A multicenter retrospective study has constructed and validated a clinical prediction nomogram designed to assess the risk of venous thromboembolism among elderly inpatients aged 65 years and older. Developed and evaluated across three medical institutions, the statistical modeling tool demonstrated that a prior history of blood transfusion, the Caprini risk assessment score, C-reactive protein levels measured as a continuous metric, coronary heart disease, and cerebrovascular disease serve as independent predictors of in-hospital venous thromboembolic events. Across both derivation and temporal validation cohorts, the resulting nomogram exhibited strong discriminatory accuracy and favorable calibration for stratifying relative risk in older hospitalized individuals.
The primary objective of the investigation was to address venous thromboembolism in the elderly inpatient demographic by isolating independent clinical risk factors, translating these variables into an accessible nomogram, and evaluating its overall predictive performance. By constructing a quantitative nomogram within the R software environment, the investigators sought to provide clinicians with a standardized, reproducible graphical scoring system to aid relative risk evaluation during acute hospital care.
To develop the predictive framework, the investigators conducted a multicenter, retrospective case–control study utilizing one-to-one frequency matching. The study population was gathered from three distinct medical facilities, encompassing one municipal hospital and two county-level hospital centers. Patient admissions spanning the two-year period from January 1, 2022, through December 31, 2023, were evaluated against rigorous inclusion and exclusion criteria. Following screening and exclusion, the derivation cohort enrolled a total of 496 elderly inpatients, comprising 248 confirmed venous thromboembolism cases and 248 frequency-matched non-venous thromboembolism control patients aged 65 years or older.
Candidate variable selection for the model combined an a priori selection approach with complementary univariate statistical screening. Variables demonstrating potential relevance were subsequently evaluated through multivariable logistic regression modeling to pinpoint factors that remained independently associated with in-hospital venous thromboembolism. The multivariable regression analysis confirmed five statistically independent clinical predictors: a documented history of blood transfusion, baseline Caprini risk score, C-reactive protein concentrations analyzed continuously, the presence of coronary heart disease, and diagnosed cerebrovascular disease. These five distinct variables were subsequently parameterized and integrated to construct the final nomogram in R.
In performance assessments within the derivation cohort, the nomogram demonstrated strong discriminatory ability, achieving an area under the receiver operating characteristic curve of 0.903, with a 95 percent confidence interval extending from 0.877 to 0.930. The investigators also determined the optimal cutoff threshold for clinical risk stratification. At this optimal operating cutoff value, the five-variable nomogram yielded a diagnostic sensitivity of 0.86 alongside a diagnostic specificity of 0.82, reflecting balanced accuracy in identifying elderly inpatients who developed venous thromboembolism while minimizing misclassification among control subjects.
To confirm model reliability over time, the research team conducted temporal validation within an entirely independent cohort of elderly inpatients admitted during the subsequent calendar year, from January through December 2024. This temporal validation cohort enrolled 206 elderly inpatients, consisting of 103 venous thromboembolism cases and 103 matched non-thromboembolism controls. In this independent validation cohort, the nomogram retained high discriminatory capacity, registering an area under the receiver operating characteristic curve of 0.891, with a 95 percent confidence interval from 0.845 to 0.936. Model calibration was evaluated using the Hosmer-Lemeshow goodness-of-fit test, which demonstrated good agreement between predicted probabilities and observed clinical outcomes in both the derivation cohort (chi-square equals 8.24, P equals 0.413) and the validation cohort (chi-square equals 7.56, P equals 0.477). In addition, the model achieved a Brier score of 0.12 within the derivation cohort, further confirming favorable predictive accuracy.
The authors concluded that the proposed nomogram successfully synthesizes five readily accessible clinical predictors into a cohesive tool for relative venous thromboembolism risk stratification among hospitalized elderly patients. By enabling earlier identification of high-risk individuals, the tool may assist clinical practitioners in implementing targeted thromboprophylaxis strategies during hospital admission. However, the researchers underscored that because the analysis was conducted retrospectively and assesses relative risk stratification, additional prospective clinical validation will be required to confirm the nomogram’s absolute risk estimation capabilities and verify its real-world clinical utility across diverse inpatient populations.
Subject of Research: Clinical nomogram for venous thromboembolism risk prediction in elderly inpatients
Article Title: Construction and validation of a nomogram for venous thromboembolism risk in elderly inpatients: a multicenter retrospective study
Article References: Xing, Y., Yan, F., Xiang, Y., Huang, L., Gong, G., Sun, J., Zhang, J., & Tian, F. (2026). Construction and validation of a nomogram for venous thromboembolism risk in elderly inpatients: a multicenter retrospective study. BMC Cardiovascular Disorders. https://doi.org/10.1186/s12872-026-06940-0
Image Credits: AI Generated
DOI: 10.1186/s12872-026-06940-0
Keywords: venous thromboembolism, nomogram, elderly inpatients, geriatric medicine, Caprini score, C-reactive protein, blood transfusion history, coronary heart disease, cerebrovascular disease, multicenter study, case-control study, risk stratification, thromboprophylaxis, temporal validation, logistic regression, receiver operating characteristic, Hosmer-Lemeshow test, Brier score
News Source: Ophelia Keating. (November 11, 2026). Five-Factor Nomogram Accurately Predicts Venous Thromboembolism Risk in Elderly Inpatients. Scienmag.



