawback to PCR and other diagnostic approaches (including other NA, serology, antigen detection) is that they all fail to determine virus infectivity: PCR sensitivity is excellent but specificity for detecting replicative virus is poor .
Cycle threshold, an easily obtainable number from RT-PCR results, had excellent ability to predict the presence of live virus. A de-escalation strategy of isolation precautions that involves re-testing critically ill patients and, if patients are RT-PCR positive with a Ct > 25, isolation precautions can be safely removed with a low risk of remaining infectious. The Ct threshold of 25 for detection of live virus in ICU patients is remarkably consistent with previous work we have done in outpatient adults and pediatric patients with COVID [3, 7].
Limitations to our study include the limited sample size (although the largest to date), and the lack of repeat testing to determine when patients transitioned to a negative culture/RT-PCR test. We felt that daily repeated tests in this patient population were overly invasive, considering how long patients may remain RT-PCR positive. Degradation of sample from collection to viral culture, due to a freeze–thaw cycle, also may have reduced our yield. However, even if this was case, the consistency of the Ct in this study, our previous work and that of van Kampen strongly suggests that the Ct can be used to predict the presence of live virus and may play a role in discontinuation of isolation precautions in patients with COVID-19 in the ICU. Our study did not look at lower respiratory tract specimens, leaving the possibility that the lower respiratory tract could still contain live virus. The discordance between upper and lower respiratory tract sample positivity is possible and has been demonstrated in other studies [14, 15]. The number of immunosuppressed patients in our study is small, and it is well known that this group can shed virus for a prolonged period. We are therefore not able to examine risk in this population due to small sample size.
Institutions that use other qRT-PCR assays and PCR targets will need to determine the threshold for live viral growth based on cycle threshold, thereby potentially limiting the applicability. Previous work has shown that the cross-platform validity for PCR of the E gene of SARS-CoV-2 is robust, suggesting that the Ct from different platforms for this gene may be used to predict the presence of live virus . The salient point, however, is that PCR detects both culturable, and non-culturable/non-infectious viral particles, and that the Ct cutoff can be defined above which no live virus can be found.
Finally, the sampling was done before the delta strain of SARS-CoV-2 became dominant in the province. Ongoing work at our institution has shown that it is extremely unlikely to grow live delta variant virus in culture from outpatient samples when the sample has a Ct greater than 25. This is consistent with previous work we have completed in outpatients who grew wild type virus, so we feel it is likely that the Ct value will hold true for variant infections in the ICU population [3, 7].
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.