A Clinical Prediction Model for Atypical Tuberculosis Manifestations Among Older Adults

Area of Science:

• Geriatric Medicine
• Infectious Diseases
• Clinical Prediction Modeling

Background:

• Active pulmonary tuberculosis (aPTB) is often under-recognized in older adults (≥75 years) due to atypical symptoms and comorbidities.
• Early identification of aPTB in this population is crucial but challenging in non-pulmonology settings.

Purpose of the Study:

• To develop and validate a clinical prediction model for the early identification of atypical aPTB in adults aged 75 years and older.
• To create a TRIPOD-compliant model to aid non-pulmonologists in recognizing aPTB.

Main Methods:

• Retrospective analysis of 5651 patients aged ≥75 years with culture-confirmed aPTB and low WHO symptom scores (<5).
• Development of a weighted scoring system based on multivariate logistic regression identifying predictors like advanced age (>85), hypoalbuminemia, cardiovascular disease, diabetes, and lower lung field involvement.
• Validation in derivation and temporal cohorts using TRIPOD standards, including discrimination, calibration, and decision curve analysis (DCA).

Main Results:

• Five key predictors were identified: age > 85 years, hypoalbuminemia, cardiovascular disease, diabetes mellitus, and lower lung field involvement.
• The developed scoring model demonstrated excellent performance (AUC 0.95-0.96, sensitivity 91-94%, specificity 97-99%) across validation cohorts.
• Nearly 70% of atypical cases had very low symptom scores (≤1), highlighting the need for predictive tools beyond typical symptoms.

Conclusions:

• Individuals over 85 years old are the strongest predictor of atypical TB, outperforming traditional frailty indicators.
• The validated scoring system offers a simple, accurate tool for early aPTB detection in non-pulmonology settings.
• Integration into electronic medical records can reduce diagnostic delays and improve health outcomes for elderly patients with aPTB.