Skip to main content
PMC home page
PLOS ONE logoLink to PLOS ONE
. 2021 Mar 26;16(3):e0249231. doi: 10.1371/journal.pone.0249231

Risk factors for in-hospital mortality in laboratory-confirmed COVID-19 patients in the Netherlands: A competing risk survival analysis

Gerine Nijman 1,2, Maike Wientjes 3, Jordache Ramjith 4, Nico Janssen 2,5, Jacobien Hoogerwerf 1,2, Evertine Abbink 1, Marc Blaauw 6, Ton Dofferhoff 7, Marjan van Apeldoorn 8, Karin Veerman 9, Quirijn de Mast 1,2, Jaap ten Oever 1,2, Wouter Hoefsloot 2,10, Monique H Reijers 2,10, Reinout van Crevel 1,2, Josephine S van de Maat 1,2,*
Editor: Francesco Di Gennaro11
PMCID: PMC7997038  PMID: 33770140

Abstract

Background

To date, survival data on risk factors for COVID-19 mortality in western Europe is limited, and none of the published survival studies have used a competing risk approach. This study aims to identify risk factors for in-hospital mortality in COVID-19 patients in the Netherlands, considering recovery as a competing risk.

Methods

In this observational multicenter cohort study we included adults with PCR-confirmed SARS-CoV-2 infection that were admitted to one of five hospitals in the Netherlands (March to May 2020). We performed a competing risk survival analysis, presenting cause-specific hazard ratios (HRCS) for the effect of preselected factors on the absolute risk of death and recovery.

Results

1,006 patients were included (63.9% male; median age 69 years, IQR: 58–77). Patients were hospitalized for a median duration of 6 days (IQR: 3–13); 243 (24.6%) of them died, 689 (69.9%) recovered, and 74 (7.4%) were censored. Patients with higher age (HRCS 1.10, 95% CI 1.08–1.12), immunocompromised state (HRCS 1.46, 95% CI 1.08–1.98), who used anticoagulants or antiplatelet medication (HRCS 1.38, 95% CI 1.01–1.88), with higher modified early warning score (MEWS) (HRCS 1.09, 95% CI 1.01–1.18), and higher blood LDH at time of admission (HRCS 6.68, 95% CI 1.95–22.8) had increased risk of death, whereas fever (HRCS 0.70, 95% CI 0.52–0.95) decreased risk of death. We found no increased mortality risk in male patients, high BMI or diabetes.

Conclusion

Our competing risk survival analysis confirms specific risk factors for COVID-19 mortality in a the Netherlands, which can be used for prediction research, more intense in-hospital monitoring or prioritizing particular patients for new treatments or vaccination.

Introduction

The pandemic of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) associated coronavirus disease 2019 (COVID-19) causes significant morbidity and mortality worldwide. The first laboratory-confirmed case of COVID-19 in the Netherlands was reported on February 27, 2020 [1]. The disease initially spread across the southern provinces and rapidly disseminated further throughout the country. Currently, the Netherlands is amidst a second wave of infections and >350 thousand confirmed cases have been reported, including >7 thousand deaths [2, 3].

Studies describing the clinical features of COVID-19 and risk factors associated with incidence and timing of poor outcome have been extensively published [4], but survival data regarding risk factors for COVID-19 mortality in north-western Europe is limited. Furthermore, to our knowledge, none of the published survival studies have considered recovery as competing risk for mortality. Not taking competing risks into account leads to biased mortality estimates and to overestimation of survival curves [5, 6]. Analyzing mortality data with a more accurate competing risk analysis adds to the growing body of evidence on disease course and risk factors of a poor COVID-19 outcome. More robust knowledge about these risk factors is crucial to inform international prediction research [4]. In the current phase of the pandemic, now specific vaccines are underway, correct identification of patients at risk of severe disease and mortality is essential to prioritize patients for vaccination. In addition, while the virus still circulates in the population, it can guide clinical decisions on close monitoring, ICU admission or selection for new treatments. Therefore, this study aims to identify risk factors for in-hospital mortality in laboratory-confirmed COVID-19 patients in the Netherlands, using a competing risk approach.

Materials and methods

In this observational, multicenter cohort study, we consecutively included adult patients with PCR-confirmed infection with SARS-CoV-2, who were admitted to one of five collaborating hospitals in the region of Gelderland and North-Brabant for at least 24 hours between March and May 2020. For the current analysis, we excluded patients if data regarding duration of hospital admission were missing for patients who died or recovered. A flow chart of inclusion is shown in S1 Fig. The study was approved by the institutional review board (IRB) of the Radboud university medical center (number 2020–2923 and 2020–6344). According to the IRB, only oral consent was required. Oral consent was obtained from all patients or their family and documented in the electronic medical records.

Data collection and definitions

We extracted routine data from the electronic medical records, including baseline patient characteristics, information on clinical presentation, diagnostics, disease course, treatment and outcome. All patient data were entered anonymously into a web-based electronic case report form (using Castor Electronic Data Capture), only using a study identifier. The key linking patient information to study ID was saved in a local, protected file in the participating hospitals and not available to the researchers performing data analyses. Detailed methods on data collection and definitions are described in our parallel publication in this issue. For the current analysis, we used the Modified Early Warning Score (MEWS) as a summary measure of abnormal vital signs, based on body temperature, blood pressure, heart rate, respiratory rate, and the Alert Voice Pain Unresponsive (AVPU) score [7]. The AVPU score was not reported in the electronic medical records, so we used ‘level of consciousness’ as a proxy in the MEWS, with ‘conscious’ indicating Alert, and ‘reduced level of consciousness’ substituting Verbal, Pain and Unresponsive. We totaled duration of intensive care unit (ICU) admission for patients who were admitted to the ICU more than once. Readmissions within 4 weeks after initial discharge were considered extensions of the initial hospitalization. Follow-up duration was set to the duration until database lock for patients still admitted at the end of the study (n = 6). We defined death as in-hospital death or palliative discharge. Patients who were discharged due to clinical improvement, for medical rehabilitation, or transfer to a nursing home were considered ‘recovered’. Patients who were transferred to a non-study hospital, who were lost to follow-up because of other reasons, or whose reason of discharge was unknown or missing were censored.

Statistical analysis

Descriptive analyses were performed. Furthermore, we used a competing risks survival analysis approach to evaluate the effect of risk factors on the time from admission to death and on the time from admission to recovery [8, 9]. According to Noordzij et al, a competing risk occurs ‘when subjects can experience one or more events or outcomes which ‘compete’ with the outcome of interest’ [9]. In our study, recovery is considered a competing risk for mortality and is taken into account as an extra outcome, whereas in standard survival analyses, patients who recover are censored. However, the latter violates the assumption of noninformative censoring, i.e. the recovered patients are not representative of those who are still admitted to the hospital in terms of their risk of dying. Censoring recovered patients induces bias and overestimation of survival curves, i.e. Kaplan Meier will estimate incidence of death with upwards biases [8, 9]. For the competing risk analysis, we estimated univariable and multivariable cause-specific hazard ratios (HRCS) for death and recovery for selected risk factors [5, 8]. These risk factors were pre-selected based on literature and expert opinion to be clinically relevant and routinely available at time of presentation, rather than based on statistical significance [10]. Furthermore, cumulative incidence probabilities were estimated using the Fine & Gray approach [5, 8]. Gray’s test was used to compare equality of cumulative incidence curves (CIFs) across subgroups [11, 12]. The proportional hazards assumption was checked by an evaluation of the Schoenfeld residuals, as shown in S2 Fig.

Multivariable HRCS were also estimated for a subpopulation of patients with a ‘non ICU admission policy’. The Netherlands executes a restrictive policy for ICU admission for patients who do not wish to be admitted to the ICU, or for whom the expected chance of recovery is thought to be too low or their overall prognosis too poor (e.g. because of comorbidity) to outweigh the potential harm of an ICU admission. The cause-specific hazard (CSH) model was fitted again for this subpopulation, to assess the influence of risk factors on the outcome in these patients. Statistical analyses were conducted in R (v1.2.1578), using the ‘survival’, ‘survminer’ and ‘cmprsk’ packages for time-to-event analyses. Laboratory values were log-transformed to adjust for their skewed distribution.

Missing data

Missing data in the variables ‘non ICU admission policy, ‘COVID treatment’, ‘oxygen supplementation’ and ‘ventilatory support’ were considered as ‘no’ or ‘none’, assuming that it would be reported in the medical records if the concerning variable applied to the patient. Furthermore, missing values in the variables ‘X-ray’, ‘CT-scan’, and ‘blood culture’ were considered as ‘not performed’, assuming that the results would be reported if patients had one of these tests. Missing values in all other variables were assumed to be missing at random (MAR) and were imputed using multiple imputation in ten datasets with the ‘MICE’ package. Analyses were performed on all ten imputed datasets and the results were pooled.

Results

We included 1,006 PCR-confirmed COVID-19 patients (see flowchart of inclusion in S1 Fig). Their median age of 69 years (IQR: 58–77), most were male and presented with fever, cough and dyspnea, as shown in Table 1. Nearly all patients had one or more comorbidities. Of all patients, 243 (24.6%) died in-hospital or were discharged for palliative care, 689 (69.9%) recovered, and 74 (7.4%) were censored. The median duration of hospital admission until death and recovery was 6 days (IQR: 3–11) and 7 days (IQR: 4–13), respectively.

Table 1. Characteristics of the study population.

Total (N = 1,006) Median (IQR) or N(%) Missing N(%)
Age (years) 69 (58–77) 0 (0.0)
Sex, male 643 (63.9) 0 (0.0)
BMI 27.6 (24.7–31.0) 278 (27.6)
≥1 comorbidity 904 (90.1) 3 (0.3)
Most frequent comorbidities a
    Cardiovascular disease (incl. hypertension)
    Hypertension
    Pulmonary disease
    Diabetes mellitus
    Solid organ malignancies
    Auto-immune disease
    Chronic kidney disease
582 (58.0)
390 (38.9)
246 (24.5)
229 (22.8)
149 (14.9)
122 (12.2)
112 (11.2)		 3 (0.3)
Immunocompromised b 213 (21.3) 4 (0.4)
Chronic use of ACE inhibitors and/or angiotensin-II receptor blockers 336 (33.4) 1 (0.1)
Chronic use of antiplatelet medication or anticoagulants 374 (37.2) 1 (0.1)
Non ICU admission policy 312 (31.0) 26 (2.6)
Symptoms
Duration of symptoms before admission (days) 7 (5–10) 85 (8.4)
Most frequent symptoms reported at admission c
    Fever
    Cough
    Shortness of breath (dyspnea)
    Fatigue
    Diarrhea
765 (76.7)
757 (75.9)
696 (69.7)
354 (35.5)
329 (33.0)		 8 (0.8)
Vital signs and physical examination at time of admission
Ill appearance 399 (56.7) 302 (30.0)
Dyspnea 336 (55.5) 401 (39.9)
Fever d 459 (47.2) 33 (3.3)
Blood pressure (mmHg) e
    Normal
    Hypotension
    Hypertension
862 (88.5)
7 (0.7)
105 (10.8)		 32 (3.2)
Tachycardia f 278 (28.5) 30 (3.0)
Tachypnea g 573 (60.1) 53 (5.3)
Hypoxia h 149 (15.1) 22 (2.2)
Modified Early Warning Score (MEWS) i 3 (2–4) 76 (7.6)
Laboratory parameters at time of admission
Hemoglobin (mmol/L) 8.6 (7.8–9.2) 32 (3.2)
White blood cell count (*109 /L) 6.7 (5.0–9.0) 31 (3.1)
Neutrophil count (*109 /L) 5.2 (3.5–7.3) 87 (8.6)
Lymphocyte count (*109 /L) 0.9 (0.6–1.2) 84 (8.3)
Thrombocytes (*109 /L) 205 (159–268) 40 (4.0)
C-Reactive Protein (mg/L) 92.0 (47.0–147.0) 28 (2.8)
Ferritin (μg/L) 796.5 (408.8–1466.2) 206 (20.5)
D-dimer (ng/L) j 920.0 (500.0–1810.0) 194 (19.3)
Lactate dehydrogenase (U/L) 359.0 (278.5–466.5) 103 (10.2)
Procalcitonin (μg/L) 0.15 (0.08–0.33) 412 (41.0)
Creatinine (μmol/L) 84.0 (68.0–106.0) 34 (3.4)
Diagnostics
Result first PCR, positive 924 (93.8) 21 (2.1)
Second PCR performed
    Of which: result second PCR positive 		218 (21.7)
133 (61.0) 		12 (1.2)
4 (1.8)
Chest X-Ray performed
    Of which: chest X-ray suggestive for COVID-19 		608 (61.6)
325 (53.5) 		19 (1.9)
208 (34.2)
CT scan performed
    Of which: CORADS classification
        1–3 (not suggestive for COVID-19)
        4–5 (suggestive for COVID-19)
    Of which: CT severity score 		500 (50.9)
44 (11.5)
337 (88.5)
12 (9–15) 		23 (2.3)
125 (25.0)
158 (31.6)
Discharge
Duration of hospital admission (days) 6 (3–13) 6 (0.6)
Reason for discharge
    Clinical improvement
    Patient deceased in-hospital
    Palliative discharge
    Transfer to a non-study hospital
    Other
    Unknown
689 (69.9)
238 (24.1)
5 (0.5)
44 (4.5)
8 (0.8)
2 (0.2)		 20 (2.0)
Readmission after initial discharge 48 (4.8) 0 (0.0)
ICU admission 207 (21.2) 30 (3.0)
Duration of hospital admission to first ICU admission (days) 2.0 (0.5–4.0) 0 (0.0)
Duration of ICU admission (days) 17 (10–27) 13 (6.3)
Invasive ventilatory support (incl. prone position) 184 (18.4) 5 (0.5)
Complications k 640 (63.6) 0 (0.0)
Open in a new tab

Abbreviations: N = number, IQR = interquartile range, BMI = body mass index

a Comorbidities present in >10% of the population

b Immunocompromised was defined as having a haematologic malignancy, stem cell or organ transplantation, auto-immune disease, HIV/AIDS and/or use of immunosuppressive medication.

c five most commonly reported symptoms

d Fever was defined as a body temperature ≥ 38.0°C.

e Hypertension was defined as: systolic blood pressure ≥140 mmHg and diastolic blood pressure ≥90 mmHg.

Hypotension was defined as: systolic blood pressure <90 mmHg and diastolic blood pressure <60 mmHg.

f Tachycardia was defined as a heart rate >100 bpm.

g Tachypnea was defined as a respiratory rate >20 bpm.

h Hypoxia was defined as a peripheral oxygen saturation <90%.

i Adjusted version of Modified Early Warning Score.

j D-dimer was tested within 24h of hospital admission.

k The most common complications included acute kidney injury (18.2%), delirium (13.0%), and new onset/episode of atrial fibrillation (9.6%).

Cumulative incidence curves

The CIF curves of the total population show that the probability of death after one, two and three weeks of hospital admission was 15.4% (95% 13.1–17.6), 20.5% (95% CI 18.0–23.0), and 22.4% (95% CI 19.8–25.0), respectively (Fig 1). The probabilities of recovery were 38.6% (95% CI 35.5–41.6), 54.1% (95% CI 51.0–57.2), and 60.3% (95% CI 57.3–63.3), respectively. Patients aged ≥70 years had a higher chance of death (p<0.001) and a lower chance of recovery (p<0.001) than patients aged <70 years (Fig 2). Males had a lower probability of recovery (p = 0.003) than females, but there was no statistical difference between both groups for death (p = 0.05) (Fig 3).

Fig 1. Cumulative incidence plot of death and recovery in the total population.

Fig 1

Open in a new tab

The probability of death conditional on not having recovered after one, two and three weeks of hospital admission was 15.4% (95% 13.1–17.6), 20.5% (95% CI 18.0–23.0), and 22.4% (95% CI 19.8–25.0), respectively. The probability of recovery conditional on not having died after one, two and three weeks of hospital admission was 38.6% (95% CI 35.5–41.6), 54.1% (95% CI 51.0–57.2), and 60.3% (95% CI 57.3–63.3).

Fig 2. Cumulative incidence plot of death and recovery in the total population, separated by age group.

Fig 2

Open in a new tab

Gray’s test indicated a significant difference between two groups for both death (p<0.001) and recovery (p<0.001). The probability of death for patients aged <70 years after one, two and three weeks of hospital admission was 3.6% (95% CI 2–5.2), 5.4% (95% CI 3.4–7.4), and 6.5% (95% CI 4.3–8.6), respectively, whereas for patients aged ≥70 years, the probability of death was 27.8% (95% CI 23.8–31.8), 36.4% (95% CI 32.1–40.7), and 39.2% (95% CI 34.8–43.5), respectively.

Fig 3. Cumulative incidence plot of death and recovery in the total population, separated by sex.

Fig 3

Open in a new tab

Gray’s test indicated a statistically significant difference between both groups for recovery (p = 0.003), but not for death (p = 0.050). The probability of death for females after one, two and three weeks of hospital admission was 12.5% (95% CI 10.0–17.1), 17.6% (95% CI 13.7–21.6), and 19.1% (95% CI 15.0–23.2), respectively, whereas for males the probability of death was 16.4% (95% CI 13.5–19.3), 22.1% (95% CI 18.9–25.4), and 24.3% (95% CI 20.9–27.6), respectively.

Univariable and multivariable SH model

HRCS from univariable and multivariable SH models are reported in Table 2. Univariable analyses showed that older age increased the risk of death: with every year increase in age, the risk of death increased with 9% (HRCS 1.09, 95% CI 1.08–1.11), and the chances of recovery decreased with 1% (HRCS 0.99, 95% CI 0.98–0.99). Of the comorbidities, cardiovascular disease, hypertension and pulmonary diseased increased risk of death. Patients with cardiovascular disease had a 99% increase in the risk of death (HRCS 1.99, 95% CI 1.50–2.65), and a 17% decrease in the chances of recovery (HRCS 0.83, 95% CI 0.72–0.97). Patients with hypertension had a 39% increase in the risk of death (HRCS 1.39, 95% CI 1.07–1.79), and a 130% increase in chances of recovery (HRCS 2.30, 95% CI 1.98–2.68). Furthermore, patients with pulmonary disease had a 42% increase in risk of death (HRCS 1.42, 95% CI 1.08–1.88).

Table 2. Univariable and multivariable cause-specific hazard ratios (HRCS) including 95% confidence intervals for death and recovery.

Univariable Multivariable
Death Recovery Death Recovery
Age (years) 1.09 (1.08–1.11) * 0.99 (0.98–0.99) * 1.10 (1.08–1.12) * 0.99 (0.98–0.99) *
Sex, male 1.15 (0.88–1.52) 0.80 (0.68–0.93) * 1.07 (0.79–1.47) 0.90 (0.75–1.08)
BMI 0.99 (0.96–1.01) 1.01 (0.99–1.03) 1.01 (0.98–1.04) 1.01 (0.99–1.03)
Diabetes Mellitus 1.23 (0.93–1.63) 0.89 (0.74–1.07) 1.17 (0.86–1.59) 0.85 (0.69–1.04)
Cardiovascular disease (incl. hypertension) 1.99 (1.50–2.65) * 0.83 (0.72–0.97) * 1.05 (0.69–1.59) 0.71 (0.54–0.93) *
Hypertension 1.39 (1.07–1.79) * 2.30 (1.98–2.68) * 0.78 (0.56–1.10) 1.15 (0.90–1.48)
Pulmonary disease 1.42 (1.08–1.88) * 1.01 (0.85–1.21) 1.33 (0.98–1.80) 0.88 (0.73–1.07)
Immunocompromised a 1.31 (0.99–1.74) 0.87 (0.72–1.05) 1.46 (1.08–1.98) * 0.76 (0.62–0.93) *
Chronic use of anticoagulant or antiplatelet medication 2.23 (1.73–2.87) * 1.01 (0.86–1.18) 1.38 (1.01–1.88) * 1.15 (0.93–1.43)
Chronic use of ACE inhibitors and/or angiotensin II receptor blockers 1.38 (1.07–1.79) * 0.97 (0.83–1.14) 0.99 (0.74–1.32) 1.09 (0.88–1.33)
Chest X-Ray
    Performed, not suggestive for COVID-19 Ref Ref Ref Ref
    Performed, suggestive for COVID-19 0.96 (0.59–1.55) 0.96 (0.71–1.28) 1.07 (0.63–1.80) 1.52 (1.05–2.19) *
    Not performed 0.76 (0.48–1.20) 1.02 (0.75–1.39) 0.90 (0.54–1.48) 1.18 (0.82–1.70)
CT scan severity score 0.99 (0.97–1.02) 0.95 (0.93–0.97) * 1.01 (0.98–1.05) 0.97 (0.95–0.99) *
Symptom duration (days) 0.98 (0.95–1.00) 1.01 (1.00–1.02) 0.98 (0.96–1.01) 1.01 (1.00–1.02) *
Symptoms, fever 0.55 (0.42–0.72) * 0.95 (0.79–1.15) 0.70 (0.52–0.95) * 1.05 (0.85–1.30)
Symptoms, dyspnea 0.75 (0.57–0.98) * 0.94 (0.80–1.11) 0.77 (0.58–1.03) 1.01 (0.84–1.21)
Modified Early Warning Score (MEWS) b 1.06 (0.99–1.14) 0.95 (0.91–1.00) * 1.09 (1.01–1.18) * 0.97 (0.93–1.02)
Neutrophil-to-lymphocyte rate c 1.58 (1.02–2.42) * 0.65 (0.50–0.85) * 0.97 (0.59–1.60) 1.01 (0.74–1.38)
Lactate dehydrogenase (U/L) c 1.67 (0.76–3.65) 0.16 (0.10–0.25) * 6.68 (1.95–22.8) * 0.25 (0.13–0.48) *
Creatinine (μmol/L) c 5.44 (3.13–9.44) * 0.63 (0.40–0.99) * 1.84 (0.87–3.90) 1.17 (0.69–1.99)
Procalcitonin (μg/L) c 1.27 (0.98–1.64) 0.78 (0.68–0.90) * 1.04 (0.76–1.41) 0.88 (0.75–1.03)
C-reactive protein (mg/L) c 1.16 (0.82–1.63) 0.61 (0.51–0.74) * 1.25 (0.79–2.00) 0.88 (0.69–1.11)
Ferritin (μg/L) c 0.69 (0.50–0.96) * 0.58 (0.49–0.70) * 0.66 (0.43–1.02) 0.77 (0.60–0.99) *
D-dimer (ng/L) d 1.10 (0.94–1.30) 0.88 (0.83–0.93) * 0.99 (0.84–1.16) 0.94 (0.87–1.00)
Open in a new tab

* Statistically significant, i.e. p<0.05.

a Immunocompromised was defined as having an hematologic malignancy, stem cell or organ transplantation, auto-immune disease, HIV/AIDS and/or use of immunosuppressive medication.

b Adjusted version of Modified Early Warning Score.

c These laboratory variables are log-transformed. HRs should be interpreted for a 10-fold increase in the concerning variable, rather than a one unit increase.

d D-dimer is determined within 24h of hospital admission.

In terms of medication, both chronic use of anticoagulants or antiplatelet medication as well as ACE inhibitors or angiotensin-II receptor blockers were associated with an increased risk of death (HRCS 2.23, 95% CI 1.73–2.87, and HRCS 1.38, 95% CI 1.07–1.79, respectively). For symptoms, both fever and dyspnea decreased risk of death. Patients with fever had a 45% decrease in risk of death (HRCS 0.55, 95% CI 0.42–0.72) and patients with dyspnea a 25% decrease in risk of death (HRCS 0.75, 95% CI 0.57–0.98). In terms of diagnostics, neutrophil-to-lymphocyte rate, creatinine levels and ferritin levels at time of admission were associated with risk of death and risk of recovery. A ten-fold increase in the neutrophil-to-lymphocyte rate was associated with a 58% increase in risk of death (HRCS 1.58, 95% CI 1.02–2.42) and a 35% decrease in the chance of recovery (HRCS 0.65, 95% CI 0.50–0.85). Furthermore, a ten-fold increase in creatinine levels at time of admission were associated with a 444% increased risk of death (HRCS 5.44, 95% CI 3.13–9.44) and a 37% decrease in chances of recovery (HRCS 0.63, 95% CI 0.40–0.99). Finally, a ten-fold increase in ferritin levels at time of admission was associated with a 31% decrease in risk of death (HRCS 0.69, 95% CI 0.50–0.96), and a 42% decrease in chances of recovery (HRCS 0.58, 95% CI 0.49–0.70).

In multivariable analyses, the following factors were associated with risk of death: age, immunocompromised state, chronic use of anticoagulants or antiplatelet medication, fever, MEWS, lactate dehydrogenase values at time of admission, and ferritin values at time of admission. Firstly, with every year increase in age, the risk of death increased with 10% (HRCS 1.10, 95% CI 1.08–1.12), and the chances of recovery decreased with 1% (HRCS 0.99, 95% CI 0.98–0.99). In other words, if in two patients all variables except for age are the same, the patient who is one year older has a 10% higher risk of dying. Furthermore, patients with immunocompromised state had a 46% increased risk of death (HRCS 1.46, 95% CI 1.08–1.98), and 24% decrease in chances of recovery (HRCS 0.76, 95% CI 0.62–0.93). Moreover, patients that used anticoagulants or antiplatelet medication had a 38% increase in risk of death (HRCS 1.38, 95% CI 1.01–1.88). Furthermore, patients with fever had a 30% decrease in risk of death (HRCS 0.70, 95% CI 0.52–0.95). In terms of diagnostics, with every unit increase of MEWS, the risk of death increased with 9% (HRCS 1.09, 95% CI 1.01–1.18). Moreover, a ten-fold increase in lactate dehydrogenase values at time of admission was associated with a 568% increase in risk of death (HRCS 6.68, 95% CI 1.95–22.8), and a 75% decrease in chances of recovery (HRCS 0.25, 95% CI 0.13–0.48). Finally, a ten-fold increase in ferritin values at time of admission was associated with a 23% decrease in chances of recovery (HRCS 0.77, 95% CI 0.60–0.99).

Non ICU admission policy

In this study, 318 (31.5%) patients had a non ICU admission policy, of whom 199 (62.6%) were male, with a median age of 79 years (IQR 74–83), and a median MEWS of 2 (IQR 2–4). Of these patients, 151 (47.5%) recovered, 158 (49.7%) died and 9 (2.8%) were censored. Multivariable analyses revealed no statistically significant associations with risk of death, but that patients with longer symptom duration had an increased chance of recovery (HRCS 1.02, 95% CI 1.00–1.04) and patients with symptoms of dyspnea had a decreased chance of recovery (HRCS 0.65, 95% CI 0.44–0.96).

Discussion

Summary of findings

In this Dutch hospital population, approximately a quarter of all COVID-19 patients died after a median hospital admission of six days during the first wave of the epidemic. Using a competing risk approach, we identified age, comorbidities, such as cardiovascular diseases, symptoms, and abnormal laboratory values as the most important risk factors in univariate analyses. After adjusting for all relevant factors at baseline, we found that higher age, immunocompromised state, chronic use of anticoagulants or antiplatelet medication, higher MEWS, and higher values of LDH at time of hospital admission were associated with increased risk of death. On the other hand, fever at time of admission was associated with lower mortality. Male sex was not a significant risk factor for mortality.

Interpretation of results and comparison to literature

Conventional survival analyses do not take competing risks into account, which leads to biased mortality estimates and to overestimation of survival curves. Using the competing risk approach, we took into account that patients who recovered were no longer at the same risk of dying than those who remained hospitalized, resulting in less biased mortality estimates. Even though death or recovery are the two possible final outcomes of the disease, the time to death and time to recovery may not the same (i.e. time to death was generally shorter than time to recovery). In addition, 10% of our patients were censored due to transfer to other hospitals. As a result, the risk factors influencing death may differ from risk factors for recovery [13]. For example, we identified strong risk factors that both increased the risk of dying and reduced the recovery risk, namely higher age, immunosuppression and high LDH values at admission. At the same time, use of anticoagulant or antiplatelet medication and a high MEWS were only associated with an increased risk of dying. Fever at admission reduced this risk, and ferritin showed equivocal results, reducing the risk of death as well as the risk of recovery.

In our study population, approximately 25% of all patients died, which is consistent with other studies [14, 15]. Our study showed that age and higher MEWS at time of admission were risk factors for in-hospital mortality, which is in line with results of many other studies [4, 16–18]. Higher blood LDH levels at time of admission increased the probability of death and decreased the chance of recovery, which is in line with a growing body of evidence [4, 19, 20]. We included both cardiovascular disease and hypertension in our model, which may be subject to collinearity. However, a sensitivity analyses excluding hypertension from the model did not change our estimates and standard errors, showing that our reported estimates are valid. We found immunosuppression as a risk factor for mortality. This has been reported previously, although the exact role of the immune system in COVID-19 is complex [16, 21, 22]. Poor outcome could be determined by a declining immune system less able to clear the virus, but lung tissue damage in severe cases could also be caused by an exaggerated immune response, rather than damage inflicted by the virus itself [23–26]. Consequently, immunocompromised patients may be protected from this type of hyperinflammation [21, 27]. We found that chronic use of anticoagulants or antiplatelet medication was associated with increased risk of death. Although this may be partially explained by the fact that these medications are used by patients with cardiovascular disease, which has been reported as an individual risk factor for in-hospital mortality [4], anticoagulant and/or antiplatelet medication remained an independent risk factor for death in our multivariable analyses. Thromboembolic events are frequently reported in association with severe COVID-19 disease and mortality [28, 29], and current guidelines suggest prophylactic anticoagulants in all hospitalized COVID-19 patients if not contraindicated. However, studies have reported conflicting results regarding the effect of anticoagulants/antiplatelet medication on COVID-19 mortality [30], ranging from a protective effect [31] to a harmful effect [32, 33], or no association [31, 34]. Prospective studies and RCTs are needed to explore the true effects of these medications in hospitalized COVID-19 patients.

Findings of a living systematic review of 23 prognostic studies about COVID-19 mortality indicated that age, immunocompromising comorbidities, composite scores of vital parameters and blood LDH are frequently reported predictors of in-hospital mortality, similar to our findings [4]. Blood ferritin levels and anticoagulants or antiplatelet medication were scarcely reported [4]. On the other hand, male sex and comorbidities, such as cardiovascular and pulmonary diseases, that were significantly associated with in-hospital mortality in our univariable but not multivariable analyses, were frequently reported in other prognostic studies. This is an important finding and shows that many of these risk factors are interacting with each other. Last, we found no increased risk of dying in patients with male sex, high BMI or diabetes, which are also risk factors that have been reported previously. Larger studies may be needed to reveal additional risk factors of clinical importance.

Strengths and limitations

A strength of this study is the large, multicenter population in an area in which infections were clustered. We analyzed mortality in COVID-19 patients using a competing risk approach, leading to more accurate risk estimates for mortality than when using conventional survival analysis. As previously explained, conventional survival analyses may have resulted in biased estimates and Kaplan-Meier curves presenting overestimated incidence of death. There were some limitations to this study. First, approximately 10% of patients in our study were censored, mainly due to frequent transfer of patients. However, this was considered within acceptable limits [35–37]. Second, our data consisted of routinely collected data, resulting in missing values in several variables. We used multiple imputation to minimize associated bias. The advantage of routinely collected data is that the study’s predictors are readily available and be used in clinical practice without extra effort. All reported predictors are variables measured at time of hospital admission, which means that these factors can be used to identify patients in need of more intensive care and monitoring in an early stage of hospitalization. Finally, data were collected in multiple centers with differences in manner of reporting in medical records and differences in clinical management. The advantage of this multicenter approach is that it increased the generalizability of our findings.

Conclusion

Using a robust, competing risk survival analysis, our study confirmed specific risk factors for in-hospital COVID-19 mortality, adding rigor to the current knowledge of risk factors. We confirmed that age, immunocompromised state, use of anticoagulants or antiplatelet medication, MEWS, blood LDH are important risk factors for death in hospitalized COVID-19 patients. The risk factors we identified are mostly in line with the growing body of evidence on COVID-19, although we did not find evidence for increased mortality risk in male patients or those with high BMI or diabetes. All identified risk factors are routinely available at time of admission. They can be used to guide clinical decisions on intense monitoring of patients in-hospital, or to select patients that may benefit most from new treatments. In addition, the more general risk factors like age and comorbidities could be used to prioritize patients for vaccinations in current times where vaccines are still scarce. Lastly, our risk factor analysis provides further input for international prediction research for mortality of COVID-19.

Supporting information

S1 Checklist. STROBE statement—checklist of items that should be included in reports of cohort studies.

(PDF)

Click here for additional data file. (72.7KB, pdf)
S1 Fig. Flowchart of inclusion.

(PDF)

Click here for additional data file. (8.6KB, pdf)
S2 Fig. Schoenfeld residuals plots.

A. Plots of the Schoenfeld residuals for the multivariable CSH model for death. B. Plots of the Schoenfeld residuals for the multivariable CSH model for recovery.

(PDF)

Click here for additional data file. (460.6KB, pdf)
S1 Dataset. Anonymized dataset.

(CSV)

Click here for additional data file. (124.7KB, csv)

Acknowledgments

We thank the RCI-COVID-19 study group for clinical input and critical feedback. Furthermore, we thank all students and nurses that helped to extract data from electronic records.

Data Availability

All relevant data are within the manuscript and its Supporting Information files.

Funding Statement

The author(s) received no specific funding for this work.

References

  • 1.Dutch National Institute for Public Health and the Environment (RIVM). Patient with novel coronavirus COVID-19 in the Netherlands 2020 [updated 27-02-2020. Available from: https://www.rivm.nl/en/news/patient-with-novel-coronavirus-covid-19-in-netherlands.
  • 2.Dutch National Institute for Public Health and the Environment (RIVM). Current information about the novel coronavirus (COVID-19) 2020 [updated 06-04-2020. Available from: https://www.rivm.nl/en/current-information-about-novel-coronavirus-covid-19.
  • 3.World Health Organization (WHO. Coronavirus (COVID-19), global situation dashboard: WHO; 2020. [Available from: https://covid19.who.int/. [Google Scholar]
  • 4.Wynants L, Van Calster B, Collins GS, Riley RD, Heinze G, Schuit E, et al. Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal. BMJ. 2020;369:m1328. 10.1136/bmj.m1328 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Schuster NA, Hoogendijk EO, Kok AAL, Twisk JWR, Heymans MW. Ignoring competing events in the analysis of survival data may lead to biased results: a nonmathematical illustration of competing risk analysis. Journal of clinical epidemiology. 2020;122:42–8. 10.1016/j.jclinepi.2020.03.004 [DOI] [PubMed] [Google Scholar]
  • 6.Oulhaj A, Ahmed LA, Prattes J, Suliman A, Alsuwaidi A, Al-Rifai RH, et al. Preprint: The competing risk between in-hospital mortality and recovery: A pitfall in COVID-19 survival analysis research. medRxiv. 2020. [Google Scholar]
  • 7.Subbe CP, Kruger M, Rutherford P, Gemmel L. Validation of a modified Early Warning Score in medical admissions. QJM: An International Journal of Medicine. 2001;94(10):521–6. [DOI] [PubMed] [Google Scholar]
  • 8.Austin PC, Lee DS, Fine JP. Introduction to the Analysis of Survival Data in the Presence of Competing Risks. Circulation. 2016;133(6):601–9. 10.1161/CIRCULATIONAHA.115.017719 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Noordzij M, Leffondré K, van Stralen KJ, Zoccali C, Dekker FW, Jager KJ. When do we need competing risks methods for survival analysis in nephrology? Nephrology Dialysis Transplantation. 2013;28(11):2670–7. 10.1093/ndt/gft355 [DOI] [PubMed] [Google Scholar]
  • 10.Steyerberg EW. Clinical prediction models: a practical approach to development, validation, and updating.: Springer US; 2019. [Google Scholar]
  • 11.Gray RJ. A Class of K-Sample Tests for Comparing the Cumulative Incidence of a Competing Risk. The Annals of Statistics. 1988;16(3):1141–54. [Google Scholar]
  • 12.Scrucca L, Santucci A, Aversa F. Competing risk analysis using R: an easy guide for clinicians. Bone marrow transplantation. 2007;40(4):381–7. 10.1038/sj.bmt.1705727 [DOI] [PubMed] [Google Scholar]
  • 13.Dignam JJ, Zhang Q, Kocherginsky M. The use and interpretation of competing risks regression models. Clin Cancer Res. 2012;18(8):2301–8. 10.1158/1078-0432.CCR-11-2097 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Wiersinga WJ, Rhodes A, Cheng AC, Peacock SJ, Prescott HC. Pathophysiology, Transmission, Diagnosis, and Treatment of Coronavirus Disease 2019 (COVID-19): A Review. JAMA. 2020;324(8):782–93. 10.1001/jama.2020.12839 [DOI] [PubMed] [Google Scholar]
  • 15.Goel S, Jain T, Hooda A, Malhotra R, Johal G, Masoomi R, et al. Clinical Characteristics and In-Hospital Mortality for COVID-19 Across The Globe. Cardiology and Therapy. 2020;9(2):553–9. 10.1007/s40119-020-00189-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Parohan M, Yaghoubi S, Seraji A, Javanbakht MH, Sarraf P, Djalali M. Risk factors for mortality in patients with Coronavirus disease 2019 (COVID-19) infection: a systematic review and meta-analysis of observational studies. The Aging Male. 2020:1–9. 10.1080/13685538.2020.1774748 [DOI] [PubMed] [Google Scholar]
  • 17.Lu L, Zhong W, Bian Z, Li Z, Zhang K, Liang B, et al. A comparison of mortality-related risk factors of COVID-19, SARS, and MERS: A systematic review and meta-analysis. Journal of Infection. 2020;81(4):e18–e25. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Hu H, Yao N, Qiu Y. Comparing Rapid Scoring Systems in Mortality Prediction of Critically Ill Patients With Novel Coronavirus Disease. Academic Emergency Medicine. 2020;27(6):461–8. 10.1111/acem.13992 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Henry BM, Aggarwal G, Wong J, Benoit S, Vikse J, Plebani M, et al. Lactate dehydrogenase levels predict coronavirus disease 2019 (COVID-19) severity and mortality: A pooled analysis. Am J Emerg Med. 2020;38(9):1722–6. 10.1016/j.ajem.2020.05.073 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Erez A, Shental O, Tchebiner JZ, Laufer-Perl M, Wasserman A, Sella T, et al. Diagnostic and prognostic value of very high serum lactate dehydrogenase in admitted medical patients. The Israel Medical Association journal: IMAJ. 2014;16(7):439–43. [PubMed] [Google Scholar]
  • 21.Fung M, Babik JM. COVID-19 in Immunocompromised Hosts: What We Know So Far. Clin Infect Dis. 2020:ciaa863. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Gao Y, Chen Y, Liu M, Shi S, Tian J. Impacts of immunosuppression and immunodeficiency on COVID-19: A systematic review and meta-analysis. The Journal of infection. 2020:S0163-4453(20)30294–2. 10.1016/j.jinf.2020.05.017 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Mehta P, McAuley DF, Brown M, Sanchez E, Tattersall RS, Manson JJ, et al. COVID-19: consider cytokine storm syndromes and immunosuppression. Lancet (London, England). 2020;395(10229):1033–4. 10.1016/S0140-6736(20)30628-0 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Coperchini F, Chiovato L, Croce L, Magri F, Rotondi M. The cytokine storm in COVID-19: An overview of the involvement of the chemokine/chemokine-receptor system. Cytokine & growth factor reviews. 2020;53:25–32. 10.1016/j.cytogfr.2020.05.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Abdulamir AS, Hafidh RRJEJoGM. The Possible Immunological Pathways for the Variable Immunopathogenesis of COVID—19 Infections among Healthy Adults, Elderly and Children. 2020;17(4). [Google Scholar]
  • 26.Prompetchara E, Ketloy C, Palaga T. Immune responses in COVID-19 and potential vaccines: Lessons learned from SARS and MERS epidemic. Asian Pacific journal of allergy and immunology. 2020;38(1):1–9. 10.12932/AP-200220-0772 [DOI] [PubMed] [Google Scholar]
  • 27.Minotti C, Tirelli F, Barbieri E, Giaquinto C, Donà D. How is immunosuppressive status affecting children and adults in SARS-CoV-2 infection? A systematic review. J Infect. 2020;81(1):e61–e6. 10.1016/j.jinf.2020.04.026 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Bavaro DF, Poliseno M, Scardapane A, Belati A, De Gennaro N, Stabile Ianora AA, et al. Occurrence of Acute Pulmonary Embolism in COVID-19-A case series. Int J Infect Dis. 2020;98:225–6. 10.1016/j.ijid.2020.06.066 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Malas MB, Naazie IN, Elsayed N, Mathlouthi A, Marmor R, Clary B. Thromboembolism risk of COVID-19 is high and associated with a higher risk of mortality: A systematic review and meta-analysis. EClinicalMedicine. 2020;29. 10.1016/j.eclinm.2020.100639 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Salah HM, Naser JA, Calcaterra G, Bassareo PP, Mehta JL. The Effect of Anticoagulation Use on Mortality in COVID-19 Infection. Am J Cardiol. 2020;134:155–7. 10.1016/j.amjcard.2020.08.005 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Wijaya I, Andhika R, Huang I. The Use of Therapeutic-Dose Anticoagulation and Its Effect on Mortality in Patients With COVID-19: A Systematic Review. Clinical and Applied Thrombosis/Hemostasis. 2020;26:1076029620960797. 10.1177/1076029620960797 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Chen F, Sun W, Sun S, Li Z, Wang Z, Yu L. Clinical characteristics and risk factors for mortality among inpatients with COVID-19 in Wuhan, China. Clinical and translational medicine. 2020;10(2):e40. 10.1002/ctm2.40 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Giacomelli A, Ridolfo AL, Milazzo L, Oreni L, Bernacchia D, Siano M, et al. 30-day mortality in patients hospitalized with COVID-19 during the first wave of the Italian epidemic: A prospective cohort study. Pharmacol Res. 2020;158:104931–. 10.1016/j.phrs.2020.104931 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Klok FA, Kruip MJHA, van der Meer NJM, Arbous MS, Gommers D, Kant KM, et al. Confirmation of the high cumulative incidence of thrombotic complications in critically ill ICU patients with COVID-19: An updated analysis. Thromb Res. 2020;191:148–50. 10.1016/j.thromres.2020.04.041 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Qin J, Shen Y. Statistical methods for analyzing right-censored length-biased data under cox model. Biometrics. 2010;66(2):382–92. 10.1111/j.1541-0420.2009.01287.x [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Demler OV, Paynter NP, Cook NR. Reclassification calibration test for censored survival data: performance and comparison to goodness-of-fit criteria. Diagnostic and Prognostic Research. 2018;2(1):16. 10.1186/s41512-018-0034-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Zhu X, Zhou X, Zhang Y, Sun X, Liu H, Zhang Y. Reporting and methodological quality of survival analysis in articles published in Chinese oncology journals. Medicine (Baltimore). 2017;96(50):e9204–e. 10.1097/MD.0000000000009204 [DOI] [PMC free article] [PubMed] [Google Scholar]

Decision Letter 0

Francesco Di Gennaro

17 Feb 2021

PONE-D-20-40238

Risk factors for in-hospital mortality in laboratory-confirmed COVID-19 patients in the Netherlands: a competing risk survival analysis.

PLOS ONE

Dear Dr. Josephine S. van de Maat

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by 28 february. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Francesco Di Gennaro

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for including your ethics statement:  "The study was reviewed by the institutional review board of the Radboud university medical center (number 2020-2923 and 2020-6344). Verbal informed consent was obtained from all patients or their family. ".   

Please amend your current ethics statement to confirm that your named institutional review board or ethics committee specifically approved this study.

Once you have amended this/these statement(s) in the Methods section of the manuscript, please add the same text to the “Ethics Statement” field of the submission form (via “Edit Submission”).

For additional information about PLOS ONE ethical requirements for human subjects research, please refer to http://journals.plos.org/plosone/s/submission-guidelines#loc-human-subjects-research.

3. In the Methods, please provide further clarifications on the following:

- Why written consent could not be obtained

- Whether the Institutional Review Board (IRB) approved use of oral consent

- How oral consent was documented

For more information, please see our guidelines for human subjects research: https://journals.plos.org/plosone/s/submission-guidelines#loc-human-subjects-research.

4. In ethics statement in the manuscript and in the online submission form, please provide additional information about the patient records/samples used in your retrospective study. Specifically, please ensure that you have discussed whether all data/samples were fully anonymized before you accessed them.

5. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

Additional Editor Comments:

dear authors follow reviewer suggestion to improve your paper

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: the data about COVID-19 risk factors and prognostic indications are far from being satisfactory. Any study that help to understand factors that influence prognosis is welcomed. The current study discusses many of these factors. It is now expected to cover all of them andeven each factor can have several sub-risk factors. For example if you discuss diabetes, type of diabetes, duration, degree of control, comorbidies and complications can be risk factors to be discussed. Therefore, each reviewer may have additonal factors to be discussed but what has been analysed in this study is comperhensive enough. However for the application of the data you mentioned in the aim that the data can be used to define priorities for vaccination and intensive treatment. I recommed you subdivide the risk factors into two groups, first the risk factors to define priorities for vaccination like age, BMI, diabetes, autoimmune disease, etc. this will help decision makers define priorities for vaccination which can be critical, especially in countries with limitted resources where mass vccinations is not feasible in the short term. Second is the risk factors for already infected subjects like duration of cough fever diarrhea, PCR, lymphocyte count, etc. Such a data will make the study useful for practical application and may be a nucleus with which other similar studies can establish a score system for risk of serious infection and risk of mortality among infected subjects.

Reviewer #2: In this well performed and interesting study, Authors used a competing risk survival analysis to evaluate predictors of death in a large cohort of Dutch patients.

Overall, the analysis is sound, robust, and well described.

A few minor concerns are listed below:

1) In multivariable model (Table 2) “cardiovascular diseases (incl. hypertension)” and “hypertension” were both included in the model; collinearity has been investigated in this case?

2) “Use of anticoagulants or antiplatelet medication” resulted connected with an increased risk of death. According to Authors hypothesis this association could be explained by the underlying cardiovascular disease.

However: i) in multivariable model cardiovascular disease lost the statistically significance, implying that the use of anticoagulants/antiplatelets was independently associated with mortality;

ii) a considerable risk of thromboembolic events was reported in course of COVID-19 (Bavaro DF, et al. Occurrence of Acute Pulmonary Embolism in COVID-19-A case series. Int J Infect Dis. 2020;98:225-226.).

In my opinion, this work (or similar ones) should be cited, and the risk of death according to use of anticoagulants should be better discussed since current guidelines suggest the use of low-molecular-weight-heparin in all hospitalized COVID-19 patients, if not contraindicated.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Aly Ahmed Abdel Rahim

Reviewer #2: No

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Mar 26;16(3):e0249231. doi: 10.1371/journal.pone.0249231.r002

Author response to Decision Letter 0

3 Mar 2021

Ref: PONE-D-20-40238

Title: Risk factors for in-hospital mortality in laboratory-confirmed COVID-19 patients in the Netherlands: a competing risk survival analysis

Nijmegen, the Netherlands

25 February 2021

Dear Francesco Di Gennaro, academic editor of PLoS ONE,

We are pleased to hear that our manuscript entitled “Risk factors for in-hospital mortality in laboratory-confirmed COVID-19 patients in the Netherlands: a competing risk survival analysis“ is considered for publication in PLoS ONE. The reviewers acknowledged the importance of our study, and mentioned that our analyses were comprehensive, robust and well described. We thank the reviewers for their valuable comments to improve the paper, we have taken the opportunity to clarify some issues raised, and have revised our manuscript according to their suggestions.

Our response to the individual review items can be found below.

We hope our manuscript is now suitable for publication in PLoS ONE.

On behalf of the co-authors,

Yours sincerely,

Gerine Nijman and Josephine van de Maat

Journal requirements:

Comment #1: Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming.

Response #1: The main body and author affiliations now meet PLoS ONE’s style requirements, including those for file naming.

Comment #2: Thank you for including your ethics statement: "The study was reviewed by the institutional review board of the Radboud university medical center (number 2020-2923 and 2020-6344). Verbal informed consent was obtained from all patients or their family. ". Please amend your current ethics statement to confirm that your named institutional review board or ethics committee specifically approved this study. Once you have amended this/these statement(s) in the Methods section of the manuscript, please add the same text to the “Ethics Statement” field of the submission form (via “Edit Submission”).

Response #2: We have adjusted the ethics statement in the methods section to clarify this and have amended the current ethics statement, Methods section, line 91-92:

“The study was approved by the institutional review board (IRB) of the Radboud university medical center (number 2020-2923 and 2020-6344).”

Comment #3: In the Methods, please provide further clarifications on the following:

- Why written consent could not be obtained

- Whether the Institutional Review Board (IRB) approved use of oral consent

- How oral consent was documented

Response #3: The institutional review board (IRB) of the Radboudumc waived the need for written consent and therefore approved the use of oral consent for this study. Oral consent was obtained from all patients or their family and documented in the electronical medical records. We have changed the methods section to clarify this, line 93-95:

“According to the IRB, only oral consent was required. Oral consent was obtained from all patients or their family and documented in the electronic medical records.”.

Comment #4: In ethics statement in the manuscript and in the online submission form, please provide additional information about the patient records/samples used in your retrospective study. Specifically, please ensure that you have discussed whether all data/samples were fully anonymized before you accessed them.

Response #4: All patient data were entered anonymously into the electronic case report forms (eCRF; Castor Electronic Data Capture). The authors that were involved in the data analysis only received coded data without traceable information leading to patients involved in this study.

We have clarified this in the methods section, line 101-102:

“All patient data were entered anonymously into a web-based electronic case report form (using Castor Electronic Data Capture), only using a study identifier. The key linking patient information to study ID was saved in a local, protected file in the participating hospitals and not available to the researchers performing data analyses.”

Comment #5: We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions. In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers.

Response #5: In our re-submission we have uploaded a supporting information file with the minimal required anonymized data to reproduce our study findings (S1 Dataset).

Reviewer #1:

The data about COVID-19 risk factors and prognostic indications are far from being satisfactory. Any study that help to understand factors that influence prognosis is welcomed. The current study discusses many of these factors. It is now expected to cover all of them and even each factor can have several sub-risk factors. For example if you discuss diabetes, type of diabetes, duration, degree of control, comorbidies and complications can be risk factors to be discussed. Therefore, each reviewer may have additonal factors to be discussed but what has been analysed in this study is comprehensive enough. However for the application of the data you mentioned in the aim that the data can be used to define priorities for vaccination and intensive treatment. I recommend you subdivide the risk factors into two groups, first the risk factors to define priorities for vaccination like age, BMI, diabetes, autoimmune disease, etc. this will help decision makers define priorities for vaccination which can be critical, especially in countries with limited resources where mass vaccinations is not feasible in the short term. Second is the risk factors for already infected subjects like duration of cough fever diarrhea, PCR, lymphocyte count, etc. Such a data will make the study useful for practical application and may be a nucleus with which other similar studies can establish a score system for risk of serious infection and risk of mortality among infected subjects.

Response: We thank the reviewer for carefully examining our manuscript. The reviewer recommends to subdivide risk factors into two groups; (1) risk factors to define priorities for vaccination, i.e. in patients without a confirmed COVID-19 infection; and (2) risk factors to define priorities for intensive treatment, i.e. in patients who are already infected.

In our study population, this subdivision is not possible, as we included a population that solely consisted of laboratory-confirmed COVID-19 patients (non-infected patients were not included). Therefore, we could not make a separate survival model to describe risk factors for patients that were not yet infected at time of inclusion. Nevertheless, we still think our results relevant to define prioritized target groups for both intensive treatment and vaccination. COVID-19 vaccination is, among other things, aimed at protection against severe disease and reducing mortality. This study helps to identify groups of patients who are at most at risk of death (once hospitalized), and in whom prevention of the disease is most urgent. Consequently, these groups of patients may be prioritized as target groups for vaccination. We agree with the reviewer that certain risk factors cannot be used for this purpose (e.g. abnormal lab values), so these factors only apply to the already hospitalized population in whom decisions need to be made on intensive monitoring and treatment.

We added this consideration to the discussion section of our revised manuscript, line 379 – 380:

“In addition, the more general risk factors like age and comorbidities could be used to prioritize patients for vaccinations in current times where vaccines are still scarce.”

Reviewer #2:

In this well performed and interesting study, authors used a competing risk survival analysis to evaluate predictors of death in a large cohort of Dutch patients. Overall, the analysis is sound, robust, and well described. A few minor concerns are listed below:

Comment #1: In multivariable model (Table 2) “cardiovascular diseases (incl. hypertension)” and “hypertension” were both included in the model; collinearity has been investigated in this case?

Response #1: We thank the reviewer for reading our manuscript and we appreciate comments and suggestions for improvement. The reviewers expressed concerns with regards to collinearity between “cardiovascular diseases (incl. hypertension)” and “hypertension”.

First of all, we would like to explain why we included both variables in the model. We pre-selected relevant risk factors for the model based on literature and clinical relevance, rather than purely on the basis of statistical significance. Several types of “cardiovascular diseases” have been associated with increased risk of COVID-19 mortality, which is why this category was of interest. However, “cardiovascular diseases” is a very broad category and many studies and doctors have expressed interest in the effect of hypertension specifically. Therefore, we considered both variables separately to be relevant in our model. We have clarified this in our methods section, line 130-132:

“These risk factors were pre-selected based on literature and expert opinion to be clinically relevant and routinely available at time of presentation, rather than based on statistical significance”.

Nonetheless, the reviewer’s concern about collinearity remains valid. To show the effect of potential collinearity between these two variables on our findings, we have performed a sensitivity analysis, leaving out the variable ‘hypertension’. The table below reports the cause-specific hazard ratios of a multivariable model where “hypertension” is left out of the analysis. When comparing these results to the results from the model including hypertension, we see that the that the estimates of all variables are stable. In addition, the standard errors remain stable and the significance has not changed. Therefore, we can conclude that including “hypertension” as a separate variable has not influenced the reliability of our estimates.

We have included these arguments in the discussion section, line 314-317:

“We included both cardiovascular disease and hypertension in our model, which may be subject to collinearity. However, a sensitivity analyses excluding hypertension from the model did not change our estimates and standard errors, showing that our reported estimates are valid.”

Multivariable including hypertension Multivariable excluding hypertension

Death Recovery Death Recovery

Age (years) 1.10 (1.08-1.12) * 0.99 (0.98-0.99) * 1.10 (1.08-1.12) * 0.99 (0.98-0.99) *

Sex, male 1.07 (0.79-1.47) 0.90 (0.75-1.08) 1.11 (0.82-1.51) 0.89 (0.74-1.07)

BMI 1.01 (0.98-1.04) 1.01 (0.99-1.03) 1.01 (0.97-1.04) 1.01 (0.99-1.03)

Diabetes Mellitus 1.17 (0.86-1.59) 0.85 (0.69-1.04) 1.14 (0.84-1.56) 0.86 (0.70-1.05)

Cardiovascular disease (incl. hypertension) 1.05 (0.69-1.59) 0.71 (0.54-0.93) * 0.90 (0.63-1.29) 0.78 (0.62-0.97) *

Hypertension 0.78 (0.56-1.10) 1.15 (0.90-1.48)

Pulmonary disease 1.33 (0.98-1.80) 0.88 (0.73-1.07) 1.30 (0.96-1.75) 0.89 (0.74-1.08)

Immunocompromised a 1.46 (1.08-1.98) * 0.76 (0.62-0.93) * 1.50 (1.11-2.02) * 0.75 (0.62-0.92) *

Use of anticoagulant or antiplatelet medication 1.38 (1.01-1.88) * 1.15 (0.93-1.43) 1.41 (1.03-1.92) * 1.13 (0.91-1.39)

Use of ACE inhibitors and/or angiotensin II receptor blockers 0.99 (0.74-1.32) 1.09 (0.88-1.33) 0.95 (0.72-1.27) 1.11 (0.90-1.36)

Chest X-Ray

Performed, not suggestive for COVID-19

Performed, suggestive for COVID-19

Not performed

Ref

1.07 (0.63-1.80)

0.90 (0.54-1.48)

Ref

1.52 (1.05-2.19) *

1.18 (0.82-1.70)

Ref

1.07 (0.63-1.81)

0.90 (0.54-1.48)

Ref

1.51 (1.05-2.18) *

1.18 (0.82-1.69)

CT scan severity score 1.01 (0.98-1.05) 0.97 (0.95-0.99) * 1.01 (0.98-1.05) 0.97 (0.95-0.99) *

Symptom duration (days) 0.98 (0.96-1.01) 1.01 (1.00-1.02) * 0.98 (0.96-1.01) 1.01 (1.01-1.02) *

Symptoms, fever 0.70 (0.52-0.95) * 1.05 (0.85-1.30) 0.69 (0.51-0.93) * 1.05 (0.85-1.30)

Symptoms, dyspnea 0.77 (0.58-1.03) 1.01 (0.84-1.21) 0.78 (0.58-1.04) 1.01 (0.84-1.21)

Modified Early Warning Score (MEWS) b 1.09 (1.01-1.18) * 0.97 (0.93-1.02) 1.08 (1.00-1.17) * 0.97 (0.93-1.02)

Neutrophil-to-lymphocyte rate c 0.97 (0.59-1.60) 1.01 (0.74-1.38) 1.00 (0.61-1.65) 1.00 (0.74-1.36)

Lactate dehydrogenase (U/L) c 6.68 (1.95-22.8) * 0.25 (0.13-0.48) * 6.67 (1.95-22.86) * 0.25 (0.13-0.49) *

Creatinine (µmol/L) c 1.84 (0.87-3.90) 1.17 (0.69-1.99) 1.78 (0.85-3.75) 1.21 (0.71-2.05)

Procalcitonin (µg/L) c 1.04 (0.76-1.41) 0.88 (0.75-1.03) 1.03 (0.76-1.39) 0.88 (0.75-1.03)

C-reactive protein (mg/L) c 1.25 (0.79-2.00) 0.88 (0.69-1.11) 1.22 (0.77-1.94) 0.89 (0.70-1.12)

Ferritin (µg/L) c 0.66 (0.43-1.02) 0.77 (0.60-0.99) * 0.66 (0.43-1.02) 0.78 (0.61-0.99) *

D-dimer (ng/L) d 0.99 (0.84-1.16) 0.94 (0.87-1.00) 0.99 (0.84-1.15) 0.94 (0.88-1.00)

Comment #2 “Use of anticoagulants or antiplatelet medication” resulted connected with an increased risk of death. According to Authors hypothesis this association could be explained by the underlying cardiovascular disease.

However: i) in multivariable model cardiovascular disease lost the statistically significance, implying that the use of anticoagulants/antiplatelets was independently associated with mortality;

ii) a considerable risk of thromboembolic events was reported in course of COVID-19 (Bavaro DF, et al. Occurrence of Acute Pulmonary Embolism in COVID-19-A case series. Int J Infect Dis. 2020;98:225-226.).

In my opinion, this work (or similar ones) should be cited, and the risk of death according to use of anticoagulants should be better discussed since current guidelines suggest the use of low-molecular-weight-heparin in all hospitalized COVID-19 patients, if not contraindicated.

Response 2i and 2ii: we appreciate this valuable comment and we agree that the effect of “anticoagulants or antiplatelet medication” cannot be explained solely by the effect of cardiovascular diseases, but that it was also an independent risk factor in our results. We recognize that there is conflicting evidence with regards to the effect of these type of medications on COVID-19 mortality. We have elaborated on this issue in more detail in the discussion section of the revised manuscript, also including the reference the reviewer has suggested.

Discussion, line 323-333:

“Although this may be partially explained by the fact that these medications are used by patients with cardiovascular disease, which has been reported as an individual risk factor for in-hospital mortality [4], anticoagulant and/or antiplatelet medication remained an independent risk factor for death in our multivariable analyses. Thromboembolic events are frequently reported in association with severe COVID-19 disease and mortality [28, 29], and current guidelines suggest prophylactic anticoagulants in all hospitalized COVID-19 patients if not contraindicated. However, studies have reported conflicting results regarding the effect of anticoagulants/antiplatelet medication on COVID-19 mortality [30], ranging from a protective effect [31] to a harmful effect [32, 33], or no association [31, 34]. Prospective studies and RCTs are needed to explore the true effects of these medications in hospitalized COVID-19 patients.”

Attachment

Submitted filename: 210303 Rebuttal letter_Survival paper.docx

Click here for additional data file. (30.9KB, docx)

Decision Letter 1

Francesco Di Gennaro

15 Mar 2021

Risk factors for in-hospital mortality in laboratory-confirmed COVID-19 patients in the Netherlands: a competing risk survival analysis.

PONE-D-20-40238R1

Dear Dr. Josephine S. van de Maat,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Francesco Di Gennaro

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

dear authors congratulations

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: thank you. conviced and satisfied with your response to the comments. the data are helpful and may be usful for stratification of the patients

Reviewer #2: (No Response)

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Aly Ahmed Abdel Rahim

Reviewer #2: No

Acceptance letter

Francesco Di Gennaro

18 Mar 2021

PONE-D-20-40238R1

Risk factors for in-hospital mortality in laboratory-confirmed COVID-19 patients in the Netherlands: a competing risk survival analysis.

Dear Dr. van de Maat:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Francesco Di Gennaro

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Checklist. STROBE statement—checklist of items that should be included in reports of cohort studies.

    (PDF)

    Click here for additional data file. (72.7KB, pdf)
    S1 Fig. Flowchart of inclusion.

    (PDF)

    Click here for additional data file. (8.6KB, pdf)
    S2 Fig. Schoenfeld residuals plots.

    A. Plots of the Schoenfeld residuals for the multivariable CSH model for death. B. Plots of the Schoenfeld residuals for the multivariable CSH model for recovery.

    (PDF)

    Click here for additional data file. (460.6KB, pdf)
    S1 Dataset. Anonymized dataset.

    (CSV)

    Click here for additional data file. (124.7KB, csv)
    Attachment

    Submitted filename: 210303 Rebuttal letter_Survival paper.docx

    Click here for additional data file. (30.9KB, docx)

    Data Availability Statement

    All relevant data are within the manuscript and its Supporting Information files.

    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES