Clinical characteristics, radiological profiles, and factors associated with mortality in COVID-19-associated pneumonia: a retrospective cohort analysis

COVID-19 pneumonia remains a major cause of morbidity and in-hospital mortality. Identifying clinical, demographic, and radiological factors independently associated with severity and mortality is essential for risk stratification, particularly in resource-limited settings where serial chest radiography is the principal imaging modality. We characterized the clinical and radiological profiles and identified independent predictors of in-hospital mortality in a cohort of hospitalized patients with COVID-19 pneumonia recruited during the first pandemic wave in Saudi Arabia. This retrospective single-center cohort study included 286 consecutive adult patients (≥ 18 years) with reverse transcription polymerase chain reaction (RT-PCR)-confirmed SARS-CoV-2 infection and clinical and radiological evidence of pneumonia, admitted to King Abdulaziz Medical City, Jeddah, between March and October 2020 (pre-vaccination, ancestral, and early-variant periods). Demographic data, symptoms, comorbidities, and findings on initial and follow-up chest radiographs (CXR) and chest computed tomography (CT) were extracted from electronic medical records using a standardized form. Univariate associations with severity and in-hospital mortality were tested using the Mann–Whitney U test, Fisher’s exact test, or the chi-squared test, with Benjamini–Hochberg false discovery rate correction within each analytic family. Factors independently associated with in-hospital mortality were identified through multivariable logistic regression with pre-specified clinical and demographic covariates and backward elimination; radiographic variables were not entered into the model. A sensitivity analysis re-included the 7 patients aged < 18 years who had been admitted under the early-pandemic institutional policy. The cohort comprised 286 adult patients (162 male, 56.6%; median age 57 years, IQR 45–70). Severity classification was available for 267 patients: 205 (76.8%) had mild-to-moderate pneumonia and 62 (23.2%) had severe pneumonia; severity was missing in 19 (6.6%). Overall in-hospital mortality was 14.3% (41/286), rising to 29.0% in the severe group versus 10.7% in the mild-to-moderate group ( p < 0.001). Imaging-confirmed interstitial lung disease (ILD) on CT was documented in 6 patients (2.1%); pulmonary embolism occurred in 2 patients (0.7%) and no venous thromboembolism was recorded. On the initial CXR, the predominant findings were consolidation (57.3%) and ground-glass opacity (GGO; 52.4%), with bilateral involvement present in 37.8%. Bilateral involvement on the initial CXR was significantly more common in patients who died than in those discharged alive (65.9% vs. 33.1%; raw p < 0.001, BH-adjusted p < 0.001), as were consolidation (78.0% vs. 53.9%) and diffuse distribution (17.1% vs. 3.3%); differences between outcome groups were larger on the follow-up CXR, with bilateral involvement in 85.4% of deceased versus 47.3% of discharged patients ( p < 0.001). In multivariable logistic regression, three variables remained independently associated with in-hospital mortality after backward elimination: advanced age (adjusted OR per 10-year increment 2.19, 95% CI 1.62–2.96; p < 0.001), male sex (aOR 2.47, 95% CI 1.07–5.71; p = 0.035), and severe pneumonia at presentation (aOR 3.89, 95% CI 1.75–8.64; p = 0.001). Hypertension and diabetes, both significantly associated with mortality in univariate analysis, and ischemic heart disease, malignancy, and general weakness did not retain independent associations after adjustment. The model showed acceptable goodness of fit (Hosmer–Lemeshow p = 0.62), good discrimination (AUROC 0.83, 95% CI 0.76–0.90), and a non-significant age × severity interaction ( p = 0.11). Estimates were essentially unchanged when the 7 pediatric patients were re-included ( n = 293, 42 deaths). In this pre-vaccination cohort of adults hospitalized with COVID-19 pneumonia in Saudi Arabia, advanced age, male sex, and severe pneumonia at presentation were independently associated with in-hospital mortality. The univariate associations of hypertension, diabetes, ischemic heart disease, and general weakness with mortality were attenuated after multivariable adjustment, consistent with confounding by age, severity, and underlying frailty. Bilateral, consolidative, and diffuse radiographic patterns at presentation differed markedly between survivors and non-survivors, although these are cross-sectional associations rather than validated predictors. These findings support an integrated, multivariable risk-stratification approach for hospitalized COVID-19 pneumonia and underscore the need for prospective revalidation in vaccinated and Omicron-era cohorts.