Biased and unbiased estimation of the average length of stay in intensive care units in the Covid-19 pandemic

Abstract

Background

The average length of stay (LOS) in the intensive care unit (ICU_ALOS) is a helpful parameter summarizing critical bed occupancy. During the outbreak of a novel virus, estimating early a reliable ICU_ALOS estimate of infected patients is critical to accurately parameterize models examining mitigation and preparedness scenarios.

Methods

Two estimation methods of ICU_ALOS were compared: the average LOS of already discharged patients at the date of estimation (DPE), and a standard parametric method used for analyzing time-to-event data which fits a given distribution to observed data and includes the censored stays of patients still treated in the ICU at the date of estimation (CPE). Methods were compared on a series of all COVID-19 consecutive cases ( n = 59) admitted in an ICU devoted to such patients. At the last follow-up date, 99 days after the first admission, all patients but one had been discharged. A simulation study investigated the generalizability of the methods' patterns. CPE and DPE estimates were also compared to COVID-19 estimates reported to date.

Results

LOS ≥ 30 days concerned 14 out of the 59 patients (24%), including 8 of the 21 deaths observed. Two months after the first admission, 38 (64%) patients had been discharged, with corresponding DPE and CPE estimates of ICU_ALOS (95% CI) at 13.0 days (10.4–15.6) and 23.1 days (18.1–29.7), respectively. Series' true ICU_ALOS was greater than 21 days, well above reported estimates to date.

Conclusions

Discharges of short stays are more likely observed earlier during the course of an outbreak. Cautious unbiased ICU_ALOS estimates suggest parameterizing a higher burden of ICU bed occupancy than that adopted to date in COVID-19 forecasting models.

Funding

Support by the National Natural Science Foundation of China (81900097 to Dr. Zhou) and the Emergency Response Project of Hubei Science and Technology Department (2020FCA023 to Pr. Zhao).

SciScore for 10.1101/2020.04.21.20073916: (What is this?)

Please note, not all rigor criteria are appropriate for all manuscripts.

Table 1: Rigor

Institutional Review Board Statement	IRB: Ethics: This study was approved by the Medical Ethics Committee, ZHWU (Clinical Ethical Approval No.2020005). Consent: The informed consent was waived by the Medical Ethics Committee for emerging infectious disease.
Randomization	not detected.
Blinding	not detected.
Power Analysis	not detected.
Sex as a biological variable	not detected.

Table 2: Resources

No key resources detected.

Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:

Nevertheless, our study also has some limitations. The study is monocentric and therefore, the extrapolation of our estimates to other settings is questionable. Moreover, cultural behaviors, modifications of triage decisions, changes in the management of patients may vary not only from one place to another, but also with time according to the pressure of this threatening epidemics on the organization of healthcare workers. However identical drawbacks would also stand for most studies reported to date. The main outcome of this study is alerting the community about three elements. First, all scientists working on COVID-19 must realize that when dealing with data relating to LOS, they should imperatively use appropriate methods devoted to the analysis of censored data. Such methods are not original, they belong to the standard tools used in the domain of survival analysis and are easily available in any statistical software. There is no reason for avoiding their usage, and the reader will find an illustrative computer code in the Supplementary Appendix 3. An additional strength of these methods – not explored in this article but illustrated in Supplementary Appendix 3 – is their ability to fit individual characteristics of patients with multivariable models to predict LOS in specific strata of the population in order to adapt to varying recruitment settings. A side result of the analyses made in the present study suggests that the death rate of COVID-19 patients in the ICU might...

Results from TrialIdentifier: No clinical trial numbers were referenced.

Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

Results from JetFighter: We did not find any issues relating to colormaps.

Results from rtransparent:

Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
No protocol registration statement was detected.

Read the original source