Background:  Non-Hodgkin lymphoma (NHL) comprises a heterogeneous group of hematological malignancies whose prognosis is shaped by the interaction between tumor biology, patient characteristics, and health system determinants. Although survival has improved markedly in high-income countries due to advances in diagnostics and targeted therapies, outcomes in low- and middle-income countries (LMICs) remain less favorable. In Colombia, population-based cancer registries provide robust information on incidence and mortality; however, detailed institutional evidence on survival patterns and prognostic factors of NHL is still scarce. Hospital-based cancer registries offer a valuable framework to integrate clinical, pathological, therapeutic, and social information and to explore survival inequalities in real-world care settings.
Methods:  A retrospective cohort study was conducted using data from a hospital-based cancer registry at a high-complexity referral institution in Cali, Colombia. Adult patients (≥18 years) with a first diagnosis of NHL between 2015 and 2022 were included and followed until December 31, 2023. Sociodemographic, clinical, histopathological, and treatment-related variables were collected following international standards. Overall survival was estimated using the Kaplan–Meier method, and differences between groups were assessed with log-rank tests. Independent prognostic factors for mortality were identified using multivariable Cox proportional hazards regression, integrating biological and social determinants. Model assumptions and internal validity were evaluated.
Results:  The cohort comprised 1,421 patients with NHL, with a median age at diagnosis of 61 years and a balanced sex distribution. B-cell lymphomas predominated (87.7%), while T-cell lymphomas accounted for 10.7%. More than half of patients were diagnosed at advanced clinical stages (III–IV), and aggressive biological behavior was common. Estimated overall survival was 81.3% at one year and 59.7% at five years, comparable to reports from other middle-income settings but lower than survival observed in high-income countries.
In multivariable analysis, age over 60 years, middle and low socioeconomic strata, T-cell lineage, aggressive tumor behavior, and disease recurrence were independently associated with increased mortality risk. Patients who never achieved a disease-free period experienced particularly poor outcomes. Although clinical stage and health insurance regime were associated with survival in unadjusted analyses, their independent effects diminished after adjustment for tumor biology and recurrence. The Cox model demonstrated good internal validity and compliance with proportional hazards assumptions.
Conclusions:  Survival from NHL in this institutional cohort reflects a complex interaction between biological aggressiveness and structural social inequalities. Beyond tumor-related factors, socioeconomic conditions significantly influenced outcomes, even within a high-complexity care setting. These findings underscore the importance of hospital-based cancer registries as essential complements to population-based registries for cancer surveillance in LMICs. Strengthening early diagnosis, ensuring continuity of treatment, and addressing inequities in access and quality of care are critical to improving NHL survival. Institutional registry data provide actionable evidence to inform clinical practice, cancer surveillance, and equity-oriented health policies in resource-constrained settings.