Tuberculin skin test: estimation of positive and negative predictive values from routine data

Area of Science:

• Infectious Diseases
• Public Health
• Diagnostic Accuracy

Background:

• Latent tuberculosis infection (LTBI) diagnosis relies on accurate testing methods.
• Tuberculin skin testing (TST) is a common diagnostic tool for LTBI.
• Establishing appropriate cut-off values for TST is crucial for reliable LTBI detection.

Purpose of the Study:

• To estimate the predictive values of TST using 2TU RT23 for LTBI detection in the Dutch population.
• To provide data-driven recommendations for TST cut-off values based on varying infection prevalence.
• To assess the utility of routinely available data for TST predictive value estimation.

Main Methods:

• Utilized smoothed distributions of TST reactions from tuberculosis patients and healthy non-BCG-vaccinated individuals.
• Estimated sensitivity and specificity of TST at various cut-off values (5 mm, 10 mm, 15 mm).
• Calculated positive and negative predictive values (PPV/NPV) based on different infection prevalence levels.

Main Results:

• Sensitivity ranged from 79.8% (15 mm) to 98.9% (5 mm).
• Specificity ranged from 95.3% (5 mm) to 97.1% (15 mm) before adjustment, and higher post-adjustment.
• A 10 mm cut-off yielded a PPV >75% only if prevalence was ≥10%; a 15 mm cut-off was proposed for lower prevalences to maintain PPV.

Conclusions:

• Estimating TST predictive values from routine data is valuable for setting appropriate cut-off values.
• The optimal TST cut-off for LTBI detection varies with population prevalence.
• Recommendations for TST cut-offs can be tailored using this methodology for different populations and tuberculin preparations.