Geographic variation in posttreatment surveillance intensity for patients with cutaneous melanoma.

BACKGROUND: We investigated whether geographic determinants could account for variation in posttreatment melanoma surveillance intensity among plastic surgeons. METHODS: A custom-designed questionnaire was mailed to U.S. and non-U.S. members of the American Society of Plastic and Reconstructive Surgeons (ASPRS). Subjects were asked how they use 14 specific follow-up modalities during years 1 to 5 and 10 following primary treatment for patients with cutaneous melanoma. Repeated-measures analysis of variance was used to compare practice patterns by TNM stage, year postsurgery, U.S. census region, metropolitan statistical area (MSA), and managed care organization (MCO) penetration rate. RESULTS: Of the 1,142 respondents, 395 were evaluable. Those who did not perform melanoma surgery or follow-up were excluded. Correlation analysis showed that mean follow-up intensity for the modalities surveyed was highly correlated across TNM stages and years postsurgery. Within MSAs, only chest radiograph utilization varied significantly. The pattern of testing varied significantly by geographic region for seven modalities (office visit, computed tomography scan of the brain and chest/abdomen, alpha-fetoprotein level, 5S-cysteinyl dopa level, abdominal ultrasonogram, bone scan); in each of these, utilization by non-U.S. surgeons exceeded utilization in any U.S. census region. The pattern of testing varied significantly by MCO penetration rate for chest radiograph (greater utilization in the lowest MCO penetration rate areas) and 5S-cysteinyl dopa level (greater utilization in the highest MCO penetration rate areas). CONCLUSIONS: The intensity of posttreatment surveillance recommended by ASPRS members caring for patients with cutaneous melanoma varies markedly. This analysis provides the first evidence that geographic factors significantly affect the surveillance strategies of clinicians following patients with cutaneous melanoma. Variation by census region was most prominent, although the size of detected differences was small.