Survival of hospitalised COVID-19 patients in Hawassa, Ethiopia: a cohort study

Abstract

The COVID-19 pandemic, caused by SARS-CoV-2, led to 622,119,701 reported cases and 6,546,118 deaths. Most studies on COVID-19 patients in hospitals are from high-income countries, lacking data for developing countries such as Ethiopia.This study assesses clinical features, demographics, and risk factors for in-hospital mortality in Hawassa, Ethiopia. The research cohort comprises 804 cases exhibiting clinical diagnoses and/or radiological findings and indicative of symptoms consistent with COVID-19 at Hawassa University Comprehensive Specialized Hospital from September 24, 2020, to November 26, 2021. In-hospital mortality rate was predicted using Cox regression. The median age was 45 years, with males making up 64.1% of the population. 173 (21.5%) fatalities occurred, with 125 (72.3%) among males. Male patients had higher mortality rates than females. Severe and critical cases were 24% and 21%. 49.1% had at least one comorbidity, with 12.6% having multiple. Common comorbidities were diabetes (15.9%) and hypertension (15.2%). The Cox regression in Ethiopian COVID-19 patients found that factors like gender, advanced age group, disease severity, symptoms upon admission, shortness of breath, sore throat, body weakness, hypertension, diabetes, multiple comorbidities, and prior health facility visits increased the risk of COVID-19 death, similar to high-income nations. However, in Ethiopia, COVID-19 patients were young and economically active. Patients with at least one symptom had reduced death risk. As a conclusion, COVID-19 in Ethiopia mainly affected the younger demographic, particularly economically active individuals. Early detection can reduce the risk of mortality. Prompt medical attention is essential, especially for individuals with comorbidities. Further research needed on diabetes and hypertension management to reduce mortality risk. Risk factors identified at admission play a crucial role in guiding clinical decisions for intensive monitoring and treatment. Broader risk indicators help prioritize patients for allocation of hospital resources, especially in regions with limited medical facilities. Government’s focus on timely testing and strict adherence to regulations crucial for reducing economic impact.

Introduction

The COVID-19 pandemic has had significant impact on global health, leading to widespread morbidity and mortality. Global mortality has already exceeded 6.7 million [1]. Major risk factors for severe outcomes from COVID-19 include elevated age, obesity,underlying cardiometabolic disease and weakened immune systems. Infants, pregnant women and smokers are also at increased risk [2].

Although the infection burden is highest in Europe, the Americas, and the West Pacific [3], COVID-19 pandemic has posed significant challenges in Africa, especially in highly populous countries like Ethiopia, where many nations lack resources including skilled personnel, financial support, and logistical capabilities to control the prevalence of the pandemic. Ethiopia is one of the most densely populated countries in Africa with a projected population of 115 million and an annual population growth rate of 2.6% [4].

In Ethiopia, COVID-19 was first detected in March 2020 in the capital city, Addis Ababa, immediately followed by outbreaks reported in various regional cities. As of February 2023, nearly 500,000 cases have been documented and more than 7500 deaths reported [1, 5]. The actual number of cases and deaths are likely much higher due to limited testing capacity and under reporting.

Studies of hospitalised COVID-19 patients in Ethiopia are needed due to the particular challenges faces by the healthcare system in Ethiopia as well as the marked differences in the age structure and prevalence of known risk factors. In Ethiopia, the COVID-19 pandemic has further strained a healthcare system already facing challenges, including limited infrastructure, healthcare worker shortages and inadequate resources [6]. Encyclopaedia Britannica notes that although the majority of hospitals in Ethiopia are in Addis Abeba, the country’s healthcare system is continuously beset by problems with equipment and drug shortages [7]. In addition, hospitals with full-time doctors are only located in major cities [7]. However, compared to Western nations like the USA, Ethiopia has a different distribution of risk variables for catastrophic COVID-19 results. With a median age of about 19 years (as opposed to a global average of 30.3 years; 38 years in the USA), the population of Ethiopia is extremely young. Less than 4% of the population is over 65 years old, as opposed to 16% in the USA [4, 8]. Cardiovascular disease and obesity are less prevalent in Ethiopia than in the West, but the prevalence is increasing [9].

Several studies of hospitalised COVID-19 patients in Ethiopia have already been conducted. A study in Addis Ababa millennium COVID-19 care center included 147 patients and reported 33.3% mortality [10] and in Eka Kotebe General Hospita included 602 patients and reported 16.9% mortality [11], while a study conducted in the Amhara region included 28,533 patients and reported 2873 (11.2%) mortality [12]. A study conducted in the Tigray region included 139 patients reported 40.3% mortality [13], and a study from Wollega, western Ethiopia include 306 patients reported 5.7% [14]. Despite this past research, there is currently a lack of information on the outcomes of COVID-19 patients hospitalized in various locations in Ethiopia. None of these investigations have been undertaken in Hawassa, a metropolis of nearly 350,000 people. As a result, this study was conducted in Hawassa’s Sidama regional state, which is home to people of various ethnic groups who coexist peacefully, to shed light on the diversity of patients of different ethnicities and address the shortcomings of previous research by taking into account all ethnic groups residing in Hawassa and its neighbours. Furthermore, with Hawassa University Comprehensive Specialty Hospital facilities nearby, communities in and around Hawassa have expressed a desire for greater services. This makes a preferred hospital, and it can take patients with a variety of comorbidities who come for a variety of treatments, increasing the likelihood of finding a predictive variable related to our COVID-19 study. This cohort study aimed to assess clinical and demographic features, along with risk factors for in-hospital mortality, among COVID-19 patients at Hawassa University Comprehensive Specialized Hospital in Ethiopia.

Materials and methods

Study design and population

A hospital-based retrospective cohort study was conducted using medical records involving 804 COVID-19 cases at Hawassa University Comprehensive Specialty Hospital, COVID-19 isolation and treatment center. The Center is located in the regional capital of Sidama regional state, Hawassa. For the analysis, we included all patients infected with COVID-19 in the analysis regardless of age. We included patients admitted to the hospital between September 24, 2020 and November 26, 2021 seeking COVID-19 treatment. We collected data until December 6, 2021, at which point patients still in hospital were censored [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24]. The selection of patients was done to include patients in whom COVID-19 was the primary reason for their stay in Hawassa University COVID-19 center and patients in whom COVID-19 was an incidental finding likely to be unrelated to their hospital stay also included. The unit of analysis was the individual patient. Since one person might have had several hospital-stays during the observation period, due to a transfer from one hospital to another, we grouped adjacent completed hospital stays into one patient. Each patient was followed up after admission for survival status and after discharge for hospitalizations for a maximum of 62 days (maximum patient data 62 days, total date from September 24, 2020 to November 26, 2021 is 428 days, or more than 1 year and 2 months).

Data collection procedures

Epidemiological, clinical, laboratory, radiological characteristics, oxygen therapy, treatment, and outcomes data were obtained using data collection forms from patients’ electronic medical records for all patients with a clinical diagnosis and/or radiological findings consistent with signs and symptoms of COVID-19 and/or laboratory-confirmed SARS-CoV-2 infection from November 24, 2020 to November 26, 2021. Additional information was also collected by a trained team of physicians by referring to the patient’s medical history card. The information recorded included in this study were demographic data, medical history, underlying comorbidities (cardiovascular conditions, trauma, renal diseases, virological conditions, diabetes, malignancy/cancer, hypertension, immuno-suppression status, and other respiratory conditions (Tuberculosis, Pulmonary Tuberculosis, Asthma, Lung abscess, Bronchi, Central Nervous System Tuberculosis (CNS-TB), and Pulmonary tuberculoma)); symptoms from onset to hospital admission (fever, cough, shortness of breath, sore throats, general body weakness, chest pain, headache, and non-specific symptoms (Hiccup, vomiting, loss of appetite, chills, myalgia, sneezing, easy fatigue, fast breathing, chest tightness, epilepsy, weight loss, epigastric pain, diarrhea, dyspepsia, and difficulty of swallowing)); The date of disease onset was defined as the day when the symptom was noticed and the date of admission was defined as the day when the patients were admitted to the hospital; Patients who have not been previously admitted to any government hospitals, private clinics, or HUCSH wards, and have arrived directly from their residences or communities to be admitted to a COVID-19 isolation and treatment center were categorized as home-isolated patients see Fig. 1 for the details.

Fig. 1

Flow chart of the data set

Inclusion and exclusion criteria

Inclusion criteria were positivity of real-time reverse transcriptase-polymerase chain reaction (RT-PCR) or a clinical diagnosis and/or radiological findings consistent with signs and symptoms, of COVID-19. Patients who attended before 24/9/2020 and after 26/11/2021 were excluded, and try to include some incomplete hospital records from patients’ follow-up cards were included.

Statistical analysis

The main objective of this study is assesses clinical and demographic characteristics, along with risk factors for in-hospital mortality in 804 COVID-19 patients during the study period and to report the results of data analyses separately by the medical center where patients came from, the stage of illness, the differences between survivors and non-survivors, clinical courses, clinical and related outcomes. Therefore, no formal assumptions were applied to drive the sample size calculation and we included the maximum number of patients who met the inclusion criteria.

Descriptive statistics are applied as mean (SD) or median (minimum, maximum) for continuous variables and number (%) for categorical variables to the general features of the data in the study. The bivariate analysis was performed between the source of reporting, covariates to see the significant association, and the Pearson chi-square test or Fisher’s exact test was used to analyze categorical variables. A Kaplan-Meier estimation technique was used to see the estimate of survival analysis for an event death of COVID-19 patients and a separate Kaplan-Meier survivor functions curve was constructed to estimate the death time based on different covariates, to see the existence of difference in death rate between categories of individual covariates. To show the significance of survival difference, the log-rank test was computed at a 5% significance level. A Univariate and Multivariate cox regression analyses used for time to death. Table was produced from chi-square test or Fisher’s exact test, Univariate and Multivariate Cox proportional hazard and Kaplan-Meier estimate curves ; the tests were two-sided with significance set at $\alpha$ less than 0$\bullet$05. The R software was applied for all analyses. To analyze the time to death, patients who are lost from follow-up at the end of the study and patients discharged/ transferred to another hospital are all considered as censored.

Results

Demographic characteristics of the study population

Hawassa University Comprehensive Specialty Hospital admitted 804 patients with confirmed COVID-19 between September 24, 2020, and November 26, 2021. Of the patients, 512 (64.1%) were male and 289 (34.9%) were female. Out of the 173 fatalities (21.5%), 125 (72.3%) were men. The average age of the patients was 44.8 years, and the elderly were the deceased. Most of the patients were under 49. The average duration of hospital stays for COVID-19 patients ranged from admission to discharge, with a range of 0.5 to 60 days. The duration from symptom onset to treatment was 16.5 days, and from symptom onset to discharge was 63 days. Patients’ average body temperature was 37.2$^{\circ }$C on admission. Dead patients had a significantly shorter hospitalization period and higher body temperature compared to censored patients. Following the identification of the statistical relationship between each predictor and the outcome variable, the following factors were found to be important to determine the patient status: body temperature, onset to admission, onset to discharge times, sex, age category, source of report, address (residence), and duration (censored or died). Additionally, a significant and higher risk of COVID-19 death was observed in males, patients from private and government hospital sources, and older age groups compared with the respective group (Table 1).

Table 1 Descriptive statistics (count, percent and proportion) of 804 COVID-19 patients and subgroups by death outcome in HUCSH, 2022

The data presented in Table 1 reveals that 69.9% of individuals admitted to Hawassa University Speciality Hospital for reasons unrelated to COVID-19 subsequently tested positive for the virus. COVID-19 has a notable impact on the younger demographic, with 40.7% of cases occurring in individuals under the age of 39. Moreover, the patients exhibited severe and critical symptoms.

Baseline clinical characteristic of patients

Ninety percent of the eight hundred and four patients had at least one symptom. Prior to being hospitalized to the COVID-19 inpatient treatment program at HUCSH, 157 (19.5%) of the symptomatic individuals had three or fewer, 451 (56.1%) had three to five, and 184 (22.9%) had five or more.

Table 2 Descriptive statistics (count, percent and proportion) for clinical covariates and subgroups by death status of 804 COVID-19 patients in HUCSH, 2022

According to Table 2, the most common symptoms were fever (57.6%), dyspnea (65.7%), cough (68.7%), and chest pain (68.9%). Aside from headaches and sore throats, other symptoms included general body weakness and vague ones including dyspepsia, quick breathing, epigastric discomfort, diarrhea, chest tightness, vomiting, chills, appetite loss, sneezing, easy weariness, myalgia, and epilepsy. Before being admitted to the inpatient treatment facility for HUCSH COVID-19, 188 individuals (23.4%) made visits to other hospitals. Following the statistical analysis of the association between each clinical predictor and the outcome variable, it was found that the presence of at least one symptom, cough, dyspnea, headache, general body weakness, chest pain, number of symptoms experienced, and disease severity status were significant factors in determining the patient’s status and increased risk of COVID-19 death when compared with the relative reference group.

Comorbidities in patients with COVID-19

The study found that 49.1% of patients had at least one medical condition/co-morbidity, with 36.2% having one co-morbidity,10.4% having two and 2.2% having three or more. Among these, 22.4% had cardiovascular defects, 5.2% experienced trauma, and 5% had other respiratory problems. The most common coexisting conditions were diabetes mellitus, hypertension, cancer, and RVI/HIV. The study found that, from the total death 115 patients (66.5%) had co-morbidities, with 32.4% having cardiovascular defects, 31.2% diabetic, and 24.3% hypertensive, with 40.5% having one and 24.4% had two or more co-morbidities. Once the statistical link between each predictor and the outcome variable was established, it became evident that the following variables were critical in establishing the patient’s status and higher risk of death: at least One Underline Co-morbidity, Hypertension, Diabetes Melitus, Cardiovascular Conditions and the Number of Co-morbidity, Table 3.

Table 3 Descriptive statistics (count, percent and proportion) for comorbidity related covariates and subgroups by death status of 804 COVID-19 patients in HUCSH, 2022

Survival estimate of time to death of COVID-19 patients

One hundred seventy-three observations have developed an event (death). The overall graph of the Kaplan-Meier survivor function depicted that the graphs decrease rapidly during the first 32 days, showing that most patients died because of COVID-19 during this time (Fig. 2) and we can see that the time-to-death is shorter than the median time.

Fig. 2

The Survival probability of the time-to-death of COVID-19 patients in HUCSH from time to admission to deceased

Predictors of time to death from COVID-19

The resulting Kaplan-Meier graphs show the probability of deaths due to COVID-19 during the study period. These results were supplemented by the log-rank test to compare the covariates. The Kaplan-Meier survival plots for the prognostic factors are presented in Figs. 3, 4 and 5. From Kaplan Meier survival curve of individual covariates, there was a significant difference in the death rate among males and females, age under 18 and above, age in a different category, number of comorbidities exists, number of symptoms, and being residing in rural and other, severity status, and patients taking second-line antibiotics before admission (Fig. 3).

Fig. 3

Kaplan-Meier curves illustrating more than 60-day comparison between the survival curves from time to death, with different demographics and the indicating variable for any one of these conditions. “NCOM” refers to the number of comorbidities; “Num.symptom” refers to the number of symptoms that occurred. “Agecat” refers to the age category

Fig. 4

Kaplan-Meier curves illustrating more than 60-day survival comparison between the survival curves from time to death, with only different symptoms variable. For any one of these conditions, Red lines for indicator variables are for those with show symptomatic. Whereas the grey lines are for those who do not have a symptom. “NSS” refers to non-specified symptoms, “ShBreath” refers to shortness of breath, “Pain” refers to unexplained pain, “GBW” refers to general body weakness, and “HFVBOS” refers to a health facility visit before onset symptoms

Fig. 5

Kaplan-Meier curves illustrating more than 60-day survival comparison between the survival curves from time to death, with only different comorbidity variables. For any one of these conditions, Red lines for indicator variables are for those that show comorbidity. Whereas the blue lines are for those who do not have comorbidity. “UCC” refers to underline comorbidity and condition, “RDB” refers to renal disease, “OResCon” refers to Other respiratory conditions, “CardioVDis” refers to cardiovascular disease, and “RVI/HIV” refers to respiratory viral infections

Significant variability was observed among the different age groups and the severity status of the disease, as evident from the Kaplan-Meier curve, where, the risk of a patient dying of COVID-19 increased with increasing age, and severity of the disease increased from mild to critical (Fig. 3). In addition, with increased number of symptoms and comorbidities the risk of dying of COVID-19. also increased. Additionally, there was a significant difference in the death rate of patients in relation to the type of symptoms on admission, where symptoms like cough ($p < 0.001$), Chest pain (p = 0.0023), shortness of breath ($p < 0.001$), Headache (p = 0.039), and general body weakness ($p < 0.001$) all had significant differences in the probability of occurrence of COVID-19-related deaths (Fig. 4). On the other hand, The Kaplan-Meier curve and log-rank test of the status time-to-death showed that there was no significant difference on showing symptoms like Fever (p = 0.48), health facility visit before onset of symptom (P = 0.37), sore throat (p = 0.054), and Non-Specified Symptoms (P = 0.088) in the probability of the occurrence of COVID-19 related deaths.

From the result of Fig. 5, the survival curves of Kaplan-Meier and the log-rank test result showed the presence of at least one comorbidity have significantly affected the status of death $p < 0.001$. Therefore, we can find that underline comorbidity was significantly associated with the death rate of COVID-19 without adjusting for other covariates. Accordingly, the Kaplan-Meier curve and log-rank test of the status time-to-death showed that there was no significant difference on other respiratory conditions (p = 0.28), renal disease (p = 0.64), RVI/HIV (P = 0.48), cancer (p = 0.076), and Trauma (0.12) in the probability of the occurrence of COVID-19 related deaths. However, a significant difference was observed in the Kaplan-Meier curve regarding hypertension ($p < 0.001$), diabetics ($p < 0.001$), and cardiac disease ($p < 0.001$) (Fig. 5).

Univariate Cox-PH for time-to-death

The results of the univariate survival analysis of time-to-death using the Cox proportional hazard model confirm that for patients of either gender, male was a significant risk factor for mortality (hazard ratio [HR = 1.49]; 95% CI [1.07-2.08]; p= 0.019). Based on the result of the patients’ source of reporting, patients from government hospitals had 2.48 times increased risk of death (hazard ratio [HR = 2.48]; 95% CI [1.69 - 3.63]; $p<0.001$), and patients whose source of reporting was from private hospitals had 2.14 times increased in the risk of death hazard ratio [HR=2.14]; 95% CI [1.42 - 3.20]; $p<0.001$). On the other hand, no significant effect was observed on the risk of death for those patients whose reports were referred home isolated patients or patients from the community compared with patients from Hawassa University Comprehensive Speciality Hospitals. While older age appeared to be a strong predictor of COVID-19 mortality. Compared with those between 39 and, 49 years old, individuals in the age groups less than 39, between 49 and 59, between 59 and 69, and over 69 years of age had 1.83 (95%CI [1.08,3.11]; p = 0.025), 1.95(95%CI [1.18,3.24]; p = 0.010), 2.81(95%CI [1.80,4.39]; $p<0.001$), and, 4.18(95%CI [2.67,6.55]; $p<0.001$) times higher risk of mortality, respectively. Based on the result of the time-to-event univariate Cox proportional hazard (Table 4), as age increases the risk of dying increases at any time during follow-up.

Patients with mild and moderate severity level had a 92% and 76% lower probability of dying from COVID-19 infection, compared to critical cases, according to a survival curve analysis. Similarly, there was no significant difference in death rates between patients with and without symptoms. Those with three to five and more than five symptoms had significantly higher risk of mortality and shorter survival time compared to those with less than three symptoms (HR = 2.59 [1.45, 4.63]; p = 0.001 and HR = 4.48 [2.47, 8.14]; $p<0.001$, respectively). Individuals with at least one comorbidity had a significantly higher risk of death than those without (HR = 1.96 [1.43, 2.69]; $p<0.001$). Those with one, two, or more than three comorbidities had a significantly higher risk of death and shorter length of staying alive compared to those without comorbidities (HR = 1.57 [1.11, 2.22; p = 0.010], HR = 2.62 [1.73, 3.98; $p<0.001$], and HR = 3.14 [1.43, 6.86; p = 0.004], respectively).

Table 4 Univariate and multivariate Cox regression analysis for time-to-event (Death) among COVID-19 patients hospitalized at HUCSH, Ethiopia, 2021

The result of time-to-death univariate analysis indicates that cough had a 2.46 (HR= 2.46; 95% CI [1.64, 3.69]; $p<0.001$), Shortness of breath a 3.73 (HR=3.73; 95% CI [2.38, 5.83]; $p<0.001$), General Body Weakness a 2.57 (HR=2.57; 95% CI [1.82, 3.65]; $p<0.001$), chest pain a 1.74 (HR=1.74; 95% CI [1.21, 2.49]; P = 0.003), and headache a 1.37 (HR=1.37; 95% CI [1.01, 1.84]; P = 0.040) times increased risk of death. Cardiovascular diseases had a 71% higher risk of death, with 1.71 (1.24-2.35, p=0.001). Furthermore, the prevalence of hypertension and diabetes increased the risk of death by 1.93 times (HR= 1.93; 95% CI [1.37, 2.74]; $p<0.001$) and (HR=2.42; 95% CI [1.75, 3.34]; $p<0.001$), respectively. When the interaction between diabetes and hypertension, shortness of breath and sore throats, general body weakness, and health facility visits before the onset of symptoms are taken into account, the risk of mortality increases significantly. Patients with shortness of breath are at a 6.97 times higher risk, diabetic patients at a 2.76 times higher risk, and hypertensive patients at a 2.11 times higher risk. The interaction effect between diabetic and hypertensive patients is estimated to increase the risk of death by 18.7%(2.76*0.43=1.187). Patients with general body weakness who had visited health facilities before admission have an increased probability of mortality by 89.2%(2.91*0.65=1.8915). Similarly, the interaction effect of shortness of breath on sore throats raised the probability of mortality by 67.3%(6.97*0.24=1.6728). However, Fever, sore throat, health facility visit, RVI/HIV, RDB, trauma, cancer, non-specified symptoms, and other respiratory conditions have not been statistically significant for the status of death. Finally, the presence of any type of comorbidity was another factor that determines the risk of death in patients with COVID-19. The presence of diabetics, hypertension, and cardiovascular disease increase the risk of death by 2.34, 1.93, and 1.71, respectively as compared to those patients who had no comorbidity (Table 4).

Multivariate analysis for predictors of time-to-death

When the p-values for all three overall tests (likelihood, Wald, and score) are calculated using reduced multivariate Cox regression, the results show that the overall model is significant, rejects the null hypothesis, and accepts at least one of the predictors with a significant effect on death status (Fig. 6). All covariate values and the Schonefield residual graph (Fig. 7) support the insignificant global test value of (0.555) obtained from the Cox proportional hazard assumption. We used a variety of plots (Figs. 8, 9 and 10) to investigate significant observations, outliers, and non-linearity. A time-to-death analysis of hospitalized COVID-19 patients found that symptoms before admission had a lower risk of mortality (HR: 0.26, p = 0.0013). Patients with severe to mild severity status had a decreased risk of death compared to critical and significantly lower risk of death at each successive level (Mild HR: 0.09, $p<0.001$, Moderate HR: 0.23, $p<0.001$, and severe HR: 0.50, $p<0.001$). Patients aged 59-69 had a 68% increased risk of death (HR: 1.68, p = 0.035), whereas those beyond 69 years had a more than doubled risk (HR: 2.21, p = 0.002) compared with (0,39]. Diabetes (HR: 1.93, P=0.010), hypertension (HR: 1.81, P=0.036), shortness of breath (HR: 4.35, $p<0.001$), sore throat (HR: 3.27, P=0.013), general body weakness (HR: 2.91, $p<0.001$), health facility visit before admission (HR: 3.37, P=0.001), and more than three comorbidities (HR: 3.09, P=0.001) were significantly associated with higher risk of death. However, the factors of age (39, 49] and (49, 59], cough, and one or two comorbidities had no significant association with COVID-19 death ($p > 0.05$). Those with a report from a government hospital had a considerably higher risk of death (HR: 1.72, p=0.011) than those with a report from Hawassa University Comprehensive Specialty hospitals. Whereas other sources showed no significant difference with the reference group.

Fig. 6

The multivariable reduced model of the time-to-death of COVID-19 patients in HUCSH

Fig. 7

Schoenfeld residual plots used to test the proportional hazard assumption

Fig. 8

Plots of dfbeta for Cox Regression fro time to death

Fig. 9

Martingale residual plot

Fig. 10

Deviance residual plot

Concerning the interaction effect of diabetics with hypertension, general body weakness with health facility visits before admission, and shortness of breath with sore throat at the time of admission were all substantially linked with the probability of death. The estimated effect of diabetes on hypertensive individuals with COVID-19 is 1.93 * 0.28 = 0.5404, indicating a 46% decrease in the risk of death (HR: 0.28, P=0.004). Patients with general body weakness who visit health facilities before admission have a 44.7% (2.91* 0.19 = 0.5529) lower risk of mortality (HR: 0.19, $p<0.001$). The interaction between shortness of breath and sore throat raised the chance of mortality by 34.9% (4.35*0.31 = 1.3485) (HR: 0.31, $p<0.001$).

Discussion

Hawassa University Comprehensive Specialty Hospital is a tertiary care hospital located in a resource-constrained country. This study examined at the clinical and demographic characteristics of COVID-19 patients admitted to Hawassa University Comprehensive Specialty Hospital between September 24, 2020, and November 26, 2021. It has also provided the critical risk variables that influence COVID-19 patients’ odds of survival. This work supplements previous observational studies [2, 11, 12, 25,26,27] conducted in resource-constrained situations where the bulk of the population is densely populated and over 56 ethnic groups live together with a wide variety of cultural practices. Unfortunately, in this study, the younger and economically active demographic is disproportionately impacted by the COVID-19 pandemic. This is likely due to their frequent participation in a variety of activities, the fact that many of them depend on daily income, their active involvement in business, and their potential disregard of government regulations (such as mask-wearing, voluntary testing, hand sanitizer use, proper hand hygiene, hugging and shaking hands, maintaining social distance, not quickly isolating infected individuals, and addressing cultural and societal issues).

The graph from Kaplan-Mayer (Figs. 3, 4 and 5) analyses showed survival curves for patients with and without comorbidity, symptoms, or demographic factors in association to the variable time-to-death. For Sex, age category, number of symptoms, number of comorbidities, source of reporting, severity status, cough, general body weakness, Shortness of Breath, headache, chest pain, at least one underline comorbidity and condition (UCC), diabetes, hypertension, and cardiovascular conditions the survival curve on the status of death shows a visual understanding that it significantly affects the time-to-death.

The results of our analysis revealed that men accounted for a higher proportion of COVID-19 patients than women, which covers a total death of 72.3%, and had a higher risk of dying due to COVID-19. The result is in line with a recent single-arm meta-analysis [28], and other previous studies conducted in Ethiopia [10,11,12,13,14, 25, 29,30,31], Middle East [32], Shenzhen (China) [33], Nigeria [34], New York City [35], Israel [36], Iran [37], Lyon (France) [15], Municipality in Bhutan [16], Netherlands [17], Sardinia (Italy) [18], Wuhan [19, 20, 38] and other places in France [39]. The overall mean (44.8) and median (45) age of hospitalized patients with COVID-19 were comparable to that found in other studies with a mean [19, 21, 25, 33, 37, 40] and median [41,42,43,44,45,46] age of about the mid-40s; and also having similar age range [3 months, 99 years] reported from a meta-analysis [46]. In our study, the mean and median age of death were 55.4 years and 60 years, respectively, which had more or less the younger value as the mean age of reported by Ge, Erjia, et al. [40] (82.2years), Zhang et al. [47] (77.0 years), Wang et al. [48] (71.0 years), and Passamonti, Francesco, et al. [49] but most deaths reported by Marschner, I. C [50] were more than 60.0 years age group. Additionally, the young and economically active population group (less than 39)years was most commonly affected by COVID-19 [13, 14, 25, 29, 30, 32, 33, 35, 40].

Regarding the history of hospital stay, looking at the time from the entry point hospital and where the patient showed symptoms, the overall mean (SD) was 5.92 (5.22) days, and the median (IQR) was 5 (IQR 3 to 7) days in Hawassa University Comprehensive Specialty Hospital. When we compare this with the studies done by different researchers from several countries, the average days it took to stay at the hospital was 5 in a meta-analysis estimate [51, 52] and another meta-analysis with a pooled mean number of days 4.92 [53], which is consistent with the median days observed in the present study. Studies from other countries also reported almost similar median duration from the time of symptom onset to admission to hospitals; 3 days in Malesia [54], 5 in Norway [55], 4.5 to 9 days in China [22, 43, 44, 47], 4 to 5 in Italy [18, 52], 6 in Brazil [23], 6 in Colombia [39], 6 to 7 in France [15, 56], 7 in Iran, Netherlands, Egypt and Morocco [17, 37, 57, 58]. Furthermore, a study also in the US supports a mean (SD) of time from symptom onset to admission in moderate and critical patients was 5.8 (4.4) and 5.3 (5.2), respectively [59]. However, a study in Germany reported double median time (10 days) from symptom onset to admission [60] compared with our study.

The overall time between admission to death in the current study was a median of 5 and a mean (SD) of 6.63(5.98) days, and the median was consistent with other similar studies in different countries, 5 to 7 in Italy [18, 60], 8 in Iran and Brazilian [37, 61], and 6 in the Netherlands [17]. On the other hand, in a study conducted in Germany, the mean and median length of stay in-hospital all-cause mortality was 12.8 (14.2) and 8.0 (4.0-16.0), respectively [59]. The overall time between admissions to death observed in the current study was generally shorter than reports from studies conducted in Germany [59] and China [20]. More than 67% of confirmed deaths were in the age category above 50 years and the risk of dying significantly increased in older age group patients’ in-hospital death [17, 22, 23, 35, 40, 55, 60, 62, 63]. Evidence from previous studies suggests that older male patients were the most susceptible to SARS-CoV-2 infection [20, 22, 23, 35, 37, 52, 63, 64], which is supported by our data. Additionally, the majority (40.7%) of patients’ ages were less than 39 years, this is supported by the study held in Wollega hospital (46.4%), in Ethiopia [14]. Moreover, in our study population, by excluding 36 censored observations, approximately 22.53% of the patients died from COVID-19. This is higher than another study 11.14% of total deaths observed in Bekoji hospital Ethiopia [25] and the study enrolled on Birjand,17.2% [37], and higher than the study carried out in the UK 31% [65], overall lethality rate in Brazil’s 41.28% [61], and a mortality rate study in Mexico 33.1% [66].

The finding in this study that at least one comorbidity occurred among 49.1% of hospital admitted COVID-19 patients was similar to the observation from Bokoji hospital (49.05%) [25] but much higher than the one from Eka Kotebe General Hospital, Addis Ababa (13.4%) [30] and 22.72% in multi-centered retrospective cohort study in SNNPR [29]. Similarly, observations from international studies have documented spectrum of occurrence of at least one comorbidity, where some reported almost comparable rate to ours (43.8% from Canada [40], and from 46.4% Wuhan, China [67]); others reported less than the rate from our study (38% from France [56], 33.3% from Wuhan, China [68], 29.6% from Germany [60], 28.7% from Sardine, Italy [18], 24.6% from the Netherlands [17], and a few others reported much higher rate (61.5% from another study in Wuhan, China [64]), with the most important comorbidities in this study were cardiovascular conditions that occurred among180 (22.4%) of the total 804 patients, as was the case from other national and international studies [15, 17, 18, 25, 35, 47, 56, 69]. More specifically, the first two most common comorbidities in this study were diabetes and hypertension, similar to what as observed from other local (Eka Kotebe General Hospital, and Millennium COVID-19 care center in Addis Ababa [10, 30]) and international [17, 20,21,22, 26, 32, 34, 35, 37, 39, 42, 47, 57, 61, 66, 69, 70] studies.

Among the patients died from COVID-19, 49.1% of them had at least one comorbidity while 2.2% had more than three comorbidities. This high proportion of comorbidities among the fatally ill patients indicates the high epidemiological impact of this disease on individuals with chronic illnesses. Some of the main comorbidities found among the deceased patients in this study were cardiovascular conditions (32.4%), diabetes (31.2%), hypertension (24.3%), other respiratory conditions (7.5%), and HIV infection (4.0%). Moreover, from the univariate analysis patients with cardiovascular conditions (71%), diabetes mellitus (142%), and hypertension (69%) had an increased risk of death than those without these diseases remaining other cases constant [17, 19,20,21, 23, 32, 34, 37, 40, 70]. In addition, as the number of comorbidities, and age increased, so did the risk of death, an observation supported by several other studies globally [11, 23, 34, 37, 39, 40, 61, 63, 69, 71]. Additionally, like most previous reports from Ethiopia [14, 30, 34], and elsewhere globally [15,16,17,18, 22, 26, 28, 32, 33, 35,36,37,38,39, 44, 47, 51, 57, 69], the most common symptoms at the time of admission to COVID-19 care centers were fever (57.6%), cough (68.7%), shortness of breath (65.7%), chest pain (68.9%), general body weakness (56.2%), headache (43.0%), and sore throat (32.2%). We also found from the univariate analysis that the risk of dying due to COVID-19 was significantly higher among patients with certain specific symptoms. For instance, cough, shortness of breath, Headache, general body weakness, and chest pain significantly increase the risk of death. These results were supported by previous studies as the most common symptoms associated with a higher risk of death from COVID-19 [21, 54, 61].

In this study, the result of multivariate cox proportional analysis shows a significant association between older age group and an increase in the risk of death. The age group 59-69 and above 69 had 1.68 and 2.21 times, respectively, higher risk of death than the age group less than 39 years old. This finding of a significant risk among the older age group was in line with observations from different studies conducted globally: in South Africa [71], Nigeria [34], China [20], Brazil [61], Norway [72], Netherlands [17], Malaysia [53], Canada [21], Ethiopia [25] and Africa region [70]. Furthermore, multivariate cox proportional analysis also revealed that the risk of death from COVID-19 was more likely higher among patients with more than three comorbidities than those with no comorbidity. Several other studies also reported that comorbidity was an independent risk factor that can increase the risk of death [19,20,21, 40, 56]. With respect to association of disease severity status with risk of death, mild, moderate, and sever illness levels had 0.09, 0.23, and 0.50 times, respectively, lower the risk of death as compared with the critical severity status. This finding shows that as the severity status increases from mild to critical level, the risk of death also increases substantially. This finding is in line with reports from different studies including reports from Meta-analyses on sub-Saharan Africa [73], Nigeria [34], and Congo [74]. According to the univariate and multivariate Cox proportional hazard regression model, shortness of breath, sore throats, general body weakness, diabetes, hypertension, and those who had visited a health facility before the onset of symptoms are at a significantly higher risk of mortality. Further data from various sources confirms that patients experiencing dyspnea (shortness of breath) [11, 24, 69], hypertension [19,20,21, 34, 37], and diabetes [11, 19,20,21, 27, 37, 40, 56, 61, 69], have a significantly higher risk of dying.

In COVID-19 patients, sore throats, generalized weakness, and shortness of breath along with sore throats are among the symptoms that raise the risk of death. On the other hand, mortality is lower in those with the interaction of diabetes and hypertension, generalized body weakness, and a history of medical visits before to the onset of symptoms. However, it remains unclear if the hospital’s care or the therapy interaction effect is the reason for this decline. However, no published research has addressed these mortality risk factors for COVID-19 patients in this particular scenario.

Conclusion

The current study identified several key characteristics that influenced the survival of COVID-19 patients admitted to Hawassa University Comprehensive Specialty Hospital between September 24, 2020, and December 6, 2021. It also includes clinical and demographic parameters related to COVID-19 patient deaths. In comparison to industrialized countries, the average age of COVID-19 patients was relatively young. This makes it very clear that local context should be considered while establishing COVID-19 preventive and control measures, rather than just implementing global perspectives. This research endeavor could potentially produce unexpected benefits in terms of strengthening and increasing information about clinical and demographic features, as well as critical determinants influencing COVID-19 patient survival. This literature may be valuable in developing models for better COVID-19 prediction and management during pandemics. As a result, the current study may contribute to the development of effective medicines to reduce COVID-19-related mortality.

Limitation of the study

Our research was conducted under certain constraints that may have introduced bias. This single-centred retrospective cohort study focused on COVID-19 cases at Hawassa University Comprehensive Specialty Hospital (HUCSH), who attended from different sources of patients, and mild cases were permitted to go back to home isolation. As a result, most home isolated patients not come to hospital on the appointed date, this causes to increasing the number of censored patients. Moreover, our dataset derived from routinely collected information, leading to missing values in electronic medical records. To address potential biases, we carefully crosschecked patient records. This takes a long time, limit the time interval of the data collection and the number of sample observations. Conversely, not incorporating data from various healthcare systems would not offer a more comprehensive assessment of the overall impact of COVID-19. However, utilizing our single-centre population in HUCSH, it is one of the largest specialty hospitals in the area serving to a diverse patient, we were able to consider various factors such as patient demographics, residency, clinical indicator and comorbidity related variables, this enhancing the robustness of our findings and the quality of the data. Our data collection is scheduled to be completed by November 26, 2021. During the study period, HUCSH’s COVID-19 isolation unit received a continuous flow of patients. So, it’s important to note that certain significant criteria that could impact mortality risks, such as bed capacity, facilities, patient-to-physician ratios, medical support levels, and hospital resource availability, were not factored into our analysis.