Determinants of in-hospital mortality in COVID-19; a prospective cohort study from Pakistan

Samreen Sarfaraz; Quratulain Shaikh; Syed Ghazanfar Saleem; Anum Rahim; Fivzia Farooq Herekar; Samina Junejo; Aneela Hussain

doi:10.1371/journal.pone.0251754

. 2021 May 27;16(5):e0251754. doi: 10.1371/journal.pone.0251754

Determinants of in-hospital mortality in COVID-19; a prospective cohort study from Pakistan

Samreen Sarfaraz ^1,^*, Quratulain Shaikh ^2,^*, Syed Ghazanfar Saleem ³, Anum Rahim ⁴, Fivzia Farooq Herekar ⁵, Samina Junejo ⁶, Aneela Hussain ⁷

Editor: Wisit Cheungpasitporn⁸

PMCID: PMC8158897 PMID: 34043674

Abstract

A prospective cohort study was conducted at the Indus Hospital Karachi, Pakistan between March and June 2020 to estimate the in-hospital mortality among hospitalized COVID-19 patients and its determinants. A total of 170 adult patients were enrolled and all-cause mortality was found to be 39% (67/170). Most non-survivors were above 60 years of age (64%) while gender distribution was quite similar in both groups (males: 77% vs 78%). Most (80.6%) non-survivors came with peripheral oxygen saturation less than 93% while 95% of them had critical disease on arrival. Use of non-invasive ventilation in emergency room was higher among non-survivors (56.7%) versus survivors (26.2%). Median Interleukin-6 levels were higher among non-survivors (78.6: IQR = 33.8–49.0) compared to survivors (21.8: IQR = 12.6–36.3). Most patients in the non-survivor group (86.6%) required invasive ventilator support during hospital stay compared to 7.8% in the survivors. The median duration of ICU stay was longer for non-survivors (9: IQR = 6–12) compared to survivors (5: IQR = 3–7) days. Univariable binary logistic regression showed that age above 60 years, oxygen saturation below 93%, Neutrophil to lymphocyte ratio above 5, procalcitonin above 2ng/ml, unit increase in SOFA score and arterial lactate levels were associated with mortality. We also found that a unit decrease in Pao2/FiO2 ratio and serum albumin were associated with mortality in our patients. Multivariable regression showed that age above 60 years (aOR = 3.4: 95% CI = 1.6–6.9), peripheral oxygen saturation below 93% (aOR = 3.5:95% CI = 1.6–7.7) and serum pro-calcitonin above 2ng/ml (aOR = 4.8; 95% CI = 1.9–12.2) were associated with higher odds of mortality when adjusted by month of admission. Most common cause of death was multisystem organ failure in 35 (56.6%) non-survivors while 22 (35.5%) died due to respiratory failure. Larger prospective studies are needed to further strengthen these findings.

Introduction

In December 2019 a highly transmissible respiratory illness caused by the severe acute respiratory syndrome coronavirus 2 (SARS-Cov-2) originated in Wuhan, China and caused a pandemic by its rapid spread [1]. It has resulted in 128,229,141 infections and 2,803,975 deaths globally as of 30^th March 2021. Pakistan ranks 31^st among the list of high burden countries with 659,116 confirmed infections and 14,256 deaths which is much lower compared to its immediate neighbours [2]. The Pakistan mortality rate of 2% [3] is comparable to that of India (1.45%) but lower than that of Iran (4.68%) and several European countries including UK (3.43%) and Italy (3.52%) [2]. The reasons for this difference in fatality is largely unknown but a multifactorial combination of viral immunogenicity, genetic makeup of the host, demography and seasonal variation may play a role in this [4]. Sind province recorded the country’s first case on 26^th February 2020 and since then has received 44% of the country’s COVID-19 confirmed cases with the largest city Karachi being worst hit [3]. The peak of infection in the first wave was reached in mid-June when on average 7000 new infections were recorded in a day and the maximum number of deaths recorded were 153 on 20^th June 2020. Major hospitals in all big cities were overwhelmed straining the health infra-structure. The Indus Hospital, Karachi emerged as a front liner with a dedicated isolation unit providing free of cost treatment to the sick COVID-19 patients requiring hospitalization. The hospital worked with the government of Sind as a diagnostic and referral center. The experience was challenging for our physicians and allied health staff. In this paper we aim to estimate the in-hospital mortality of COVID-19 in our hospital and study its determinants. To this date there is insubstantial published data on in-hospital mortality from Pakistan. As Pakistan has already entered the second wave now, these data will help in risk stratification and management of COVID-19 patients.

Methods

This prospective cohort study was conducted at the Indus Hospital’s COVID isolation unit. The Indus Hospital (TIH) Karachi is a 300 bedded tertiary care hospital set up with public private partnership which provides free of cost services to the people. The COVID unit was initially a 20 bedded dedicated COVID facility established in March 2020, later expanded to 56 beds. All COVID-19 (Nasopharyngeal PCR positive) patients aged 18 years and above admitted between 19th March and 7th June 2020 were included. Patients were enrolled into the study as soon as they were admitted in the COVID unit and were followed till death or discharge from hospital.

Demographic information, clinical presentation, laboratory abnormalities including inflammatory markers and imaging results were recorded. The primary outcome was all-cause in-hospital mortality while we also compared length of stay, occurrence of in-hospital complications, use of inotropic support and of mechanical ventilation among survivors and non survivors.

Patients were categorized as per WHO definitions into asymptomatic (COVID-19 PCR positive but with no clinical manifestation attributable to COVID-19), mild (symptomatic without evidence of pneumonia), moderate (with clinical signs of pneumonia and oxygen saturation ≥ 90% on room air), severe (signs of pneumonia with respiratory rate ≥ 30 breaths/min or SpO2 < 90% on room air) and critical (development of Acute Respiratory Distress Syndrome, septic shock or multi organ dysfunction) [5]. The study was approved by the institutional IRB under IRD_IRB_2020_04_002.

Data were recorded on a REDCap electronic data capture tool hosted at (The Indus Hospital) [6] and analyzed on Stata 14 [7]. Continuous variables were summarized by mean (SD) or median (IQR) as appropriate. Distribution of categorical variables were expressed as percentages of the various categories (%). Univariable binary logistic regression was used to determine association of predictors with mortality. Purposeful selection method [8] was applied for building the multivariable logistic regression model and likelihood ratio tests were used to select the final model. Interaction was not tested in the final model. Multicollinearity was assessed between predictors in the final model. Goodness of fit was tested using classification table and area under curve. Crude and adjusted Odds ratios and 95% confidence intervals were reported.

Results

The emergency department of Indus hospital, Karachi received 11,855 suspected COVID-19 patients out of which 3,851 tested positive for COVID-19 PCR (positivity rate 32.5%) from 19th March to 7th June 2020 (Fig 1). The median age of those who tested positive for COVID-19 PCR was 35 (IQR = 26–50) years. The teleconsultation service set up for COVID-19 at TIH, managed most (3,422; 88.8%) of these asymptomatic to mild spectrum patients at home through a robust algorithm based system of symptomatic treatment, follow up and counselling. The rest of the (n = 384/3851: 9.9%) patients needed admission but only 193 (50%) got admitted due to inavailability of beds on arrival, others were referred to other isolation units in the city. Seven patients who were admitted at TIH were less than 18 years old hence removed from this analysis. Some (n = 16) patients were either asymptomatic or had mild disease and needed inpatient care due to other indications like hemodialysis or emergency surgical intervention (obstructed hernia, acute appendicitis). Eventually, 170 participants were included in the cohort. The all-cause in-hospital mortality was 39% (67/170).

Clinical presentation

Those who died had a mean age of 61 (± 12.57) years compared to those who survived 53 (± 13) years (not shown in the table). A similar gender distribution was observed in both groups Table 1 (males = 77.6% in non-survivors compared to 78.6%). Median oxygen saturation in peripheral blood measured through a pulse oximeter was 86% (IQR = 71.7%-90.2%) among non-survivors versus 91.5% (IQR = 85.0% -95.2%) in survivors (not shown in the table). Most patients among non-survivors had critical disease 95.5% versus 70.9% among survivors. Use of non-invasive ventilation (NIV) in emergency room was higher among non-survivors (56.7%) versus survivors (26.2%). The Sequential Organ Failure Assessment (SOFA) [9] score was higher among non-survivors (median = 6: IQR = 5–8.2) compared to survivors (median = 4: IQR = 4–6).

Table 1. Clinical characteristics of study participants by mortality.

	Non-Survivors	Survivors
	n(%)	n(%)
Age (years)
≤60	24 (35.8)	64 (62.1)
>60	43 (64.2)	39 (37.9)
Gender
Male	52 (77.6)	81 (78.6)
Female	15 (22.4)	22 (21.4)
Symptoms
Fever	54 (80.6)	86 (63.5)
Cough	40 (59.7)	62 (60.2)
SOB	57 (85.1)	83 (80.6)
Runny nose	1 (1.5)	3 (2.9)
Sore Throat	3 (4.5)	4 (3.9)
Chest Pain	5 (7.5)	2 (1.9)
Fatigue	8 (11.9)	5 (4.9)
Diarrhea	4 (6.0)	8 (7.8)
Vomiting	2 (3.0)	10 (9.7)
Others	13 (19.4)	26 (25.2)
Duration of symptoms^* (days)	6 (4–7)	7 (4–10)
Comorbid conditions
None	13 (19.4)	38 (36.9)
Hypertension	35 (52.2)	39 (37.9)
Diabetes	35 (52.2)	41 (39.8)
Liver Disease	1 (1.5)	0 (0)
Lung Disease	3 (4.5)	3 (2.9)
Renal Disease	6 (9.0)	7 (6.8)
Heart Disease	7 (10.4)	10 (9.7)
Other	12 (17.9)	26 (25.2)
Number of comorbid conditions
< = 2	43 (79.6)	45 (69.2)
3 or more	11 (20.4)	20 (30.8)
Clinical Signs at presentation
Systolic Blood pressure^* (mm/Hg)	144 (123–159)	135 (122–149)
Diastolic Blood pressure^* (mm/Hg)	80 (68.5–95)	79 (72.5–89)
Pulse^**/minute	104.0 ± 22.0	101.1 ±20.3
Respiratory Rate^*/m	32 (26–38)	28 (23–32)
Temperature^* °F	98.6 (98–99.0)	98.4 (98–98.6)
Oxygen Saturation (%)
≥93%	13 (19.4)	46 (44.7)
<93%	54 (80.6)	57 (51.4)
Glasgow Coma Scale	15 (15–15)	15 (15–15)
Disease Severity
Moderate	0 (0)	18 (17.5)
Severe	3 (4.5)	12 (11.7)
Critical	64 (95.5)	73 (70.9)
Non-invasive ventilation in emergency room
Yes	38 (56.7)	27 (26.2)
No	29 (43.3)	76 (73.8)
Clinical Severity Scores
SOFA score^*	6 (5–8.2)	4 (4–6)
CURB-65 score^*	2 (1–3)	1 (0–2)
MuLBSTA^*	7.5 (6–9.5)	5 (2–8)
Month of admission
March	4 (5.9)	2 (1.9)
April	22 (32.8)	28 (27.4)
May	32 (47.8)	46 (45.1)
June	9 (13.4)	26 (25.5)

Open in a new tab

*Median (Q1-Q3)

**Mean±SD.

Laboratory parameters are described in detail in Table 2. Neutrophil to lymphocyte ratio (NLR) was at least five in 86.6% of non-survivors compared to 70.9% among survivors. Median PaO2/Fio2 ratio was lower among non-survivors (156.8: IQR = 77–233.3) than survivors (215.5: IQR = 141.8–304.7). Inflammatory markers including C-Reactive protein (CRP), d-dimer, lactate dehydrogenase (LDH), Ferritin, Interleukin-6 (IL-6) were all higher among non-survivors compared to survivors. Serum pro-calcitonin levels were higher among non-survivors compared to survivors.

Table 2. Baseline laboratory parameters by survival status.

	Non-Survivors Median (IQR)	Survivors Median (IQR)
Hemoglobin(gm/dl)	12.7 (11.1–14.1)	13.10 (11.6–14.1)
WBC Count (x10E9/L)	12.0 (8.3–15.1)	9.8 (7.2–13.9)
Platelet Count (x10E9/L)	198.0 (147.0–296.0)	239.0 (190.0–331.0)
Absolute neutrophil count (ANC) (x10E9/L)	10.3 (7.6–13.9)	7.9 (5.5–11.5)
Absolute Lymphocyte count (ALC) (x10E9/L)	1.32 (0.53–1.32)	1.07 (0.74–1.69)
Neutrophil to Lymphocyte ratio (NLR)^*
<5	9 (13.4)	30 (29.1)
≥5	58 (86.6)	73 (70.9)
Arterial pH	7.4 (7.3–7.4)	7.4 (7.4–7.4)
PCO2 (mmHg)	31.3 (27.5–35.8)	31.9 (29.0–34.2)
PO2 (mmHg)	50.2 (41.3–64.7)	64.3 (53.8–77.8)
PaO2/FiO2 Ratio	156.8 (77–233.3)	215.5 (141.8–304.7)
Urea (mg/dl)	44.0 (32.0–67.5)	34.0 (24.0–51.0)
Creatinine (mg/dl)	1.2 (0.9–1.8)	1.0 (0.8–1.3)
Sodium (mg/dl)	136.0 (133.0–139.0)	137 (133.0–139)
Potassium(mg/dl)	4.0 (3.6–4.6)	4.2 (3.9–4.6)
Bicarbonate (mg/dl)	19.5 (16.0–21.7)	20 (18–22)
Total Bilirubin (mg/dl)	0.6 (0.4–0.9)	0.6 (0.4–0.8)
SGPT (U/L)	43.5 (22.0–90.2)	31.0 (15.5–52.0)
Arterial Lactate (mmol/L)	2.3 (1.8–3.3)	1.3 (1.1–2.0)
Serum Albumin (g/L)	3.2 (2.6–3.4)	3.5 (3.1–3.7)
Prothrombin Time (sec)	11.2 (10.4–12.8)	11.2 (10.7–11.7)
APTT (sec)	29.8 (26.8–32.4)	29.9 (23.9–33.7)
C-Reactive Protein mg/L	192.2 (86.9–337.0)	111.3 (60.7–220.3)
D-dimer ng/ml	1860.5 (728.7–5767.5)	1277.0 (622.0–5688.0)
Lactate Dehydrogenase (LDH) U/L	549.5 (469–855)	476 (355–694)
Ferritin ng/ml	1315 (532.7–1675.5)	1227.0 (292.7–1675.5)
Interleukin-6 (IL-6) pg/ml	78.6 (33.8–490.0)	21.8 (12.6–36.3)
Procalcitonin ng/ml	0.6 (0.2–2.2)	0.2 (0.1–0.6)
Troponin ng/ml	16 (6.7–53.5)	7 (4–18)
Blood culture^*
Positive	5 (8.3)	7 (8.3)
Negative	55 (91.7)	77 (91.7)
Chest x-ray^*
Unilateral Radiologic Findings	3 (4.5)	4 (3.9)
Bilateral Radiologic Findings	59 (88.1)	85 (82.5)
Multilobar Infiltrates	36 (53.7)	28 (27.2)
Consolidation	32 (47.8)	32 (31.1)
Pleural Effusion	3 (4.5)	4 (3.9)
Others	7 (10.4)	7 (6.8)
Normal	0 (0)	10 (9.7)

Open in a new tab

*Frequency (%).

Hospital course and outcome

Details of experimental therapies given to patients are depicted in the Table 3. Most patients in the non-survivor group required invasive ventilator support during hospital stay (86.6% vs 7.8%) while median duration of ICU stay was longer (9: IQR = 6–12) compared to survivors (5: IQR = 3–7) days. Most common cause of death was multisystem organ failure in 35 (56.6%) non-survivors while 22 (35.5%) died due to respiratory failure.

Table 3. Hospital course by mortality.

	Non-Survivors n (%)	Survivors n (%)
Treatment
Methyl Prednisolone
Yes	55 (82.1)	82 (82.0)
No	12 (17.9)	18 (18.0)
Antibiotics
Yes	63 (94.0)	82 (81.2)
No	4 (6.0)	19 (18.8)
Anticoagulation
Therapeutic doses	43 (64.2)	35 (35.0)
Prophylactic doses	18 (26.9)	57 (76.0)
None	6 (9.0)	8 (8.0)
Hydroxychloroquine (HCQ)
No	27 (40.3)	56 (54.4)
Yes	40 (59.7)	47 (45.6)
Azithromycin (AZT)
No	30 (44.8)	67 (65.0)
Yes	37 (55.2)	36 (35.0)
Tocilizumab (TCZ)
Single Dose	13 (19.4)	23 (22.3)
Two Doses	8 (11.9)	1 (1.0)
Not Given	46 (68.7)	79 (76.7)
IVIG
Single Dose	14 (20.9)	7 (6.8)
Multiple Doses	5 (7.5)	0 (0)
Not Given	48 (71.6)	96 (93.2)
Admission to ICU
Yes	61 (93.8)	34 (35.8)
No	4 (6.2)	61 (64.2)
Invasive Ventilation
Yes	58 (86.6)	8 (7.8)
No	9 (13.4)	95 (92.2)
In-Hospital complications
None	10 (14.9)	71 (68.9)
Cardiac	17 (25.4)	10 (9.7)
Nosocomial Infection	20 (29.9)	11 (10.7)
CNS	3 (4.5)	2 (1.9)
Septic Shock	30 (44.8)	2 (1.9)
MODS	23 (34.3)	0 (0)
AKI	39 (58.2)	7 (6.8)
Thromboembolism	6 (9.0)	1 (1.0)
Barotrauma	3 (4.5)	1 (1.0)
DIC	9 (13.4)	0 (0)
Sever Hyperglycemia	4 (6.0)	3 (2.9)
Electrolyte Imbalance	11 (16.4)	4 (3.9)
Other	7 (10.4)	7 (6.8)
Length of Hospital stay^* (days)	9 (6–12)	13 (11–15)
-Length of ICU Stay^* (days)	9 (6–12)	5 (3–7)
Days of Intubation^*	6 (3–9)	4.5 (2–7)

Open in a new tab

*Median (IQR).

Predictors of in-hospital all-cause mortality

Univariable binary logistic regression (Table 4) showed that age greater than 60 years, peripheral oxygen saturation less than 93%, use of NIV in emergency room, serum procalcitonin level of higher than 2 ng/ml, NLR of at least 5, high clinical severity score (SOFA, CURB 65, MuLBSTA) and high arterial lactate levels were associated with higher odds of mortality. Lower Pao2/Fio2 ratio and low serum albumin levels were associated with higher odds of mortality in our data.

Table 4. Univariable binary logistic regression for predictors of in-hospital all-cause mortality.

	Crude OR	95% CI	p-value
Age
≤60 years	1
>60 years	2.9	1.5–5.6	0.001
Gender
Female	1
Male	0.9	0.4–2.0	0.874
Oxygen saturation
≥93%	1
<93%	3.3	1.6–6.9	0.005
SOFA	1.4	1.2–1.7	0.001
CURB-65	1.8	1.3–2.5	0.001
MuLBSTA	1.2	1.1–1.4	0.000
(NLR)
<5	1
≥5	2.6	1.2–6.0	0.020
Pao2/Fio2 ratio	0.5	0.4–0.8	0.000
Arterial lactate (mmol/L)	1.5	1.1–2.0	0.012
Albumin g/L	0.3	.13- .66	0.003
Procalcitonin
≤ 2 ng/ml	1
>2 ng/ml	3.6	1.6–8.1	0.002
NIV in Emergency room
No	1
Yes	3.6	1.9–7.1	0.000

Open in a new tab

NLR-neutrophil to lymphocyte ratio, NIV-Non-invasive ventilation.

Multivariable binary logistic regression (Table 5) showed that those who were older than 60 years had 3.5 times (95% CI = 1.6–7.8) the odds of mortality compared to those who were less than or equal to 60 years when adjusted for other variables in the model including month of admission. Patients who had peripheral oxygen saturation measured through pulse oximeter less than 93% had 3.5 (95% CI = 1.6–7.8) times higher odds of mortality compared to those with oxygen saturation of at least 93% in the presence of other variables in the model. Serum procalcitonin levels of more than 2 ng/ml were associated with 4.8 (95% CI = 1.9–12.4) times the odds of mortality after adjustment for other variables in the model. Final model was adjusted for month of admission. Receiver operating curve (ROC) analysis revealed that the model has a sensitivity of 60.9% and specificity of 80.2% in predicting mortality among hospitalized COVID-19 patients and the correct classification rate was 72%. The Area under the curve (AUC) of the model was 0.76.

Table 5. Multivariable Logistic regression model of all-cause mortality in COVID-19 patients.

	Adjusted Odds ratios^*	95% CI	p-value
Age
≤ 60 years	1
˃ 60 years	3.4	1.6–6.9	0.001
Oxygen saturation
≥ 93%	1
< 93%	3.5	1.6–7.7	0.002
Procalcitonin
≤2 ng/ml
˃2 ng/ml	4.8	1.9–12.2	0.001

Open in a new tab

*Adjusted for month of admission.

Discussion

In this prospective cohort study, we report the clinical attributes and risk factors associated with all-cause mortality among hospitalized COVID-19 patients. We found the all-cause mortality to be 39% which is disproportionately high in those who were ventilated (58/66: 88%). It is important to highlight here that our most of our patients suffered from critical disease (approximately 80.5% of total admitted) with 38% needing non-invasive ventilation (NIV) in emergency room to manage respiratory failure. ICU admission was needed for 59% (95/160). Approximately, 38.8% (66/170) required mechanical ventilation (MV) hence our patient population appears to be sicker compared to the only other unpublished data from the city [10]. We believe, being a private referral center they may have admitted milder spectrum patients for the purpose of isolation and monitoring. On the contrary, as mentioned before, our center is a philanthropic, free of cost referral centre for the underprivileged with limited bed capacity. Hence, admission was strictly reserved for sick patients requiring in-hospital management. The poor survival in ventilated cases, apart from the natural disease process, may be due to little knowledge of the pathogenic mechanism of respiratory injury and its management in the initial days. Most patients were managed with early invasive ventilator support to avoid fatigue and potential risk of aerosolization of COVID-19 with NIV [11]. Gradual understanding of the disease process has shifted the management strategy from early mechanical ventilation towards use of NIV till tolerated as suggested by the National and International COVID guidelines [3, 5]. Global mortality from COVID-19 varies widely (20% -97%) [12, 13] depending on ICU facilities, ventilator performance, experience of ICU team, patient and disease characteristics, geographic area, seasonality and duration of follow up [4]. High ventilator mortality has been reported even from the best centers in Wuhan (97%), New York (88%), UK (67%) and Italy (53.4%) [14–17] questioning the role of invasive ventilation in COVID-19 management especially in lower middle income countries [18]. Ventilator induced lung injury due to barotrauma, volutrauma, atelectrauma, oxytrauma and infections further jeopardize the outcome of COVID-19 patients [19]. Hospitalized patients with COVID-19 have 5 times higher reported mortality than those with influenza pneumonia [20].

Non-survivors in our study showed worse clinical profile with low peripheral oxygen saturation, respiratory rate and raised inflammatory parameters at presentation. This indicates that they were already in the late phases of Cytokine Release Syndrome (CRS) at admission [21]. The median time to hospitalization from onset of symptoms was similar for both survivors and non-survivors (6 days vs 7 days) consistent with reported literature [22]. However, why some patients were more prone to develop CRS by day 7 is not clear. It is important to note that the overall COVID-19 cohort presenting to our hospital (3851 patients) was nearly 20 years younger (median age 35 vs 58 years) than the subgroup who got admitted [23]. Older age was also found to be associated with higher mortality in our data as reported globally [24, 25]. Age above 60 years has consistently shown to be relevant to mortality since the beginning of the pandemic and our results were consistent in identifying this as a predictor of mortality among others.

Neutrophilia and thrombocytopenia were more pronounced among non-survivors with NLR of at least 5 was seen more commonly among non-survivors than survivors (Table 2). Wu C et al found a significant association between neutrophilia, lymphopenia (peripheral CD3, CD4, and CD8 T-cell counts decreased) and development of Acute Respiratory Distress Syndrome (ARDS) [26]. As observed in literature, CRP, D-dimer, LDH, Ferritin and IL-6 were all higher among non-survivors in our cohort [26]. Non-survivors in our cohort developed acute kidney injury, sepsis and multi organ dysfunction syndrome (MODS) more frequently than survivors. Unit increase in SOFA score and CURB 65 was associated with higher odds of mortality in our data. Since SOFA tends to reflect the effect on multiple organ systems, it has proved to be a better predictor of mortality in COVID-19 in previous literature [27]. These were not retained in the final model.

Use of experimental therapies including steroids, antibiotics, anticoagulation, Hydroxychloroquine (HCQ) during the first wave was similar in both survivors and non-survivors. Among them, Hydroxychloroquine despite inhibiting viral replication in vitro [28] did not prove beneficial in RCTs [29], rather proved to be toxic.This led to the U.S. FDA revoking its emergency use authorization in June 2020 [30]. Use of therapeutic anticoagulation, two doses of Tocilizumab and Azithromycin was more frequent among non-survivors in our data. Azithromycin is the only effective oral drug for treatment of XDR salmonella [31] and COVID-19 pandemic has led to its mass injudicious use, both over the counter and prescription driven, which may increase antimicrobial resistance in the long run. Studies have failed to demonstrate any benefit of AZT alone [32] or in combination with HCQ [33] with a risk of compounding cardiac toxicity by QTc prolongation [34]. Antibiotics were started in 87.6% of our patients suspecting respiratory bacterial co-infection although only 7% initial blood cultures were positive (most patients were not producing sputum) and markers like WBC count and procalcitonin were not elevated in the majority. Literature reports low rates of secondary infection with COVID-19 (only 8% in a review of 9 studies) with paradoxical high consumption (72%) of broad spectrum antibiotics [35]. We believe COVID-19 specific antibiotic stewardship guidance is essential to stop the rampant over- use of antibiotics especially in LMIC countries like Pakistan where antimicrobial resistance (AMR) is already high.

Steroids have shown benefit in severe and critical COVID-19 in the RECOVERY trial [36] and a recent meta-analysis of 7 trials conducted on 1703 patients showed a reduction in 28-day mortality compared with standard care or placebo (32% vs 40%, OR = 0.66, 95%CI = 0.53–0.82) [37]. However, most experience is with dexamethasone and not methyl prednisolone (MP) although MP has been recommended as an alternate to dexamethasone in a dose of 32mg/d in current guidelines [38]. Our patient population was given MP but in higher doses (40mg q8hrs) which may have resulted in the observed hyperglycemia, secondary infections and electrolyte imbalance in our data. Moreover, a randomized trial on severe COVID-19 patients in Brazil did not show any mortality benefit of MP at 28 days as compared to placebo (37% vs 38%) [39]. Some observational studies have reported mortality benefit of TCZ [40, 41] but RCTs failed to demonstrate any difference in survival when compared to placebo or usual care [42]. Data from our center also did not show any survival benefit of TCZ [43].

Among other predictors, a peripheral oxygen saturation at presentation below 93% and use of assisted ventilation in emergency room was associated with mortality in our data. These indicators of severity of the COVID pneumonia are now universally used in guidelines [3, 5]. A high NLR is predictive of disease severity as shown by Yang et al [44] and Nalbant et al [45] among many others. Our data also shows that an NLR of at least 5 is associated with 2.6 (95% CI: 1.2–6.0) times the odds of mortality. Bacterial superinfection in COVID-19 can trigger a cascade of multiorgan failure through sepsis and further decrease the probability of recovery. Although baseline procalcitonin was low in the overall cohort (median = 0.4 ng/ml), levels above 2 ng/ml were associated with 3.6 (95% CI: 16.-8.1) times the odds of mortality in our patients. It is uncertain whether this suggests secondary bacterial infection or hyper-inflammation as studies have suggested that raised procalcitonin as a marker of bacterial infection tends to lose specificity as COVID progresses [25, 26]. Pao2/FiO2 ratio has been shown to predict mortality as it relates to the severity of lung involvement and impairment of gas exchange [46]. Our data shows a 50% reduction in mortality with unit increase in Pao2/FiO2 ratio (OR = 0.5:95% CI = 0.4–0.8). Serum albumin is a marker of nutritional status and usually used to depict chronic undernutrition. However, it has shown a strong association with mortality in COVID-19 as suggested by Rica et al [47]. Unit increase in serum albumin in our patients was associated with 70% decreased odds of mortality (OR = 0.3; 95% CI = 0.13–0.66). Factors which remain significant in the final model were age above 60 years, SpO2 less than 93% and procalcitonin above 2 ng/ml adjusted for month of admission to account for differences in management strategies and the potential change in virus genotype over the those months.

This is the first prospective cohort study from Pakistan on in-hospital mortality of COVID-19 patients. Detailed clinical history, laboratory parameters and therapeutics have been compared. There is no attrition as patients were followed till discharge or death. Despite the limitations of a small sample our data revealed some important predictors of mortality in our study population. The study is limited by data from a single-center with critically ill COVID-19 patients which may introduce a selection bias and inflate the mortality. Hence, results from this study may help in the risk stratification and management of similar critically ill patients only. We cannot say whether patients who were referred out were sicker than our cohort or vice versa and hence results otherwise would have changed. The readers should be mindful of measurement errors in laboratory parameters as single readings were taken and validation systems vary across laboratories. The sample size is limited because of the number of beds available at the unit during the first wave. We plan to add more subjects in future as we are currently seeing the second wave. A larger multi-center cohort study from various hospitals of the country would help to further validate the findings of our study.

Acknowledgments

We would like to thank Indus Hospital Research Center for assistance in database designing and data collection.

Data Availability

Data file is available from the Mendeley database (DOI:10.17632/j6jz4vk6r6.2).

Funding Statement

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

References

1.WHO. Timeline: WHO’s COVID-19 response 2019. [Available from: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/interactive-timeline/#]. [Google Scholar]
2.Worldometers. COVID-19 Coronavirus Pandemic. 2020. [Google Scholar]
3.Pakistan Go. Health Advisory 2020. [Available from: https://covid.gov.pk/stats/pakistan]. [Google Scholar]
4.Banik A, Nag T, Chowdhury SR, Chatterjee R. Why Do COVID-19 Fatality Rates Differ Across Countries? An Explorative Cross-country Study Based on Select Indicators. Global Business Review. 2020:0972150920929897. [Google Scholar]
5.Organization WH. Clinical management of COVID-19: interim guidance, 27 May 2020. World Health Organization; 2020. [Google Scholar]
6.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. Journal of biomedical informatics. 2009;42(2):377–81. 10.1016/j.jbi.2008.08.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.College Station TSL. StataCorp. 2015. Stata Statistical Software: Release 14. 2015. [Google Scholar]
8.Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in logistic regression. Source code for biology and medicine. 2008;3(1):1–8. 10.1186/1751-0473-3-17 [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Vincent J-L, De Mendonça A, Cantraine F, Moreno R, Takala J, Suter PM, et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Critical care medicine. 1998;26(11):1793–800. 10.1097/00003246-199811000-00016 [DOI] [PubMed] [Google Scholar]
10.Nasir N, Habib K, Khanum I, Khan N, Muhammad ZA, Mahmood SFJAP. Predictors of severity and mortality of COVID-19 at a tertiary care center in a Lower-middle income country. 2020. [Google Scholar]
11.Tran K, Cimon K, Severn M, Pessoa-Silva CL, Conly J. Aerosol generating procedures and risk of transmission of acute respiratory infections to healthcare workers: a systematic review. PloS one. 2012;7(4):e35797. 10.1371/journal.pone.0035797 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Wang Y, Lu X, Li Y, Chen H, Chen T, Su N, et al. Clinical Course and Outcomes of 344 Intensive Care Patients with COVID-19. Am J Respir Crit Care Med. 2020;201(11):1430–4. 10.1164/rccm.202003-0736LE [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Twigg III HL, Khan SH, Perkins AJ, Roberts S, Sears C, Rahman O, et al. Mortality Rates in a Diverse Cohort of Mechanically Ventilated Patients With Novel Coronavirus in the Urban Midwest. 2020;2(8). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Grasselli G, Greco M, Zanella A, Albano G, Antonelli M, Bellani G, et al. Risk factors associated with mortality among patients with COVID-19 in intensive care units in Lombardy, Italy. 2020;180(10):1345–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Audit IJA. ICNARC report on COVID-19 in critical care. 2020;10:2020. [Google Scholar]
16.Richardson S, Hirsch JS, Narasimhan M, Crawford JM, McGinn T, Davidson KW, et al. Presenting characteristics, comorbidities, and outcomes among 5700 patients hospitalized with COVID-19 in the New York City area. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Yang X, Yu Y, Xu J, Shu H, Xia J, Liu H, et al. Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study. Lancet Respir Med. 2020;8(5):475–81. 10.1016/S2213-2600(20)30079-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Dondorp AM, Hayat M, Aryal D, Beane A, Schultz MJ. Respiratory Support in COVID-19 Patients, with a Focus on Resource-Limited Settings. Am J Trop Med Hyg. 2020;102(6):1191–7. 10.4269/ajtmh.20-0283 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Slutsky AS, Ranieri VM. Ventilator-induced lung injury. N Engl J Med. 2013;369(22):2126–36. 10.1056/NEJMra1208707 [DOI] [PubMed] [Google Scholar]
20.Cates J, Lucero-Obusan C, Dahl RM, Schirmer P, Garg S, Oda G, et al. Risk for In-Hospital Complications Associated with COVID-19 and Influenza—Veterans Health Administration, United States, October 1, 2018–May 31, 2020. 2020;69(42):1528. [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Lee DW, Santomasso BD, Locke FL, Ghobadi A, Turtle CJ, Brudno JN, et al. ASTCT consensus grading for cytokine release syndrome and neurologic toxicity associated with immune effector cells. Biology of Blood and Marrow Transplantation. 2019;25(4):625–38. 10.1016/j.bbmt.2018.12.758 [DOI] [PubMed] [Google Scholar]
22.Wang D, Hu B, Hu C, Zhu F, Liu X, Zhang J, et al. Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus–infected pneumonia in Wuhan, China. 2020;323(11):1061–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
23.WSJ. Why Youthful, Conservative Pakistan Is a Coronavirus Bright Spot 2020. [Available from: https://www.wsj.com/articles/why-youthful-conservative-pakistan-is-a-coronavirus-bright-spot-11596297600. [Google Scholar]
24.Wu Z, McGoogan JMJJ. Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention. 2020;323(13):1239–42. [DOI] [PubMed] [Google Scholar]
25.Petrilli CM, Jones SA, Yang J, Rajagopalan H, O’Donnell L, Chernyak Y, et al. Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study. 2020;369. [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Wu C, Chen X, Cai Y, Xia J, Zhou X, Xu S, et al. Risk Factors Associated With Acute Respiratory Distress Syndrome and Death in Patients With Coronavirus Disease 2019 Pneumonia in Wuhan, China. JAMA Intern Med. 2020;180(7):934–43. 10.1001/jamainternmed.2020.0994 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Liu S, Yao N, Qiu Y, He C. Predictive performance of SOFA and qSOFA for in-hospital mortality in severe novel coronavirus disease. The American Journal of Emergency Medicine. 2020;38(10):2074–80. 10.1016/j.ajem.2020.07.019 [DOI] [PMC free article] [PubMed] [Google Scholar]
28.Yao X, Ye F, Zhang M, Cui C, Huang B, Niu P, et al. In Vitro Antiviral Activity and Projection of Optimized Dosing Design of Hydroxychloroquine for the Treatment of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). Clin Infect Dis. 2020;71(15):732–9. 10.1093/cid/ciaa237 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Medicine RCGJNEJo. Effect of hydroxychloroquine in hospitalized patients with Covid-19. 2020;383(21):2030–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
30.FDA U. Coronavirus (COVID-19) update: FDA revokes emergency use authorization for chloroquine and hydroxychloroquine. June; 2020. [Google Scholar]
31.Qureshi S, Naveed AB, Yousafzai MT, Ahmad K, Ansari S, Lohana H, et al. Response of extensively drug resistant Salmonella Typhi to treatment with meropenem and azithromycin, in Pakistan. PLoS neglected tropical diseases. 2020;14(10):e0008682. 10.1371/journal.pntd.0008682 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Furtado RHM, Berwanger O, Fonseca HA, Correa TD, Ferraz LR, Lapa MG, et al. Azithromycin in addition to standard of care versus standard of care alone in the treatment of patients admitted to the hospital with severe COVID-19 in Brazil (COALITION II): a randomised clinical trial. Lancet. 2020;396(10256):959–67. 10.1016/S0140-6736(20)31862-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Cavalcanti AB, Zampieri FG, Rosa RG, Azevedo LC, Veiga VC, Avezum A, et al. Hydroxychloroquine with or without Azithromycin in Mild-to-Moderate Covid-19. 2020;383(21):2041–52. 10.1056/NEJMoa2019014 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Rosenberg ES, Dufort EM, Udo T, Wilberschied LA, Kumar J, Tesoriero J, et al. Association of Treatment With Hydroxychloroquine or Azithromycin With In-Hospital Mortality in Patients With COVID-19 in New York State. JAMA. 2020;323(24):2493–502. 10.1001/jama.2020.8630 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Rawson TM, Moore LSP, Zhu N, Ranganathan N, Skolimowska K, Gilchrist M, et al. Bacterial and Fungal Coinfection in Individuals With Coronavirus: A Rapid Review To Support COVID-19 Antimicrobial Prescribing. Clin Infect Dis. 2020;71(9):2459–68. 10.1093/cid/ciaa530 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Horby P, Lim WS, Emberson JR, Mafham M, Bell JL, Linsell L, et al. Dexamethasone in hospitalized patients with Covid-19-preliminary report. 2020. [Google Scholar]
37.Sterne JA, Murthy S, Diaz JV, Slutsky AS, Villar J, Angus DC, et al. Association between administration of systemic corticosteroids and mortality among critically ill patients with COVID-19: a meta-analysis. JAMA. 2020;324(13):1330–41. 10.1001/jama.2020.17023 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.WHO/-nCoV/Corticosteroids/.1 WHOJWrn. Corticosteroids for COVID-19. Living Guidance. 2020. [Google Scholar]
39.Jeronimo CMP, Farias MEL, Val FFA, Sampaio VS, Alexandre MAA, Melo GC, et al. Methylprednisolone as adjunctive therapy for patients hospitalized with COVID-19 (Metcovid): a randomised, double-blind, phase IIb, placebo-controlled trial. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Malgie J, Schoones JW, Pijls BG. Decreased mortality in COVID-19 patients treated with Tocilizumab: a rapid systematic review and meta-analysis of observational studies. Clin Infect Dis. 2020. 10.1093/cid/ciaa1445 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Gupta S, Wang W, Hayek SS, Chan L, Mathews KS, Melamed ML, et al. Association between early treatment with tocilizumab and mortality among critically ill patients with COVID-19. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Stone JH, Frigault MJ, Serling-Boyd NJ, Fernandes AD, Harvey L, Foulkes AS, et al. Efficacy of Tocilizumab in Patients Hospitalized with Covid-19. N Engl J Med. 2020;383(24):2333–44. 10.1056/NEJMoa2028836 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Sarfaraz S, Shaikh Q, Iftikhar S, Herekar FF, Saleem SG, Kanani F. Is Tocilizumab an effective therapy for Severe COVID-19: a retrospective cohort study. 2020. [DOI] [PubMed] [Google Scholar]
44.Yang A-P, Liu J-P, Tao W-Q, Li H-m. The diagnostic and predictive role of NLR, d-NLR and PLR in COVID-19 patients. International immunopharmacology. 2020;84:106504. 10.1016/j.intimp.2020.106504 [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Nalbant A, Kaya T, Varim C, Yaylaci S, Tamer A, Cinemre H. Can the neutrophil/lymphocyte ratio (NLR) have a role in the diagnosis of coronavirus 2019 disease (COVID-19)? Revista da Associação Médica Brasileira. 2020;66(6):746–51. 10.1590/1806-9282.66.6.746 [DOI] [PubMed] [Google Scholar]
46.Santus P, Radovanovic D, Saderi L, Marino P, Cogliati C, De Filippis G, et al. Severity of respiratory failure at admission and in-hospital mortality in patients with COVID-19: a prospective observational multicentre study. BMJ open. 2020;10(10):e043651. 10.1136/bmjopen-2020-043651 [DOI] [PMC free article] [PubMed] [Google Scholar]
47.de la Rica R, Borges M, Aranda M, Del Castillo A, Socias A, Payeras A, et al. Low albumin levels are associated with poorer outcomes in a case series of COVID-19 patients in Spain: a retrospective cohort study. Microorganisms. 2020;8(8):1106. [DOI] [PMC free article] [PubMed] [Google Scholar]

PLoS One. doi: 10.1371/journal.pone.0251754.r001

Decision Letter 0

Wisit Cheungpasitporn

19 Feb 2021

PONE-D-21-00091

Determinants of in-hospital mortality in COVID-19; a prospective cohort study from Pakistan

PLOS ONE

Dear Dr. Quratulain Shaikh,

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.

==============================

ACADEMIC EDITOR: The reviewers have raised a number of points which we believe major modifications are necessary to improve the manuscript, taking into account the reviewers' remarks. Please consider and address each of the comments raised by the reviewers before resubmitting the manuscript. This letter should not be construed as implying acceptance, as a revised version will be subject to re-review.

==============================

Please submit your revised manuscript by Apr 05 2021 11:59PM. 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.

We look forward to receiving your revised manuscript.

Kind regards,

Wisit Cheungpasitporn, MD

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) We note that you have stated that you will provide repository information for your data at acceptance. Should your manuscript be accepted for publication, we will hold it until you provide the relevant accession numbers or DOIs necessary to access your data. If you wish to make changes to your Data Availability statement, please describe these changes in your cover letter and we will update your Data Availability statement to reflect the information you provide.

[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: No

Reviewer #2: Partly

Reviewer #3: Partly

**********

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

Reviewer #1: No

Reviewer #2: Yes

Reviewer #3: No

**********

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: No

Reviewer #2: No

Reviewer #3: 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

Reviewer #3: 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 paper is well writtem and as noted by the authors in the introduction, these data could help in risk stratification and management of COVID-19 patients. The main issue of the paper resides in the fact that the treatments assignments are not randomized so that any conclusion regarding the association between treatment and survival is dubious. There are also issues in the presentation of the statistical analysis that should be (easily) corrected.

I would recommend to the authors to state clearly the assumptions they want to investigate and a major revision of the statistical analysis before any resubmission.

Statistical aspects

-------------------

It is not clear whether the study followed a pre-speficied statistical analysis plan. For example, was the criteria that led to the exclusion of patients under 15 set before the start of the data analysis? If so, were the various pre-defined steps registered anywhere?

p11 l 99 "Among the rest 384 (9.9%) needed admission but due to limited bed availability in the isolation unit, only 193 (50%) were admitted while the rest were referred to other isolation units in the city."

What was the rationale underlying the choice of patients referred to other isolation units?

Also it appears later in the text and table 1 that some of the 186 patients admitted were asymptomatic (n=2) or presenting mild symptoms (n=13). Please expain.

The statistical analysis reported at lines 110-142 table 1&2 is supported by various p-values. Those p-values have to be taken with a grain of salt since we are in the classical setting of multiplicity or multiple testing. With a number of hypotheses tested large enough, there will always be some significant p values by chance alone. Therefore, it would be good to report which p-values are significant after Benjamini-Hochberg correction.

Table 3 and l 144-158 the assignment of treatment being non randomized, it is difficult to interpet any association between treatment and outcome.

Besides, table 3 report various frequencies of treatment conditionnal on survival status. Even in case of an RCT, such numbers would be difficult to interpret. What is required indeed is a frequency of survival conditionnal on treatment.

For those two reasons, a conclusion like l. 155 "There was no mortality benefit of using methylprednisolone in our patients (p>0.05)" is not grounded at all.

This table should be left as is and would stand as a purely descriptive report of the pathways of cares.

A table of frequency of survival conditional on treatment should be provided, but again, any attempt to draw conclusion about the observed association should be avoided.

l 160 Please explain better what is meant by "Univariable binary logistic regression (Table 4) at a cut-off of p<0.10, showed that ...".

Do you mean that you implemented a logistic regression for all variables and present here only the results for variables with the lowest p-values?

Table 4: this table mixes variables of different types: demographic parameters, clinical characteristics at admission, baseline lab parameters and treatments. The treatment being non-randomized, reporting and commenting a p-value for treatment variables makes little sense. Please also add a column with the significance obtained by the Bonferroni-Hochberg procedure.

l 170-175: can you please rephrase/explain what multivariate logistic regression was implemented and what was the goal/rationale for it. Again, it seems to mix variables of different types.

I understand the idea of implementing a multivariate logistic regression to predict the survival status of of a patient at admission, but such a an approach should involve not treatment variables.

What is the goal of the multivariate regression involving all the variables?

Specific comments

-----------------

p2 Funded/unfunded study statement missing

p4 Human Subject Research (involving human

participants and/or tissue) missing

p9 l52 "Our mortality rate of 2% (3) is comparable to the India (1.45%) but lower"

-> "The Pakistan mortality rate of 2% (3) is comparable to that of India (1.45%) but lower than that of Iran (4.68% )"

p9 l58 "The 58 peak of infection in the first wave reached in..."

-> The 58 peak of infection in the first wave was reached in...

p10 Do I understand correctly that asymptomatic patients or with mild symtoms were included in this study?

p10 l88 "variables were expressed as"

-> variables were summarized by

p 20 l88 "Categorical variables were expressed as number (%)."

-> "Distribution of categorical variables were expressed as percentages of the various categories."

p10 l89 "Student’s t-test or Mann Whitney U test were used to compare continuous data. Chi square or Fischer’s Exact test was used to compare categorical data."

This statement is a bit vague at this step

p10 l91 "All variables with p-value <0.10 were considered for multivariable binary logistic regression model in order of significance." I do not not understand the last bit of this sentence "in order of significance."

p11 l101 "Seven patients who were admitted at TIH were less than years old hence removed from this analysis." Please explain why.

p11 l113 "fatigue was common in non-survivors and vomiting was seen in survivors (p>0.05)"

It is not clear to which test this pvalue refers to. Please clarify.

Reviewer #2: The information included in this article is of critical and immediate importance. While the sample size may be relatively small, it is only through publication of work such as this that we can build the literature to see the bigger picture. Because of its importance, and the critical eye on science in this area, it is crucial that the results be communicated effectively and scientifically.

In that vein, I would ask the authors to take a second look at the overall manuscript and clean up small errors and inconsistencies. Most importantly, the statistical interpretation of the data needs improvement. There is an over-reliance on p-values of more or less than a threshold and not nearly enough attention paid to the data itself, which is the more important interpretation. The p-value tells us only whether what we are seeing is likely to be a random result; it cannot speak to the magnitude or importance of an effect.

Introduction

- Could use another run through by the authors to tighten up language and remove small errors; but is intelligible.

- The end of the introduction needs to be more direct about what is being estimated. From Line 64, you aren’t estimating in-hospital mortality of Covid-19 in Pakistan, but in the Indus Hospital’s COVID isolation unit.

Methods

- The Methods section needs to be re-organized into paragraphs, currently the information jumps around topically from one sentence to the next making it hard to follow. Suggest:

o Line 74, put IRB approval on its own at the end

o Line 81, a description of the types of patients admitted should immediately follow Line 73 where you describe who was included in the study

o In a new paragraph, Lines 75-87 (apart from the above) describe how the patients were recorded and followed and make a single coherent paragraph

o From Line 87 to the end (“Data was recorded…”) this should be a new paragraph on the data analysis.

Results

- Line 99 – how was it decided who needed admission? What was the criteria? I think it is moderate to severe, but this should be stated more plainly.

- A brief background on the teleconsultation service and its role, could go in introduction or methods

- Line 99 – in Figure 1 it is indicated 9.97 needed admission, this should be rounded to 10.0 not 9.9.

- Percentages should be displayed consistently – some are rounded to whole numbers and other to 00.0%. If significant digits, that would be fine, too, but that isn’t consistent either. Even if the number is, for example, 36.0% the display should be consistent.

- Lines 101-102: the 7 patients <18 & excluded from the analysis need to be included on Figure 1 with a final box for the included cohort for analysis.

- Figure 1 – what does LAMA mean?

- Table 1 (and all others) put a space between n & (%), e.g. 8 (12.1%); true also of mean (SD) and median (IQR)

- Table 1 – headers of n (%) are not applicable to continuous variables – needs to be stated in the table what 61.4 (54.1-69.1) are. The use of ^ and ^^ is non-standard and not clear enough. It appears that even the authors got mixed up (see next point).

- Table 1 / Line 103-104 – the text says that 61.4 is a median, but the table says 61.4 is a mean, using the ^^ nomenclature defined at the bottom of the table.

- Line 103-104 and 110-111 is the same sentence repeated

- Line 111 [incorrect numbers, conclusion]– the statement that males are more likely to die because there are 77.6% males in the non-survivors compared to 76.5% males in the survivors, p<0.05 is nonsensical and at odds with what is presented in Table 1 where there is a p-value of 0.859 comparing the distribution of gender in the survivors to the distribution of gender in the non-survivors.

- Lines 112-113, what does the p>0.05 refer to? Why not put the actual p-value? But more importantly, I am unsure the authors are correctly interpreting what each p-value in Table 1 represents. In the case of the Symptoms, the p-value in Table 1 using the chi-square method only says that across ALL of the symptoms presented among the survivors and non-survivors, there is not a statistically significant difference in this distribution. It does NOT relate to which are the three most common symptoms in both groups. Also, the statement about the 3 most prevalent symptoms does not need a p-value.

- Lines 114-115, again, I cannot decipher what this p-value is supposed to represent or where it came from.

- Line 118 – I find the conclusion that temperature among non-survivors was higher suspect based on the data presented and that fact that the p-value is marked by a “D” which is not defined by the authors. Also, if this finding is real, then address the fact that differences were only by a few tenths of a degree.

- Tables – there are some errant “a” and “b” in the tables which appear to have no discernable meaning.

Not continuing line by line, the authors need to consult a statistician on the interpretation of p-values. Furthermore, their utility in this kind of paper is not very high. The authors would do better not to try to repeat all of the information in the tables in the results, but only to highlight the findings of most interest. Each sentence in the results does not need a p-value. Results such as prevalence of certain symptoms are of interest on their own and do not neatly adhere to a p-value framework.

Table 4 - For binary variables, the data provided in prior tables is sufficient to re-run the univariable logistic regression models. Having done this for a few, there may be transcribing errors. I used R and not Stata, but for example, for Gender (male) the OR I get is 1.07 after appropriate rounding of 1.067, the 95% CI I get is 0.53 – 2.22 after appropriate rounding of 0.528 and 2.218. These differences could be due to different statistical programs being run. However, there are also inconsistent numbers of digits displayed for the 95% CIs and the p-values.

Figure 2 - The depiction of multivariable logistic regression coefficients (ORs) as a forest plot is not helpful. It would be better to display this information in a table. Forest plots such as the one included are typically used to evaluate effect estimates across different studies, not within a single model.

Interpretation – the authors need to re-assess their interpretation of their OR results for both the univariable and multivariable models. All values with a significant p-value are not “associated with high odds of mortality”. A large effect size along with a significant p-value can be used to establish an association with a high odds of mortality, but a statistically significant p-value alone is not enough. More attention should be paid to the meaning of the OR and its interpretation and less to p-values. NOTE: in the abstract the phrase ‘higher odds of mortality’ is used – this is much more accurate.

Discussion

The discussion appears to be more carefully written. It focuses on important findings and does not over extend the interpretation of p-values (e.g. "The final model shows that the factors associated with mortality are…" which is an accurate interpretation). It could still go further if effect size (i.e. the value of the OR matters!) are interpreted.

Can the authors address any potential bias from patients being referred to other centers? Were more severe patients retained? Or is it assured that these were random depending only on available beds at the time of entry?

Can the authors address the reliability of the data? Was there any missing data for patients included – i.e. this being real world data could tests performed have failed to be recorded? You included many continuous laboratory values, what is the likelihood of measurement error? Measurement error is likely to reduce the observed associated with mortality. Finally, is it possible for the categorical values to have any important misclassification? As it is expected this information will be used in larger meta-analyses and further research, inclusion of any such limitations may be invaluable.

Abstract

The abstract is also over-reliant on p-value thresholds. It also says that those who died were more likely to be males, implying that they were more likely to be males than those who did not die – which is not supported by the data (see above).

Reviewer #3: I appreciate the intent of this research and the need for it to be both publicly available and scrutinized in the details before publication. There are some minor typographic concerns as well as a general need to define terms and acronyms used throughout the manuscript (at least a supplement or in first use – see GCS for example). More substantively, there are two general issues and then some more specific minor issues of note related to the statistical results presented that listed below.

First, there are many p-values reported (at least 70 unique tests), (Table 1: 16, Table 2: 40, Table 3: 14, Table 4 (some redundant but different tests used, 45) with no control for inflated Type I error rates when “fishing” across this many different response/predictor variables. At least some of these are spurious detections without some sort of control like Bonferroni or False Discovery Rate adjustments (at least within a table or suite of similar sort of tests). I would suggest doing this for the limitation of chance false discoveries and because it would simplify the final reporting of results, if only those that are selected to have small p-values are discussed. Related to this is the redundancy of testing for survivor/not vs each variable as a response (Chi-square/Fisher’s are non-directional but the others are not) and then repeating the same tests in the univariate logistic regression models with survivor/not as the response. The summary of characteristics that Tables 1 to 3 provide is interesting, but some of those tests match the Table 4 result exactly (Gender for example with a p-value of 0.859 – also note that the reported result in the Abstract about Gender is wrong if this is the correct p-value). It is also important to remember that only when p-values are 1 is there is no evidence of a difference. For example, line 133 has “There was no difference in the arterial pH” – the p-value was 0.421 (the estimated medians and quartiles match exactly which doesn’t make sense with that p-value, so should be checked) which suggests weak evidence of a difference but not “no difference”. Now if the medians do match exactly, then this statement is true but I don’t think that was support for it. Finally, there is the issue of selection bias in the final reported results. If you use p-values to select your final results to discuss, you overstate precision and understate the size of your p-values. Methods for adjusting for this are complex but there is a danger to model selection using p-values and then reporting the p-values as if you knew those were the predictors you were going to look at. Caution should be incorporated in the limitations discussion in both being too definitive in what was identified as being the only predictors that could be important and that what was identified really are the important predictors – this was a very exploratory analysis to consider so many different predictors. Additionally, no interactions were considered, so the assumption of no interactions (or at least no interactions explored) should be noted as that could change results dramatically. I am not saying they have to be, but that is an assumption.

Second, there are four serious issues in the logistic regression models that need to be addressed before this might be publishable as the results hinge on these aspects of the work.

One is that there appears either to be separation issue or something close to it in both some of the univariate logistic regression models and in the reported final model. In Stata, there is supposed to be a check for this (see https://stats.idre.ucla.edu/other/mult-pkg/faq/general/faqwhat-is-complete-or-quasi-complete-separation-in-logisticprobit-regression-and-how-do-we-deal-with-them/ for example). But the estimates are large and the CIs are very wide, which can indicate this issue, at least in some software. Were there any warning messages/automatic modifications when running the Stata logistic regression models? If so, how was that handled? As an example of one of these that I find suspicious for this issue, the final model contained oxygen saturation with an OR CI from -100 to 300. And Figure 2 does not match the numbers reported on line 173 of 1.2 to 105.9 (but the margin labels do, so there is something wrong with the lines displayed in this plot at a minimum).

Another issue in the multivariable logistic regression is that the model building/selection process is unclear. It appears to be a forward selection process of starting with the predictor with the smallest p-value and adding others sequentially. But no details are given on what happens to get the final model from this. I was left to assume that each predictor was tried and then not kept in the model if the p-value was not small after it was added? But after the first predictor was added, you really need to try each one again one at a time to see if its addition is warranted given that new predictor and then pick the one with the smallest p-value each time. This is especially important in the possible presence of multicollinearity of predictors that is likely here (was that checked and what sort of results did it provide?). At a minimum the steps that were taken need to be clarified. Since so much rides on what is in the final model, this process should be carefully documented. With so many predictors, it is challenging to do and report the model building process. The previous suggestion of a multiple testing correction could even mean missing important predictors in this step of the process, so the variable screening before model building might not be the best approach. And with potential separation issues, I would not necessarily recommend starting with all predictors in the model and then dropping them sequentially as important predictors can get large p-values in these situations in some software (although a step-down process could be simpler to implement than step up methods).

An additional assumption in the logistic regression models is that quantitative predictors are linearly related to the response on the logit scale. Was this checked? Not all software makes this possible and if not, then you should note this as an assumption you might make. I could imagine there are some predictors, like pH, that having either too high or too low levels could lead to adverse outcomes. Visualization of the survival response versus the quantitative predictors can help with this (and with the potential for separation issues too) and identify variables that might be candidates for polynomial treatment in the models.

Finally, an issue with all of the logistic regression models is that you have multiple cohorts of subjects being combined. You show this in Figure 1 with the 4 months of subjects and discuss it on lines 188 to 193 which discuss changing standards of care. At least controlling for month as a categorical variable or random effect in all the logistic regression models would help to address this as a potential source of the differences in the survival outcome. With more recent focus on different variants of COVID, is it possible that this played some role in changing survival rates over time (if they were present)? The month variable would likely best not be subject to model refinement but serve as a control variable for considering all the other results.

More minor but important issues:

I am checking the box that data will be available as the authors promise that. Some Plos One authors confuse sharing summary information with sharing the entire data set as used in the models. I am assuming that is the case here. I think this is critical as others work to create meta-analysis of studies like these to synthesize local study results such as this and also because different modeling choices here could lead to very different conclusions.

Abstract should be changed if any of the reported results are modified, which I think likely should happen.

Line 83: Did the patients admitted for other reasons have COVID too? It was unclear from this sentence.

Line 102: Why were the less than 15 year olds dropped? This relates to the limitations on line 257 and should be part of that discussion that these results are limited to subjects over 15.

Line 89: How did you decide between t-test and Mann-Whitney U? Between the parametric Chi-squared test and Fisher’s Exact test?

Table 1: It would great to add some details on each test. A column for the numerical value of the test statistic and another for its distribution (Chi-square (df) or t(df) or “exact”/permutation for the other two versions). This adds quite a bit to the table but can help the reader fully understand what you did. There are some other issues in Table 1: First, the Age^^ suggests it should be mean +- SD but those appear to be medians (Q1 – Q3). Second is how the median results are presented – you are not reporting the IQR, you are reporting Q1 and Q3 with a minus sign between them that is confusing. I would suggest either switching to reporting the IQR = Q3-Q1 or use a comma between them. For Symptoms and Comorbid conditions – are these categories really mutually exclusive? Maybe “Others” means more than one and just wasn’t labelled well – or is it really “Other” things no in the list. I would suggest turning this into a suite of binary variables for each aspect, such as “Fever” or “No Fever” and then each can be explored for relationships to survived/not. Some may not be suitable for use in the tests (like Runny Nose that occurred a total of 5 times or Liver disease that occurred 1 time), but those rare conditions are likely going to cause you issues in the current version with small expected cell counts that violate the assumptions of the parametric Chi-squared test that was used at least for Symptoms. Table 2’s “CXR findings” and Table 3’s “In Hospital complications” also seem like they might not be mutually exclusive categories? After turning these variables into multiple binary variables (possibly), some might be useful in the logistic regression models/persist through the model refinement stages?

Line 136: The directional interpretations (higher/lower) sound like you did one-sided tests. Is that true? If not, then try to report the p-values first and then discuss direction of differences to avoid this confusion.

Line 160: It might be easier to discuss the results if you organize the results by groups – like for categorical predictors separate from quantitative ones? And where possible use the units of the predictors for quantitative variables and have “higher odds” not “high odds”. Maybe even a plot of these results sorted by p-values like in Figure 2 could aid the discussion of them?

Table 4: As noted above, some of these are redundant with previous tests or test a similar null hypothesis (variable vs survived/not) but use a different method. It seems to make some of the earlier work redundant if this is what you really care about – and it seems like this is what you care about most. For oxygen saturation, why was the binary version with a cutoff at 93% used instead the quantitative version of it (both were used before). For the multi-category (analysis of deviance?) tests for things like “Use of Tocilizumab”, some formatting change is needed as the IVIG row looks like part of the previous multi-group test. I am assuming most are from z-tests but some are from Chi-squared tests and all are conditional on all other aspects of the models? Again a little detail on the results can help the reader understand what you did to get those.

Lines 170-175: I think this needs to be expanded. These are your main results. And it is important if your final model contains a suite of predictors to note that these results are conditional/controlled for other variables in the model – as results are different when other variables are not present. And see previous notes on the multivariable logistic regression. Adding visualizations of survived/not versus each of the final model chosen predictors could help the reader understand these results.

Line 265: I think a citation to the software used should be included. Would you be posting your code for the analyses or just the data set? Reproducible research methods that provide code run with output would help readers with other questions about results have access to that although it is not a journal requirement.

General note: I am not sure on PlosOne’s standing on this, but Oxford commas would help in the many lists as some sentences use more complex constructions that get confusing by this absence when it could be used.

Figure 1: The “Referred Out” is the sample analyzed except for the under 15 year olds? How many were in the different months? This would get clarified if month of initial hospitalization were used as a predictor.

There are some minor typographic issues that could be resolved in another round of edits to improve readability and grammatical correctness in a few spots but it is not a poorly written paper.

**********

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: No

Reviewer #2: Yes: Christen M Gray

Reviewer #3: Yes: Mark C Greenwood

[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 May 27;16(5):e0251754. doi: 10.1371/journal.pone.0251754.r002

Author response to Decision Letter 0

31 Mar 2021

We have redone the analysis in the light of reviewer comments- The comments were very comprehensive and detailed hence all responses are given in the Response to reviewers document. It may not be possible to rewrite the details here. Kindly guide us if the journal requires us to do so.

Attachment

Submitted filename: Response to reviewers.docx

Click here for additional data file.^{(39.7KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0251754.r003

Decision Letter 1

Wisit Cheungpasitporn

19 Apr 2021

PONE-D-21-00091R1

Determinants of in-hospital mortality in COVID-19; a prospective cohort study from Pakistan

PLOS ONE

Dear Dr. Quratulain Shaikh,

==============================

ACADEMIC EDITOR: The reviewers have still raised a number of points which we believe major modifications are necessary to improve the manuscript, taking into account the reviewers' remarks. Please consider and address each of the comments raised by the reviewers before resubmitting the manuscript. This letter should not be construed as implying acceptance, as a revised version will be subject to re-review.

==============================

Please submit your revised manuscript by Jun 03 2021 11:59PM. 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 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. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Wisit Cheungpasitporn, MD

Academic Editor

PLOS ONE

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

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 #3: (No Response)

**********

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

Reviewer #1: Yes

Reviewer #3: Yes

**********

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

Reviewer #1: Yes

Reviewer #3: Yes

**********

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

Reviewer #1: No

Reviewer #3: Yes

**********

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

Reviewer #1: Yes

Reviewer #3: Yes

**********

6. Review Comments to the Author

Reviewer #1: l. 28: " A prospective cohort study was conducted [...] to describe the predictors of mortality among hospitalized COVID-19 patients."

"describe" is a really vague objective but using the word predictor suggests that the paper is about P(survival given age).

but at l. 30 "Most non-survivors were above 60 years of age (64%)"

this is about P(Age given survival)

This ambiguity should be resolved.

The method section concerning the multivariate logistic regression could provide more details, in particular regarding the set of candidate explanatory variables.

It is claimed that "all data are fully available without restriction" but the link provided to access the data does not work.

Reviewer #3: Thank you for working to address reviewer comments. It is better but still needs some work to document what was done and why with the variables used and statistical modeling, and the writing is a bit rough in places.

For the response related to Table 4, “Moreover, Linearity assumptions were confirmed for continuous variables and those not fulfilling the assumption were not considered for univariable regression.”

• I did not intend for you to abandon useful predictors because they might show nonlinear relationships. There are options for transformations or polynomials to help with these situations. The papers you reference discuss using plots to check this for at least one way to qualitatively assess for clear missed curvature. This can not confirm no violation of the linearity assumption but can fail to show a problem needs to be addressed.

Potential bias in referrals: “However, the referral was at random depending only on bed availability.”

• This sounds more like haphazard assignment if it is related to openings in beds rather than using a truly random mechanism, like a coin flip or randomization software. It still could have bias because of this. This wording is not used in the paper.

For the revised model selection methods, I like that you have refined this process. I don’t see information on the steps taken (what was in the model to start and what were the removal steps and those p-values – based on the papers (not cited –see below) I suspect they are bigger than 0.25 but the information on the intermediate steps is part of the evidence story) and those should be reported, not just the final model.

There is no clear reason for many of the binary splits in predictor variables used – over 60 years? Over 93% oxygen, Procalcitonin over 2 ng/ml? Were these selected to optimally relate to the response or are the standard cutoffs or were they arbitrarily chosen? It seems like each needs a quick reason for the splits or the variables could have just been used as quantitative predictors (with linearity checks) – is that why they were split because of nonlinearity seen in the relationship to the empirical logits? Or are these standard splits? Or were they are the median or mean of the predictors? This should be explained in the article.

More specific feedback on new version:

You should decide where you are placing references, before or after periods (I think before is best) and be consistent. It is inconsistent throughout the manuscript.

Abstract:

There is a missing a space before “Most”. Extra parenthesis before 21.8. I am not sure what the group is that is being referred to in the vs 7.8% - was that survivors that had 7.8%?

This sentence is in need of at least some punctuation to be more clear: “…, Neutrophil to lymphocyte ratio above 5, procalcitonin above 2ng/ml, unit increase in SOFA score and arterial lactate levels while unit decrease in Pao2/FiO2 ratio and serum albumin were associated with mortality in our patients.”

Data availability:

There is a link posted but it is not active. It looks like this is just waiting until the publication is accepted to become active or the link posted was in error. I appreciate the move to archiving and posting the data set and assume this will get sorted out if the paper is accepted. I am clicking that the data are available under this understanding of the posted inactive link.

Line 69: Change “Till date” to “To this date” or “Up to the time of writing”

Line 71: …, these data or …, these results or …, this information. But data should be a plural word (Plos one may have a policy on this).

Line 90: Data were…

Line 94: I had not run into the term “Purposeful selection…” so you need to cite the two papers here that you are using for the model selection template that are discussed in the response to reviewers. And maybe change “advanced” to “multivariable”. And “likelihood ratio test was” to “likelihood ratio tests were”. You should document the general steps you took to implement this method – it is not a well-enough known process and some of the aspects are arbitrary choices (what p-value cut-off and how the variables are considered for re-entry) and so should be documented for the reader. And here or later on you should note that no interactions among predictors were considered and results might be very different if they were included in the model refinement process.

Line 103: The rest of … and no comma after but and cut “while” and inavailability not in availability?

Line 114: Similar gender distributions were or A similar gender distribution was… in both groups in Table 1

Line 119: First mention of “SOFA score” – no definition or citation for it. It and any other acronyms need to be defined on first use with a citation for its source/definition at a minimum if not explained in the text.

Table 1 and 2: Technically the IQR is the result of calculating Q3-Q1 and a single number. I am assuming you are presenting the (first quartile – third quartile) and the table labels at least need to document that or you can actually report the IQR.

Line 133: capitalize Table

Table 3: There is a ^ symbol but I am not sure what it means as it seems to not be defined?

Table 4: For the Oxygen saturation >93%, a 1 is needed for the Crude OR to be consistent with other categorical predictor presentations. Why is NLR in parentheses and what is it – it is first mentioned on line 143 and not defined. Similarly for CURB 65 and MuLBSTA? Based on the text, it seems like you only tested these predictors versus the response – so you never explored any of the ones? Or you did but are not reporting them? The evidence weakens if you are reporting selected results here. You should document how you chose to report just these predictors.

Line 147: you never defined which predictors were included in the full model – was it all of the Table 4 ones or did you drop gender since it was the only one with a larger p-value reported from the univariate work? Or was it all of the variables discussed in Tables 1, 2, and 3? You should report what what in the initial model and (maybe not required but it would be nice) the order of the variables dropped and possibly re-added in the “purposive” selection process from the papers you mentioned. Also, you note that month was included – but was it used as a quantitative or categorical predictor – and did you see differences based on it too? I am assuming you kept it in all models regardless of its p-value, but making that clear is still needed and a reader might still be interested in changes in survival rates across months. So maybe start with including month in Table 1, 2, or 3 and then in a univariate model in Table 4 and again in Table 5. And if you clarify this early on, you do not need Line 153 with “Final model was adjusted…” – if you keep that, it should be “The final model…” or “All multivariable models were adjusted…”.

Line 154: I don’t usually think of using ROCs to get sensitivity and specificity – that is only one place in ROC at a cutoff of 0.5 for assigning as success/failure. But you can keep this wording if you feel strongly about it.

Table 5: Include the p-values for these terms? The purposive selection should mean all are less than 0.25 but you should still report the p-values that led to their retention in the model.

Line 181: Another acronym not defined (LMIC); Line 187: What is CRS and how is it defined?; Line 198: ARDS?.

Line 200: MODS was defined and that is helpful.

Line 201: Maybe a caveat should be added to these two results to note that they did not end up in the final model. So something like: …, but these predictors were not retained in our final model. And then something to compare what was different about your study/methods than in citation (25) that might have led to different results on these predictors?

Line 207: I know it is the Food and Drug Administration here, but maybe the U.S. FDA?

Line 212: HCQ is Hydroxychloroquine? Not defined again. QTc? And many more than follow – I’ll let the authors carefully check the rest of these.

Line 230: Is this a new paragraph? It is a very long paragraph if not. Maybe think about structure of all these results? Start with reviewing the three predictors retained and then revisit some other intriguing but ultimately abandoned predictors from the univariate results?

Line 259: We plan to add more subjects in the future… or We plan to add more observations in the future…

Line 269: Missing bracket in citation.

Line 279: No details on citation 8 are provided other than year and title.

Please check all other citations carefully as it seems many are missing critical details. For example, I also saw that line 363 had no details on citation 41 other than year and title.

**********

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: Gilles Guillot

Reviewer #3: Yes: Mark Greenwood

PLoS One. 2021 May 27;16(5):e0251754. doi: 10.1371/journal.pone.0251754.r004

Author response to Decision Letter 1

25 Apr 2021

A detailed reply to reviewer's comments is available in the "Response to reviewers" document.

Attachment

Submitted filename: Response to reviewers.docx

Click here for additional data file.^{(27.8KB, docx)}

PLoS One. doi: 10.1371/journal.pone.0251754.r005

Decision Letter 2

Wisit Cheungpasitporn

3 May 2021

Determinants of in-hospital mortality in COVID-19; a prospective cohort study from Pakistan

PONE-D-21-00091R2

Dear Dr. Quratulain Shaikh,

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,

Wisit Cheungpasitporn, MD

Academic Editor

PLOS ONE

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

Reviewer #4: All comments have been addressed

Reviewer #5: All comments have been addressed

**********

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

Reviewer #4: Yes

Reviewer #5: Yes

**********

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

Reviewer #4: I Don't Know

Reviewer #5: Yes

**********

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

Reviewer #4: Yes

Reviewer #5: Yes

**********

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

Reviewer #4: Yes

Reviewer #5: Yes

**********

6. Review Comments to the Author

Reviewer #4: It appears that all comments have been appropriately responded to. I have no further comments and recommend publication.

Reviewer #5: This is an interesting study with a huge number of patients and a pleasant outcome. Authors have satisfied the comments of the reviewers

**********

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 #4: Yes: Paul W Davis

Reviewer #5: No

PLoS One. doi: 10.1371/journal.pone.0251754.r006

Acceptance letter

Wisit Cheungpasitporn

19 May 2021

PONE-D-21-00091R2

Determinants of in-hospital mortality in COVID-19; a prospective cohort study from Pakistan

Dear Dr. Shaikh:

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. Wisit Cheungpasitporn

Academic Editor

PLOS ONE

Associated Data

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

Supplementary Materials

Attachment

Submitted filename: Response to reviewers.docx

Click here for additional data file.^{(39.7KB, docx)}

Attachment

Submitted filename: Response to reviewers.docx

Click here for additional data file.^{(27.8KB, docx)}

Data Availability Statement

Data file is available from the Mendeley database (DOI:10.17632/j6jz4vk6r6.2).

[pone.0251754.ref001] 1.WHO. Timeline: WHO’s COVID-19 response 2019. [Available from: https://www.who.int/emergencies/diseases/novel-coronavirus-2019/interactive-timeline/#]. [Google Scholar]

[pone.0251754.ref002] 2.Worldometers. COVID-19 Coronavirus Pandemic. 2020. [Google Scholar]

[pone.0251754.ref003] 3.Pakistan Go. Health Advisory 2020. [Available from: https://covid.gov.pk/stats/pakistan]. [Google Scholar]

[pone.0251754.ref004] 4.Banik A, Nag T, Chowdhury SR, Chatterjee R. Why Do COVID-19 Fatality Rates Differ Across Countries? An Explorative Cross-country Study Based on Select Indicators. Global Business Review. 2020:0972150920929897. [Google Scholar]

[pone.0251754.ref005] 5.Organization WH. Clinical management of COVID-19: interim guidance, 27 May 2020. World Health Organization; 2020. [Google Scholar]

[pone.0251754.ref006] 6.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. Journal of biomedical informatics. 2009;42(2):377–81. 10.1016/j.jbi.2008.08.010 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref007] 7.College Station TSL. StataCorp. 2015. Stata Statistical Software: Release 14. 2015. [Google Scholar]

[pone.0251754.ref008] 8.Bursac Z, Gauss CH, Williams DK, Hosmer DW. Purposeful selection of variables in logistic regression. Source code for biology and medicine. 2008;3(1):1–8. 10.1186/1751-0473-3-17 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref009] 9.Vincent J-L, De Mendonça A, Cantraine F, Moreno R, Takala J, Suter PM, et al. Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Critical care medicine. 1998;26(11):1793–800. 10.1097/00003246-199811000-00016 [DOI] [PubMed] [Google Scholar]

[pone.0251754.ref010] 10.Nasir N, Habib K, Khanum I, Khan N, Muhammad ZA, Mahmood SFJAP. Predictors of severity and mortality of COVID-19 at a tertiary care center in a Lower-middle income country. 2020. [Google Scholar]

[pone.0251754.ref011] 11.Tran K, Cimon K, Severn M, Pessoa-Silva CL, Conly J. Aerosol generating procedures and risk of transmission of acute respiratory infections to healthcare workers: a systematic review. PloS one. 2012;7(4):e35797. 10.1371/journal.pone.0035797 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref012] 12.Wang Y, Lu X, Li Y, Chen H, Chen T, Su N, et al. Clinical Course and Outcomes of 344 Intensive Care Patients with COVID-19. Am J Respir Crit Care Med. 2020;201(11):1430–4. 10.1164/rccm.202003-0736LE [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref013] 13.Twigg III HL, Khan SH, Perkins AJ, Roberts S, Sears C, Rahman O, et al. Mortality Rates in a Diverse Cohort of Mechanically Ventilated Patients With Novel Coronavirus in the Urban Midwest. 2020;2(8). [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref014] 14.Grasselli G, Greco M, Zanella A, Albano G, Antonelli M, Bellani G, et al. Risk factors associated with mortality among patients with COVID-19 in intensive care units in Lombardy, Italy. 2020;180(10):1345–55. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref015] 15.Audit IJA. ICNARC report on COVID-19 in critical care. 2020;10:2020. [Google Scholar]

[pone.0251754.ref016] 16.Richardson S, Hirsch JS, Narasimhan M, Crawford JM, McGinn T, Davidson KW, et al. Presenting characteristics, comorbidities, and outcomes among 5700 patients hospitalized with COVID-19 in the New York City area. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref017] 17.Yang X, Yu Y, Xu J, Shu H, Xia J, Liu H, et al. Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study. Lancet Respir Med. 2020;8(5):475–81. 10.1016/S2213-2600(20)30079-5 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref018] 18.Dondorp AM, Hayat M, Aryal D, Beane A, Schultz MJ. Respiratory Support in COVID-19 Patients, with a Focus on Resource-Limited Settings. Am J Trop Med Hyg. 2020;102(6):1191–7. 10.4269/ajtmh.20-0283 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref019] 19.Slutsky AS, Ranieri VM. Ventilator-induced lung injury. N Engl J Med. 2013;369(22):2126–36. 10.1056/NEJMra1208707 [DOI] [PubMed] [Google Scholar]

[pone.0251754.ref020] 20.Cates J, Lucero-Obusan C, Dahl RM, Schirmer P, Garg S, Oda G, et al. Risk for In-Hospital Complications Associated with COVID-19 and Influenza—Veterans Health Administration, United States, October 1, 2018–May 31, 2020. 2020;69(42):1528. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref021] 21.Lee DW, Santomasso BD, Locke FL, Ghobadi A, Turtle CJ, Brudno JN, et al. ASTCT consensus grading for cytokine release syndrome and neurologic toxicity associated with immune effector cells. Biology of Blood and Marrow Transplantation. 2019;25(4):625–38. 10.1016/j.bbmt.2018.12.758 [DOI] [PubMed] [Google Scholar]

[pone.0251754.ref022] 22.Wang D, Hu B, Hu C, Zhu F, Liu X, Zhang J, et al. Clinical characteristics of 138 hospitalized patients with 2019 novel coronavirus–infected pneumonia in Wuhan, China. 2020;323(11):1061–9. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref023] 23.WSJ. Why Youthful, Conservative Pakistan Is a Coronavirus Bright Spot 2020. [Available from: https://www.wsj.com/articles/why-youthful-conservative-pakistan-is-a-coronavirus-bright-spot-11596297600. [Google Scholar]

[pone.0251754.ref024] 24.Wu Z, McGoogan JMJJ. Characteristics of and important lessons from the coronavirus disease 2019 (COVID-19) outbreak in China: summary of a report of 72 314 cases from the Chinese Center for Disease Control and Prevention. 2020;323(13):1239–42. [DOI] [PubMed] [Google Scholar]

[pone.0251754.ref025] 25.Petrilli CM, Jones SA, Yang J, Rajagopalan H, O’Donnell L, Chernyak Y, et al. Factors associated with hospital admission and critical illness among 5279 people with coronavirus disease 2019 in New York City: prospective cohort study. 2020;369. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref026] 26.Wu C, Chen X, Cai Y, Xia J, Zhou X, Xu S, et al. Risk Factors Associated With Acute Respiratory Distress Syndrome and Death in Patients With Coronavirus Disease 2019 Pneumonia in Wuhan, China. JAMA Intern Med. 2020;180(7):934–43. 10.1001/jamainternmed.2020.0994 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref027] 27.Liu S, Yao N, Qiu Y, He C. Predictive performance of SOFA and qSOFA for in-hospital mortality in severe novel coronavirus disease. The American Journal of Emergency Medicine. 2020;38(10):2074–80. 10.1016/j.ajem.2020.07.019 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref028] 28.Yao X, Ye F, Zhang M, Cui C, Huang B, Niu P, et al. In Vitro Antiviral Activity and Projection of Optimized Dosing Design of Hydroxychloroquine for the Treatment of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2). Clin Infect Dis. 2020;71(15):732–9. 10.1093/cid/ciaa237 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref029] 29.Medicine RCGJNEJo. Effect of hydroxychloroquine in hospitalized patients with Covid-19. 2020;383(21):2030–40. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref030] 30.FDA U. Coronavirus (COVID-19) update: FDA revokes emergency use authorization for chloroquine and hydroxychloroquine. June; 2020. [Google Scholar]

[pone.0251754.ref031] 31.Qureshi S, Naveed AB, Yousafzai MT, Ahmad K, Ansari S, Lohana H, et al. Response of extensively drug resistant Salmonella Typhi to treatment with meropenem and azithromycin, in Pakistan. PLoS neglected tropical diseases. 2020;14(10):e0008682. 10.1371/journal.pntd.0008682 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref032] 32.Furtado RHM, Berwanger O, Fonseca HA, Correa TD, Ferraz LR, Lapa MG, et al. Azithromycin in addition to standard of care versus standard of care alone in the treatment of patients admitted to the hospital with severe COVID-19 in Brazil (COALITION II): a randomised clinical trial. Lancet. 2020;396(10256):959–67. 10.1016/S0140-6736(20)31862-6 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref033] 33.Cavalcanti AB, Zampieri FG, Rosa RG, Azevedo LC, Veiga VC, Avezum A, et al. Hydroxychloroquine with or without Azithromycin in Mild-to-Moderate Covid-19. 2020;383(21):2041–52. 10.1056/NEJMoa2019014 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref034] 34.Rosenberg ES, Dufort EM, Udo T, Wilberschied LA, Kumar J, Tesoriero J, et al. Association of Treatment With Hydroxychloroquine or Azithromycin With In-Hospital Mortality in Patients With COVID-19 in New York State. JAMA. 2020;323(24):2493–502. 10.1001/jama.2020.8630 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref035] 35.Rawson TM, Moore LSP, Zhu N, Ranganathan N, Skolimowska K, Gilchrist M, et al. Bacterial and Fungal Coinfection in Individuals With Coronavirus: A Rapid Review To Support COVID-19 Antimicrobial Prescribing. Clin Infect Dis. 2020;71(9):2459–68. 10.1093/cid/ciaa530 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref036] 36.Horby P, Lim WS, Emberson JR, Mafham M, Bell JL, Linsell L, et al. Dexamethasone in hospitalized patients with Covid-19-preliminary report. 2020. [Google Scholar]

[pone.0251754.ref037] 37.Sterne JA, Murthy S, Diaz JV, Slutsky AS, Villar J, Angus DC, et al. Association between administration of systemic corticosteroids and mortality among critically ill patients with COVID-19: a meta-analysis. JAMA. 2020;324(13):1330–41. 10.1001/jama.2020.17023 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref038] 38.WHO/-nCoV/Corticosteroids/.1 WHOJWrn. Corticosteroids for COVID-19. Living Guidance. 2020. [Google Scholar]

[pone.0251754.ref039] 39.Jeronimo CMP, Farias MEL, Val FFA, Sampaio VS, Alexandre MAA, Melo GC, et al. Methylprednisolone as adjunctive therapy for patients hospitalized with COVID-19 (Metcovid): a randomised, double-blind, phase IIb, placebo-controlled trial. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref040] 40.Malgie J, Schoones JW, Pijls BG. Decreased mortality in COVID-19 patients treated with Tocilizumab: a rapid systematic review and meta-analysis of observational studies. Clin Infect Dis. 2020. 10.1093/cid/ciaa1445 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref041] 41.Gupta S, Wang W, Hayek SS, Chan L, Mathews KS, Melamed ML, et al. Association between early treatment with tocilizumab and mortality among critically ill patients with COVID-19. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref042] 42.Stone JH, Frigault MJ, Serling-Boyd NJ, Fernandes AD, Harvey L, Foulkes AS, et al. Efficacy of Tocilizumab in Patients Hospitalized with Covid-19. N Engl J Med. 2020;383(24):2333–44. 10.1056/NEJMoa2028836 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref043] 43.Sarfaraz S, Shaikh Q, Iftikhar S, Herekar FF, Saleem SG, Kanani F. Is Tocilizumab an effective therapy for Severe COVID-19: a retrospective cohort study. 2020. [DOI] [PubMed] [Google Scholar]

[pone.0251754.ref044] 44.Yang A-P, Liu J-P, Tao W-Q, Li H-m. The diagnostic and predictive role of NLR, d-NLR and PLR in COVID-19 patients. International immunopharmacology. 2020;84:106504. 10.1016/j.intimp.2020.106504 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref045] 45.Nalbant A, Kaya T, Varim C, Yaylaci S, Tamer A, Cinemre H. Can the neutrophil/lymphocyte ratio (NLR) have a role in the diagnosis of coronavirus 2019 disease (COVID-19)? Revista da Associação Médica Brasileira. 2020;66(6):746–51. 10.1590/1806-9282.66.6.746 [DOI] [PubMed] [Google Scholar]

[pone.0251754.ref046] 46.Santus P, Radovanovic D, Saderi L, Marino P, Cogliati C, De Filippis G, et al. Severity of respiratory failure at admission and in-hospital mortality in patients with COVID-19: a prospective observational multicentre study. BMJ open. 2020;10(10):e043651. 10.1136/bmjopen-2020-043651 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pone.0251754.ref047] 47.de la Rica R, Borges M, Aranda M, Del Castillo A, Socias A, Payeras A, et al. Low albumin levels are associated with poorer outcomes in a case series of COVID-19 patients in Spain: a retrospective cohort study. Microorganisms. 2020;8(8):1106. [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Determinants of in-hospital mortality in COVID-19; a prospective cohort study from Pakistan

Samreen Sarfaraz

Quratulain Shaikh

Syed Ghazanfar Saleem

Anum Rahim

Fivzia Farooq Herekar

Samina Junejo

Aneela Hussain

Roles

Abstract

Introduction

Methods

Results

Fig 1. Flow and outcomes of COVID-19 suspects at the Indus Hospital Emergency (19th March-7th June 2020).

Clinical presentation

Table 1. Clinical characteristics of study participants by mortality.

Table 2. Baseline laboratory parameters by survival status.

Hospital course and outcome

Table 3. Hospital course by mortality.

Predictors of in-hospital all-cause mortality

Table 4. Univariable binary logistic regression for predictors of in-hospital all-cause mortality.

Table 5. Multivariable Logistic regression model of all-cause mortality in COVID-19 patients.

Discussion

Acknowledgments

Data Availability

Funding Statement

References

Decision Letter 0

Wisit Cheungpasitporn

Roles

Author response to Decision Letter 0

Decision Letter 1

Wisit Cheungpasitporn

Roles

Author response to Decision Letter 1

Decision Letter 2

Wisit Cheungpasitporn

Roles

Acceptance letter

Wisit Cheungpasitporn

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

Fig 1. Flow and outcomes of COVID-19 suspects at the Indus Hospital Emergency (19^th March-7^th June 2020).