Abstract
BACKGROUND: Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), produces outcomes ranging from mild illness to death. Although in-hospital prognostic factors are well described, long-term mortality data are scarce, particularly from middle-income countries with relatively young populations. We therefore quantified one-year all-cause mortality and identified independent predictors in a large Turkish cohort support follow-up planning and prevention for high-risk groups. METHODS: We retrospectively included all adults (≥18 years) with polymerase chain reaction-confirmed SARS-CoV-2 infection who presented to Haydarpaşa Numune Training and Research Hospital between 1 March 2020 and 31 January 2021, covering the pre-vaccination period in Türkiye. Demographics, smoking status, and ICD-10 coded comorbidities were extracted from hospital's electronic medical record system and national electronic health database. Follow-up was complete: 98.9% had 365 days; the remaining 168 had 360-364 days and were administratively censored, with no losses to follow-up. Missing smoking data was imputed by multiple imputations with chained equations. One-year survival status was obtained from the death registry. Risk factors were examined by Cox proportional hazards models after verifying assumptions. RESULTS: Among 14,975 patients (median age 40 years, interquartile range 28.5-52.2; 50.8% male), 357 deaths occurred within 365 days, giving one-year mortality of 2.4% (95% CI [2.1-2.6]%). Non-survivors were older (71.8 vs 39.5 years), more often male (63% vs 50.5%), and had higher rates of type 2 diabetes mellitus, ischemic heart disease, heart failure, chronic kidney disease, and cancer. In multivariable analysis each additional year of age increased risk by 10%. Other independent predictors were male sex, type 2 diabetes mellitus, ischemic heart disease, heart failure, chronic kidney disease, and cancer. Smoking was inversely associated with mortality after inverse-probability weighting. Model discrimination was excellent. Only cancer showed minor time-dependency, mainly within the first 60 days of follow-up; two-thirds of cancer-related deaths occurred by day 60. Bootstrap optimism correction confirmed robustness. CONCLUSION: One-year mortality after COVID-19 in this relatively young Turkish population was low yet clinically meaningful. Persistent excess risk among older adults, men, and patients with cancer or major cardiometabolic disease supports structured post-acute surveillance, prioritized vaccination boosters, and aggressive management of underlying conditions. The inverse association with smoking is probably due to residual or unmeasured confounding and should not influence practice. These findings refine long-term risk stratification and can guide resource allocation as COVID-19 becomes an endemic threat external validation that incorporates vaccination status and viral variants is warranted.