BACKGROUND: Inter-rater reliability between breast surgical oncologists and reconstructive surgeons using cosmesis scales, and the correlation between their observations and patients' own subjective assessments, is poorly understood. METHODS: Patients undergoing BCS in a prospective trial rated their cosmetic outcome on a Likert scale (poor/fair/good/excellent) at the postoperative and 1-year time points; photographs were also taken. Three breast surgical oncologists (not involved in these cases) and two reconstructive surgeons were asked to independently rate cosmesis using the Harvard/NSABP/RTOG scale. RESULTS: Overall, 55 and 17 patients had photographs and Likert self-evaluations at the postoperative and 1-year time points, respectively. There was poor agreement between surgeon and patient ratings postoperatively [kappas - 0.042 (p = 0.659), 0.069 (p = 0.226), and 0.076 (p = 0.090) for the breast surgical oncologists; and 0.018 (p = 0.689) and 0.112 (p = 0.145) for the reconstructive surgeons], and poor interobserver agreement between surgeons of the same specialty (kappa - 0.087, 95% confidence interval [CI] - 0.091 to - 0.082, p = 0.223 for breast surgical oncologists; and kappa - 0.150, 95% CI - 0.157 to - 0.144, p = 0.150, for reconstructive surgeons). At 1 year, the interobserver agreement between breast surgical oncologists was better (kappa 0.507, 95% CI 0.501-0.512, p < 0.001); however, there was still poor correlation between the reconstructive surgeons (kappa - 0.040, 95% CI - 0.049 to - 0.031, p = 0.772). Agreement between surgeon and patient ratings remained poor at this time point [kappas - 0.115 (p = 0.477), 0.177 (p = 0.245), and 0.101 (p = 0.475) for breast surgical oncologists; and 0.335 (p = 0.037) and -0.118 (p = 0.221) for reconstructive surgeons]. CONCLUSION: Despite gradation scales for measuring cosmesis after BCS, high levels of agreement between surgeons is lacking and these do not always reflect patients' subjective assessments.
BACKGROUND: Inter-rater reliability between breast surgical oncologists and reconstructive surgeons using cosmesis scales, and the correlation between their observations and patients' own subjective assessments, is poorly understood. METHODS:Patients undergoing BCS in a prospective trial rated their cosmetic outcome on a Likert scale (poor/fair/good/excellent) at the postoperative and 1-year time points; photographs were also taken. Three breast surgical oncologists (not involved in these cases) and two reconstructive surgeons were asked to independently rate cosmesis using the Harvard/NSABP/RTOG scale. RESULTS: Overall, 55 and 17 patients had photographs and Likert self-evaluations at the postoperative and 1-year time points, respectively. There was poor agreement between surgeon and patient ratings postoperatively [kappas - 0.042 (p = 0.659), 0.069 (p = 0.226), and 0.076 (p = 0.090) for the breast surgical oncologists; and 0.018 (p = 0.689) and 0.112 (p = 0.145) for the reconstructive surgeons], and poor interobserver agreement between surgeons of the same specialty (kappa - 0.087, 95% confidence interval [CI] - 0.091 to - 0.082, p = 0.223 for breast surgical oncologists; and kappa - 0.150, 95% CI - 0.157 to - 0.144, p = 0.150, for reconstructive surgeons). At 1 year, the interobserver agreement between breast surgical oncologists was better (kappa 0.507, 95% CI 0.501-0.512, p < 0.001); however, there was still poor correlation between the reconstructive surgeons (kappa - 0.040, 95% CI - 0.049 to - 0.031, p = 0.772). Agreement between surgeon and patient ratings remained poor at this time point [kappas - 0.115 (p = 0.477), 0.177 (p = 0.245), and 0.101 (p = 0.475) for breast surgical oncologists; and 0.335 (p = 0.037) and -0.118 (p = 0.221) for reconstructive surgeons]. CONCLUSION: Despite gradation scales for measuring cosmesis after BCS, high levels of agreement between surgeons is lacking and these do not always reflect patients' subjective assessments.