Descriptive statistics were used to summarize study population characteristics. Treatment patterns over the first and second years following the index date, and for the full duration of postindex follow-up, were assessed as the proportion of patients treated with gynecologic procedures and/or prescribed pharmacologic therapies of interest that were reimbursed by insurance. Pharmacologic therapies of interest were hormonal treatments (oral and nonoral contraceptives), including intrauterine devices (IUDs, except ParaGard®/copper IUD), estrogen, progestin, aromatase inhibitors, elagolix, danazol, leuprolide, or any luteinizing hormone-releasing hormone agonists.
Also evaluated were the use of tranexamic acid and pain medicines, including narcotic (prescribed for ≥30 days) and prescription non-narcotic analgesics. Not available for analysis were over-the-counter products not captured in medical claims and prescriptions not reimbursed by the payer. Gynecologic procedures of interest were hysterectomy, operative laparoscopy, myomectomy, oophorectomy, ablation of the endometrium and/or fibroids, excision, and salpingectomy. Finally, data were collected for pharmacologic treatments of interest (hormonal or analgesic) received by patients in the year preceding the index date.
Patients in both cohorts who underwent hysterectomy within 1 year postindex date were further stratified by age. Logistic regression models were constructed to determine factors associated with specific treatments (hysterectomy and hormonal therapy) in patients with UF-HMB and UF-only. To isolate these findings to patients who received hysterectomy due to UF, the regression analysis excluded patients with a claim for endometriosis (ICD-9 617.X or ICD-10 N80.X). This exclusion was applied because of the potential for confounding due to concomitant comorbidity. The variables included in the logistic regression were factors that could contribute to treatment decision-making and that could be captured in claims data. These were age, abnormal bleeding, anemia, fatigue, infertility, pain, prior- and post-UF diagnosis use of medications, including hormonal treatment, non-narcotic, or narcotic analgesic treatment, and inpatient or outpatient diagnosis site. Data were analyzed using SAS/STAT(r) software, version 15.1 (2016 SAS Institute Inc., Cary, NC, USA).
Also evaluated were the use of tranexamic acid and pain medicines, including narcotic (prescribed for ≥30 days) and prescription non-narcotic analgesics. Not available for analysis were over-the-counter products not captured in medical claims and prescriptions not reimbursed by the payer. Gynecologic procedures of interest were hysterectomy, operative laparoscopy, myomectomy, oophorectomy, ablation of the endometrium and/or fibroids, excision, and salpingectomy. Finally, data were collected for pharmacologic treatments of interest (hormonal or analgesic) received by patients in the year preceding the index date.
Patients in both cohorts who underwent hysterectomy within 1 year postindex date were further stratified by age. Logistic regression models were constructed to determine factors associated with specific treatments (hysterectomy and hormonal therapy) in patients with UF-HMB and UF-only. To isolate these findings to patients who received hysterectomy due to UF, the regression analysis excluded patients with a claim for endometriosis (ICD-9 617.X or ICD-10 N80.X). This exclusion was applied because of the potential for confounding due to concomitant comorbidity. The variables included in the logistic regression were factors that could contribute to treatment decision-making and that could be captured in claims data. These were age, abnormal bleeding, anemia, fatigue, infertility, pain, prior- and post-UF diagnosis use of medications, including hormonal treatment, non-narcotic, or narcotic analgesic treatment, and inpatient or outpatient diagnosis site. Data were analyzed using SAS/STAT(r) software, version 15.1 (2016 SAS Institute Inc., Cary, NC, USA).
