Distribution and diagnostic modeling of osteoporosis and comorbidities across demographic factors: A cross-sectional study of 2224 female patients

Abstract

Objective

This study investigates the distribution of osteoporosis (OP) and its associated comorbidities across different demographic factors. Furthermore, this study seeks to develop a statistically-based diagnostic model leveraging demographic and health indicators to provide personalized risk assessments for OP.

Methods

A retrospective analysis was conducted on the demographic data, health profiles, and bone density measurements of 2224 female patients. Key variables associated with OP were identified using chi-square tests. Feature selection was refined through Lasso regression and recursive feature elimination (RFE), which guided the development of a logistic regression-based dynamic nomogram. This model was subsequently implemented on the Shiny platform for personalized online OP risk assessments.

Results

Among 2224 female patients, 801 (36.0 %) were diagnosed with OP. Women aged 70 and older exhibited a significantly higher prevalence of OP compared to younger age groups (OR = 5.83, 95 % CI: 1.74–19.61, P = 0.004), and this remained significant in the multivariable analysis (OR = 5.18, 95 % CI: 1.19–22.52, P = 0.028). Later age at menarche was associated with increased OP risk (OR = 1.31, 95 % CI: 1.09–1.57, P = 0.004), persisting in multivariable analysis (OR = 1.25, 95 % CI: 1.03–1.52, P = 0.023). In rheumatoid arthritis (RA) patients, higher education reduced OP risk, with secondary education (OR = 0.09, P = 0.024) and college education (OR = 0.04, P = 0.009) showing protective effects. Diabetic patients who were unmarried or had non-traditional marital statuses showed increased OP risk (univariate OR = 2.73, P = 0.006; multivariate OR = 2.34, P = 0.029). Among nonalcoholic fatty liver disease (NAFLD) patients, age at menopause was significantly linked to OP risk (univariate OR = 1.04, P = 0.012). The prediction model showed strong performance (AUC = 0.720), and the dynamic nomogram on the Shiny platform provided effective personalized OP risk assessments.

Conclusion

Age and age at menarche are significant risk factors for OP, with later menarche increasing the risk. In RA patients, higher education levels were associated with a lower risk of OP. In contrast, unmarried or non-traditional marital statuses increased OP risk among diabetic patients. Additionally, age at menopause was found to be a significant factor for OP risk in NAFLD patients. The prediction model developed in this study, with an AUC of 0.720, provides a reliable method for personalized OP risk assessment through a dynamic nomogram. These findings highlight the crucial role of demographic factors in predicting OP risk and underscore the importance of personalized treatment strategies for effective OP prevention and management.