An Early Warning Score to predict ICU admission in COVID-19 positive patients

      The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) that causes coronavirus disease 2019 (COVID-19) poses multiple challenges to our healthcare systems. A particular challenge is the surge of hospital admissions with a significant fraction requiring transfer to intensive care units (ICU) because of respiratory failure.
      • Grasselli G.
      • Pesenti A.
      • Cecconi M.
      Critical Care Utilization for the COVID-19 Outbreak in Lombardy.
      ,
      • Guan W.
      • Ni Z.
      • Hu Y.
      • et al.
      Clinical Characteristics of Coronavirus Disease 2019 in China.
      Early recognition of patients requiring ICU admission is a critical step in the management of COVID-19 patients. We read with interest the communication in this Journal from Su and colleagues who investigated the utility of clinical scoring systems to predict ICU requirement in patients with COVID-19.
      • Su Y.
      • Tu G.-W.
      • Ju M.-J.
      • et al.
      Comparison of CRB-65 and quick sepsis-related organ failure assessment for predicting the need for intensive respiratory or vasopressor support in patients with COVID-19.
      Scoring at admission may however be fraught with heterogeneity due to the timing of presentation. Here, we asked whether a monitoring tool on the wards could help identify patients that would require intensive care up to 36 hours in advance. Early warning scores have been developed as composite scores to quantify patient's deterioration.
      • Prytherch D.R.
      • Smith G.B.
      • Schmidt P.E.
      • Featherstone P.I.
      ViEWS—Towards a national early warning score for detecting adult inpatient deterioration.
      We reviewed data from 36 consecutive PCR-positive COVID-19 patients admitted to the medical wards of the Lausanne University Hospital between March 2, 2020 and March 17, 2020 and examined whether a modified version of the Early Warning Score (EWS) described by Prytherch et al.
      • Prytherch D.R.
      • Smith G.B.
      • Schmidt P.E.
      • Featherstone P.I.
      ViEWS—Towards a national early warning score for detecting adult inpatient deterioration.
      could contribute to an early identification of COVID-19 patients requiring ICU admission. All variables described by Prytherch et al. were included except for the AVPU variable, which is only documented in a subset of departments at our hospital. Physiological variables were analyzed during a 12 to 36-hour period prior to ICU admission (ICU group) or prior to the most abnormal respiratory variables (i.e. FiO2 or respiratory rate) (non-ICU group) defined as t0. EWS was calculated at 12-hour intervals prior t0. Among the 36 patients, 9 were excluded for the following reasons: incomplete or single set of physiological variables (7 patients), pregnancy or immediate ICU admission (1 patient each). Nine required ICU admission and 17 did not. The median age (range) of patients was 74 yr (39-86) in the ICU group and 65.5 yr (26-83) in the non-ICU group (p=0.793). Mean duration of symptoms prior to t0 was 7.8 days in the ICU-group and 7.6 days in the non-ICU group. Risk factors
      • Guan W.
      • Ni Z.
      • Hu Y.
      • et al.
      Clinical Characteristics of Coronavirus Disease 2019 in China.
      associated with severe COVID-19 were present in 80.8% of the patients (ICU group: 77.7% and non-ICU group: 82.3%). Fig. 1 shows the evolution of the EWS over a study period up to 36 hours prior t0 in the two groups of patients. The median EWS was significantly higher in a time-dependent manner in ICU group than in the non-ICU group (p<0.0001) as assessed by mixed effects model
      • McLean R.A.
      • Sanders W.L.
      • Stroup W.W.
      A Unified Approach to Mixed Linear Models.
      . At t0 or t-12 hours, an EWS greater than 7 predicted ICU admission with sensitivities and specificities of 87% and 93% and 94% and 78%, respectively (AUROC 0.98 and 0.88, respectively).
      Fig. 1
      Fig. 1Evolution of the Early Warning Score (EWS) over time in the two groups of patients. EWS was assessed at 12-h intervals over a 36-hour time period. Time zero refers to the time at which the last set of physiological variables was recorded prior to ICU admission (ICU group, red) or to the worst respiratory variables (non-ICU group, blue). Median (range) EWS was 1 (0-3) vs 7 (5-8) (p= 0.25) at -36-h; 4 (0-9) vs 7 (3-11) (p = 0.1875) at -24-h, 2 (0-8) vs 9 (0-11) (p = 0.0078) at –12-h and 9 (7-12) vs 4 (0-8) (p= 0.0078) at -0h (Wilcoxon test). Number of patients per treatment group at each time point (-36h: 3 vs. 11; -24h: 7 vs 12, -12h: 9 vs 15, 0h: 8 vs 17).
      To read this article in full you will need to make a payment

      Purchase one-time access:

      Academic and Personal
      One-time access price info
      • For academic or personal research use, select 'Academic and Personal'
      • For corporate R&D use, select 'Corporate R&D Professionals'

      Subscribe:

      Subscribe to Journal of Infection
      Already a print subscriber? Claim online access
      Already an online subscriber? Sign in
      Institutional Access: Sign in to ScienceDirect

      References

        • Grasselli G.
        • Pesenti A.
        • Cecconi M.
        Critical Care Utilization for the COVID-19 Outbreak in Lombardy.
        Jama. 2020; 323https://doi.org/10.1001/jama.2020.4031
        • Guan W.
        • Ni Z.
        • Hu Y.
        • et al.
        Clinical Characteristics of Coronavirus Disease 2019 in China.
        New Engl J Med. 2020; https://doi.org/10.1056/nejmoa2002032
        • Su Y.
        • Tu G.-W.
        • Ju M.-J.
        • et al.
        Comparison of CRB-65 and quick sepsis-related organ failure assessment for predicting the need for intensive respiratory or vasopressor support in patients with COVID-19.
        J Infect. 2020; https://doi.org/10.1016/j.jinf.2020.05.007
        • Prytherch D.R.
        • Smith G.B.
        • Schmidt P.E.
        • Featherstone P.I.
        ViEWS—Towards a national early warning score for detecting adult inpatient deterioration.
        Resuscitation. 2010; 81: 932-937
        • McLean R.A.
        • Sanders W.L.
        • Stroup W.W.
        A Unified Approach to Mixed Linear Models.
        Am Statistician. 1991; 45: 54