The coronavirus disease SARS-CoV-2 (COVID-19) was reported for the first time in China (Wuhan) in December 2019. After a few months, it became a global emergency, as more than 245 million positive cases were affected and 45 million deaths were caused worldwide, with a mortality rate of approximately 2%. ¹ This disease was classified as a pandemic by the World Health Organization (WHO) in March 2020.

Tunisia is one of the most affected countries with a high rate of patients with acute respiratory syndrome coronavirus. In fact, 1145686 cases and 29249 deaths were reported as of September 2022. ²

The clinical features of COVID-19 range from asymptomatic to critical illness and death. Some cases may experience respiratory failure that requires mechanical ventilation and organ failure requiring admission to the intensive care unit (ICU). A literature review in 2020 showed that clinical predictors of disease severity ranged from demographic factors (such as ‘older age’), comorbidities (such as cardiovascular disease, chronic kidney disease, chronic lung diseases (particularly COPD), diabetes mellitus, hypertension, immunosuppression and obesity) and severe hypoxemia. ³ A good understanding of these possible clinical factors associated with COVID-19 severity is helpful for clinicians in identifying patients who are at high risk and require prioritized treatment to prevent disease progression and adverse outcomes.

Thus, this study was conducted to identify independent predictive factors for hospital mortality of COVID-19 patients in a Tunisian tertiary care center from September 2020 to September 2022.

This study represents one of the first large case studies with confirmed COVID-19 cases at a regional level and provides the most relevant novelty about the region of Sousse.

In the present work, data from hospitalized COVID-19-positive patients in a Tunisian university hospital were analyzed from the first case hospitalization.

This study discusses independent predictive factors for hospital mortality of COVID-19 patients in a Tunisian tertiary care center since September 2020. Many studies were conducted in the same context, but the study period is different, as the majority analyzed data of some months or a maximum of one year, while our study included a longest period. ^{12– 14}

To identify the main predictive factors affecting the survival probability for COVID-19, survival analysis methods were used. Such analytic models can work with specific problems, with a data type that waits for the event to occur. In our study, the event mentioned was ‘death.’ We used Cox regression, which is the most appropriate method to analyze survival.

Data extracted from methods with survival analysis are robust. ¹⁵

These data are helpful for making the right decision about each patient’s case according to the available treatment and medical instruments. Thus, this work will contribute at different levels, such as finding the survival probability for each patient, finding the effect of each feature on survival probability by calculating p value for each feature, and finally enhancing the COVID-19 system. ¹⁰

Many other works have studied clinical characteristics, outcomes, and predictive factors for mortality in hospitalized patients with COVID-19, ^{16– 19} but few of them used survival analysis.

Moreover, our study focuses on the positive effect of vaccination against death and length of stay. Thus, the insights gained from this study have the potential to enhance people’s reassurance, thereby addressing vaccine hesitancy and dispelling conspiracy beliefs.

On the other hand, this study involved a large sample size that is needed to conflict with valid studies and representative results.

One of the major strengths of our work was the exhaustivity of our data. Patients were enrolled and registered upon their hospitalization, and all their information was directly collected. This significantly minimized, or even eliminated, the bias of loss to follow-up in our study.

From the first COVID-19 hospitalization in September 2020 until September 2022, 1978, hospitalizations due to COVID-19 were followed with a sex ratio of 1.2. The mean age and the median length of stay were 55.65± 21.39 years and 8 days, respectively. The most common comorbidity among the population was hypertension (42.7%).

A similar study during 2020 in Sofia Bulgaria included 138 hospitalized COVID-19 patients. The mean age was 52.9 years, and males were the dominant sex (sex ratio: = 1.6). The mean age was significantly different compared to other studies. ²⁰

Another study in early 2020 showed a higher mean age (65.5 years). ²¹

In other studies, the mean age was 70.5 (±15.7, 30–101) years.

This difference can be related to the study population or the specific populations on which the research was conducted. Some studies have focused on a specific comorbidity in hospitalized COVID-19 patients, such as the characteristics of COVID-19 among hospitalized patients suffering from diabetes. ²²

The current findings regarding the sex ratio align with emerging research that indicates a higher susceptibility for severe outcomes, including death, in men with COVID-19 compared to women. ²³

Hypertension as the most frequent comorbidity was similar to the findings of many previous studies ²⁴

In our study, the reported death rate varied from 9.7% in November 2020 to 21.1% in September 2022, which is close to the fatality rate found in a study in Mexico and the United States (12%). ^{25, 26} However, this rate was lower than the reported death rate in an Italian hospital (33.3%). ²⁷ This can be justified, as Italy has registered the most important mortality rates in the world, especially in the first period of the disease. It can also be justified due to differences in thresholds for hospitalization. A third possible explanation for variation in country/regional-specific case-fatality rates is the different strategies used for COVID-19 PCR testing. ²⁸ Additionally, mortality rate differences can relate to the types of hospitals (public or private hospitals), and this can be explained by the difference in socioeconomic profiles. ²⁹

In our study, the median survival time was 30 days (95% CI [23.9; 36.1]). At the end of the follow-up, 417 deaths were recorded. These findings came online with the results of a similar study.

The median survival time was 12 days (95% CI 11.82–12.18). At the end of the follow-up, 21 408 deaths were recorded. ³⁰

The overall survival rate in our study (78.9%) was higher than that in a study by Bonfada et al., which analyzed the survival of elderly patients hospitalized in the ICU (66.64%). ³¹

The median survival time was close to that reported in Brazil (19 days) and higher than that reported in Ethiopia (9 days), the USA (10 days) and Mexico (9 days). ^{30, 32, 33}

When searching for the different predictive factors in the ‘Cox’ model, we found similar studies using the survival multilevel analysis conducted in Brazil, India, South India, Sweden and China. ^{30, 33– 38}

Our study identified three statistically significant covariates sex, hypertension, and ICU admission - in our Cox proportional hazards model, all of which were separately identified in previous studies. Notably, gender’s significance in our model aligns with the study from Sweden, hypertension’s relevance matches findings from the study in Wuhan, and ICU admission’s impact mirrors conclusions drawn in the Brazilian study. These consistent results across different studies underscore the importance of these covariates in predicting the risk of death in our research context.

Concerning sex-related factors, our analysis has confirmed a higher risk for male patients compared to females. The different prevalence of infection among patients according to sex was mentioned in other studies. ²⁶

The reduced susceptibility of females to viral infections may be attributed to their inherent protection from factors associated with the X chromosome and sex hormones, both of which play crucial roles in both innate and adaptive immunity. ³⁹ Notably, SARS-CoV-2, the virus responsible for COVID-19, utilizes ACE2 receptors located on pulmonary endothelial cells for entry, and it is worth noting that the ACE2 gene is situated on the X chromosome. ⁴⁰ This genetic aspect could explain the higher prevalence of COVID-19 in men than in women. Additionally, in premenopausal women, the ovaries produce estradiol. However, after menopause, the ovaries significantly decrease estradiol production, equating with men.

ICU admission emerged as a crucial factor strongly associated with 80.1% of deaths among confirmed cases of COVID-19.

These results were in line with other studies that concluded that the ICU was a risk factor for disease severity, especially among men, as men admitted to the ICU were 5.42 and 1.39 times more likely to die, respectively, than those who did not enter these units. ⁴¹

Other covariates that demonstrated statistical significance in various separate studies were also found to be statistically significant in our univariate survival analysis. These covariates include age, ⁴² comorbidities, ⁴³ diabetes, endocrine diseases, obesity, and the need for invasive ventilatory support. The present study confirmed that age above 65 years was significantly associated with death in patients with COVID-19 in univariate survival analysis (52.8% vs 47.2%; p<10-3).

In a similar study using survival analysis, it was found that the hazard ratio increased with increasing age. For instance, in the group aged above 65 years, the patient is at 39.3 times higher risk of dying overall than a patient in a younger age group. ⁴⁴

The significant comorbidities associated with mortality in our univariate analysis were hypertension, diabetes, endocrine diseases, and obesity. Similarly, comorbidities found in a Mexican study were hypertension, diabetes and obesity in COVID-19–positive patients. ²⁵

Our findings can be explained by the high prevalence of diabetes, hypertension, and obesity in Tunisia. On the other hand, comorbidities are the main risk factor for hospitalization. Thus, the highest prevalence of these hospitalized patients already suffers from a comorbidity, and more importantly, the prevalence of hypertension in positive patients was higher than the population mean. ⁴⁵

This observation could be elucidated by considering the detrimental impacts of comorbidities on the body’s autoimmune response and its capacity to manage metabolic stress, which are characteristic features of systemic diseases. Comorbidities, such as diabetes, hypertension, and cardiovascular diseases, often introduce a state of chronic inflammation and metabolic dysfunction within the body. These conditions can compromise the immune system’s ability to mount an effective defense against pathogenic agents, including viruses such as SARS-CoV-2. In essence, the presence of comorbidities not only weakens the body’s innate and adaptive immune responses but also creates an environment that is more conducive to the replication and spread of pathogens, potentially contributing to the increased susceptibility and severity of diseases such as COVID-19 in individuals with underlying health conditions. ⁴⁶

Even studies that did not use survival analysis showed that older patients, men, and those with preexisting comorbidities such as hypertension, diabetes, cardiac diseases, renal diseases, respiratory diseases, neurological diseases and other health problems were the highly prevalent characteristics in our study. This comes in line with previous records in the literature, where these characteristics were associated with higher mortality rates and greater risk of negative health outcomes. ^{10, 15, 16}

Tunisia was one of the most affected countries with COVID-19 disease until vaccination became a potential resolution of the threat.

Our study aimed to analyze the survival of patients admitted to a Tunisian tertiary care center (Sahloul University Hospital) to identify independent predictive factors for hospital COVID-19 mortality.

This work is the first large case study with confirmed COVID-19 cases at a regional level and provides the most relevant novelty about the region of Sousse.

The survival analysis made this work unique thanks to the robust results that it produces. This analysis showed three independent predictive factors associated with higher COVID-19 mortality: ICU hospitalization, hypertension, and sex.

The corresponding author attests that all listed authors meet the journals authorship criteria and that no others meeting the criteria have been omitted.

Conceptualization, A. N, H.G; Methodology, H. G and A.N.; Software, H. G, A.N., and R.B.; Validation, H.S.L, and M.B.R; Formal Analysis, A.N.; Investigation A. N, H.G and R.B.; Data Curation A. N, R.B. and H.G. Writing—Original Draft Preparation, A.N.; Writing—Review and Editing, H. G, B. R, M.B.R and H.S.L.; Visualization B. S, H.B.H, B. K and A.B.C; Supervision M. K, H.G, W. N, H.T; Project Administration H. T, H.S.L, W. N and M.K. All authors approve the final manuscript submitted and agree to be personally accountable for the work.