Abstract
Background
Coronavirus disease 2019 (COVID-19) has significantly increased mortality worldwide, with advanced age and comorbidity burden has been identified as a major risk factors. However, the combined effect of age and comorbidity burden on COVID-19 mortality remains unclear. This study aimed to assess the independent and interactive effects of age and comorbidity burden on in-hospital mortality from COVID-19.
Methods
A retrospective cohort study was conducted among 642 laboratory-confirmed COVID-19 patients admitted to a tertiary care hospital in Belagavi, Karnataka, India, between October 2020 and September 2021. Bivariate analysis, multivariable logistic regression, and interaction analysis were performed to examine associations with mortality.
Results
Patients aged ≥ 50 years had significantly higher odds of mortality compared to those aged < 50 years (AOR = 2.01, 95% CI: 1.27–3.16, p = 0.002). Dyspnea (AOR = 2.88, 95% CI: 1.85–4.48, p < 0.001) and kidney disease (AOR = 6.56, 95% CI: 2.39–18.02, p < 0.001) were significant predictors of mortality. Blood glucose (per 10 mg/dL increase; AOR = 1.03, 95% CI: 1.01–1.05, p < 0.001) and pulse rate (per bpm increase; AOR = 1.02, 95% CI: 1.01–1.03, p < 0.001) were also significantly associated with mortality. Multimorbidity was not independently associated with mortality in the main model (AOR = 1.63, 95% CI: 0.67–4.00, p = 0.278). In the interaction model, no statistically significant interaction between age and multimorbidity was observed. However, predicted probabilities suggested a higher mortality risk among older patients with multimorbidity compared to other groups.
Conclusion
Advanced age, dyspnea, kidney disease, elevated blood glucose, and pulse rate are significant predictors of COVID-19 mortality. While multimorbidity may contribute to mortality risk, no strong evidence of interaction between age and comorbidity burden was observed. These findings highlight the importance of comprehensive risk stratification using multiple clinical factors in hospitalized COVID-19 patients.
Clinical trial
Not applicable.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12879-026-13497-y.
Keywords: COVID-19, Mortality, India, Multimorbidity, Age
Introduction
The emergence of Coronavirus Disease 2019 (COVID-19), caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), resulted in substantial global morbidity and mortality and imposed unprecedented challenges on public health systems worldwide [1–4]. Identifying predictors of severe disease, particularly mortality, emerged as a major priority for clinical practice management and public health planning from the outset of the pandemic [5]. By 2025, India had reported the second-highest number of confirmed cases in the world, with 45,041,748 COVID-19 infections and the third-highest number of COVID-19 deaths [6, 7].
Epidemiological studies rapidly identified advanced age as one of the strongest independent predictors of COVID-19 mortality [8–10]. Mortality rates increase sharply with age, particularly among older adults, likely due to immunosenescence, dysregulated inflammatory responses, and a higher prevalence of underlying chronic health conditions [11, 12]. Apart from age, the presence of comorbidities has been significantly linked to adverse COVID-19 outcomes [13].
Comorbidities, including cardiovascular disease, diabetes mellitus, chronic respiratory disease (COPD), chronic kidney disease, obesity, and cancer, have been frequently implicated in increased hospitalization rates, intensive care admission, and mortality [14–16]. Moreover, the burden of comorbidity is commonly quantified as the number of coexisting chronic conditions (i.e., multimorbidity). Evidence indicates that individuals with multiple comorbidities have a substantially higher risk of adverse outcomes than those with none or only one condition (i.e., no multimorbidity) [17, 18].
While the independent effects of age and comorbidity on COVID-19 mortality are well established, evidence on their combined and interactive effects remains limited. It is unclear whether age and comorbidity burden interact synergistically to amplify mortality risk or whether their effects are primarily additive. Clarifying this interaction has meaningful implications for clinical risk stratification, triage, and targeted management of high-risk patients.
Additionally, using data from a tertiary care hospital in Belagavi, South India, an underrepresented setting in global literature, this study provides context-specific evidence on how age and comorbidity jointly influence mortality risk through multivariable logistic regression and interaction analysis. Therefore, this study examined the relative influence of age and comorbidity burden on COVID-19 mortality using a retrospective cohort design.
Methods
Study design, population, and setting
This retrospective cohort study was conducted from October 2020 to September 2021 at a tertiary care hospital in Belagavi, Karnataka, India.
A total of 1,322 patients with laboratory-confirmed COVID-19 (RT-PCR or rapid antigen test) were screened. Of these, 240 patients were younger than 18 years and were excluded. Among the remaining 1,082 adult patients (≥ 18 years), 400 were excluded due to incomplete or missing medical records, uncertain diagnosis, or absence of outcome data.
Given that missing data were not random and a substantial proportion of cases were excluded, the results may be subject to selection bias and should be interpreted with caution, particularly regarding generalizability. Missingness among predictor variables was generally low, whereas missing outcome data were the primary reason for case exclusion. The extent of missing data for each variable is summarized in Supplementary Table S1. To assess potential selection bias, included and excluded adult patients were compared using available variables (age and sex), as presented in Supplementary Table S2.
Finally, a total of 642 adult patients with laboratory-confirmed COVID-19 and complete clinical records were included in the analysis. All patients who met the inclusion criteria were enrolled consecutively during the study period, helping to minimize selection bias and improve statistical power. Patients with incomplete medical records, missing outcome data, or uncertain COVID-19 diagnosis were excluded from the analysis (Fig. 1).
Fig. 1.
Flowchart of study inclusion and exclusion criteria
Data collection
A standardized pre-tested data extraction form was used to collect relevant variables from patients’ medical records. Two trained research assistants independently abstracted data from hospital charts under the supervision of the principal investigator (N.A.), and the extracted data were verified by a clinical consultant (A.M.). The data collectors underwent one week of training on data extraction techniques, coding procedures, and verification standards to ensure accuracy and consistency. Additionally, 10% of the records were randomly rechecked for accuracy and completeness. Any discrepancies were resolved through discussion among the research team.
The primary outcome was in-hospital mortality, defined as death occurring during the hospital admission period, and categorized as survivor and non-survivor status. Due to the retrospective nature of the study and the absence of post-discharge follow-up, 28-day mortality could not be assessed; therefore, the analysis was restricted to mortality occurring during the hospital admission.
Predictor variables were selected based on clinical relevance and prior evidence of association with COVID-19 outcomes. These included sociodemographic variables (sex and age), clinical characteristics, comorbidities, symptoms, vital signs, and laboratory findings. Age was categorized at 50 years, as mortality risk increased substantially beyond this threshold. This cut-off is also consistent with prior regional evidence identifying ≥ 50 years as a clinically relevant threshold for increased COVID-19 severity and mortality in Indian and Asian populations [19, 20]. Additionally, dichotomization was used to facilitate interpretability and interaction modeling.
The comorbidity burden was defined based on the number of pre-existing chronic conditions documented in the patient’s clinical records, excluding COVID-19 infection. Multimorbidity was defined as the presence of two or more chronic conditions in the same individual, while patients with zero or one chronic condition were categorized as having no multimorbidity. This approach was adopted due to the absence of standardized comorbidity indices (e.g., Charlson or Elixhauser index) in the dataset. The comorbid conditions included diabetes mellitus, hypertension, kidney disease, thyroid disorders, cardiac disease, pulmonary disease, liver disease, and hematological disorders, as documented at the time of hospital admission.
Although this approach does not account for disease severity or differential weighting of conditions, it provides a practical and commonly used measure of overall comorbidity burden in retrospective datasets [17, 18].
Hospitalization length of stay was categorized as < 7 days and ≥ 7 days based on the median value and prior evidence linking prolonged hospitalization with worse COVID-19 outcomes [21]. Symptoms such as cough, fever, and dyspnea were coded as binary variables (present/absent) based on documentation at admission. These symptoms were either self-reported or clinically observed and recorded by the treating physician. Vital signs and blood glucose levels were analyzed as continuous variables. Additional symptoms (e.g., weakness, vomiting, abdominal pain, diarrhea, headache, loss of appetite, body ache, chest pain, and myalgia) and other clinical variables (e.g., temperature, respiratory rate, and less frequent comorbidities such as neurological, thyroid, cardiac, pulmonary, liver, and hematological conditions) were also collected and analyzed, but are presented in the Supplementary Materials to improve clarity and avoid overloading the main tables (Supplementary Tables S3–S4).
Blood glucose values were obtained from hospital laboratory records at the time of admission; however, the dataset did not specify whether the measurements were random or fasting.
Therefore, these values likely reflect both pre-existing diabetes and stress-induced hyperglycemia in acutely ill patients. Accordingly, blood glucose levels were interpreted as indicators of metabolic stress rather than diagnostic measures of diabetes status.
Statistical analysis
All statistical analyses were performed using Stata/MP version 17.0 (StataCorp LLC, College Station, TX, USA). Missing data were handled using complete-case (listwise) deletion. Multiple imputation was considered for handling missing data; however, it was not used because a substantial proportion of excluded cases had missing outcome data (mortality status), for which imputation is typically not recommended due to the risk of introducing bias and model-dependent assumptions. Additionally, key clinical variables were missing in a non-random manner during the early phase of the pandemic, further limiting the validity of imputation approaches.
Continuous variables were summarized using mean and standard deviation (SD) for normally distributed data, and median and interquartile range (IQR) for non-normally distributed data. Categorical variables were presented as frequencies and percentages and compared using the chi-square test or Fisher’s exact test, as appropriate.
Multivariable logistic regression analysis was performed to identify independent predictors of in-hospital mortality, with results presented as adjusted odds ratios (AORs) and 95% confidence intervals (CIs). Variables were selected based on clinical relevance and prior evidence, including demographic factors, comorbidities, and key clinical severity markers (such as oxygen saturation and vital signs), rather than relying solely on univariate statistical significance.
Hospital length of stay was used in descriptive and bivariate analyses only, and it was excluded from the multivariable model to avoid reverse causation bias, as prolonged hospitalization may reflect survival duration rather than a baseline risk factor.
Interaction analysis was performed to assess the potential effect modification between age and comorbidity burden on COVID-19 mortality. To avoid multicollinearity and overlapping constructs, individual comorbidities were not included simultaneously with the multimorbidity variable in the interaction model.
Sensitivity analyses were conducted using alternative age cutoffs (≥ 60 years), which resulted in smaller subgroups and reduced statistical power for interaction analyses (Supplementary Tables S6–S8). Additionally, age was also analyzed as a continuous variable to assess the robustness of the findings (Supplementary Table S9).
Further analyses, including descriptive statistics, crude (unadjusted) analyses, multicollinearity diagnostics, and sensitivity analyses, are presented in the Supplementary Materials (Supplementary Tables S3–S10).
Results
Baseline characteristics of the study population
Table 1 shows that 642 hospitalized COVID-19 patients were included in the analysis, with a mean age of 52.59 ± 16.75 years. Of these, 44.08% were aged < 50 years and 55.92% were aged ≥ 50 years. The ≥ 50 years group had a higher mean age (64.79 ± 9.66 years) compared to the < 50 years group (37.12 ± 9.48 years).
Table 1.
Baseline characteristics of the study population stratified by age and comorbidity status
| Variable | Overall (N = 642) |
Age < 50 (n = 283) |
Age ≥ 50 (n = 359) |
No multimorbidity (n = 478) |
Multimorbidity (n = 164) |
|---|---|---|---|---|---|
| Age (years), mean ± SD | 52.59 ± 16.75 | 37.12 ± 9.48 | 64.79 ± 9.66 | 49.33 ± 16.50 | 62.09 ± 13.58 |
| Sex, n (%) | |||||
| Male | 444 (69.16) | 195 (68.90) | 249 (69.36) | 330 (69.04) | 114 (69.51) |
| Female | 198 (30.84) | 88 (31.10) | 110 (30.64) | 148 (30.96) | 50 (30.49) |
| Length of hospital stay, n (%) | |||||
| < 7 days | 326 (50.78) | 135 (47.70) | 191 (53.20) | 234 (48.95) | 92 (56.10) |
| ≥ 7 days | 316 (49.22) | 148 (52.30) | 168 (46.80) | 244 (51.05) | 72 (43.90) |
| Signs and symptoms, n (%) | |||||
| Fever | 304 (47.35) | 146 (51.59) | 158 (44.01) | 242 (50.63) | 62 (37.80) |
| Dyspnea | 335 (52.18) | 152 (53.71) | 183 (50.97) | 252 (52.72) | 83 (50.61) |
| Cough | 76 (11.84) | 33 (11.66) | 43 (11.98) | 46 (9.62) | 30 (18.29) |
| Comorbidity burden, n (%) | |||||
| No multimorbidity | 478 (74.45) | 253 (89.40) | 225 (62.67) | 478 (100) | — |
| Multimorbidity | 164 (25.55) | 30 (10.60) | 134 (37.33) | — | 164 (100) |
| Comorbidity type, n (%) | |||||
| Hypertension | 197 (30.69) | 41 (14.49) | 156 (43.45) | 51 (10.67) | 146 (89.02) |
| Diabetes mellitus | 187 (29.13) | 40 (14.13) | 147 (40.95) | 44 (9.21) | 143 (87.20) |
| Kidney disease | 23 (3.58) | 5 (1.77) | 18 (5.01) | 7 (1.46) | 16 (9.76) |
| Vital signs, median (IQR) | |||||
| SBP (mmHg) | 120 (20) | 120 (20) | 130 (30) | 120 (20) | 130 (30) |
| DBP (mmHg) | 80 (10) | 80 (10) | 80 (20) | 80 (10) | 80 (20) |
| Pulse rate (bpm) | 88 (20) | 88 (22) | 88 (20) | 88 (21) | 88 (20) |
| SpO₂ (%) | 95 (7) | 95 (7) | 95 (7) | 95 (7) | 95 (7) |
| Blood glucose (mg/dL), median (IQR) | 144 (123) | 131 (97) | 158 (134) | 131 (104) | 185 (146) |
| COVID-19 outcome, n (%) | |||||
| Survivors | 495 (77.10) | 238 (84.10) | 257 (71.59) | 389 (81.38) | 106 (64.63) |
| Non-survivors | 147 (22.90) | 45 (15.90) | 102 (28.41) | 89 (18.62) | 58 (35.37) |
Values are presented as mean ± standard deviation (SD), median (interquartile range [IQR]), or frequency (percentage), as appropriate. IQR: interquartile range; mm Hg: millimeters of mercury; bpm: beats per minute; SBP: systolic blood pressure; DBP: diastolic blood pressure; SpO₂: peripheral oxygen saturation
Males constituted the majority of the study population across all groups (69.16% overall). The median length of hospital stay was 6 days (IQR: 6 days) with no substantial differences across age or comorbidity groups.
The most common symptoms were dyspnea (52.18%), followed by fever (47.35%). Dyspnea was slightly more frequent among patients aged < 50 years (53.71%) compared to those aged ≥ 50 years (50.97%).
Multimorbidity was more prevalent among patients aged ≥ 50 years (37.33%) compared to those < 50 years (10.60%). Hypertension (30.69%) and diabetes mellitus (29.13%) were the most prevalent comorbidities, both of which were more common in patients aged ≥ 50 years (43.45% and 40.95%, respectively) and among those with multimorbidity (89.02% and 87.20%, respectively).
Vital signs were broadly comparable across groups. The median SpO2 was 95% (IQR: 7%), and median systolic and diastolic blood pressures were 120 mmHg (IQR: 20 mmHg) and 80 mmHg (IQR: 10 mmHg), respectively.
The overall median blood glucose level was 144 mg/dL (IQR: 123 mg/dL), with higher levels observed in patients aged ≥ 50 years (158 mg/dL; IQR: 134 mg/dL) compared to those aged < 50 years (131 mg/dL; IQR: 97 mg/dL).
The overall in-hospital mortality rate was 22.90%. Mortality was higher among patients aged ≥ 50 years (28.41%) compared to those aged < 50 years (15.90%). Similarly, mortality was higher among patients with multimorbidity (35.37%) compared to those without multimorbidity (18.62%).
A comparison of age and sex between included and excluded adult patients showed no substantial differences between the two groups (Supplementary Table S2), suggesting that no major differences in age and sex were observed between the groups; however, the possibility of residual selection bias cannot be excluded.
Bivariate analysis of factors associated with in-hospital mortality
Table 2 shows significant bivariate associations between selected variables and COVID-19 mortality. Older age (≥ 50 years, p < 0.001), dyspnea (p < 0.001), comorbidity status (p < 0.001), and specific comorbidities such as hypertension (p = 0.009), diabetes mellitus (p < 0.001), and kidney disease (p < 0.001) were significantly associated with mortality. Non-survivors had significantly higher median blood glucose levels compared to survivors (175 mg/dL vs. 132 mg/dL, p < 0.001), higher pulse rates (p < 0.001), and lower diastolic blood pressure (p = 0.005) compared to survivors. In contrast, sex, length of hospital stays, the majority of symptoms, and vital signs did not show a statistically significant association with mortality.
Table 2.
Comparison of clinical and demographic variables between survivors and non-survivors of COVID-19 patients
| Variables | Survivors (N = 495) |
Non-survivors (N = 147) |
p-value |
|---|---|---|---|
| Age | |||
| < 50 | 238 (84.10) | 45 (15.90) | < 0.001 |
| ≥ 50 | 257 (71.59) | 102 (28.41) | |
| Sex | |||
| Male | 340 (76.85) | 104 (23.42) | 0.635 |
| Female | 155 (78.28) | 43 (21.72) | |
| Length of hospital stay | |||
| < 7 days | 251 (76.99) | 75 (23.01) | 0.947 |
| ≥ 7 days | 244 (77.22) | 72 (22.78) | |
| Signs and symptoms | |||
| Fever | |||
| Yes | 233 (76.64) | 71 (23.36) | 0.793 |
| No | 262 (77.51) | 76 (22.49) | |
| Dyspnea | |||
| Yes | 234 (69.85) | 101 (30.15) | < 0.001 |
| No | 261 (85.02) | 46 (14.98) | |
| Cough | |||
| Yes | 61 (80.26) | 15 (19.74) | 0.485 |
| No | 434 (76.68) | 132 (23.32) | |
| Comorbidity burden | |||
| No multimorbidity | 389 (81.38) | 89 (18.62) | |
| Multimorbidity | 106 (64.63) | 58 (35.37) | < 0.001 |
| Comorbidity type | |||
| Hypertension | |||
| Yes | 139 (70.56) | 58 (29.44) | 0.009 |
| No | 356 (80.00) | 89 (20.00) | |
| Diabetes mellitus | |||
| Yes | 126 (67.38) | 61 (32.62) | < 0.001 |
| No | 369 (81.10) | 86 (18.90) | |
| Kidney disease | |||
| Yes | 8 (34.78) | 15 (65.22) | < 0.001 |
| No | 487 (78.68) | 132 (21.32) | |
| Vital signs, median (IQR) | |||
| SBP (mm Hg) | 120 (20) | 120 (26) | 0.880 |
| DBP (mm Hg) | 80 (12) | 76 (10) | 0.005 |
| Pulse rate (bpm) | 86 (18) | 96 (29) | < 0.001 |
| SpO2 (%) | 95 (7) | 94 (9) | 0.143 |
| Blood glucose (mg/dL) | 132 (106) | 175 (162) | < 0.001 |
Values are presented as frequency (percentage) or median (interquartile range [IQR]), as appropriate. P-values were calculated using the chi-square test or Fisher’s exact test, as appropriate. Continuous variables were compared using the Mann–Whitney U test
Multivariable logistic regression analysis of factors associated with in-hospital mortality
Table 3 illustrates multivariable logestic regression analysisi associted with in-hospital mortality The final analytic sample size used in all multivariable models was N = 642, following complete-case exclusion of missing data across all covariates.
Table 3.
Multivariable logistic regression analysis of factors associated with COVID-19
| Variable | AOR | 95% CI | p-value |
|---|---|---|---|
| Age (ref: <50) | |||
| ≥ 50 | 2.01 | 1.27–3.16 | 0.002 |
| Sex (ref. Male) | |||
| Female | 0.87 | 0.55–1.36 | 0.560 |
| Dyspnea (ref: No) | |||
| Yes | 2.88 | 1.85–4.48 | < 0.001 |
| Multimorbidity (ref: No) | |||
| Yes | 1.63 | 0.67–4.00 | 0.278 |
| Hypertension (ref: No) | |||
| Yes | 0.98 | 0.49–1.97 | 0.969 |
| Diabetes mellitus (ref: No) | |||
| Yes | 1.01 | 0.51–1.99 | 0.965 |
| Kidney disease (ref: No) | |||
| Yes | 6.56 | 2.39–18.02 | < 0.001 |
| DBP (per mmHg) | 0.97 | 0.95–0.99 | 0.007 |
| SpO2 (%) | 0.98 | 0.97-1.00 | 0.287 |
| Pulse rate (per bpm) | 1.02 | 1.01–1.03 | < 0.001 |
| Blood glucose (per 10 mg/dL) | 1.03 | 1.01–1.05 | < 0.001 |
AOR: Adjusted odds ratio, DBP: Diastolic blood pressure
Patients aged ≥ 50 years had higher odds of mortality compared to those aged < 50 years (AOR = 2.01, 95% CI: 1.27–3.16; p = 0.002). When modeled as a continuous variable, age remained significantly associated with mortality (OR = 1.02 per year increase, 95% CI: 1.01–1.03, p < 0.001) (see Supplementary Table S9).
Sex was not significantly associated with in-hospital mortality, although females showed slightly lower odds of mortality compared to males (AOR = 0.87, 95% CI: 0.55–1.36, p = 0.560).
The presence of dyspnea was associated with higher odds of mortality (AOR = 2.88, 95% CI: 1.85–4.48, p < 0.001). Patients with multimorbidity had slightly higher odds of mortality compared to patients without multimorbidity, although this was not statistically significant (AOR = 1.63, 95% CI: 0.67–4.00, p = 0.278). Hypertension and diabetes were not significantly associated with mortality (AOR = 0.98, 95% CI: 0.49–1.97, p = 0.969), (AOR = 1.01, 95% CI: 0.51–1.99, p = 0.965), respectively. Thus, while both conditions were frequent among non-survivors, neither hypertension nor diabetes independently predicted mortality after adjusting for age and other comorbidities.
Kidney disease was strongly associated with mortality (AOR = 6.56, 95% CI: 2.39–18.02, p < 0.001). Lower diastolic blood pressure was significantly associated with increased mortality risk (AOR = 0.97 per 1 mmHg, 95% CI: 0.95–0.99, p = 0.007). Oxygen saturation (SpO₂) was not significantly associated with in-hospital mortality (AOR = 0.98, 95% CI: 0.97–1.00, p = 0.287).
Every 1 beat per minute (bpm) increase in pulse rate raised mortality odds by 1.02 (AOR = 1.02, 95% CI: 1.01–1.03, p < 0.001). Blood glucose was associated with higher odds of mortality per 10 mg/dL increase (AOR = 1.03, 95% CI: 1.01–1.05, p < 0.001).
Multicollinearity was assessed using variance inflation factors (VIF). All variables showed VIF values below 5, indicating no evidence of significant multicollinearity in the model (Supplementary Table S5).
To further examine potential effect modification, an interaction model was fitted based on the main multivariable model (Table 3), with the addition of an interaction term between age and comorbidity burden.
Interaction effects of age and multimorbidity on in-hospital mortality
The interaction analysis presented in Table 4 illustrates the association between age, comorbidity burden, and COVID-19 mortality. The interaction model was based on the same complete-case analytic sample.
Table 4.
Multivariable interaction model examining the association between age, comorbidity burden, and COVID-19 mortality
| Variable | AOR | 95% CI | p-value |
|---|---|---|---|
| Age (ref: <50) | |||
| ≥ 50 | 2.83 | 1.60–5.03 | < 0.001 |
| Multimorbidity (ref: No) | |||
| Yes | 2.74 | 1.33–5.66 | 0.006 |
| Age ≥ 50 × Multimorbidity | 0.47 | 0.19–1.14 | 0.097 |
| Dyspnea (ref: No) | |||
| Yes | 2.78 | 1.81–4.27 | < 0.001 |
| Pulse rate (per bpm) | 1.02 | 1.01–1.03 | < 0.001 |
| Diastolic blood pressure (per mmHg) | 0.97 | 0.95–0.99 | 0.007 |
| Blood glucose (per 10 mg/dL) | 1.03 | 1.01–1.05 | 0.001 |
AOR: Adjusted odds ratio, DBP: Diastolic blood pressure
The interaction between age and multimorbidity was not statistically significant (p > 0.05), indicating no strong evidence of multiplicative interaction.
To facilitate interpretation, predicted probabilities of in-hospital mortality were estimated for each combination of age and multimorbidity status. Mortality was lowest among younger patients without multimorbidity, higher among younger patients with multimorbidity and older patients without multimorbidity, and highest among older patients with multimorbidity, demonstrating a graded increase in mortality risk across groups (Supplementary Table S11).
Figure 2 shows the estimated probability of death in hospitalized patients with COVID-19 stratified by age group (< 50 years and ≥ 50 years) and the burden of comorbidities (No multimorbidity vs. multimorbidity). The analysis was carried out using a multivariable logistic regression model with an interaction between age and multimorbidity. The graph depicts an increase in the probability of death with increasing age among patients without multimorbidity as well as those with multimorbidity. Additionally, within each age category, patients with multimorbidity had a higher predicted probability of death compared to those without multimorbidity. Although the plotted lines appear approximately parallel, this visual pattern should be interpreted cautiously, as a lack of statistical interaction does not necessarily imply biological independence; the interaction term was not statistically significant.
Fig. 2.
Predicted probability of COVID-19 mortality by age group and comorbidity burden
Discussion
This study evaluated the determinants of in-hospital mortality among COVID-19 patients, with a particular focus on the role of age and comorbidity burden. Our findings show that patients aged ≥ 50 years had significantly higher odds of mortality compared to younger individuals (AOR = 2.01, 95% CI: 1.27–3.16; p = 0.002), consistent with prior literature demonstrating age as a significant predictor of worse disease outcomes [11, 12, 22]. Immunosenescence, alterations in cytokine responses, and the higher burden of chronic conditions in older individuals are probable mechanisms for such susceptibility [22].
Multimorbidity showed higher crude mortality proportions but was not independently associated with mortality after adjustment (AOR = 1.63; p = 0.278). This may reflect residual confounding, heterogeneity in comorbidity patterns, or the influence of stronger predictors such as age and organ-specific conditions. This is partially consistent with prior findings that multimorbidity is implicated in increased risks of adverse outcomes, including admission to the ICU and death [17, 18]. Notably, patients with multimorbidity accounted for a higher proportion of death (35.37%) compared to those without multimorbidity (18.62%), underscoring the clinical relevance of comorbidity burden in this cohort.
Dyspnea was among the strongest predictors of mortality, with an almost threefold increase in odds of death (AOR, 2.88; p < 0.001). This indicates substantial respiratory compromise and reflects extensive pulmonary involvement, correlating with progression to acute respiratory distress syndrome and fatal outcomes, consistent with prior reports [3, 4].
Kidney disease was another major determinant of mortality and was associated with substantially increased odds of mortality (AOR = 6.56, 95% CI: 2.39 − 18.02, p < 0.001). Renal impairment may aggravate systemic inflammatory responses, and may impair drug clearance clearance, leading to impaired fluid and electrolyte balance, which may complicate COVID-19 management [23]. This highlights the need for enhanced clinical monitoring or even prioritized interventional approaches toward patients with renal comorbidities.
Blood glucose levels were also associated with increased mortality risk. When modeled per 10 mg/dL increase, higher glucose levels were significantly associated with mortality, enhancing the clinical interpretability of the effect size. Although the magnitude of the association was modest, it suggests that larger elevations in blood glucose may meaningfully impact patient outcomes, potentially reflecting metabolic stress or uncontrolled glycemic status in severely ill patients.
It is important to distinguish between baseline risk factors and acute severity markers in interpreting these findings. Variables such as age and chronic kidney disease represent underlying patient vulnerability, whereas dyspnea, pulse rate, blood pressure, and admission glucose reflect the severity of illness at presentation. Therefore, the latter should be interpreted as indicators of disease severity rather than independent causal determinants.
Hypertension and diabetes were individually associated with mortality in unadjusted analyses; however, in multivariable analyses and interaction models, the association of mortality with these variables weakened. This attenuation suggests that their effects may have been confounded or mediated by stronger predictors such as age and renal dysfunction, or overall comorbidity burden, as suggested in recent meta-analyses [14, 16]. Although these conditions were prevalent among patients who died, they did not independently predict mortality after adjustment, suggesting that these factors should be viewed as components of the broader multimorbidity profile linked to COVID-19 outcomes rather than as independent predictors in this cohort.
Furthermore, the large confidence intervals of some of the predictors, such as kidney disease and multimorbidity, reflect uncertainty in the estimates due to small subgroup sizes, emphasizing the need for cautious interpretation and validation in larger multicenter studies.
Advanced age was independently associated with increased mortality in hospitalized COVID-19 patients, while multimorbidity showed higher crude mortality but was not independently associated after adjustment. The interaction analysis did not demonstrate a statistically significant multiplicative interaction between age and multimorbidity (p > 0.05). However, post-estimation analyses using predicted probabilities demonstrated a clear gradient in mortality risk across age–multimorbidity groups, with the highest risk observed among older patients with multimorbidity. These findings suggest that although a formal statistical interaction was not detected, the combined presence of older age and multimorbidity is clinically associated with an increased mortality risk.
This study was conducted over a period spanning multiple phases of the COVID-19 pandemic (October 2020 to September 2021), during which mortality risk may have varied due to evolving treatment protocols, healthcare system capacity, and the introduction of vaccination. As data on calendar time, treatment regimens, and vaccination status were not available, these factors were not accounted for in the analysis and may have influenced the observed associations.
These findings contribute to the growing body of evidence by providing a single-center analysis of the independent and combined effects of age and comorbidity burden on COVID-19 mortality in a South Indian tertiary care hospital, with implications for clinical risk stratification and healthcare planning in similar contexts.
Strengths and limitations
This study has several strengths. It utilized a relatively large cohort of hospitalized COVID-19 patients and applied robust multivariable and interaction analyses to assess both independent and combined effects of age and comorbidity burden. Additionally, the use of detailed clinical and laboratory data, along with transparent handling of missing data, strengthens the validity of the findings. However, several limitations should be considered.
This single-center retrospective design may limit the generalizability of the findings to other patient populations and healthcare settings. Additionally, a substantial proportion of patients were excluded due to incomplete records or missing outcome data, which may have introduced selection bias. Patients with missing data may have differed systematically from those included in the analysis, potentially affecting the validity of the findings. Since a significant percentage of the excluded cases lacked mortality status and several important covariates, multiple imputation was not performed because reliable imputation was not feasible. Consequently, a complete-case analysis was used, and the results should be interpreted with caution. Although missingness in predictor variables was relatively low, complete-case analysis may still introduce bias if the missingness mechanism is not completely at random.
Age was analyzed as a dichotomous variable (≥ 50 vs. <50), which may have resulted in loss of information and reduced statistical power. This approach may also obscure potential nonlinear relationships between age and mortality risk. However, sensitivity analyses using alternative age cutoffs yielded consistent findings, supporting the robustness of the results.
Moreover, standardized comorbidity severity scores (i.e., Charlson or Elixhauser indices) were not available in the dataset. As a result, comorbidity burden was categorized as no multimorbidity or multimorbidity based on the number of chronic conditions recorded in the medical records rather than by severity. However, this simplified classification does not account for the severity or differential impact of individual conditions, which may lead to misclassification or attenuation of the true effect of comorbidity burden on mortality. As a result, the observed associations may underestimate or obscure the relative contribution of more severe or high-risk conditions. Future studies using standardized indices could estimate the impact of comorbidity burden with greater accuracy and comparability. Furthermore, due to small subgroup sizes, many odds ratios (especially those for kidney disease and multimorbidity) had wide confidence intervals, indicating statistical uncertainty. As with all retrospective analyses, residual confounding cannot be fully ruled out.
Conclusion
Advanced age, dyspnea, kidney disease, and clinical indicators such as elevated blood glucose levels and pulse rate are significant predictors of COVID-19 mortality. Although multimorbidity was associated with higher crude mortality, it was not independently associated with mortality after adjustment, and no strong evidence of a statistical interaction between age and multimorbidity was observed. These findings highlight the importance of comprehensive risk stratification using multiple clinical factors in hospitalized patients. Future research should explore longitudinal outcomes and the potential impact of targeted management of comorbid conditions on mortality.
Electronic Supplementary Material
Below is the link to the electronic supplementary material.
Acknowledgements
The authors acknowledge the administration and staff of KLE’s Dr. Prabhakar Kore Hospital & Medical Research Center, Belgaum, Karnataka, India, for their support with record selection and data extraction. N.A. also wishes to express her sincere gratitude to the Kuwaiti Women Philanthropic Team for their generous support and encouragement throughout her academic journey.
Abbreviations
- AOR
Adjusted odds ratio
- ARDS
Acute respiratory distress syndrome
- BP
Blood pressure
- bpm
Beats per minute
- CI
Confidence interval
- COVID-19
Coronavirus disease 2019
- DBP
Diastolic blood pressure
- ICU
Intensive care unit
- IQR
Interquartile range
- mg/dL
Milligrams per deciliter
- RT-PCR
Reverse transcription polymerase chain reaction
- SARS-CoV-2
Severe acute respiratory syndrome coronavirus 2
- SBP
Systolic blood pressure
- SD
Standard deviation
- SpO₂
Peripheral capillary oxygen saturation
- VIF
Variance inflation factor
- WHO
World Health Organization
Author contributions
Conceptualization, data collection, and data analysis were conducted by NA. Manuscript writing and revision were done by NA, AM, and MA.
Funding
None.
Data availability
The corresponding author will provide the data supporting the study’s findings upon a reasonable request.
Declarations
Ethics approval and consent to participate
Ethical clearance of this study was obtained from the Institutional Human Ethics Committee, Jawaharlal Nehru Medical College, KLE Academy of Higher Education and Research, Belagavi, Karnataka, India (Ref. No.: MDC/DOME/243). Informed consent was waived as this study involved secondary analysis of anonymized retrospective data. The study protocol was approved by the committee. The study was conducted in accordance with institutional guidelines and the ethical principles of the Declaration of Helsinki. Permission for access to and use of de-identified patient records was granted by the Clinical Services Administrator, Medical Director, and Chief Executive Officer of KLE Hospital, as well as from the Medical Records Department. During analysis, all direct identifiers were removed, and the data were fully anonymized to protect participant confidentiality.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Zhu N, Zhang D, Wang W, Li X, Yang B, Song J, et al. A Novel Coronavirus from Patients with Pneumonia in China, 2019. N Engl J Med. 2020;382:727–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.World Health Organization. Clinical management of severe acute respiratory infection (SARI) when COVID-19 disease is suspected. Geneva: WHO; 2020. p. 1–19. Available from: https://www.who.int/docs/default-source/coronaviruse/clinical-management-of-novel-cov.pdf.
- 3.Huang C, Wang Y, Li X, Ren L, Zhao J, Hu Y et al. Clinical features of patients infected with 2019 novel coronavirus in Wuhan, China. Lancet [Internet]. 2020;395(January):497–506. Available from: https://www.thelancet.com/article/S0140-6736(20)30183-5/fulltext. [DOI] [PMC free article] [PubMed]
- 4.Wiersinga WJ, Rhodes A, Cheng AC, Peacock SJ, Prescott HC. Pathophysiology Transmission, Diagnosis, and Treatment of Coronavirus Disease 2019 (COVID-19) A Review. JAMA - J Am Med Assoc. 2020;324(8):782–93. [DOI] [PubMed] [Google Scholar]
- 5.Remuzzi A, Remuzzi G. Health Policy COVID-19 and Italy: what next? Lancet [Internet]. 2020;395(10231):1225–8. Available from: 10.1016/S0140-6736(20)30627-9. [DOI] [PMC free article] [PubMed]
- 6.Mathieu E, Ritchie H, Rodés-Guirao L, Appel C, Gavrilov D, Charlie Giattino J, Hasell BM, et al. Our world in data. COVID-19 Pandemic. 2025 [cited 2025 Apr 1]. Available from: https://ourworldindata.org/coronavirus.
- 7.Jindal R, Gupta M, Khan FR, Chaudhry G. Prevalence of co-morbidities and its association with mortality in Indian patients with COVID-19: a meta-analysis. Indian J Anaesth [Internet]. 2022;66(6):399–418. Available from: https://journals.lww.com/10.4103/ija.ija_845_21. [DOI] [PMC free article] [PubMed]
- 8.Onder G, Rezza G, Brusaferro S. Case-Fatality Rate and Characteristics of Patients Dying in Relation to COVID-19 in Italy. JAMA - J Am Med Assoc. 2020;323(18):1775–6. [DOI] [PubMed] [Google Scholar]
- 9.Richardson S, Hirsch JS, Narasimhan M, Crawford JM, Mcginn T, Davidson KW. Presenting Characteristics, Comorbidities, and Outcomes Among 5700 Patients Hospitalized With COVID-19 in the New York City Area. JAMA - J Am Med Assoc. 2020;323(20):2052–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Al-Aghbari N, Maldar A, Angolkar M, Khurseed R, Shrestha S, Saikam P. Inflammatory Markers for Prognosis of COVID-19 Mortality in Hospitalized Indian Patients: A Single-Center Retrospective Study. Univ Sci Technol J Med Sci. 2023;1(1):1–7. [Google Scholar]
- 11.Bonanad C, García-Blas S. The effect of age on mortality in patients with COVID-19: a meta-analysis with 611,583 subjects. JAMDA J [Internet]. 2020;21(7):915–918. Available from: https://pubmed.ncbi.nlm.nih.gov/32674819/. [DOI] [PMC free article] [PubMed]
- 12.Mueller AL, Mcnamara MS, Sinclair DA. Sinclair. Why does COVID-19 disproportionately affect older people? Aging (Albany NY) [Internet]. 2020;12(10):9959–9981. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC7288963/. [DOI] [PMC free article] [PubMed]
- 13.Guan W jie, Liang W hua, Zhao Y, Liang H rui, Chen Z sheng, Li Y min, et al. Comorbidity and its impact on 1590 patients with COVID-19 in China: a nationwide analysis. Eur Respir J. 2020;55(5):2000547. [DOI] [PMC free article] [PubMed]
- 14.Zhou F, Yu T, Du R, Fan G, Liu Y, Liu Z, et al. Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study. Lancet [Internet]. 2020;395(10229):1054–1062. Available from: https://pubmed.ncbi.nlm.nih.gov/32171076/. [DOI] [PMC free article] [PubMed]
- 15.Ssentongo P, Ssentongo AE, Heilbrunn ES, Ba DM, Chinchilli VM. Association of Cardiovascular Disease and 10 Other Pre-existing Comorbidities with COVID- 19 Mortality: A Systematic Review and Meta- Analysis. PLoS ONE. 2020;15(8):e0238215. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Williamson EJ, Walker AJ, Bhaskaran K, Bacon S, Morton CE, Curtis HJ, et al. OpenSAFELY: factors associated with COVID-19 death in 17 million patients. Nature [Internet]. 2021;584(7821):430–6. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC7611074/. [DOI] [PMC free article] [PubMed]
- 17.Lagadinou M, Salomou EE, Zareifopoulos N, Marangos M, Gogos C, Velissaris D, et al. Prognosis of COVID-19: changes in laboratory parameters. Le Infez Med [Internet]. 2020;28(suppl 1):89–95. Available from: http://www.ncbi.nlm.nih.gov/pubmed/32532944. [PubMed]
- 18.Cho SI, Yoon S, Lee HJ. Impact of comorbidity burden on mortality in patients with COVID-19 using the Korean health insurance database. Sci Rep [Internet]. 2021;11(6375):1–9. Available from: 10.1038/s41598-021-85813-2. [DOI] [PMC free article] [PubMed]
- 19.Padmaprakash KV, Vardhan V, Thareja S, Muthukrishnan J, Raman N, Kumar K, et al. Clinical characteristics and clinical predictors of mortality in hospitalised patients of COVID 19: an Indian study. Med J Armed Forces India [Internet]. 2021;77(Suppl C):S319–32. Available from: 10.1016/j.mjafi.2021.01.009. [DOI] [PMC free article] [PubMed]
- 20.Singh P, Bhaskar Y, Verma P, Rana S, Goel P, Kumar S, et al. Impact of comorbidity on patients with COVID-19 in India: a nationwide analysis. Front public Heal. 2022;10:1027312. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lucijanic M, Marelic D, Stojic J, Markovic I, Sedlic F, Kralj I, et al. Predictors of prolonged hospitalization of COVID-19 patients. Eur Geriatr Med. 2023;14(3):511–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Chen N, Zhou M, Dong X, Qu J, Gong F, Han Y, et al. Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study. Lancet [Internet]. 2020;395(10223):507–513. Available from: https://pubmed.ncbi.nlm.nih.gov/32007143/. [DOI] [PMC free article] [PubMed]
- 23.Michael B, Giuseppe H, Henry BM. Chronic Kidney disease is associated with severe coronavirus disease 2019 (COVID-19) infection. Int Urol Nephrol [Internet]. 2020;52(6):1193–4. Available from: 10.1007/s11255-020-02451-9. [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The corresponding author will provide the data supporting the study’s findings upon a reasonable request.