Establishing a standard method for analysing case detection delay in leprosy using a Bayesian modelling approach.

Area of Science:

• Epidemiology
• Biostatistics
• Infectious Disease Modeling

Background:

• Leprosy (caused by Mycobacterium leprae) causes preventable disability, with case detection delay being a key indicator of transmission and prevention.
• Standardized methods for analyzing leprosy case detection delay data are lacking, hindering progress in control efforts.
• Understanding and modeling detection delays are crucial for effective leprosy intervention strategies.

Approach:

• Evaluated two leprosy case detection delay datasets using Bayesian models.
• Assessed log-normal, gamma, and Weibull distributions to determine the best fit for detection delay variability.
• Utilized leave-one-out cross-validation and incorporated covariates like age, sex, and leprosy subtype.

Key Points:

• A log-normal distribution best described leprosy case detection delays in both datasets.
• Multibacillary (MB) leprosy patients experienced significantly longer delays (1.57 times) than paucibacillary (PB) leprosy patients.
• Data from the PEP4LEP study showed longer delays (1.51 times) compared to self-reported data from a systematic review.

Conclusions:

• The developed log-normal model provides a standardized method for comparing leprosy case detection delay datasets.
• This modeling approach can be applied to studies measuring the reduction of case detection delay, including the PEP4LEP study.
• Recommends utilizing this modeling strategy for leprosy and other skin-neglected tropical diseases (skin-NTDs) to test probability distributions and covariate effects.