| To Register SMDM Homepage | |
|
|
|
||||
Purpose: Several treatment alternatives for endometriosis pain are similar in effectiveness but differ considerably in their risk profiles and quality of life outcomes. The objective was to determine the distribution of physicians’ treatment recommendations for endometriosis pain and how they vary by physician characteristics.
Methods: We conducted a national mailed survey of a random sample of gynecologists, consisting of generalists (n=112) and endometriosis specialists (n=248). The questionnaire presented a hypothetical patient with confirmed endometriosis who seeks relief from severe pelvic pain after first-line therapies have failed. Respondents rank-ordered eight treatments from the most to the least preferred and provided a rating on how likely they were to recommend each alternative (on a 5-point scale). We summarized the rank orderings and calculated the percentage of respondents who provided relatively high recommendation ratings (“probably” or “definitely” recommend) for each treatment. Logistic regression analyses (of the odds of a high recommendation rating) were used to explore associations between recommendations and physician characteristics (generalist/specialist, sex, age, race, U.S./foreign medical school, geographic region). Each model’s ability to discriminate between high versus low recommendation was characterized by the area under the receiver operating characteristic curve (C-statistic).
Results: The percentage of respondents providing a high recommendation rating [ranking treatment as most preferred] were as follows: GnRH agonist 83.2% [65.5%]; laparoscopy 43.4% [23.0%]; progestin 34.9% [5.4%]; “other” 30.6% [9.1%]; alternative therapies 16.3% [4.0%]; danazol 12.3% [1.9%]; pain treatment center 11.9% [1.9%]; hysterectomy 2.1% [0.7%]. Specialists were significantly more likely than generalists to recommend laparoscopy (OR=2.0; p=0.01), alternative therapies (OR=2.3; p=0.03), and “other” treatments (12.2%; p=0.03). Female physicians were more likely to recommend GnRH agonists (OR=2.0; p=0.08), progestin (OR=1.7; p=0.05), alternative therapies (OR=2.0; p=0.05), and “other” treatments (OR=3.3; p=0.08). Finally, younger physicians were significantly more likely to recommend laparoscopy (OR=1.9; p=0.04), and less likely to recommend danazol (OR=0.3; p=0.08) and alternative therapies (OR=0.4; p=0.09). There were no statistically significant associations between treatment recommendation and race, medical school, or geographic region. The accuracy of the eight models in discriminating between high versus low recommendation (C-statistics) ranged from 0.61 for progestin to 0.76 for “other” treatment.
Conclusions: We found substantial variations in physicians’ treatment recommendations for endometriosis pain. Furthermore, there was evidence of significant physician-related effects on recommendations, suggesting the possibility of unwarranted sources of variation.
See more of Oral Concurrent Session A - Patient and Physician Behavior/Preferences 1
See more of The 26th Annual Meeting of the Society for Medical Decision Making (October 17-20, 2004)