Evaluating photographic scales of facial pores and diagnostic agreement of tests using latent class models.

ABSTRACT

BACKGROUND: Ordinal severity scales illustrated by photographs have been widely developed to help dermatologists in evaluating skin problems or improvements. Numerous scales have been published, and none of them were used for assessing facial pores. METHODS: A five-point photographic scale of facial pores was formulated, and photographs of pores on nasal ala from 128 female volunteers were acquired. Five dermatologists with similar experiences rated the 128 photographs independently using the reference photographs. Latent Class Models (LCM) were used to analyze the data. Firstly, we hypothesized that the conditional probabilities of the five dermatologists were identical to build the first LCM and without the restriction to formulate the second LCM. Conditional probability and posterior probability were also calculated. RESULTS: The five-point scales were ambiguous as the raters actually had difficulties in distinguishing between some adjacent categories. Adjacent categories were pooled for reanalyzing, and the model fitted well. CONCLUSIONS: The newly developed photographic scale of Chinese facial pores should be redefined to improve their quality and reproducibility in future studies. Standardized scales for the measurement of aging and response to cosmetic therapy were essential for assessing diagnostic experiment. The LCM can effectively deal with diagnostic test of agreement and reproducibility.