with epilepsy Use of antiepileptic drugs and risk of fractures: Case-control study among patients

— Objective: To study the association between use of antiepileptic drugs (AEDs) and risk of fractures. Methods: The authors obtained data from the General Practice Research Database (GPRD). A case–control study was nested within a cohort of patients with active epilepsy. Cases were patients with a first fracture after cohort entry. Up to four controls were matched to each case by practice, sex, year of birth, timing of first epilepsy diagnosis, index date, and duration of GPRD history. Cumulative exposure to AEDs was assessed by summing the duration of all AED prescriptions. A distinction was made between AEDs that induce the hepatic cytochrome P-450 enzyme system and AEDs that do not. Medical conditions and drugs known to be associated with bone metabolism or falls were evaluated as potential confounders. Conditional logistic regression analysis was used to calculate odds ratios (ORs) and 95% CIs. Results: The study population comprised 1,018 cases and 1,842 matched controls. The risk of fractures increased with cumulative duration of exposure ( p for trend (cid:1) 0.001), with the strongest association for greater than 12 years of use: adjusted OR 4.15 (95% CI 2.71 to 6.34). Risk estimates were higher in women than in men. There was no difference between users of AEDs that induce and AEDs that do not induce the hepatic cytochrome P-450 system. Conclusions: Long-term use of AEDs was associated with an increased risk of fractures, especially in women. More research on mechanisms of AED-induced bone breakdown and female vulnerability to the effects of AEDs on bone health is warranted. We assessed whether patients were current users, recent users, or past users of AEDs on the index date. Patients were consid- ered as current users when the theoretical end date lasted to at least the 30-day window before the index date. Recent users were patients who used AEDs in the 6 months before the index date but were not current users. Former users had a last prescription end- ing 6 months or more before the index date. AEDs were also according to their capacity to induce the hepatic cyto- chrome P-450 system (appendix). We differentiated between patients who had used only enzyme-inducing AEDs (EIAEDs), patients who had used only non–enzyme-inducing AEDs (NE-IAEDs), and patients who had used both types of AEDs. We calcu- lated the cumulative exposure to EIAEDs and NEIAEDs up to the index date to evaluate the effect of duration of use. When patients used multiple AEDs concomitantly, the exposure duration of all AEDs were summed. Covariate definitions and measures. Potential confounders in this study were illnesses and medications that are known to be associated with falls or fractures. Medical conditions included diabetes mellitus, rheumatoid arthritis, hyperthyroidism, congestive heart failure, hypertension, anemia, depression, Parkinson disease, psychotic disorders, dementia, cerebrovascular accidents, urinary incontinence, and chronic obstructive pulmonary disease. were evaluated in a 6-month period before the index date. Medications assessed included use of nonsteroidal anti-inflammatory drugs, methotrexate, hormone replacement therapy, diuretics, anxiolytics/hypnotics, antipsychotics, antidepressants, antiparkinsonian drugs, systemic and inhaled glucocorticoids, bronchodilators, opiates, blockers, angiotensin-converting enzyme inhibitors, calcium channel blockers, and al- pha blockers. on smoking status of and body mass index (cid:1) (cid:1) kg/m , Confounding by disease patients with more severe epilepsy are more likely to fall and more likely to use more (and higher dosed) AEDs compared with patients who epilepsy of a less severe nature. As proxies for disease severity, we assessed the number of epilepsy medical codes in the year before the index date, information on the prescribing of rectal/parenteral benzodiazepines, and the number of different AEDs used on the index date. Data analysis. Conditional logistic regression analysis was used to calculate odds ratios (ORs) and 95% CIs. Covariates were included in the final multivariate model when they were retained in a backward analysis with a p value of 0.20 for inclusion. Pre- scriptions for rectal benzodiazepines and the number of epilepsy-related medical codes were always included in the model. The primary analysis included all AEDs. The association between duration of AED use and risk of fractures was assessed, where expo- sure duration was initially included as a continuous variable and later categorized in years. We stratified according to age and sex and the number of drugs taken during the study period (one drug vs more than one drug).