Background: Chronic diseases continue to be the leading cause of morbidity and mortality in the United States; therefore, advanced statistical models are needed to both quantify the natural history of these outcomes and factor time-to-event components into data analysis. The field of survival analysis, a bedrock in longitudinal data analysis, which is essential for estimating hazards, prognosis, and population health measures. Objective of the review: This study used a systematic approach to synthesize the utilization of survival analysis methods in U.S.-based chronic disease research, identify trends in methodology, appraise methodological quality, and recommend ways to improve it. Methods: A search was performed in peer-reviewed studies from 2010 to 2024 using Google Scholar, PubMed, Embase, and Web of Science according to PRISMA 2020 guidelines that employed survival models on chronic disease outcomes in U.S. populations. Disease focus, survival outcomes model used, data sources, model assumptions, and reporting practices were extracted from studies eligible for inclusion. The risk of bias was described regarding the Quality in Prognostic Studies tool and synthesized using either narrative or frequency tables, as appropriate. Results: Forty-six studies out of 278 records screened met the inclusion criteria. The most prevalent method was the Cox hazards model (78%), followed by Kaplan-Meier estimators, parametric models, and a few applications of machine learning-based survival models. The majority of studies used national datasets (e.g., SEER, NHANES) and few reported evaluating model assumptions or external validation. The most common deficiencies were in handling time-dependent covariates and censoring. Conclusion: Survival analysis is fundamental in chronic disease research; however, its application frequently lacks methodological discipline. Amplified use of best-available models, thorough diagnostics, and transparent reporting are necessary to meet the advancing expectations of precision medicine and an ageing population. In addition, future research is needed to support innovations in methods and capacity building of biostatistics in applied research.
Kizza et al. (Tue,) studied this question.