Abstract
Authors: Baser O, Xie L, Du J.
OBJECTIVES: Menopause symptoms are a good disease severity proxy for menopausal women, but are not available in claims database. We applied probabilistic linkage to add symptoms recorded in a registry database to claims data, and compare the healthcare costs among women with various symptoms.
METHODS: Women age 45 or older who used estrogen only hormone therapy (HT) were selected from a large US claims database (41/2005-09/30/2008). Another group who used estrogen only HT with a menopause diagnosis was selected from the University of Michigan Women’s Registry Database. Logistic regression was used to calculate the propensity score for each patient controlling for osteoporosis, gynecological disorders/procedures, genital infection, bladder/pelvic floor support problem, gynecology system cancer, breast condition, gut condition, hormone disorder, nerve problem, and other individual comorbidities such as rheumatoid disease, depression, and blood clotting. Patients with the closest propensity score from each group were matched, and menopause symptoms for registry patients were added to the claims database records. After repeating probabilistic linkage 250 times, the mean and 95% confidence interval (CI) of healthcare costs during the follow-up period were calculated.
RESULTS: 80 patients from each population were matched after probabilistically linking 20,020 claims database patients with 83 registry database patients. The average cost of patients with at least one symptom was much higher than for patients without symptoms ($13,570 [95% CI: $13,459-$13,680] vs. $3,391 [95%CI: $3,345-$3,436], p-value0.001). Cost differences were mainly from inpatient, physician visit, and pharmacy costs. Among patients with menopause symptoms, those with hot flashes had the highest costs ($10,127), followed by memory loss ($1,653), vaginal dryness ($864), reduced libido ($568), and=mood swings ($358).
CONCLUSIONS: Women with menopause symptoms incur higher health care costs than those without This study suggests symptoms are important determinants of health care expenses and their impact can be assessed by linking registry and claims databases.
For a full text PDF please complete the form below and we will be in contact shortly!
