Key Points
Question
What is the risk of incident osteoporosis following incident adult-onset epilepsy and antiseizure medication (ASM) use?
Findings
In this cohort study of 6275 patients, time to incident osteoporosis was 41% faster following incident adult-onset epilepsy independent of ASM use. The time to osteoporosis was 23% faster with exposure to non–enzyme-inducing ASMs and 9% faster for enzyme-inducing ASMs, independent of epilepsy.
Meaning
In this study, both the development of epilepsy and use of ASMs, irrespective of enzyme-inducing capacity, were associated with increased hazards for osteoporosis; these findings suggest that routine screening and prophylaxis should be considered in all people with epilepsy.
This cohort study uses population-based data to quantify and model the independent hazards for osteoporosis associated with epilepsy and antiseizure medications.
Abstract
Importance
Both epilepsy and enzyme-inducing antiseizure medications (eiASMs) having varying reports of an association with increased risks for osteoporosis.
Objective
To quantify and model the independent hazards for osteoporosis associated with incident epilepsy and eiASMS and non-eiASMs.
Design, Setting, and Participants
This open cohort study covered the years 1998 to 2019, with a median (IQR) follow-up of 5 (1.7-11.1) years. Data were collected for 6275 patients enrolled in the Clinical Practice Research Datalink and from hospital electronic health records. No patients who met inclusion criteria (Clinical Practice Research Datalink–acceptable data, aged 18 years or older, follow-up after the Hospital Episode Statistics patient care linkage date of 1998, and free of osteoporosis at baseline) were excluded or declined.
Exposure
Incident adult-onset epilepsy using a 5-year washout and receipt of 4 consecutive ASMs.
Main Outcomes and Measures
The outcome was incident osteoporosis as determined through Cox proportional hazards or accelerated failure time models where appropriate. Incident epilepsy was treated as a time-varying covariate. Analyses controlled for age, sex, socioeconomic status, cancer, 1 or more years of corticosteroid use, body mass index, bariatric surgery, eating disorders, hyperthyroidism, inflammatory bowel disease, rheumatoid arthritis, smoking status, falls, fragility fractures, and osteoporosis screening tests. Subsequent analyses (1) excluded body mass index, which was missing in 30% of patients; (2) applied propensity score matching for receipt of an eiASM; (3) restricted analyses to only those with incident onset epilepsy; and (4) restricted analyses to patients who developed epilepsy at age 65 years or older. Analyses were performed between July 1 and October 31, 2022, and in February 2023 for revisions.
Results
Of 8 095 441 adults identified, 6275 had incident adult-onset epilepsy (3220 female [51%] and 3055 male [49%]; incidence rate, 62 per 100 000 person-years) with a median (IQR) age of 56 (38-73) years. When controlling for osteoporosis risk factors, incident epilepsy was independently associated with a 41% faster time to incident osteoporosis (time ratio [TR], 0.59; 95% CI, 0.52-0.67; P < .001). Both eiASMs (TR, 0.91; 95% CI, 0.87-0.95; P < .001) and non-eiASMs (TR, 0.77; 95% CI, 0.76-0.78; P < .001) were also associated with significant increased risks independent of epilepsy, accounting for 9% and 23% faster times to development of osteoporosis, respectively. The independent associations among epilepsy, eiASMs, and non-eiASMs remained consistent in propensity score–matched analyses, cohorts restricted to adult-onset epilepsy, and cohorts restricted to late-onset epilepsy.
Conclusions and Relevance
These findings suggest that epilepsy is independently associated with a clinically meaningful increase in the risk for osteoporosis, as are both eiASMs and non-eiASMs. Routine screening and prophylaxis should be considered in all people with epilepsy.
Introduction
Active epilepsy affects almost 50 million people worldwide, with age-standardized disability-adjusted life-years of 182.6 per 100 000 population.1 Although seizures account for a substantial proportion of this burden, epilepsy-associated comorbidities also exert significant influence. Osteoporosis is of particular interest given its association with high rates of fractures, morbidity, and mortality.2
The mechanisms behind the association between epilepsy and osteoporosis remain opaque. Considerable focus has centered on the putative role of antiseizure medications (ASMs). Both enzyme-inducing ASMs (eiASMs) and non-eiASMs have been found to be associated with increased risks of fractures in a meta-analysis of case-control and cohort studies.3 Certain ASMs, especially carbamazepine, are associated with enhanced vitamin D metabolism and accelerated bone turnover, but studies have often yielded conflicting results regarding the effects of ASMs, even for eiASMs, such as carbamazepine, on bone density.4 Importantly, no studies performed to date have examined these associations at a population level with a specific focus on a clinical diagnosis of osteoporosis.
Our objectives were to quantify the hazard of osteoporosis following incident epilepsy and ASM exposure in the general population. We also sought to determine whether the hazard related to eiASMs is elevated compared with non-eiASMs in people with incident adult-onset epilepsy and late-onset epilepsy.
Methods
Study Design
This retrospective open cohort study was performed as part of the Carbohydrates, Lipids and Biomarkers of Traditional and Emerging Cardiometabolic Risk Factors (CALIBER) resource.5,6 CALIBER, led from the UCL Institute of Health Informatics, is a research resource providing validated electronic health record phenotyping algorithms and tools for national structured data sources.7,8 The CALIBER resource contains UK nationally linked structured electronic health record data from primary care, hospital care, and a cause-specific mortality registry up to March 31, 2019.5,9
Primary care records of patients 18 years or older were obtained from Clinical Practice Research Datalink (CPRD), which uses Read codes,10 version 2 for medical events and the British National Formulary to document prescriptions.11,12 These data are linked to the Hospital Episode Statistics (HES) database, which contains secondary care and administrative data. Coding in HES uses the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) and the Office of Population Censuses and Surveys Classification of Interventions and Procedures, version 4, terminology. The study was approved by the Medicines and Healthcare Products Regulatory Agency (UK) Independent Scientific Advisory Committee [17_064RA3], under section 251 (National Health Service Act 2006). Patient consent was waived due to the deidentified nature of the data. This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
Case Ascertainment and Study Population
We used the Secure Anonymised Information Linkage (SAIL) databank Epilepsy Only case definition for epilepsy.13 This definition has a sensitivity of 88% and specificity of 98% for identifying people with epilepsy in the SAIL data set. The SAIL data set is a data source comparable to CPRD9,13 (eAppendix in Supplement 1). We extracted an incident cohort of patients using a 5-year washout from the CPRD-acceptable data quality date,9 which comprises an epoch over which a patient cannot have been assigned any codes for seizures or epilepsy, thus maximizing chances of incident diagnoses. Because the risks of osteoporosis are particularly elevated in older adults, we also extracted a cohort of patients with incident late-onset epilepsy, which was restricted to those diagnosed at 65 years or older.14
Exposure and Outcome Definitions
Incident epilepsy was the primary exposure. Medication exposure was defined by receiving 4 consecutive prescriptions for an ASM. Furthermore, we defined exposure to eiASM as all 4 prescriptions being for an eiASM (median [IQR] time between first and fourth prescriptions, 4 [3-7] months). We defined exposure to non-eiASMs as at least the fourth prescription being for a non-eiASM. We chose 4 consecutive prescriptions given the assumption that the patient must be tolerating the medication with enough efficacy to merit continued use over multiple visits. We considered carbamazepine, eslicarbazepine, oxcarbazepine, phenobarbital, phenytoin, primidone, rufinamide, and topiramate to be eiASMs.15 In a sensitivity analysis, we categorized ASMs as strong inducers (carbamazepine, phenobarbital, phenytoin, and primidone),16 weak inducers (topiramate and oxcarbazepine),16 and non-eiASMs. The outcome (end point of the analysis) was incident osteoporosis as defined and validated in the CALIBER data resource.17 This definition comprises diagnostic and investigational Read, medication, and ICD-10 codes.6 To identify the independent associations, age, sex, socioeconomic status (defined using the Index of Multiple Deprivation 2015, which is divided into deciles with 1 being the lowest socioeconomic status and 10 being the highest18), cancer, 1 or more years of corticosteroid use, body mass index (BMI) (measured as weight in kilograms divided by height in meters squared), bariatric surgery, eating disorders, hyperthyroidism, inflammatory bowel disease, rheumatoid arthritis, and smoking history19,20 were included as covariates in regression models. Data on race and ethnicity are self-reported in CALIBER but are available for only approximately 30% of participants and were therefore not collected for this study. We also adjusted for falls and fragility fractures with the rationale that people with epilepsy may be more prone to both due to seizures, thus prompting investigations that could reveal previously undiagnosed osteoporosis. Likewise, we controlled for osteoporosis screening (dual-energy x-ray absorptiometry, quantitative computed tomography, or serum bone turnover markers performed at any point in patients without osteoporosis or prior to the diagnosis of osteoporosis in those who developed incident disease). Failure to include fragility fractures, falls, and osteoporosis screening could lead to inflated estimates of an association due to ascertainment bias if not accounted for in the analyses. All covariates were defined using the Health Data Research UK’s CALIBER Phenotype Library portal phenotypes.6,7,17,21
Statistical Analysis
We used χ2 (categorical) and Kruskal-Wallis (continuous) tests to compare demographic and clinical characteristics between patients with and without incident epilepsy. In the primary analysis, we determined the hazard of incident osteoporosis using Cox proportional hazards models only after testing proportional hazards assumptions with the scaled Schoenfeld residuals.22 If the model failed the proportional hazards assumption, we used an accelerated failure time model, reporting a time ratio (TR). The TR is the ratio of the time it takes for an exposed person to develop the condition of interest compared with an unexposed person. Thus, a TR of less than 1.0 means that an exposed person develops the condition quicker. For these analyses, the index date was the CPRD-acceptable data quality date. All people were followed until incident osteoporosis or last follow-up. Incident epilepsy was treated as a time-varying covariate, whereby all people start as unexposed. The person then moves from the unexposed to exposed group on the date that they develop epilepsy, thus mitigating risk of immortal time bias.23
In an a priori sensitivity analysis, we repeated the primary analysis but omitted BMI since it was missing in 2 417 807 people (30%). We then repeated the primary analysis using a propensity score–matched model. Propensity scores for eiASM exposure were developed using the MatchIt package, version 3.0.2, in R (R Foundation for Statistical Computing).24 For the propensity score, we considered the following as factors that could influence eiASM prescription: age at diagnosis, sex, age-by-sex interaction term (to account for young females of childbearing age), hypertension, diabetes, dyslipidemia, atrial fibrillation, cirrhosis, chronic kidney disease, cancer, steroid use, depression, anxiety, psychosis, suicidal ideation, obesity (BMI > 29.9), and socioeconomic status.25 We evaluated matching performance using mean SE and graphical distributions of propensity scores. We used k-nearest neighbors to match exposed and unexposed patients based on individual-level propensity scores. To directly compare the hazard of eiASMs with non-eiASMs, we then repeated the primary analysis but isolated the population to only 6275 people with incident adult-onset epilepsy, using a dichotomous ASM variable (0 = non-eiASM, 1 = eiASM) in regression models. Finally, we repeated these steps on a preexisting cohort in which all 1048 patients with incident late-onset epilepsy in the database were matched 1:10 to control patients based on age, sex, and general population practice, and the same methodology was applied as per the primary analysis. For this model, the index date was that on which the patient turned age 65 years.
Analyses were performed between July 1 and October 31, 2022, and in February 2023 for revisions. All statistical analyses were performed using Stata, version 16.1 (StataCorp LLC) and R, version 3.6.2 software. We considered a 2-sided P < .05 to be significant.
Results
Patient Population
We identified 18 410 930 patients, of whom 16 071 111 (87%) had a CPRD-acceptable data quality date. Of these patients, 10 916 166 were 18 years or older, and 8 267 161 had follow-up after the HES patient care linkage date of 1998. A total of 8 095 441 were free of osteoporosis prior to the CPRD-acceptable data quality date and incident epilepsy date. Of these patients, 51 123 (0.6%) had prevalent epilepsy. Prevalent cases were removed, leaving 6275 patients with incident epilepsy (3220 female [51%] and 3055 male [49%]; incidence rate, 62 per 100 000 person-years; and incidence proportion, 78 per 100 000 people). Compared with the 8 089 166 control patients, the 6275 patients with incident epilepsy were older (median [IQR] age, 56 [38-73] vs 42 [30-60] years) and were statistically more likely to be exposed to an ASM (6275 [100%] vs 269 251 [3%]; P < .001); to be taking an eiASM (1137 [18%] vs 36 406 [1%]; P < .001); to have a higher BMI (median [IQR], 25.4 [22.5-29.0] vs 24.7 [22.0-28.1]; P < .001); to have cancer (913 [15%] vs 494 294 [6%]; P < .001), hyperthyroidism (103 [2%] vs 74 425 [1%]; P < .001), or inflammatory bowel disease (100 [2%] vs 60 663 [1%]; P < .001); to have used corticosteroids consecutively for at least 1 year (722 [12%] vs 220 891 [3%]; P < .001); to have vitamin D deficiency (133 [2%] vs 66 334 [1%]; P < .001); or lower serum calcium levels (median [IQR], 2.28 [2.21-2.36] vs 2.31 [2.24-2.38] mmol/L; P = .001); and to have had bariatric surgery (16 [<1%] vs 7219 [<1%]; P < .001) (Table 1). Median duration of ASM exposure was 8.9 years (IQR, 4.8-14.8 years).
Table 1. Demographic and Clinical Characteristics of the General Population and Patients With Incident Adult-Onset Epilepsy Identified Between 1998 and 2019.
| No. (%) | P valuea | |||
|---|---|---|---|---|
| Full cohort (N = 8 095 441) | No epilepsy (n = 8 089 166) | Incident epilepsy (n = 6275) | ||
| Age, y, median (IQR) | 42 (30-60) | 42 (30-60) | 56 (38-73) | <.001 |
| Sex | ||||
| Female | 4 198 306 (52) | 4 195 086 (52) | 3220 (51) | .39 |
| Male | 3 897 135 (48) | 3 894 080 (48) | 3055 (49) | |
| IMD, median (IQR) | 5 (4-6) | 5 (4-6) | 5 (4-6) | .42 |
| Former or current smoker | 2 904 250 (36) | 2 902 182 (36) | 2068 (33) | <.001 |
| ASM | 275 526 (3) | 269 251 (3) | 6275 (100) | <.001 |
| eiASM | 37 543 (1) | 36 406 (1) | 1137 (18) | <.001 |
| BMI, median (IQR) | 24.7 (22.0-28.1) | 24.7 (22.0-28.1) | 25.4 (22.5-29.0) | <.001 |
| Eating disorder | 53 666 (1) | 53 604 (1) | 62 (1) | .001 |
| Cancer | 495 207 (6) | 494 294 (6) | 913 (15) | <.001 |
| Hyperthyroidism | 74 528 (1) | 74 425 (1) | 103 (2) | <.001 |
| Inflammatory bowel disease | 60 763 (1) | 60 663 (1) | 100 (2) | <.001 |
| Rheumatoid arthritis | 55 052 (1) | 54 976 (1) | 76 (1) | <.001 |
| ≥1 Consecutive years of corticosteroid use | 221 613 (3) | 220 891 (3) | 722 (12) | <.001 |
| Vitamin D deficiency | 66 467 (1) | 66 334 (1) | 133 (2) | <.001 |
| Serum Ca2+, mmol/L, median (IQR) | 2.31 (2.24-2.38) | 2.31 (2.24-2.38) | 2.28 (2.21-2.36) | .001 |
| Bariatric surgery | 7235 (<1) | 7219 (<1) | 16 (<1) | <.001 |
| Osteoporosis screeningb | 901 312 (11) | 899 301 (11) | 2011 (32) | <.001 |
| Imaging | 145 968 (2) | 145 548 (2) | 420 (7) | <.001 |
| Bone turnover markers | 793 677 (28) | 791 950 (10) | 1727 (28) | <.001 |
| Fragility fracture | 432 187 (5) | 431 460 (5) | 727 (12) | <.001 |
| Incident osteoporosis | 142 203 (2) | 141 715 (2) | 488 (8) | <.001 |
Abbreviations: ASM, antiseizure medication; BMI, body mass index (measured as weight in kilograms divided by height in meters squared); eiASM, enzyme-inducing antiseizure medication; IMD, Index of Multiple Deprivation (divided into deciles; the higher the number, the less socially deprived the area).
Between incident adult-onset epilepsy and no epilepsy.
Constitutes imaging (dual-energy x-ray absorptiometry and quantitative computed tomography) and serum bone turnover markers and fragility fractures (both of which prompt further investigations) performed at any point in patients without osteoporosis or prior to the diagnosis of osteoporosis in patients who developed incident disease.
Hazard of Osteoporosis in the General Population
Survival analysis using a Cox proportional hazards model treating incident epilepsy as a time-varying covariate failed the proportional hazards assumption (global P < .001). In addition, an interaction between age and cancer was uncovered. The hazard of osteoporosis in patients with cancer was unexpectedly lower than in patients without cancer, which may be explained by the relative hazard between older and younger patients with cancer being anticipated to be close to 1, while that between older and younger patients without cancer being expected to be greater than 1. Thus, the ratio of the hazard ratios (HRs) of osteoporosis in older vs younger patients with cancer and the HR of older vs younger patients without cancer is expected to be less than 1. This expectation was confirmed by computing the HR associated with an age-by-cancer interaction term, which was significantly less than 1 (0.96; 95% CI, 0.96-0.97; P < .001); thus, when adjusting for the interaction term, cancer was significantly associated with incident osteoporosis.
We therefore used an adjusted accelerated failure time model with a Weibull distribution, treating incident epilepsy as a time-varying covariate and incorporating an interaction term between age and cancer. Incident epilepsy was associated with a 41% faster time to incident osteoporosis (TR, 0.59; 95% CI, 0.52-0.67; P < .001) (Tables 2 and 3; Figure). Median time from epilepsy diagnosis to incident osteoporosis was 2.5 years (IQR, 1.0-5.0 years). Both eiASMs (TR, 0.91; 95% CI, 0.87-0.95; P < .001) and non-eiASMs (TR, 0.77; 95% CI, 0.76-0.78; P < .001) were also associated with statistically significant increased risks independent of epilepsy, accounting for 9% and 23% faster times to incident osteoporosis, respectively. The association between non-eiASM use and osteoporosis remained consistent when valproic acid was excluded from this group (TR, 0.77; 95% CI, 0.75-0.78; P < .001). Median time from ASM exposure to incident osteoporosis was 4.1 years (IQR, 1.5-8.7 years). These estimates remained consistent in a sensitivity analysis in which BMI was omitted due to 30% missing values (TR, 0.54 [95% CI, 0.48-0.62], 0.90 [95% CI, 0.86-0.94], and 0.76 [95% CI, 0.74-0.77] for epilepsy, eiASM, and non-eiASM, respectively; all P < .001). In a second sensitivity analysis, strong inducers (TR, 0.91; 95% CI, 0.87-0.95; P < .001) and non-eiASMs (TR, 0.74; 95% CI, 0.73-0.76; P < .001) were associated with accelerated times to osteoporosis, but weak inducers were not (TR, 0.89; 95% CI, 0.79-1.01; P = .09). This finding could be a result of underpowering, since only 4% of ASM prescriptions were for weak inducers.
Table 2. Results of the Accelerated Failure Time Model for Incident Osteoporosis Stratified by an Incident Diagnosis of Epilepsya.
| Characteristic | Time ratio (95% CI) | P value |
|---|---|---|
| Incident epilepsy | 0.59 (0.52-0.67) | <.001 |
| eiASMb | 0.91 (0.87-0.95) | <.001 |
| Non-eiASMb | 0.77 (0.76-0.78) | <.001 |
| Age | 0.95 (0.95-0.96) | <.001 |
| Female sex | 0.29 (0.29-0.30) | <.001 |
| IMD | 1.06 (1.06-1.07) | <.001 |
| Bariatric surgery | 1.20 (1.01-1.43) | .04 |
| Cancer | 0.11 (0.10-0.12) | <.001 |
| 1-y Corticosteroids | 0.56 (0.55-0.57) | <.001 |
| Eating disorder | 0.56 (0.53-0.60) | <.001 |
| Body mass index | 1.03 (1.02-1.03) | <.001 |
| Hyperthyroidism | 0.99 (0.96-1.02) | .54 |
| Inflammatory bowel disease | 0.63 (0.61-0.65) | <.001 |
| Rheumatoid arthritis | 0.61 (0.60-0.63) | <.001 |
| Former or current smoker | 0.91 (0.89-0.91) | <.001 |
| Falls | 0.86 (0.85-0.87) | <.001 |
| Fragility fracture | 0.32 (0.32-0.33) | <.001 |
| Osteoporosis screeningc | 0.57 (0.57-0.58) | <.001 |
| Age-by-cancer interaction | 1.02 (1.02-1.03) | <.001 |
Abbreviations: eiASM, enzyme-inducing antiseizure medication; IMD, Index of Multiple Deprivation.
The index date is that of the Clinical Practice Research Datalink (CPRD)–acceptable data quality date. Incident epilepsy is treated as a time-varying covariate with a 5-year washout from the CPRD-acceptable data quality date. Time ratios are the ratio of the time to incident osteoporosis in patients exposed vs those unexposed for each variable. Time ratios for continuous variables relate to each 1-unit increment in specific measure.
Compared with no ASM use.
Constitutes imaging (dual-energy x-ray absorptiometry and quantitative computed tomography) and serum bone turnover markers performed at any point in patients without osteoporosis or prior to the diagnosis of osteoporosis in patients who developed incident disease.
Table 3. Nelson-Aalen Cumulative Hazard for Incident Osteoporosis Following a Diagnosis of Incident Epilepsy Controlling for Antiseizure Medication (ASM) Use and Common Osteoporosis Risk Factors.
| Time, y | No epilepsy | Incident epilepsya |
|---|---|---|
| 0 | 0.00 | 0.00 |
| 5 | 0.02 | 0.03 |
| 7 | 0.03 | 0.05 |
| 10 | 0.04 | 0.07 |
| 15 | 0.06 | 0.13 |
Incident epilepsy is treated as a time-varying covariate.
Figure. Nelson-Aalen Cumulative Hazard Graph of the Risk of Incident Osteoporosis Following Incident Diagnosis of Adult-Onset Epilepsy.
An initial 5-year washout was applied to maximize the chances that the epilepsy is incident. Incident diagnosis of epilepsy is treated as a time-varying covariate; therefore, everyone is unexposed for the first 5 years of follow-up. Patients remain unexposed until they develop epilepsy, at which exact point they switch from the unexposed to the exposed curve. The number of people with incident epilepsy thus increases as more develop the condition and then decreases as people either develop osteoporosis or are censored at last follow-up. The orange line indicates incident epilepsy with shading indicating the 95% CIs. The blue line indicates control patients without epilepsy with shading indicating the 95% CIs.
Hazard of Osteoporosis in an eiASM Propensity Score–Matched Cohort
A 5:1 subcohort of control patients (n = 193 225) to patients exposed to 4 consecutive eiASMs (n = 38 645) was generated through propensity score matching on measures that could estimate prescription behavior. The mean standard difference between matched eiASM and non-eiASM groups was less than or equal to 0.01 for all variables apart from age, age-by-sex interaction, psychosis, and suicidal ideation l (eTable 1 in Supplement 1), and propensity scores were evenly distributed between the matched eiASM and non-eiASM groups (eFigures 1 and 2 in Supplement 1).
The model again failed the proportional hazards assumption (global test P < .001), so we proceeded with an accelerated failure time model using incident epilepsy as a time-varying covariate and incorporating an interaction term between age and cancer. In this propensity score–matched analysis, incident epilepsy was associated with incident osteoporosis (TR, 0.72; 95% CI, 0.47-0.97; P = .011) (eTable 2 in Supplement 1), as were non-eiASMs (TR, 0.80; 95% CI, 0.74-0.86; P < .001) and eiASMs (TR, 0.89; 95% CI, 0.84-0.94; P < .001).
Hazard of Osteoporosis Secondary to eiASMs vs Non-eiASMs in Patients With Incident Epilepsy
When restricting the cohort to the 6275 patients with incident adult-onset epilepsy, we had to omit bariatric surgery due to the relative lack of cases. The survival model failed proportional hazards assumptions (global P = .04). An accelerated failure time model incorporating an age-by-cancer interaction term revealed that the risk of incident osteoporosis did not differ significantly between eiASMs and non-eiASMs (TR, 1.01; 95% CI, 0.86-1.18; P = .87) (Table 4). The TR estimates for eiASMs vs non-eiASMs did not change when BMI was omitted (1.01; 95% CI, 0.87-1.18; P = .84).
Table 4. Results of the Accelerated Failure Time Model for Incident Osteoporosis in a Cohort Restricted to 6275 People With Incident Epilepsya.
| Characteristic | Time ratio (95% CI) | P value |
|---|---|---|
| eiASMb | 1.01 (0.86-1.18) | .87 |
| Age | 0.98 (0.97-0.98) | <.001 |
| Female sex | 0.59 (0.51-0.67) | <.001 |
| IMD | 1.04 (1.01-1.07) | .003 |
| Cancer | 0.35 (0.18-0.68) | .002 |
| 1-Year corticosteroids | 0.67 (0.58-0.77) | <.001 |
| Eating disorder | 0.89 (0.49-1.61) | .71 |
| Body mass index | 1.02 (1.01-1.03) | <.001 |
| Hyperthyroidism | 1.13 (0.78-1.65) | .50 |
| Inflammatory bowel disease | 0.72 (0.53-0.99) | .049 |
| Rheumatoid arthritis | 0.67 (0.50-0.90) | .007 |
| Former or current smoker | 0.93 (0.83-1.05) | .29 |
| Falls | 0.82 (0.72-0.92) | .002 |
| Fragility fracture | 0.46 (0.40-0.53) | <.001 |
| Osteoporosis screeningc | 0.78 (0.70-0.88) | <.001 |
| Age-by-cancer interaction | 1.01 (1.01-1.02) | .001 |
Abbreviations: eiASM, enzyme-inducing antiseizure medication; IMD, Index of Multiple Deprivation.
The index date is that of the Clinical Practice Research Datalink (CPRD)–acceptable data quality date. Incident epilepsy is treated as a time-varying covariate with a 5-year washout from the CPRD-acceptable data quality date. Time ratios are the ratio of the time to incident osteoporosis in patients exposed vs those unexposed for each variable. Time ratios for continuous variables relate to each 1-unit increment in specific measure. Bariatric surgery was excluded due to no cases.
Compared with non-eiASM use.
Constitutes imaging (dual-energy x-ray absorptiometry and quantitative computed tomography) and serum bone turnover markers performed at any point in patients without osteoporosis or prior to the diagnosis of osteoporosis in patients who developed incident disease.
Hazard of Osteoporosis in Incident Late-Onset Epilepsy
This analysis involved 11 143 patients (1048 with incident late-onset epilepsy and 10 095 control patients). The survival model for late-onset epilepsy failed proportional hazards assumptions (global P < .001). Using an accelerated failure time model incorporating an age-by-cancer interaction term, late-onset epilepsy was associated with an elevated risk of incident osteoporosis (TR, 0.81; 95% CI, 0.72-0.90; P < .001), as were non-eiASMs (TR, 0.90; 95% CI, 0.83-0.97; P = .007). The association was not statistically significant for eiASMs (TR, 0.95; 95% CI, 0.81-1.11; P = .56) (eTable 3 in Supplement 1), which may have been due to underpowering, since only 298 patients (3%) in this analysis were exposed to eiASMs compared with 1696 (15%) who were exposed to non-eiASMs and 9149 (82%) who lacked exposure to any ASMs.
Discussion
This cohort study shows a clear and robust association between incident adult-onset epilepsy and incident osteoporosis, independent of medications, common risk factors, fragility fractures, and falls. The findings also showed clear associations between both eiASM and non-eiASM use and incident osteoporosis, independent of incident epilepsy. This risk continued to increase over the 15 years following an epilepsy diagnosis (Table 3; Figure). These associations remained consistent in propensity score–matched analyses, cohorts restricted to incident adult-onset epilepsy, and all cases of incident late-onset epilepsy. Finally, the hazard associated with late-onset epilepsy was comparable to that of adult-onset epilepsy, indicating that the risk does not appear to be further elevated in an older population.
These findings corroborate and build upon prior studies. Few studies have reported the hazard associated with epilepsy alone, especially at a population level, instead focusing on ASM use. The prevailing theory has been that eiASMs convey a higher risk than non-eiASMs due to interactions with the cytochrome P450 system, which could accelerate metabolism of vitamin D, leading to compensatory increases of parathyroid hormone. Parathyroid hormone releases active vitamin D metabolites in an attempt to replenish stores, but the undesirable consequence is increased bone turnover.4 However, prior studies have reported conflicting results on the outcomes of ASMs. Some indicate that strong inducers, like carbamazepine, decrease bone density and bone turnover,26 while others have reported no association.27 Consistent with our results, prior literature has highlighted that even non-eiASM use, such as valproate28 and levetiracetam,29 is associated with low bone mass density, with effect sizes comparable to or even greater than eiASMs.30 The risk attributable to eiASMs appears comparable to non-eiASMs in Alzheimer disease (HR, 1.10; 95% CI, 0.77-1.57),31 a finding consistent with our analyses in a general population with and without epilepsy.
Incident adult-onset epilepsy, independent of ASM use, appears to accelerate time to osteoporosis by approximately 41% compared with the general population. The underlying mechanisms linking epilepsy and osteoporosis have been insufficiently studied compared with the role of ASMs. Population-based investigations have indicated that people with epilepsy are less likely to participate in physical activity and have lower rates of fruit and vegetable consumption, which are known risk factors for osteoporosis.20,32,33 More sedentary behaviors and remaining homebound may also lead to reduced sun exposure in people with epilepsy.34 Unintentional injuries and falls are common in epilepsy35 and could increase the risk for subsequent osteoporosis.20
Our findings highlight the importance of risk mitigation in all people with epilepsy. However, a recent large randomized controlled trial revealed that daily supplementation with 2000 IU of vitamin D did not reduce the risk of bone loss over 2 years or fractures over 5 years compared with placebo, even when calcium was concomitantly administered.36,37 Thus, although vitamin D deficiency and lower serum calcium were more common in people with epilepsy (Table 1), controlling for supplementation may have had a minimal effect on our conclusions. There may still be benefit in patients with low baseline free 25-hydroxyvitamin D levels,37 meaning that efforts to study routine supplementation in all people with epilepsy are paramount given their risks of vitamin D deficiency. The outcomes of supplemental calcium on fracture risk have been varied, and caution should be taken when recommending daily doses of 1000 to 1500 mg.19 Algorithms for routine screening through laboratory work and bone mass densitometry should be validated and deployed for all people with epilepsy, given that early detection could improve outcomes.38
Strengths and Limitations
Strengths of this study include its large, population-based cohort design that is representative of the general UK populace. The epilepsy clinical phenotype has a high sensitivity and specificity,13 while the outcome and covariate definitions have all been validated as part of the CALIBER program.17 We captured all event data at multiple levels of primary and hospital care. The effect size associated with epilepsy, eiASMs, and non-eiASMs remained robust and consistent through multiple analyses, including standard survival analyses using time-varying methodology, propensity score–matched models, adjustment for covariates, and within specific subpopulations.
There are also some limitations. The assumption was that people taking 4 consecutive ASMs continued with this class of medication. We cannot guarantee that all prescriptions were enduring, but previous analyses applying similar methodology have validated the assumption.25 Additionally, median prescription duration (time from first to last documented prescription) was almost 9 years, suggesting adherence. Our analyses lack granular data on epilepsy type, seizure type, and seizure frequency. Using a population-based study, though, we captured the full spectrum of disease, and by adjusting for falls, fragility fractures, and enhanced screening, we have accounted for the major features of epilepsy and seizures that can confer risk or increase detection of osteoporosis. Theoretically, such an adjustment could have led to greater detection rates for osteopenia in people with epilepsy, leading to disproportionate use of prophylactic treatment compared with the control population. However, in a general population of older women, median time for 10% of the those with mild and moderate osteopenia to transition to osteoporosis was 17.3 and 4.7 years, respectively.39 Hence, intervention may not prevent osteoporosis at a rate to nullify results in our population and, in the unlikely case that it did, would mean our results represent a conservative estimate of even higher effect sizes. However, this limitation is critical to note since, theoretically, the effect sizes for both epilepsy and eiASMs could be even higher than we report given that both may prompt intensive screening strategies in modern clinics.
Conclusions
The findings of our cohort study show robust and clinically meaningful independent associations between incident epilepsy and both eiASM and non-eiASM use with incident osteoporosis, further consolidating the need for enhanced vigilance and consideration of prophylaxis for all people with epilepsy. People with epilepsy are more likely to be deficient in vitamin D,40 and those who are deficient may still benefit from routine supplementation.37 Randomized controlled trials studying this approach universally in epilepsy are expediently required, as are evidence-based algorithms for routine osteoporosis screening irrespective of ASM type.
eTable 1. Summary of Balance for Propensity-Matched Cohorts Based on Exposure to Enzyme-Inducing Antiseizure Medications
eTable 2. Results of the Accelerated Time Failure Model for Incident Osteoporosis Stratified by an Incident Diagnosis of Epilepsy Following Propensity Matching for Exposure to an Enzyme-Inducing Antiseizure Medication
eTable 3. Results of the Accelerated Time Failure Model for Incident Osteoporosis Stratified by an Incident Diagnosis of Late-Onset Epilepsy in Those 65 or Older
eFigure 1. Distribution of Propensity Scores Among the Matched Cohort Exposed to Enzyme-Inducing Antiseizure Medications, the Matched Unexposed, and the Unmatched Unexposed
eFigure 2. Histograms of Propensity Scores Between the Original and Matched Cohorts Based on Enzyme-Inducing Antiseizure Medication Exposure
eAppendix.
Data Sharing Statement
References
- 1.Beghi E, Giussani G, Nichols E, et al. ; GBD 2016 Epilepsy Collaborators . Global, regional, and national burden of epilepsy, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Neurol. 2019;18(4):357-375. doi: 10.1016/S1474-4422(18)30454-X [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bliuc D, Nguyen ND, Milch VE, Nguyen TV, Eisman JA, Center JR. Mortality risk associated with low-trauma osteoporotic fracture and subsequent fracture in men and women. JAMA. 2009;301(5):513-521. doi: 10.1001/jama.2009.50 [DOI] [PubMed] [Google Scholar]
- 3.Shen C, Chen F, Zhang Y, Guo Y, Ding M. Association between use of antiepileptic drugs and fracture risk: a systematic review and meta-analysis. Bone. 2014;64:246-253. doi: 10.1016/j.bone.2014.04.018 [DOI] [PubMed] [Google Scholar]
- 4.Mintzer S. Metabolic consequences of antiepileptic drugs. Curr Opin Neurol. 2010;23(2):164-169. doi: 10.1097/WCO.0b013e32833735e7 [DOI] [PubMed] [Google Scholar]
- 5.CALIBER. UCL Institute of Health Informatics. Accessed February 1, 2023. https://www.ucl.ac.uk/health-informatics/caliber [Google Scholar]
- 6.A reference catalogue of human diseases. HDRUK Phenotype Library. Accessed February 1, 2023. https://www.caliberresearch.org [Google Scholar]
- 7.Denaxas S, Gonzalez-Izquierdo A, Fitzpatrick N, Direk K, Hemingway H. Phenotyping UK electronic health records from 15 million individuals for precision medicine: the CALIBER Resource. Stud Health Technol Inform. 2019;262:220-223. [DOI] [PubMed] [Google Scholar]
- 8.Denaxas SC, George J, Herrett E, et al. Data resource profile: cardiovascular disease research using linked bespoke studies and electronic health records (CALIBER). Int J Epidemiol. 2012;41(6):1625-1638. doi: 10.1093/ije/dys188 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ghosh RE, Crellin E, Beatty S, Donegan K, Myles P, Williams R. How Clinical Practice Research Datalink data are used to support pharmacovigilance. Ther Adv Drug Saf. 2019;10:2042098619854010. doi: 10.1177/2042098619854010 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Chisholm J. The Read clinical classification. BMJ. 1990;300(6732):1092-1092. doi: 10.1136/bmj.300.6732.1092 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.García Rodríguez LA, Pérez Gutthann S. Use of the UK General Practice Research Database for pharmacoepidemiology. Br J Clin Pharmacol. 1998;45(5):419-425. doi: 10.1046/j.1365-2125.1998.00701.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Joint Formulary Committee , ed. British National Formulary (BNF) 57. Pharmaceutical Press; 2009. [Google Scholar]
- 13.Fonferko-Shadrach B, Lacey AS, White CP, et al. Validating epilepsy diagnoses in routinely collected data. Seizure. 2017;52:195-198. doi: 10.1016/j.seizure.2017.10.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Josephson CB, Engbers JDT, Sajobi TT, et al. Towards a clinically informed, data-driven definition of elderly onset epilepsy. Epilepsia. 2016;57(2):298-305. doi: 10.1111/epi.13266 [DOI] [PubMed] [Google Scholar]
- 15.Johannessen SI, Landmark CJ. Antiepileptic drug interactions—principles and clinical implications. Curr Neuropharmacol. 2010;8(3):254-267. doi: 10.2174/157015910792246254 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Perucca E. Clinically relevant drug interactions with antiepileptic drugs. Br J Clin Pharmacol. 2006;61(3):246-255. doi: 10.1111/j.1365-2125.2005.02529.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kuan V, Denaxas S, Gonzalez-Izquierdo A, et al. A chronological map of 308 physical and mental health conditions from 4 million individuals in the English National Health Service. Lancet Digit Health. 2019;1(2):e63-e77. doi: 10.1016/S2589-7500(19)30012-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Jordan H, Roderick P, Martin D. The Index of Multiple Deprivation 2000 and accessibility effects on health. J Epidemiol Community Health. 2004;58(3):250-257. doi: 10.1136/jech.2003.013011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Compston JE, McClung MR, Leslie WD. Osteoporosis. Lancet. 2019;393(10169):364-376. doi: 10.1016/S0140-6736(18)32112-3 [DOI] [PubMed] [Google Scholar]
- 20.NIH Consensus Development Panel on Osteoporosis Prevention, Diagnosis, and Therapy . Osteoporosis prevention, diagnosis, and therapy. JAMA. 2001;285(6):785-795. doi: 10.1001/jama.285.6.785 [DOI] [PubMed] [Google Scholar]
- 21.Denaxas S, Gonzalez-Izquierdo A, Direk K, et al. UK phenomics platform for developing and validating electronic health record phenotypes: CALIBER. J Am Med Inform Assoc. 2019;26(12):1545-1559. doi: 10.1093/jamia/ocz105 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Grambsch PM, Therneau TM. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika. 1994;81(3):515-526. doi: 10.1093/biomet/81.3.515 [DOI] [Google Scholar]
- 23.Suissa S. Immortal time bias in pharmaco-epidemiology. Am J Epidemiol. 2008;167(4):492-499. doi: 10.1093/aje/kwm324 [DOI] [PubMed] [Google Scholar]
- 24.Ho D, Imai K, King G, Stuart EA. MatchIt: nonparametric preprocessing for parametric causal inference. J Stat Softw. 2011;42(8):1-28. doi: 10.18637/jss.v042.i08 [DOI] [Google Scholar]
- 25.Josephson CB, Wiebe S, Delgado-Garcia G, et al. Association of enzyme-inducing antiseizure drug use with long-term cardiovascular disease. JAMA Neurol. 2021;78(11):1367-1374. doi: 10.1001/jamaneurol.2021.3424 [DOI] [PubMed] [Google Scholar]
- 26.Mintzer S, Boppana P, Toguri J, DeSantis A. Vitamin D levels and bone turnover in epilepsy patients taking carbamazepine or oxcarbazepine. Epilepsia. 2006;47(3):510-515. doi: 10.1111/j.1528-1167.2006.00460.x [DOI] [PubMed] [Google Scholar]
- 27.Pack AM, Morrell MJ, Randall A, McMahon DJ, Shane E. Bone health in young women with epilepsy after one year of antiepileptic drug monotherapy. Neurology. 2008;70(18):1586-1593. doi: 10.1212/01.wnl.0000310981.44676.de [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Sheth RD, Wesolowski CA, Jacob JC, et al. Effect of carbamazepine and valproate on bone mineral density. J Pediatr. 1995;127(2):256-262. doi: 10.1016/S0022-3476(95)70304-7 [DOI] [PubMed] [Google Scholar]
- 29.Beniczky SA, Viken J, Jensen LT, Andersen NB. Bone mineral density in adult patients treated with various antiepileptic drugs. Seizure. 2012;21(6):471-472. doi: 10.1016/j.seizure.2012.04.002 [DOI] [PubMed] [Google Scholar]
- 30.Ensrud KE, Walczak TS, Blackwell TL, Ensrud ER, Barrett-Connor E, Orwoll ES; Osteoporotic Fractures in Men (MrOS) Study Research Group . Antiepileptic drug use and rates of hip bone loss in older men: a prospective study. Neurology. 2008;71(10):723-730. doi: 10.1212/01.wnl.0000324919.86696.a9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Pisa F, Reinold J, Lavikainen P, et al. Hip fracture risk in antiepileptic drug initiators and non-initiators with Alzheimer’s disease. Clin Epidemiol. 2021;13:295-307. doi: 10.2147/CLEP.S278306 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Roberts JI, Patten SB, Wiebe S, Hemmelgarn BR, Pringsheim T, Jetté N. Health-related behaviors and comorbidities in people with epilepsy: changes in the past decade. Epilepsia. 2015;56(12):1973-1981. doi: 10.1111/epi.13207 [DOI] [PubMed] [Google Scholar]
- 33.Qiu R, Cao WT, Tian HY, He J, Chen GD, Chen YM. Greater intake of fruit and vegetables is associated with greater bone mineral density and lower osteoporosis risk in middle-aged and elderly adults. PLoS One. 2017;12(1):e0168906. doi: 10.1371/journal.pone.0168906 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Svalheim S, Røste LS, Nakken KO, Taubøll E. Bone health in adults with epilepsy. Acta Neurol Scand Suppl. 2011;124(191):89-95. doi: 10.1111/j.1600-0404.2011.01551.x [DOI] [PubMed] [Google Scholar]
- 35.Beghi E, Cornaggia C; RESt-1 Group . Morbidity and accidents in patients with epilepsy: results of a European cohort study. Epilepsia. 2002;43(9):1076-1083. doi: 10.1046/j.1528-1157.2002.18701.x [DOI] [PubMed] [Google Scholar]
- 36.LeBoff MS, Chou SH, Ratliff KA, et al. Supplemental vitamin D and incident fractures in midlife and older adults. N Engl J Med. 2022;387(4):299-309. doi: 10.1056/NEJMoa2202106 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.LeBoff MS, Chou SH, Murata EM, et al. Effects of supplemental vitamin D on bone health outcomes in women and men in the VITamin D and OmegA-3 TriaL (VITAL). J Bone Miner Res. 2020;35(5):883-893. doi: 10.1002/jbmr.3958 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Sheth RD, Harden CL. Screening for bone health in epilepsy. Epilepsia. 2007;48(suppl 9):39-41. doi: 10.1111/j.1528-1167.2007.01401.x [DOI] [PubMed] [Google Scholar]
- 39.Gourlay ML, Fine JP, Preisser JS, et al. ; Study of Osteoporotic Fractures Research Group . Bone-density testing interval and transition to osteoporosis in older women. N Engl J Med. 2012;366(3):225-233. doi: 10.1056/NEJMoa1107142 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Teagarden DL, Meador KJ, Loring DW. Low vitamin D levels are common in patients with epilepsy. Epilepsy Res. 2014;108(8):1352-1356. doi: 10.1016/j.eplepsyres.2014.06.008 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Summary of Balance for Propensity-Matched Cohorts Based on Exposure to Enzyme-Inducing Antiseizure Medications
eTable 2. Results of the Accelerated Time Failure Model for Incident Osteoporosis Stratified by an Incident Diagnosis of Epilepsy Following Propensity Matching for Exposure to an Enzyme-Inducing Antiseizure Medication
eTable 3. Results of the Accelerated Time Failure Model for Incident Osteoporosis Stratified by an Incident Diagnosis of Late-Onset Epilepsy in Those 65 or Older
eFigure 1. Distribution of Propensity Scores Among the Matched Cohort Exposed to Enzyme-Inducing Antiseizure Medications, the Matched Unexposed, and the Unmatched Unexposed
eFigure 2. Histograms of Propensity Scores Between the Original and Matched Cohorts Based on Enzyme-Inducing Antiseizure Medication Exposure
eAppendix.
Data Sharing Statement