Characteristics of mechanically ventilated COVID-19 patients in the Al-Ahsa Region of Saudi Arabia: a retrospective study with survival analysis.

BACKGROUND: About 5-10% of coronavirus disease 2019 (COVID-19) infected patients require critical care hospitalization and a variety of respiratory support, including invasive mechanical ventilation. Several nationwide studies from Saudi Arabia have identified common comorbidities but none were focused on mechanically ventilated patients in the Al-Ahsa region of Saudi Arabia.
OBJECTIVES: Identify characteristics and risk factors for mortality in mechanically ventilated COVID-19 patients.
DESIGN: Retrospective chart review
SETTING: Two general hospitals in the Al-Ahsa region of Saudi Arabia PATIENTS AND METHODS: We included mechanically ventilated COVID-19 patients (>18 years old) admitted between 1 May and 30 November 2020, in two major general hospitals in the Al-Ahsa region, Saudi Arabia. Descriptive statistics were used to characterize patients. A multivariable Cox proportional hazards (CPH) model was used exploratively to identify hazard ratios (HR) of predictors of mortality. MAIN OUTCOME MEASURES: Patient characteristics, mortality rate, extubation rate, the need for re-intubation and clinical complications during hospitalization. SAMPLE SIZE AND CHARACTERISTICS: 154 mechanically ventilated COVID-19 patients with median (interquartile range) age of 60 (22) years; 65.6% male.
RESULTS: Common comorbidities were diabetes (72.2%), hypertension (67%), cardiovascular disease (14.9%) and chronic kidney disease (CKD) (14.3%). In the multivariable CPH model, age >60 years old (HR=1.83, 95% CI 1.2-2.7, P=.002), CKD (1.61, 95% CI 0.9-2.6, P=.062), insulin use (HR=0.65, 95% CI 0.35-.08, P<.001), and use of loop diuretics (HR=0.51, 95% CI 0.4, P=.037) were major predictors of mortality.
CONCLUSION: Common diseases in mechanically ventilated COVID-19 patients from the Al-Ahsa region were diabetes, hypertension, other cardiovascular diseases, and CKD in this exploratory analysis. LIMITATIONS: Retrospective, weak CPH model performance. CONFLICTS OF INTEREST: None.

INTRODUCTION

In just two and a half months since the first clusters of pneumonia cases were reported in Wuhan, China, the World Health Organization (WHO) announced the newly discovered coronavirus disease 2019 (COVID-19) as a global pandemic.[1] The causative agent, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), belongs to a large group of viruses called coronaviruses, commonly found in animals and humans. It can cause mild symptoms like rhinorrhea, headache, fever, cough, and sore throat that last for several days in immunocompetent people and can progress to symptoms of lower respiratory infections like acute bronchitis or pneumonia.[2] The most prevalent symptoms of this viral infection are fever, cough, and dyspnea, with a typical incubation period of 5 days and an illness duration of more than 8 days.[3,4] Males are more at risk than females and one-third of patients having comorbidities such as hypertension, cardiovascular diseases, and diabetes.[5]

It is estimated that 20% of hospitalized patients require admission to an intensive care unit (ICU), with a case fatality rate (CFR) of over 13%.[4] Common complications from the infection include acute respiratory distress syndrome (ARDS), acute cardiac injury, acute kidney injury, disseminated intravascular coagulation (DIC) and sepsis. Additionally, in-hospital death is greater in elderly patients, males, and patients with co-morbidities.[6] Several studies from different countries have identified independent risk factors as being associated with higher mortality, like age and male sex.[7] However, the rank of comorbidities, which harbor the highest risk for being intubation and death, have varied among critically ill patients (hypertension,[8] heart failure, body mass index greater than 40, diabetes,[9-11] coronary artery disease, cerebrovascular disease,[12] smoking, chronic obstructive pulmonary disease,[13] and simultaneous occurrence of three or more comorbidities).[14] Several nationwide studies from Saudi Arabia have identified common comorbidities in intensive care patients (hypertension, diabetes, ischemic heart disease).[15,16]

In this study, we examined the clinical characteristics of COVID-19 mechanically ventilated patients admitted to two major hospitals in the Al-Ahsa region. We used survival analysis exploratively to investigate predictors of mortality. The findings of this study may be useful to local practitioners who require such information in this population.

METHODS

This was a retrospective study conducted via medical record review. Patients were included if they were 18 years of age or older, had a COVID-19 diagnosis confirmed with SARS CoV-2 by real time reverse transcriptase polymerase chain reaction, and were admitted to King Fahad Hospital Hofuf (KFH) and Prince Saud Bin Jalawi Hospital (PSBJH) between 1 May 2020, and 30 November 2020. Only COVID-19 patients admitted to the ICU or inpatient ward with a mechanical ventilator were included. The list of critical cases was obtained from medical records. We excluded patients who died or recovered before being connected to a mechanical ventilator during their hospital stay as well as those who had incomplete files. This study was approved by both King Faisal University (IRB# 2020-05-10) and King Fahad Hospital Hofuf (IRB# H-05-H5-065) Institutional Review Board Human Subjects Protection Programs.

Demographic characteristics and other clinical variables such as comorbid conditions, medication used during hospitalization, symptoms at presentation, COVID-19 regimen used and clinical complications were collected from the electronic and non-electronic medical records and managed using Research Electronic Data Capture (REDCap) software.[17] The patient list was obtained from the medical records staff. There was no formal sample size calculation. We screened all patient charts at our disposal using convenience sampling.[18] The primary outcome was to identify the characteristics of patients that required mechanical ventilation in the Al-Ahsa region. We report the mortality rate, extubation rate, need for re-intubation and clinical complications during hospitalization. We also explored predictors of mortality.

We present continuous data as median with interquartile range (IQR). Categorical data are presented as frequencies and percentages. We used survival analysis by fitting a Kaplan-Meier (KM) model for the mortality outcome. Two KM models were fitted: one when time 0 was the time of admission, and the second when time 0 was initiation of mechanical ventilation. We censored the data at the time of follow up if no event occurred (i.e., patient discharged alive). We plotted survival probability indicating the median time of survival. We then implemented univariable Cox proportional hazards (CPH), a semiparametric model, to identify important and pertinent clinical predictors of mortality. The results of the univariable analysis guided our approach to narrowing down important variables to be included in the multivariable CPH, including the use of clinical importance in our modeling. Starting with a full model, we implemented backward elimination to obtain the best model based on the Akaike information criterion (AIC). Some variables were retained in the final model despite their non-significance due to improving the overall model fit. The proportional hazard assumption was tested using Schoenfeld residuals, with plots and statistical tests of independence between residuals and time. The time-varying effect assumption was tested using the “timereg” package, which uses a resampling method with 500 simulations to test time-varying effects.[19] The package provides two statistical tests for time-varying coefficients (the Kolmogorov-Smirnov test and the Cramér–von Mises criterion). We checked Martingale residuals for the linearity assumption. In case of violation, the variable was transformed into a categorical variable. The variables included in the final model were reported with hazards ratios along with 95% confidence intervals (CI). We quantified the predictive ability of the model using the index of discrimination (C index), the concordance between the predicted and observed survival.[20] This was calculated from Kendall's τ and Somers’ rank correlation (Dxy). C index values close to 1 indicate excellent prediction ability, and values <0.5 indicate the model is not better than chance. To internally validate the model, a bootstrapping bias corrected (overfitting corrected) estimate of the C index was obtained.[21] We emphasize that modeling survival in this cohort was exploratory and not meant to inform clinical practice. All statistical analyses were performed using the R Core Team (2020) software (R Foundation for Statistical Computing, Version 4.0.1, Vienna, Austria). The following R packages were used in our analyses: survival, survminer, ggplot2, timereg, MASS and rms.[22-24]

RESULTS

Of 1976 patients admitted for COVID-19 in the two hospitals during the study period only 167 patients were identified as critical cases, of which 154 patients were included in our analysis. Twelve were excluded for various reasons (). The majority (72%) were admitted to KFH, and most were from the Eastern region of Saudi Arabia (77.1%). The median age (IQR) was 60 (22) years old (range, 49 to 71 years). Male patients represented 65.6% of the cohort. Diabetes (72.7%) and hypertension (67%) were the most prevalent comorbidities, followed by other cardiovascular diseases (14.9%) and chronic kidney disease (CKD) (14.3%) (). All patients with diabetes were using insulin, either alone or in combination with other medications. Acetaminophen (74%) and loop diuretics (59.7%) were commonly used. Presenting symptoms and regimen used for COVID-19 management are listed in . Four among 154 cases had hematological diseases, including two with sickle cell disease patients (SCD), one had both SCD and glucose 6 phosphate dehydrogenase deficiency, (G6PD) and one case had only G6PD. Most patients died (n=138, 89.6%) (). Thirteen (8.4%) needed reintubation. Twenty (13%) were fully extubated. The most common complication was ARDS (89%) followed by acute kidney injury (38.3%), sepsis (33.1%) and multiorgan failure (20.8%).

Patient selection flowchart.

The median survival of the cohort was 13 days (95% CI:11 to 16) and the median survival after mechanical ventilation was 8 days (95% CI: 6 to 9). The 50% probability of survival was at day 13 (). From the date of mechanical ventilation the 50% probability of survival was at day 8 (). Factors associated with mortality were age greater than 60 years old and chronic kidney disease (). Factors associated with a decrease in mortality were insulin use and loop diuretic use. The model had a corrected C index of 0.59 (weak performance). The KM curves for survival for these variables are plotted in ().

Survival probability date of admission (dashed line is median time; shadowed areas are confidence intervals).

Survival probability from date of starting mechanical ventilation (dashed line is median time; shadowed areas are confidence intervals).

Survival probability by age group (dashed line is median time; shadowed areas are 95% confidence intervals).

Survival probability by chronic kidney disease (dashed line is median time; shadowed areas are 95% confidence intervals).

Survival probability by insulin use (dashed line is median time; shadowed areas are 95% confidence intervals).

Survival probability by use of loop diuretics (dashed line is median time; shadowed areas are 95% confidence intervals).

DISCUSSION

The aim of our study was to understand the characteristics of mechanically ventilated COVID-19 patients who were admitted to two of the tertiary care hospitals in Saudi Arabia's Al-Ahsa region. The study included mechanically ventilated COVID-19 patients with a median age of 60 years old. Advanced age is a known risk factor for critical illness in COVID-19 patients.[25-28] Our study cohort were both advanced-aged and critically ill, which mandated invasive mechanical ventilation, and hence the higher overall mortality rate in our cohort. The study by Khan and Alruthia et al[29] from Saudi Arabia found that mechanical ventilated patients have five times the risk of death compared to non-mechanically ventilated patients. The Khan and Alruthia study was conducted using the Health Electronic Surveillance Network (HESN) database from the Saudi Ministry of Health (MoH) for COVID-19 patients. The majority of the patients in their study were from Madina (68.8%); only 82 patients (5.8%) of the patients were from Al-Ahsa. Their results were comparable to our cohort. The survival probability of the patients in the Khan et al study was estimated to be less than 50% after day 10 of admission. Although not emphasized in the report, the overall mortality in mechanically ventilated patients was greater than 80% at the end of the follow up. Another nationwide study by Almutair et al conducted in 20 tertiary hospitals in Saudi Arabia examined critically ill COVID-19 patients admitted to the intensive care units. As in our study, diabetes (52.4%) and hypertension (46%) were the most common comorbidities. In line with our findings, the mortality rate among the recipients of mechanical ventilation was high at 82.7%.[15] A large study by Alhumaid et al[30] reported clinical features and prognostic factors of 1014 COVID-19 patients from six hospitals in Al-Ahsa between 1 March 2020 and 30 July 2020, a 2-month overlap with our study and found the same common comorbidities as in our study. However, Alhumaid et al did not report mortality rates among patients on mechanical ventilation.

A study from Mexico reported high mortality at 73.7% among the mechanically ventilated COVID-19 population.[31] On the other hand, a Spanish study reported a better survival probability in mechanically ventilated COVID-19 patients (>50% after 180 days of follow up).[32] The meta-analysis of 69 studies by Lim et al reported a CFR of 45% (95% CI: 39 to 52%) among mechanically ventilated COVID-19.[33] The difference in CFRs reported in the literature might be related to quality of care, setting and/or regional differences. The discussion on how to improve mechanically ventilated outcomes has already started in Saudi Arabia. The National Approach to Standardize and Improve Mechanical Ventilation (NASAM) project was initiated in 2019 with a goal to improve patient care by applying evidence-based practices.[34] Although the project did not include mechanically ventilated COVID-19 patients prior to the pandemic, improvement of mechanical ventilation was the goal of this project. A 2-year study (2016 to 2018) by Ismaeil et al from Saudi Arabian hospitals reported an overall mortality of 37% among adults who were mechanically ventilated,[35] which is lower than in studies of COVID-19 patients in Saudi Arabia.[15]

The impact of chronic diseases on COVID-19 prognosis is well established.[36,37] Conditions such as heart disease, diabetes, cancer, chronic renal disease, and obesity raise the risk of COVID-19-related severe illness.[38] Consistent with earlier findings, common comorbid conditions in our mechanically ventilated cohort were diabetes (72.7%), hypertension (61.7%), cardiac disease (14.9%) and CKD (14.3%). This indicates that conditions such as diabetes and hypertension are common risk factors in COVID-19 patients undergoing mechanical ventilation.[39,40] Our exploratory analysis confirms advanced age and CKD as independent risk factors for death.

As we limited our analysis to mechanical ventilation, we found a few findings similar to findings from the study by Alhumaid et al.[30] For example, the use of favipiravir was high in our cohort (64.9%) vs their report of 54.6% in the ICU setting. We report higher use of tocilizumab (50% vs 29.7%) dexamethasone (50% vs 34.8%) and lopinavir/ritonavir (39% vs 16.1%). Lastly, the exploratory analysis revealed unexpected findings. Insulin use was a predictor of lower mortality in our multivariable model. This finding does not agree with that in the published literature. The meta-analysis by Yang et al[41] found that insulin use was associated with a two-fold increase in the risk of mortality (odds ratio=2.10; 95% CI: 1.51-2.93). In addition to the wide confidence interval in the multivariable CPH model for this factor (HR=0.65, 95% CI 0.439-0.975, P=.037) the KM model did not reveal any significance in survival () in our cohort (P=.24). This would cast doubt on the variable being a significant predictor. The addition of more important variables in the ICU settings such as oxygenation level, severity scores and hemoglobin A1c may render this variable unimportant.[42] This incidental finding may be the result of not adjusting for important confounders. The use of explanatory modeling that can establish causal effects may be better than using prediction modeling in this case.[43] On the other hand, a mouse study showed that insulin therapy is associated with downregulation of ACE2 receptors.[44] Whether this effect can result in clinical benefits for COVID-19 patients is unknown. The use of loop diuretics as a predictor of low mortality was interesting as well. Mechanistically and in addition to its diuretic effects, furosemide may have an effect on dyspnea due to its effect on the release of pro-inflammatory cytokines (interleukin-9, IL-8 and tumor necrosis factor.[45] The role of furosemide as a potential COVID-19 treatment is being investigated.[46] Our study also reported a high hazard ratio for patients having CKDs who must undergo mechanical ventilation. This confirms earlier findings which reported a high ICU admission and mortality rate among renally ill UK COVID-19 patients, indicating that CKD is a risk factor for in-hospital mortality in COVID-19 patients.[47]

Although the study's principal objectives were met, there are several limitations that must be addressed. Due to the retrospective nature of the study, there is a risk of bias as we relied solely on the records accessible at the hospital. Manual collection of data was difficult. The availability of electronic medical record systems would be beneficial for future projects. Mechanical ventilation takes on medical wards and the ICU, but ICU beds may be unavailable. We did not explore whether the site of ventilation had an effect on mortality. Moreover, the impact of vaccination status cannot be explored as vaccination was unavailable during the study period. Exploratory CPH analysis was not informative due to the weak performance of the model. Enhancing the performance of the predictability of the model would require additional variables and a larger sample size. Considering powerful variables such as Acute Physiology and Chronic Health Evaluation II (APACHE-II) or Sequential Organ Failure Assessment score (SOFA) might enhance the predictability of our model.

In conclusion, advanced age and CKDs were independent risk factors for the need of mechanical ventilation in COVID-19 patients. A high prevalence of hypertension and diabetes mellitus were also found among our study population. Further, large-scale collaborative studies involving a number of regions of Saudi Arabia will help in determining the characteristics of COVID-19 patients that need mechanical ventilation, which may effectively help in controlling the infection and its devastating impact in our country.

Table 1.

Baseline characteristics of mechanically ventilated COVID-19 patients (n=154).

King Fahad Hospital Hofuf	111 (72.07)
Prince Saudi Bin Jalawi Hospital	43 (27.9)
Age (years) (nnedian, IQR, range)	60 (22), 49–71
Female	53 (34.4)
Ethnicity
Middle eastern	120 (77.1)
South/Southeast or Central Asian	32 (20.7)
African	2 (1.3)
Weight (kg) (median, IQR, range)	79 (23.5), 41–168
Comorbidities
Diabetes	112 (72.7)
Hypertension	95 (61.7)
Established cardiovascular disease[a]	23 (14.9)
Chronic kidney disease	22 (14.3)
Asthma	9 (5.8)
Atrial fibrillation	8 (5.2)
Chronic obstructive pulmonary disease	4. (2.6)
Cancer	3 (1.9)
Sickle cell disease	3 (1.9)
Stroke	2 (1.3)
Dyslipidemia	2 (1.3)
Glucose-6-phosphate-dehydrogenase deficiency	2 (1.3)
Venous thromboembolism	1 (0.6)
Autoimmune disease	1 (0.6)
Medications
Acetaminophen	114 (74.0)
Insulin	112 (72.7)
Loop diuretics	92 (59.7)
Calcium channel blocker	61 (39.6)
Statin	44 (28.6)
Aspirin	41 (26.6)
Beta-blocker	27 (17.5)
Angiotensin converting enzyme inhibitor	12 (7.8)
Thiazide diuretics	6 (3.9)
Angiotensin Receptor Blocker	5 (3.2)
Non-steroidal anti-inflammatory drug	2 (1.3)
Symptoms at presentation
Dyspnea	122 (79.2)
Fever	97 (63.0)
Cough	93 (60.4)
Diarrhea	6 (3.9)
COVID-19 regimen
Dexamethasone	109 (70.8)
Favlpiravir	100 (64.9)
Azithromycin alone	77 (50.0)
Tocillzumab	77 (50.0)
Lopinavir/rltonavir	60 (39.0)
Hydroxychloroquine/azithromycin	29 (18.8)
Hydroxychloroquine alone	17 (11.0)

Data are n (%) unless noted otherwise.

Established cardiovascular disease: a documented history of stable angina, unstable angina, percutaneous coronary intervention (PCI), coronary artery bypass graft surgery, or myocardial infarction (Ml). Heart failure and cerebrovascular disease included transient ischemic attack (TIA) or stroke.

Table 2.

Clinical outcomes and complications (n=154).

Clinical outcomes
Death	138 (89.6)
Fully extubated	20 (13)
Need for re-intubation	13 (8.4)
Discharge alive	16 (10.4)
Clinical connplications
Acute respiratory distress syndronne	137 (89.0)
Acute kidney injury	59 (38.3)
Sepsis	51 (33.1)
Multi-organ failure	32 (20.8)
Pulmonary embolism	4 (2.6)
Cardiac injury	4 (2.6)
Disseminated intravascular coagulation	2 (1.3)
Stroke	2 (1.3)

Data are n (%)

Table 3.

Multivariable Cox proportional hazards model and survival analysis.

Variable	Hazard ratio (95% Cl), P value	Median survival in days (95% Cl), P value
Age > 60	1.83 (1.239–2.706), P=.002	7 (5–9) vs 9 (7–13), P=.034
Chronic kidney disease	1.61 (0.977–2.638), P=.062	7 (4–11) vs 8 (6–10), P=.099
Insulin use	0.65 (0.439–0.975), P=.037	8 (7–10) vs 7 (5–9), P=.24
Loop diuretics use	0.51 (0.348–0.766). P<.001	9 (7–12) vs 6 (4–10), P=.035

Backward elimination selection process methodology to select the best predictors for mortality (started from full model). Median survival calculated from a Kaplan-Meier model. Log rank test for P value. Time (0) mechanical ventilation day. Corrected C index: 0.59