Master of Science, The Ohio State University, 2011, Public Health
The first objective of this study was to identify inconsistencies in implementation of diagnostic algorithms for gonorrhea/chlamydia (GC/CT) and bacterial vaginosis/trichomoniasis (BV/TV) diagnosis in female sex workers (FSWs) in Madagascar. Using data from clinical charts, we arrived at algorithm-based diagnoses and compared these to diagnoses recorded by clinicians implementing the same algorithms. The second objective was to identify predictors of inconsistencies between algorithm and clinician diagnoses using generalized estimating equation (GEE) logistic regression modeling. Data were collected between 2003 and 2009 and included 45,021 visits from 13,080 FSWs. We identified disagreement between algorithm and clinician diagnoses of GC/CT in 19.39% of visits and BV/TV in 5.95%. The Kappa measures of agreement for GC/CT and BV/TV were 0.56 and 0.52 respectively. Healthy vaginal pH (below 4.5), FSWs in all age categories, and presence of cervicitis or cervical motion tenderness were each identified as significant predictors of discordance in clinician and algorithm diagnosis.
Committee: J. R. Wilkins III DrPH (Advisor); Abigail Norris Turner PhD (Committee Member)
Subjects: Biostatistics; Epidemiology; Health Sciences; Medicine; Public Health; Statistics; Womens Studies