Abstract
Objective
There is poor evidence supporting the use of any pharmacologic treatments for neuropsychiatric disorders following traumatic brain injury (TBI), especially among older adults. Informed by our recent characterization of psychotropic medication use among Medicare beneficiaries with TBI, the objective of this study was to compare the risk of several adverse events associated with use of the three most commonly used classes of antidepressants following TBI in this population.
Methods
We conducted a retrospective cohort study using administrative claims data from US Medicare beneficiaries hospitalized with TBI between 2006 and 2010 (n = 30,886). We assessed monthly selective serotonin reuptake inhibitor (SSRI), serotonin-norepinephrine reuptake inhibitor (SNRI), and tricyclic antidepressant (TCA) use. We identified adverse events associated with these drug classes that were available in administrative claims data from studies in TBI and non-TBI populations: seizures, hemorrhagic stroke, ischemic stroke, gastrointestinal bleed, hyponatremia, and fractures. We made comparisons between antidepressant classes to assess excess risk of each adverse event using discrete time analysis and controlling for potential confounders.
Results
SSRIs were the most commonly used of the antidepressant classes, followed by SNRIs and TCAs. We observed a total of 23,021 adverse events. Ischemic stroke was the most frequent (8296 events). Hemorrhagic stroke (1706 events) and seizures (1841) were least often observed. Compared with TCAs, SSRI use was associated with an increased risk of hemorrhagic stroke (risk ratio 2.47; 95% confidence interval 1.30–4.70). No other antidepressant class comparisons were associated with increased risk of adverse events.
Conclusion
Compared with SSRIs, use of SNRIs and TCAs following hospitalization for TBI among Medicare beneficiaries was not associated with an increased risk of any of the studied adverse events. Compared to TCAs, SSRI use was associated with increased risk of hemorrhagic stroke. This information may help guide patients and prescribers in selecting antidepressants for older adults following TBI.
1. Introduction
Traumatic brain injury (TBI) is a significant health problem among older adults that results in over 486,000 emergency department visits and 123,585 hospitalizations annually in the United States [1]. Neuropsychiatric sequelae are common following TBI and are associated with poorer cognitive and functional recovery [2–4]. This is of particular importance among older adults whose outcomes are worse compared to younger adults with similar injury severity [5, 6].
There is poor evidence supporting the use of any pharmacologic treatment for neuropsychiatric disorders following TBI, especially among older adults, among whom multimorbidity and polypharmacy are common [7–9]. Evidence from a few small trials and treatment of neuropsychiatric disorders among individuals without TBI currently guide treatment decisions for individuals with TBI [7–10].
Individuals with TBI are at heightened risk of stroke and seizure, complicating prescribing for neuropsychiatric disorders [11–13]. In non-TBI populations, selective serotonin reuptake inhibitors (SSRIs) have been associated with increased risk of hemorrhagic stroke and seizures [14–18]. There is also evidence of increased risk of ischemic stroke associated with SSRI use among older adults [15]. Tricyclic antidepressants (TCAs) can also lower the seizure threshold [19, 20]. In addition to their association with seizures, many TCAs have strong anticholinergic effects which can exacerbate cognitive dysfunction following TBI, and are on the Beer’s list of medications to be avoided among older adults [21]. Serotonin-norepinephrine reuptake inhibitors (SNRIs) have been observed to be associated with increased risk of stroke and hyponatremia [15, 22, 23]. Little is known about the risk of adverse events specifically among individuals with TBI; however, there is evidence of increased risk of seizure associated with TCA use [24]. Concerns about the safety of psychotropic medications in individuals with TBI could lead to reluctance in prescribing and possible undertreatment of neuropsychiatric disturbances following TBI [9, 25–27].
We recently characterized psychotropic medication use before and after hospitalization for TBI among older adults as a preliminary step toward conducting safety analyses of medications used to treat neuropsychiatric disturbances following TBI [26]. Older adults hospitalized with TBI had a high prevalence of psychotropic medication use both before and after injury, yet were less likely to receive indicated pharmacologic treatment for newly diagnosed neuropsychiatric disturbances following TBl [25, 26]. Based on these results, the objective of this study was to quantify the comparative risk of adverse events associated with use of the three most commonly used classes of antidepressants following TBI: SSRIs, SNRIs, and TCAs [26]. This information can help guide patients and prescribers in selecting antidepressants for indicated use following TBI in older adults.
2. Methods
We analyzed Medicare administrative claims data obtained from the US Centers for Medicare & Medicaid Services (CMS) Chronic Condition Data Warehouse (CCW). All older Medicare beneficiaries hospitalized with TBI during 2006–2010 were eligible. Inclusion criteria were age ≥ 65 years at TBI, first TBI, and survival to hospital discharge. We used the US Centers for Disease Control and Prevention’s case definition for TBI, comprising International Classification of Disease, 9th Revision Clinical Modification (ICD-9-CM) codes 800.xx, 801.xx, 803.xx, 804.xx, 850.xx–854.1x, 950.1–950.3, or 959.01 in any position on an inpatient claim, to define cases [28, 29]. These ICD-9-CM codes have previously been reported to have high sensitivity and a positive predictive value to detect severe TBI and correctly identify TBI-related hospitalizations, respectively [30, 31]. We required 12 months continuous Medicare Parts A, B, and D with no Part C (Medicare Advantage) coverage pre-TBI to permit capture of baseline comorbidities and antidepressant medication use during the 3 months prior to TBI and required 24 months continuous coverage post-TBI hospitalization to capture antidepressant medication use and adverse events. Because prevalence of antidepressant use is typically high in this population, we did not exclude individuals who used antidepressants before TBI.
2.1. Antidepressant Medications
For this study, we examined the antidepressant classes SSRIs, SNRIs, and TCAs. We searched Medicare Part D prescription drug event files for evidence of antidepressant medication use in any of these classes (see the electronic supplementary material, Appendix 1). We created 30-day months pre- and post-TBI hospitalization and defined antidepressant use in each month as (1) a filled prescription for any antidepressant or (2) a proportion of days covered (number of daily doses in the prescription/number of days in the month) for any antidepressant > 0. Beneficiaries contributed person-months (PM) of antidepressant use for each month in which they had evidence of antidepressant medication use. We made no requirement for length of time on medication and allowed beneficiaries to contribute PM of antidepressant use as long as they met the criteria. Administrative claims do not provide information on indication for prescribed medications; therefore, this analysis focused on use of antidepressant medications regardless of indication. Furthermore, examination of different antidepressant dosages would have made comparisons practically impossible; therefore, our exposure variable was use during the month.
Conducting class-to-class comparisons between the different antidepressants at the PM level required exclusion of PM contributed to two or more classes simultaneously. We also excluded individuals who switched from one class to another during the follow-up period. Therefore, the analyses focus on beneficiaries who are exclusively using one class of antidepressants or another during the post-TBI period.
Use of antidepressants during the pre-TBI period could impact risk of adverse events in the post-TBI period. To control for this, we created an indicator variable for antidepressant use by class during the 3 months prior to TBI.
2.2. Adverse Events
Literature on the treatment of psychiatric disturbances following TBI has been limited by small sample sizes, preventing assessment of adverse events [8, 10]. Therefore, we identified serious adverse events associated with antidepressant use based on studies conducted in both TBI and non-TBI populations. TCAs have been associated with increased risk of seizures; SSRIs have been associated with increased risk of fractures, seizures, hemorrhagic and ischemic stroke, gastrointestinal bleeds, and hyponatremia; and SNRIs have been associated with increased risk of stroke and hyponatremia [15–17, 19, 20, 22–24, 32]. Adverse events available in administrative data were identified by searching inpatient and outpatient claims after TBI for the following ICD-9-CM codes: seizures (345.xx), gastrointestinal bleeding (531.xx, 532.xx, 533.xx, 534.xx, 578.xx), hemorrhagic stroke (430.xx–432.xx), ischemic stroke (433.xx, 434.xx, 435.xx, 437.0x, 437.1x), hyponatremia (276.1), and fractures (800.00–829.00). While these events do not represent all possible adverse events associated with antidepressant use, they were selected on the basis of our ability to capture them in administrative data.
2.3. Covariates
We obtained demographic and enrollment characteristics from the CCW enrollment files. Baseline comorbidities at TBI hospitalization were determined using the CMS CCW flagged comorbid conditions [33]. These 27 chronic conditions are identified on inpatient, skilled nursing, home health, or outpatient claims using ICD-9–based algorithms defined by the CMS. We used the date of first diagnosis for each flagged condition to create time-varying comorbidity diagnoses throughout the study period. In addition, we created indicator variables for alcohol dependence and abuse (ICD-9-CM 291.xx, 303.xx, 305.0x, 571.0x, 571.2x, 571.3x) and substance dependence and abuse (ICD-9-CM 304.x, 305.1x–305.9x).
Diagnosis and treatment of neuropsychiatric disorders in the pre-TBI period could influence receipt of antidepressant medications and the potential for adverse events. Therefore, we created indicator variables for pre-TBI diagnoses by searching inpatient and outpatient claims for the following ICD-9-CM codes: 296.2, 296.3, 300.4, 311 (depression); 300.0x (anxiety); 298.0 (psychosis). We searched the post-TBI period using the same ICD-9-CM codes to create variables indicating new diagnoses of depression or anxiety post-TBI for those without those diagnoses pre-TBI.
2.4. Data Analysis
We assessed distributions of demographic and clinical variables among beneficiaries in the study population overall and by total number of PM during follow-up contributed to SSRI, SNRI, and TCA use. We examined differences in the distribution of covariates between PM contributed to each of the studied antidepressant classes in a pairwise fashion, and tested these comparisons using Chi square or Student’s t tests.
We assessed monthly counts of adverse events by antidepressant use during the 2 years following TBI. We calculated monthly rates/100 persons and plotted these. An individual could have contributed more than one type of adverse event during a month as well as have repeats of the same adverse event in different months. We excluded the first 2 months post-TBI hospitalization to avoid counting follow-up visits for the TBI.
Discrete time analysis was used to make class-to-class comparisons between the different antidepressant classes (SSRI vs. SNRI, SSRI vs. TCA, SNRI vs. TCA) and risk of each of the six defined adverse events (seizures, gastrointestinal bleeding, hemorrhagic stroke, ischemic stroke, hyponatremia, and fractures). Discrete time analysis is a time-to-event analysis suited for use with data containing many ties and permits both exposures and covariates to vary over time, making it ideal for use with longitudinal data [34]. To ensure temporality, we lagged our exposure variables by 1 month. As with the monthly rates, we excluded the first 2 months following hospitalization to avoid counting follow-up visits related to fractures, seizures, and stroke as ‘new’ adverse events.
For each class-to-class comparison, we modeled the unadjusted association between use of one of the two studied antidepressant classes and time to first adverse event using generalized estimating equations with a binary distribution and a complementary log–log link [30]. The resulting exponentiated effect estimates can be interpreted as risk ratios (RRs). We added potential confounders identified from the literature or bivariate analysis to our regression model and kept variables whose type III p value was < 0.001. After establishing these models for the SSRI/SNRI comparison, we used the same model for class-to-class comparison for each adverse event, changing only the pre-TBI antidepressant use terms. RRs and 95% confidence intervals (CI) are reported. Each adverse event was modeled separately for each class-to-class comparison, resulting in 18 unadjusted and 18 adjusted models. In the event of a statistically significant association at p = 0.05, we applied the Holm correction for multiple comparisons to assess statistical significance at the penalized p value [35].
To determine if use of any antidepressant versus no antidepressant was associated with increased risk of an adverse event, we modeled time to first adverse event of any type as a function of lagged use of any antidepressant. To determine whether adverse event diagnoses related to the TBI hospitalization were inflating estimates even though we excluded the first 2 months post-TBI, we conducted sensitivity analyses. We excluded the 3rd month post-hospitalization from our analyses and compared with prior estimates. Finally, switching from one antidepressant class to another may happen due to the occurrence of an adverse event. To determine if switching antidepressants was associated with increased risk of adverse events, we ran a separate analysis on individuals we excluded from our main analysis because they switched from one antidepressant class to another during the study period.
This study was approved by the Institutional Review Board of the University of Maryland, Baltimore. All analyses were performed with SAS version 9.4 (SAS Institute, Cary, NC).
3. Results
There were 30,886 Medicare beneficiaries hospitalized for TBI during 2006–2010 meeting our inclusion criteria. Of these, 15,733 (51%) used one of the studied antidepressants at least once during the 2 years following injury. In the total 741,264 PM of follow-up, SSRI use was identified in 173,347 PM (23%), SNRI use was identified in 36,955 PM (5%), and TCA use was identified in 21,248 PM (3%) (Table 1).
Table 1.
Characteristics of Medicare beneficiaries hospitalized with traumatic brain injury (TBI) and of person-months (PM) associated with antidepressant use during the 2 years following injury, n = 30,886
| Characteristic | N = 30,886a | SSRI, PM = 173,347 | SNRI, PM = 36,955 | TCA, PM = 21,248 |
|---|---|---|---|---|
|
| ||||
| Age, mean (SD) | 79.7 (7.7) | 79.9 (7.6) | 78.3 (7.3) | 77.8 (7.3) |
| Age group | ||||
| 65–74 years | 8670 (28) | 27.1% | 33.7% | 37.1% |
| 74–84 years | 13,183 (43) | 42.1% | 45.3% | 42.7% |
| > 84 years | 9033 (29) | 30.8% | 21.1% | 20.3% |
| Female | 21,034 (68) | 76.4% | 79.5% | 78.4% |
| Race | ||||
| White | 26,691 (86) | 91.1% | 92.8% | 90.5% |
| Black | 1753 (6) | 3.8% | 2.6% | 3.5% |
| Other | 2442 (8) | 5.1% | 4.7% | 6.1% |
| Comorbid conditions | ||||
| ADRD | 10,057 (33) | 57.8% | 55.7% | 37.9% |
| Alcohol dependence | 1887 (6) | 6.7% | 8.0% | 5.7% |
| Anemia | 21,043 (68) | 81.5% | 82.6% | 78.1% |
| Atrial fibrillation | 7660 (25) | 29.0% | 28.0% | 22.3% |
| Chronic kidney disease | 7655 (25) | 34.9% | 37.0% | 33.4% |
| COPD | 10,731 (35) | 45.1% | 48.6% | 43.8% |
| Diabetes | 12,732 (41) | 47.4% | 53.1% | 50.3% |
| Heart failure | 14,339 (46) | 58.9% | 59.3% | 54.3% |
| Hip fracture | 1650 (5) | 15.4% | 15.2% | 12.5% |
| ISHD | 20,714 (67) | 75.7% | 76.9% | 73.6% |
| Neurologic disease | 4265 (14) | 23.7% | 22.8% | 16.2% |
| Rheumatoid/osteoarthritis | 20,503 (66) | 78.1% | 83.7% | 81.2% |
| Stroke | 11,678 (38) | 53.4% | 53.7% | 46.3% |
| Psychosis | 4827 (16) | 30.4% | 33.9% | 22.4% |
| Use in 3 months pre-TBI | ||||
| SSRI | 7579 (25) | 71.7% | 22.2% | 26.7% |
| SNRI | 1645 (5) | 3.3% | 66.1% | 7.8% |
| TCA | 1400 (5) | 5.6% | 6.8% | 74.0% |
| Discharge to skilled nursing | 656 (2) | 2.9% | 4.1% | 2.5% |
| Pre-TBI diagnosis of | ||||
| Depression | 5791 (19) | 36.6% | 46.7% | 28.0% |
| Anxiety | 3367 (11) | 17.5% | 21.1% | 15.9% |
| New diagnosis post-TBI | ||||
| Depression | 6281 (20) | 35.7% | 33.6% | 23.1% |
| Anxiety | 3940 (13) | 18.3% | 21.1% | 17.7% |
| OREC | ||||
| Age | 27,084 (88) | 85.6% | 83.2% | 81.6% |
| Disability | 3756 (12) | 14.4% | 16.8% | 18.1% |
| ESRD | 49 (< 1) | <1% | <1% | <1% |
ADRD Alzheimer’s disease and related dementias, COPD chronic obstructive pulmonary disease, ESRD end-stage renal disease, ISHD ischemic heart disease, OREC original reason for Medicare entitlement, SD standard deviation, SNRI serotonin-norepinephrine reuptake inhibitor, SSRI selective serotonin reuptake inhibitor, TCA tricyclic antidepressant
Values are n (%) unless otherwise stated
The average age of the sample was 79.7 [standard deviation (SD) 7.7] years (Table 1). The sample was primarily white (86%) and female (68%). Beneficiaries had a high burden of comorbidity, of note, Alzheimer’s disease (33%), anemia (68%), heart failure (46%), ischemic heart disease (67%) and stroke (38%). Twenty-five percent used SSRIs, 5% used TCAs, and 5% used SNRIs in the 3 months preceding TBI. The prevalence of individual antidepressants is listed in Appendix 2 (see the electronic supplementary material).
Almost all pairwise comparisons of baseline characteristics conducted between the different antidepressant classes were significant at the p < 0.001 level. Therefore, to conserve space, we did not present these comparisons nor the corresponding p values. Overall, comorbid burden was higher in the PM attributed to study antidepressants compared to the overall population. Among antidepressant users, PM attributed to TCAs had the youngest average age (77.8 years, SD 7.3) and the lowest prevalence of Alzheimer’s disease (37.9%).
Including only first event in each adverse event category, there were 1841 seizures, 3457 gastrointestinal bleeds, 1706 hemorrhagic strokes, 8296 ischemic strokes, 2535 cases of hyponatremia, and 5186 fractures in the sample over the 2 years following TBI (Table 2). Table 2 also presents observed numbers of adverse events when specific antidepressants were used divided by total person time contributed to the antidepressant class, without regard for whether it was contributed before or after the event of interest. These proportions should not be interpreted as risk. Rates of hemorrhagic stroke and fractures were elevated during the first 3 months following TBI and gradually decreased before leveling out by 12 months post-injury (Fig. 1). Rates of ischemic stroke were also elevated post-TBI, but leveled out by 3 months post-injury. Rates of other adverse events remained level over the duration of follow-up. Rates of any adverse event were higher among beneficiaries who used antidepressants (Fig. 2).
Table 2.
First adverse events during the 2 years following hospitalization for traumatic brain injury (TBI), first 2 months following TBI excluded, as a proportion of person-months (PM) of antidepressant use
| Adverse Event | Total | No antidepressant, PM = 438,781 (65%) | SSRI, PM = 163,212(24%) | SNRI, PM = 24,756 (5%) | TCA, PM = 19,734(3%) |
|---|---|---|---|---|---|
|
| |||||
| Seizure, n (%) | 1841 | 1105 (0.25) | 530 (0.32) | 116 (0.47) | 53 (0.27) |
| Gastrointestinal bleeding, n (%) | 3457 | 2203 (0.50) | 826 (0.51) | 222 (0.90) | 116 (0.59) |
| Hemorrhagic stroke, n (%) | 1706 | 1065 (0.24) | 431 (0.26) | 96 (0.39) | 41 (0.21) |
| Ischemic stroke, n (%) | 8296 | 5243 (1.19) | 2142 (1.31) | 466 (1.88) | 243 (1.23) |
| Hyponatremia, n (%) | 2535 | 1595 (0.36) | 632 (0.39) | 143 (0.58) | 89 (0.45) |
| Fracture, n (%) | 5186 | 2934 (0.67) | 1527 (0.94) | 392 (1.58) | 181 (0.92) |
SNRI serotonin-norepinephrine reuptake inhibitor, SSRI selective serotonin reuptake inhibitor, TCA tricyclic antidepressant
Fig. 1.
Monthly rates of adverse events/100 Medicare beneficiaries during the 2 years following hospitalization for traumatic brain injury, regardless of antidepressant use, n = 30,886
Fig. 2.
Rates of adverse events stratified by antidepressant use among Medicare beneficiaries during the 2 years following hospitalization for traumatic brain injury, n = 30,886
Unadjusted and adjusted associations between a specified antidepressant class as compared to another and each of the six adverse events are presented in Table 3. Appendix 3, in the electronic supplementary material, contains adjustment variables for each model. In unadjusted analyses, there were few significant differences between antidepressant classes with regard to risk of adverse events. Following adjustment for potential confounding variables, one association remained significant. Compared to TCAs, SSRI use was associated with increased risk of hemorrhagic stroke (RR 2.47; 95% CI 1.30–4.70). This association remained significant when accounting for multiple comparisons.
Table 3.
Unadjusted and adjusted risk ratios and 95% confidence intervals of antidepressant comparisons and risk of adverse events among Medicare beneficiaries with traumatic brain injury, n = 30,886
| Adverse event | SSRI vs. SNRI | SSRI vs. TCA | SNRI vs. TCA |
|---|---|---|---|
|
| |||
| Seizures | |||
| Reference drug | SNRI | TCA | TCA |
| Unadjusted | 0.90 (0.71–1.13) | 1.49 (1.00–2.21) | 1.16 (0.83–1.62) |
| Adjusted | 0.87 (0.58–1.30) | 1.12 (0.65–1.93) | 0.91 (0.52–1.62) |
| Gastrointestinal bleeds | |||
| Reference drug | SNRI | TCA | TCA |
| Unadjusted | 0.78 (0.65–0.93) | 0.95 (0.73–1.23) | 1.10 (0.86–1.40) |
| Adjusted | 0.88 (0.64–1.21) | 1.38 (0.93–2.06) | 0.74 (0.48–1.14) |
| Hemorrhagic stroke | |||
| Reference drug | SNRI | TCA | TCA |
| Unadjusted | 0.95 (0.73–1.24) | 1.69 (1.05–2.70)a | 1.60 (1.05–2.46) |
| Adjusted | 1.17 (0.71–1.92) | 2.47 (1.30–4.70)b | 1.39 (0.59–3.25) |
| Ischemic stroke | |||
| Reference drug | SNRI | TCA | TCA |
| Unadjusted | 1.00 (0.88–1.12) | 1.21 (1.02–1.44) | 1.17 (0.99–1.38) |
| Adjusted | 0.92 (0.73–1.15) | 0.99 (0.76–1.31) | 0.98 (0.73–1.33) |
| Hyponatremia | |||
| Reference drug | SNRI | TCA | TCA |
| Unadjusted | 0.99 (0.79–1.25) | 0.93 (0.69–1.25) | 0.82 (0.62–1.10) |
| Adjusted | 1.16 (0.77–1.75) | 0.73 (0.46–1.16) | 0.72 (0.44–1.19) |
| Fracture | |||
| Reference drug | SNRI | TCA | TCA |
| Unadjusted | 0.79 (0.70–0.91) | 1.09 (0.89–1.33) | 1.18 (0.98–1.43) |
| Adjusted | 0.96 (0.75–1.24) | 1.14 (0.85–1.53) | 1.01 (0.71–1.43) |
SNRI serotonin-norepinephrine reuptake inhibitor, SSRI selective serotonin reuptake inhibitor, TCA tricyclic antidepressant
See Appendix 2 in the electronic supplementary material for adjustment variables for each model
p = 0.03
p = 0.004
The unadjusted association between lagged use of any antidepressant and time to first adverse event of any kind suggested an increased risk (RR 1.19; 95% CI 1.14–1.24). However, after adjustment for potential confounding variables, the observed effect was attenuated (RR 0.95; 95% CI 0.90–1.00). Excluding the 3rd month post-TBI hospitalization did not change effect estimates significantly.
Among individuals who switched from one antidepressant class to another, we observed an increased risk of seizures among individuals using SSRIs compared to SNRIs (RR 2.30; 95% CI 1.35–3.92) that remained significant when multiple comparisons were accounted for. No other comparisons resulted in increased risk of adverse events.
4. Discussion
In this national study of Medicare beneficiaries hospitalized with TBI, use of SNRIs and TCAs was not associated with an increased risk of several adverse events when compared to SSRIs.
TCAs are not recommended for use in older adults, because of their anticholinergic and sedating effects [21]. Furthermore, TCAs have been associated with reduced seizure threshold among individuals with TBI [33, 36–39]. Nonetheless, we did not observe an increased risk of seizure associated with TCA or any antidepressant use. It is possible that selection bias resulted in a population of TCA users that was at lower risk for seizure. Supporting this hypothesis, there is evidence that rates of seizure increase with older age, yet TCA users were younger on average than users of other antidepressants [15].
There was a significantly increased risk of hemorrhagic stroke associated with use of SSRIs compared with TCAs. This association was also observed for SNRIs, but was not statistically significant, possibly because of fewer SNRI users. SSRIs decrease platelet aggregation and consequently increase risk of bleeding [36]. Prior studies also reported an increased risk of hemorrhagic stroke among users of SSRIs, but this study is the first to make comparisons between different classes of antidepressants among older individuals with TBI [15, 16].
Prior studies have suggested that SSRIs and SNRIs were associated with increased risk of hyponatremia, but although effect estimates were protective in this study, they were not statistically significant [15, 22, 23]. In a head-to head comparison, Coupland et al. reported that use of SSRIs and other antidepressants was associated with increased risk of hyponatremia compared to TCAs among a group of older adults diagnosed with depression [15]. Ascertainment bias may have been a factor in previous studies. This would occur when ascertainment of sodium levels is associated with use of a specific antidepressant. Several studies reported a risk of hyponatremia associated with SSRI use; hence, this may have led providers to test serum sodium levels of individuals on SSRIs more frequently [15, 22, 23].
This study has limitations that should be considered. Our study population comprised Medicare beneficiaries hospitalized with TBI. Compared to beneficiaries who were not hospitalized with TBI, these individuals likely had more severe injury or a higher level of complexity due to comorbid illness burden. Consequently, results from this study may not generalize to all older adults with TBI. Nonetheless, this population is more likely to be at higher risk of adverse events and less likely to receive indicated treatment for depression or anxiety following TBI [25]. An important limitation of administrative claims data is the lack of TBI severity information. Because this was a hospitalized population, milder cases may be underrepresented. Administrative claims do not provide information on indication for medications, although we were able to control for diagnoses of depression. Further, wide variation in dosing would have made class-to-class comparisons impossible; therefore, we compared use and not dosage. Finally, we can only observe filled prescriptions, but have no information on medication adherence.
In an observational study, selection bias can be introduced because participants are not randomized to treatment, and receipt of a particular treatment may be associated with the outcome [40]. This bias can be reduced using inverse probability of treatment weights (IPTWs) [41]. We were concerned about the potential for selection bias in this study, but we chose not to calculate IPTWs because of the complexity of the analysis. Rather, we used covariate adjustment to reduce bias due to non-random assignment of treatment. In general, estimates that were statistically significant in the unadjusted models were attenuated with adjustment, suggesting that we were able to minimize the impact of selection bias using covariate adjustment and also that our approach resulted in more conservative estimates. Administrative claims data cannot capture all side effects that are important to patients, such as sexual dysfunction or changes in appetite. Instead, we focused on potentially life-threatening adverse events that were available in administrative claims data. We did not have information on arrhythmias, a potentially important side effect of TCAs in patients with existing heart disease [42, 43]. Finally, we did not have information on use of other medications that could be associated with the adverse events we examined. This could be a source of residual confounding if use of these medications differs between people on the different antidepressants.
Despite these limitations, this study has many strengths. This large national study is the first to conduct comparisons between the three most commonly used antidepressant classes among older adults following TBI to assess risk of serious adverse events. This is important because prior studies have suggested that older adults diagnosed with depression after TBI are less likely to receive indicated treatment, in part because providers have concerns about adverse events [25–27]. Until now, this information was lacking, leaving providers to rely on evidence from idiopathic illness [7–9]. Our study population had a significant burden of comorbid illness, providing ‘real world’ evidence on adverse events associated with antidepressant use following TBI. Healthcare providers may use the results from this study to guide treatment selection based on the safety of antidepressant medications used following TBI among older adults, regardless of whether the patient has previously used antidepressants. Despite the lack of information on cardiac side effects, this study provides clinicians with guidance to appropriately evaluate and monitor elderly patients with TBI prior to and after starting an antidepressant. For example, caution should be exercised when starting SSRIs if a person is on blood thinners. Hyponatremia should be ruled out prior to starting, and sodium levels should be monitored on a regular basis after initiation of SSRIs. Careful consideration should be given to use of SSRIs in persons with prior history of strokes, aneurysms and or seizures. It is well known that SSRIs have a distinct profile of cytochrome P450 inhibition [44]. Clinicians should carefully choose SSRIs that have minimal drug–drug interactions and avoid or reduce the dose when the person is on medications that have an inhibitory effect on metabolism.
5. Conclusion
Compared with SSRIs, use of SNRIs and TCAs following hospitalization for TBI among Medicare beneficiaries was not associated with an increased risk of any of the studied adverse events. SSRI use was associated with an increased risk of hemorrhagic stroke when compared with use of TCAs, but not SNRIs.
Supplementary Material
Electronic supplementary material The online version of this article (https://doi.org/10.1007/s40266-018-0570-2) contains supplementary material, which is available to authorized users.
Key Points.
Among older adults hospitalized for traumatic brain injury (TBI), ischemic stroke and fracture were the most common of the six adverse events examined.
Selective serotonin reuptake inhibitor users had a higher risk of hemorrhagic stroke compared with tricyclic antidepressant users, but no other class comparison was associated with increased risk of adverse events.
Healthcare providers may use the results from this study to guide antidepressant treatment selection among older adults following TBI.
Acknowledgements
The authors would like to thank Susan dosReis, PhD, University of Maryland School of Pharmacy, for her thorough review of this paper. Dr. dosReis has no conflicts of interest.
Funding
Dr. Albrecht was supported by Agency for Healthcare Quality and Research grant 1K01HS024560. Dr. Rao was supported by DOD grant W81XWH-13–1-0469.
Footnotes
Conflict of interest Dr. Mullins has received grants from Bayer, Novartis, and Pfizer and consulting income from Bayer, Janssen/J&J, Novo Nordisk, Pfizer, Regeneron, and Sanofi. Dr. Perfetto is employed by the National Health Council in Washington, DC, which accepts membership dues and sponsorships from a variety of organizations and companies. For the full list of members and sponsors, please see NHCouncil.org. Drs. Albrecht and Rao declare no conflicts of interest.
References
- 1.Taylor CA, Bell JM, Breiding MJ, Xu L. Traumatic brain injury-related emergency department visits, hospitalizations, and deaths—United States, 2007 and 2013. Morbidity and mortality weekly report surveillance summaries (Washington, DC: 2002). 2017;66(9):1–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Rapoport MJ, Kiss A, Feinstein A. The impact of major depression on outcome following mild-to-moderate traumatic brain injury in older adults. J Affect Disord. 2006;92(2–3):273–6. [DOI] [PubMed] [Google Scholar]
- 3.Albrecht JS, Kiptanui Z, Tsang Y, Khokhar B, Liu X, Simoni-Wastila L, et al. Depression among older adults after traumatic brain injury: a national analysis. Am J Geriatr Psychiatry. 2015;23(6):607–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Jorge RE, Robinson RG, Starkstein SE, Arndt SV. Influence of major depression on 1-year outcome in patients with traumatic brain injury. J Neurosurg. 1994;81(5):726–33. [DOI] [PubMed] [Google Scholar]
- 5.Rapoport MJ, Herrmann N, Shammi P, Kiss A, Phillips A, Feinstein A. Outcome after traumatic brain injury sustained in older adulthood: a one-year longitudinal study. Am J Geriatr Psychiatry. 2006;14(5):456–65. [DOI] [PubMed] [Google Scholar]
- 6.Utomo WK, Gabbe BJ, Simpson PM, Cameron PA. Predictors of in-hospital mortality and 6-month functional outcomes in older adults after moderate to severe traumatic brain injury. Injury. 2009;40(9):973–7. [DOI] [PubMed] [Google Scholar]
- 7.Fann JR, Hart T, Schomer KG. Treatment for depression after traumatic brain injury: a systematic review. J Neurotrauma. 2009;26(12):2383–402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Warden DL, Gordon B, McAllister TW, Silver JM, Barth JT, Bruns J, et al. Guidelines for the pharmacologic treatment of neurobehavioral sequelae of traumatic brain injury. J Neurotrauma. 2006;23(10):1468–501. [DOI] [PubMed] [Google Scholar]
- 9.Rao V, Koliatsos V, Ahmed F, Lyketsos C, Kortte K. Neuropsychiatric disturbances associated with traumatic brain injury: a practical approach to evaluation and management. Semin Neurol. 2015;35(1):64–82. [DOI] [PubMed] [Google Scholar]
- 10.Lee HB, Lyketsos CG, Rao V. Pharmacological management of the psychiatric aspects of traumatic brain injury. Int R Psychiatry (Abingdon, England). 2003;15(4):359–70. [DOI] [PubMed] [Google Scholar]
- 11.Albrecht JS, Liu X, Smith GS, Baumgarten M, Rattinger GB, Gambert SR, et al. Stroke incidence following traumatic brain injury in older adults. J Head Trauma Rehabil. 2015;30(2):62–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Englander J, Cifu DX, Diaz-Arrastia R. Information/education page. Seizures and traumatic brain injury. Arch Phys Med Rehabil. 2014;95(6):1223–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Yeh CC, Chen TL, Hu CJ, Chiu WT, Liao CC. Risk of epilepsy after traumatic brain injury: a retrospective population-based cohort study. J Neurol Neurosurg Psychiatry. 2013;84(4):441–5. [DOI] [PubMed] [Google Scholar]
- 14.Bloechliger M, Ceschi A, Ruegg S, Kupferschmidt H, Kraehenbuehl S, Jick SS, et al. Risk of seizures associated with antidepressant use in patients with depressive disorder: follow-up study with a nested case-control analysis using the Clinical Practice Research Datalink. Drug Saf. 2016;39(4):307–21. [DOI] [PubMed] [Google Scholar]
- 15.Coupland C, Dhiman P, Morriss R, Arthur A, Barton G, Hippisley-Cox J. Antidepressant use and risk of adverse outcomes in older people: population based cohort study. BMJ (Clinical research ed). 2011;2(343):d4551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hackam DG, Mrkobrada M. Selective serotonin reuptake inhibitors and brain hemorrhage: a meta-analysis. Neurology. 2012;79(18):1862–5. [DOI] [PubMed] [Google Scholar]
- 17.Hill T, Coupland C, Morriss R, Arthur A, Moore M, Hippisley-Cox J. Antidepressant use and risk of epilepsy and seizures in people aged 20 to 64 years: cohort study using a primary care database. BMC Psychiatry. 2015;17(15):315. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Renoux C, Vahey S, Dell’Aniello S, Boivin JF. Association of Selective serotonin reuptake inhibitors with the risk for spontaneous intracranial hemorrhage. JAMA Neurol. 2017;74(2):173–80. [DOI] [PubMed] [Google Scholar]
- 19.Alldredge BK. Seizure risk associated with psychotropic drugs: clinical and pharmacokinetic considerations. Neurology. 1999;53(5 Suppl 2):S68–75. [PubMed] [Google Scholar]
- 20.Pisani F, Oteri G, Costa C, Di Raimondo G, Di Perri R. Effects of psychotropic drugs on seizure threshold. Drug Saf. 2002;25(2):91–110. [DOI] [PubMed] [Google Scholar]
- 21.The American Geriatrics Society 2015 Beers Criteria Update Expert Panel. American Geriatrics Society 2015 updated Beers criteria for potentially inappropriate medication use in older adults. J Am Geriatr Soc. 2015;63(11):2227–46. [DOI] [PubMed] [Google Scholar]
- 22.Jacob S, Spinler SA. Hyponatremia associated with selective serotonin-reuptake inhibitors in older adults. Ann Pharmacother. 2006;40(9):1618–22. [DOI] [PubMed] [Google Scholar]
- 23.Movig KL, Leufkens HG, Lenderink AW, van den Akker VG, Hodiamont PP, Goldschmidt HM, et al. Association between antidepressant drug use and hyponatraemia: a case-control study. Br J Clin Pharmacol. 2002;53(4):363–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Wroblewski BA, McColgan K, Smith K, Whyte J, Singer WD. The incidence of seizures during tricyclic antidepressant drug treatment in a brain-injured population. J Clin Psychopharmacol. 1990;10(2):124–8. [DOI] [PubMed] [Google Scholar]
- 25.Albrecht JS, Kiptanui Z, Tsang Y, Khokhar B, Smith GS, Zuckerman IH, et al. Patterns of depression treatment in Medicare beneficiaries with depression after traumatic brain injury. J Neurotrauma. 2015;32(16):1223–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Albrecht JS, Mullins DC, Smith GS, Rao V. Psychotropic medication use among Medicare beneficiaries following traumatic brain injury. Am J Geriatr Psychiatry. 2017;25(4):415–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Albrecht JS, O’Hara LM, Moser KA, Mullins CD, Rao V. Perception of barriers to the diagnosis and receipt of treatment for neuropsychiatric disturbances after traumatic brain injury. Arch Phys Med Rehabil. 2017;98(12):2548–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Marr AL, Coronado VG, editors. Central nervous system injury surveillance data submission standards—2002. Atlanta: US Department of Health and Human Services, CDC; 2004. [Google Scholar]
- 29.Thurman DJ, Sniezek JE, Johnson D, Greenspan A, Smith SM. Guidelines for surveillance of central nervous system injury. Atlanta: US Department of Health and Human Services, Public Health Service, CDC, National Center for Injury Prevention and Control; 1995. [Google Scholar]
- 30.Carroll CP, Cochran JA, Guse CE, Wang MC. Are we underestimating the burden of traumatic brain injury? Surveillance of severe traumatic brain injury using centers for disease control International classification of disease, ninth revision, clinical modification, traumatic brain injury codes. Neurosurgery. 2012;71(6):1064–70 (discussion 70). [DOI] [PubMed] [Google Scholar]
- 31.Rates of hospitalization related to traumatic brain injury—nine states, 2003. MMWR Morb Mortal Weekly Rep. 2007;56(8):167–70. [PubMed] [Google Scholar]
- 32.Richards JB, Papaioannou A, Adachi JD, Joseph L, Whitson HE, Prior JC, et al. Effect of selective serotonin reuptake inhibitors on the risk of fracture. Arch Intern Med. 2007;167(2):188–94. [DOI] [PubMed] [Google Scholar]
- 33.Dinan TG, Mobayed M. Treatment resistance of depression after head injury: a preliminary study of amitriptyline response. Acta Psychiatr Scand. 1992;85(4):292–4. [DOI] [PubMed] [Google Scholar]
- 34.Pd A Survival analysis using S AS®: a practical guide. Cary: SAS Institute; 1995. [Google Scholar]
- 35.Holm S A simple sequentially rejective multiple test procedure. Scand J Statist. 1979;6(2):65–70. [Google Scholar]
- 36.de Abajo FJ. Effects of selective serotonin reuptake inhibitors on platelet function: mechanisms, clinical outcomes and implications for use in elderly patients. Drugs Aging. 2011;28(5):345–67. [DOI] [PubMed] [Google Scholar]
- 37.Fann JR, Uomoto JM, Katon WJ. Sertraline in the treatment of major depression following mild traumatic brain injury. J Neuropsychiatry Clin Neurosci. 2000;12(2):226–32. [DOI] [PubMed] [Google Scholar]
- 38.Perino C, Rago R, Cicolini A, Torta R, Monaco F. Mood and behavioural disorders following traumatic brain injury: clinical evaluation and pharmacological management. Brain Inj. 2001;15(2):139–48. [DOI] [PubMed] [Google Scholar]
- 39.Turner-Stokes L, Hassan N, Pierce K, Clegg F. Managing depression in brain injury rehabilitation: the use of an integrated care pathway and preliminary report of response to sertraline. Clin Rehabil. 2002;16(3):261–8. [DOI] [PubMed] [Google Scholar]
- 40.Seeger JD, Williams PL, Walker AM. An application of propensity score matching using claims data. Pharmacoepidemiol Drug Saf. 2005;14(7):465–76. [DOI] [PubMed] [Google Scholar]
- 41.Cole SR, Hernan MA. Constructing inverse probability weights for marginal structural models. Am J Epidemiol. 2008;168(6):656–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Glassman AH. Cardiovascular effects of antidepressant drugs: updated. J Clin Psychiatry. 1998;59(Suppl 15):13–8. [PubMed] [Google Scholar]
- 43.Glassman AH, Roose SP, Bigger JT Jr. The safety of tricyclic antidepressants in cardiac patients. Risk-benefit reconsidered. JAMA. 1993;269(20):2673–5. [PubMed] [Google Scholar]
- 44.Hemeryck A, Belpaire FM. Selective serotonin reuptake inhibitors and cytochrome P-450 mediated drug-drug interactions: an update. Curr Drug Metab. 2002;3(1):13–37. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.