Association of Receiving Multiple, Concurrent Fracture-Associated Drugs With Hip Fracture Risk

Key Points Question Is the concurrent receipt of 2 fracture-associated drugs associated with increased risk of hip fracture? Findings In this cohort study, 11 million person-years of Medicare data revealed that the receipt of multiple fracture-associated drugs was common among older US residents. Concurrent receipt of 2 or more such drugs was associated with a more than 2-fold increase in hip fracture risk, with some specific combinations appearing especially hazardous, including commonly prescribed drugs, such as opioids, antidepressants, and sedatives. Meaning If confirmed, results of this cohort study suggest caution when combining fracture-associated medications, especially when use is discretionary, alternatives exist, or baseline fracture risk is high.


Cohort Creation:
Because of the two-year Parts A and B coverage requirement, the minimum age was 67 years at the start of Part D drug receipt observation.
Patients with cancer (except non-melanoma of the skin) were conservatively excluded due to risk of pathologic fracture that may be unrelated to drug exposure and imperfectly identified as a cancer-related fracture in claims. Patients with advanced renal disease were also excluded due to risk of fracture related to renal failure and the possibility of receiving drugs during dialysis that would not appear in our claims data. We excluded hospice patients because some medications are included in the hospice per diem payments and thus may not be apparent in our claims data.
We excluded beneficiaries with vertebral fractures at any time during the look-back or observation because the temporal relationship between occurrence and diagnosis of vertebral fractures can vary tremendously. Vertebral fractures are often found incidentally on images obtained for reasons unrelated to fracture. All patients with a past fragility fracture were conservatively excluded from this study because we aimed to study first fragility fractures rather than a beneficiary's second or third fragility fracture. Prior fracture is an important fracture risk factor and undoubtedly an effect modifier; this small, high-risk, population must be studied separately by methods appropriate for smaller populations with an event as the key cohort inclusions criterion. 1 Additionally, fragility fracture survivors have a distinctly high fracture risk and their prescriptions may be managed distinctly as a result; their vulnerability to additive adverse drug effects may be different as well. We will study them separately.
Patients censored due to disenrollment in fee-for-service A, B and D plans were permitted to reenter the study as long as they met the inclusion criteria.

Drug Exposure:
For fracture associated drugs, the LexiComp Basic database was used to obtain the active ingredient and product information for each National Drug Code (NDC) apparent in the Part D Event file. Exposure was assumed to begin on the date of prescription dispensing. We used the "days supply" variable to set exposure duration. Overlapping supplies were carried forward (e.g. a fill occurring 80 days after a 90-day fill resulted in a 10-day supply carried forward). For drugs not refilled at the end of the days-supply, discontinuation was assigned after a time elapse of 120% of most recent days-supply, assuming imperfect adherence.
Other drug exposures that were classified as non-fracture associated drug (Non-FAD)  Table 1 below for the Ultimate Parent Enhanced Therapeutic Classification Names that contributed to the top 20 most common non-FADs. For example, Cardiovascular Therapy Agents included angiotensin converting enzyme-inhibitors, statins, and angiotensin II receptor blockers, and Analgesic, Anti-inflammatory or Antipyretic drugs included non-narcotic analgesics and NSAID analgesics (COX Non-Specific).

Parkinson's and Huntington's Diseases
Some diagnoses of alcohol abuse will be missing due to redaction of data data associated with addiction care for the years (2013)(2014)   Estimates are from Cox regression models. Total person years of observation = 7,126,266 for women, 4,160,502 for men. Total Drug Use = fracture associated drugs (FADs) + non-FADS. For FADs, status of receipt of 21 single drug (1 FAD) and 210 concurrent drug pairs (2 FAD) is classified for each person, for each day of observation. 3+ FAD is a classification of person days with 3 or more concurrent FADs. The same approach was taken to identify 552 non-FAD exposures: 1 non-FAD expsoure (1 Non-FAD), 2 concurrent non-FAD exposures (2 Non-FAD), 3+ Non-FAD is a classification of person days with 3 or more concurrent Non-FADs. Model A= Sensitivity analysis excluding any drug exposure variable. Model B= Continuous exposure variable for total drug use. Model C1=Categorized exposure indicator variables for total drug use ( 1,2,3+ drugs versus none). Model C2=Exposure indicator variables for concurrent use of fracture associated drugs (FAD 1,2,3+ versus none) as well as all other drug use (Non-FAD 1,2,3+ versus none). Model C3= Model C2 stratified on age category (5 year increments except for the first and the last). All models were stratified on sex and adjusted for time varying patient factors; Medicaid eligibility and Medicare Part D low income subsidy are two poverty indicators. Long-term care indicator is defined as 50% or more annual prescription fills dispensed by a long-term care pharmacy. The three categories of drugs protective of fracture are: 1. osteoporosis treatments: oral and injected bisphosphonates, calcitonin, denosumab, parathyroid hormone, 2. Estrogens-systemic estrogens, selective estrogen receptor modulators, and 3. beta-blockers. The "Tobacco/Chronic Obstructive Lung Disease" co-morbidity includes patients with Chronic Obstructive Lung Disease diagnosis and/or tobacco use diagnosis.  Exposure indicator variables for fracture associated drug (FAD) use reflect person days with any of 21 individual FADs and 210 possible two-way combination of these drugs (included as a specific exposure in most cases, but aggregated to "FAD pairs with <100 PY" if cohort specific observation < 100 person years). "3+FAD" is an indicator variable for person days with three-or-more FADs. The reference exposure was zero FADs. Non-FADs are 552 drug ingredients that are not FADs aggregated into status of receipt of any 1 non-FAD expsoure (1 Non-FAD), 2 concurrent non-FAD exposures (2 Non-FAD) or 3+ Non-FAD exposure classified for each person, for each day of observation. Medicaid eligibility and Medicare Part D low income subsidy are two poverty indicators. Long-term care is defined as 50% or more of annual prescriptions dispensed by a long-term care pharmacy. The three categories of drugs protective of fracture are: 1. osteoporosis treatments: oral and injected bisphosphonates, calcitonin, denosumab, parathyroid hormone, 2. Estrogens-systemic estrogens, selective estrogen receptor modulators, and 3. beta-blockers. SSRI/SNRI = combined drug group of selective serotonin reuptake inhibitors and selective norepinephrine reuptake inhibitors. PPI= proton pump inhibitors, FGAP = first generation antipsychotic. SGAP = second generation antipsychotic. TCA= tricyclic antidepressants. H2RA is Histamine-2 Receptor Antagonists. TZD = Thiazolidinediones. H1 Blockers = 1st Generation Antihistamines. CAAH = Central Acting Antihypertensives. Sedative Hypnotics include only non-benzodiazepine drugs. 4.20E-28 Exposure indicator variables for fracture associated drug (FAD) use reflect person days with any of 21 individual FADs and 210 possible two-way combination of these drugs (included as a specific exposure in most cases, but aggregated to "FAD pairs with <100 PY" if cohort specific observation < 100 person years). "3+FAD" is an indicator variable for person days with three-or-more FADs. The reference exposure was zero FADs. Non-FADs are 552 drug ingredients that are not FADs aggregated into status of receipt of any 1 non-FAD expsoure (1 Non-FAD), 2 concurrent non-FAD exposures (2 Non-FAD) or 3+ Non-FAD exposure classified for each person, for each day of observation. Medicaid eligibility and Medicare Part D low income subsidy are two poverty indicators. Long-term care is defined as 50% or more of annual prescriptions dispensed by a long-term care pharmacy. The three categories of drugs protective of fracture are: 1. osteoporosis treatments: oral and injected bisphosphonates, calcitonin, denosumab, parathyroid hormone, 2. Estrogenssystemic estrogens, selective estrogen receptor modulators, and 3. beta-blockers. SSRI/SNRI = combined drug group of selective serotonin reuptake inhibitors and selective norepinephrine reuptake inhibitors. PPI= proton pump inhibitors, FGAP = first generation antipsychotic. SGAP = second generation antipsychotic. TCA= tricyclic antidepressants. H2RA is Histamine-2 Receptor Antagonists. TZD = Thiazolidinediones. H1 Blockers = 1st Generation Antihistamines. CAAH = Central Acting Antihypertensives. Sedative Hypnotics include only non-benzodiazepine drugs. The Tobacco/Chronic Obstructive Lung Disease is an aggregate indicator of these diagnoses. Bold if False Discovery Rate corrected P value < 0.05. Italicized if population-level impact criteria were met.