Premium
Screening approach for identifying candidate drugs and drug‐drug interactions related to hip fracture risk in persons with Alzheimer disease
Author(s) -
Tolppanen AnnaMaija,
Taipale Heidi,
Koponen Marjaana,
Tanskanen Antti,
Lavikainen Piia,
Paananen Jussi,
Tiihonen Jari,
Hartikainen Sirpa
Publication year - 2017
Publication title -
pharmacoepidemiology and drug safety
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.023
H-Index - 96
eISSN - 1099-1557
pISSN - 1053-8569
DOI - 10.1002/pds.4232
Subject(s) - medicine , drug , pharmacology , hip fracture , osteoporosis
Abstract Purpose To assess whether a “drugome‐wide” screen with case‐crossover design is a feasible approach for identifying candidate drugs and drug‐drug interactions. Methods All community‐dwelling residents of Finland who received a clinically verified Alzheimer disease diagnosis in 2005 to 2011 and experienced incident hip fracture (HF) afterwards (N = 4851). Three scenarios were used to test the sensitivity of this approach (1) hazard period 0 to 30 and control period 31 to 61 days before HF, (2) hazard period 0 to 30 and control period 336 to 366 days before HF, and (3) hazard period 0 to 14 and control period 16 to 30 days before HF. Results Nine, 44, and 5 drugs were associated with increased HF risk and 8, 23, and 4 with decreased risk in scenarios 1, 2, and 3, respectively. Six drugs were identified with scenario 1 only and 54 and 1 with scenarios 2 and 3, respectively. Only six drugs (metoprolol, simvastatin, trimethoprim, codeine combinations, fentanyl, and paracetamol) were associated with HF in all scenarios, four with 1 and 2 (cefalexin, buprenorphine, olanzapine, and memantine), and one with 1 and 3 (enalapril) or 2 and 3 (ciprofloxacin). The direction of associations was the same in all/both scenarios. The interaction results were equally versatile, with hydroxocobalamin*oxazepam being the only interaction observed in all scenarios. Conclusions Case‐crossover analysis is a potential approach for identifying candidate drugs and drug‐drug interactions associated with adverse events as it implicitly controls for fixed confounders. The results are highly dependent on applied hazard and control periods, but the choice of periods can help in targeting the analyses to different phases of drug use.