Factors associated with antidepressant use following hospitalization for acute stroke

Abstract

Objective:

Depression is the most common mental health complication after stroke, leading to poor post-stroke outcomes. Few observational studies have reported factors associated with antidepressant use in these patients. Therefore, this study assessed patient and clinical characteristics associated with receiving antidepressant prescriptions at discharge following hospital admission for acute stroke.

Methods:

Identified were patients admitted with acute stroke or transient ischemic attack from the institutional stroke registry (October 1, 2019, through October 31, 2020). Electronic health record review was used to assess antidepressant prescription at discharge and predictors such as demographics, health behaviors, clinical comorbidities, and stroke prognostic characteristics.

Results:

Out of 803 patients with stroke, 220 (27.4%) received an antidepressant prescription at discharge, including 82 (12.8%) patients who were started on antidepressants at discharge. A multivariable model indicated that sex (female vs male OR = 1.53; 95% CI = 1.09, 2.16), race (Black vs White: OR = 0.67; 95% CI = 0.47, 0.96), number of clinical comorbidities (OR = 1.16; 95% CI = 1.03, 1.32), history of depression (OR = 4.91; 95% CI = 2.59, 9.31), and discharge disposition (to inpatient rehabilitation facility vs home: OR = 1.52; 95% CI = 1.01, 2.30) were significantly associated with receipt of antidepressant prescriptions among all users. Discharge disposition (to an inpatient rehabilitation facility) was the only variable associated with initiating an antidepressant prescription at discharge (OR = 2.92; 95% CI = 1.62, 5.28).

Conclusion:

Antidepressant prescriptions for poststroke patients were more frequent among females, White adults, and those discharged to an inpatient rehabilitation facility. These findings describing antidepressant prescription patterns underscore the need for specific guidelines aiding antidepressant use after stroke.

Introduction

Stroke remains among the leading causes of morbidity and mortality in the United States (US).¹ Having a stroke increases the risk of developing neuropsychiatric conditions.^2–6 The prevalence of depression ranges from 28%−34% in the acute post-stroke phase,^2,3 and is associated with an array of adverse post-stroke outcomes.^7–10 Consequently, it is essential to manage these complications promptly to ensure an optimum post-stroke quality of life and potentially prevent adverse outcomes.

Antidepressants are an effective means of managing and preventing depressive symptoms in patients with stroke.^11–13 Following stroke, 13% of patients meet the diagnostic threshold requiring treatment for depression,¹⁴ with the prevalence of antidepressant use in patients for post-stroke depression ranging from 32% - 75% within the first year of stroke recovery.^15,16 However, little is known about the prevalence of antidepressant use in the acute post-stroke phase and outside the scope of co-occurring post-stroke depression. Similarly, the few observational studies reporting the prevalence of antidepressant use in the post-stroke setting often have not reported factors associated with antidepressant use in these patients. Those that have, focused on the sub-acute to chronic phases of stroke recovery.^16,17

The Scientific Statement on post-stroke depression from the American Heart Association suggests that the evidence on the efficacy of antidepressant use is considered low-level, hence there are currently no set guidelines for depressive symptom prevention and treatment in patients after stroke,¹⁸ deciding to prescribe antidepressants following stroke reliant on individual clinical decision making. Identifying patient characteristics associated with receiving an antidepressant prescription in the acute post-stroke setting, particularly upon stroke discharge, could highlight current prescription behaviors in this population, aid in the development of clinical guidelines regarding the management of depression, and identify gaps in equitable access to post-stroke care. Thus, our study aims to assess the prevalence of antidepressant prescriptions and identify patient characteristics associated with a higher prevalence of antidepressant prescriptions in the acute post-stroke phase.

Methods

Study design and setting

Using the stroke registry data from a tertiary health care center in the Southeastern US, this cross-sectional study evaluated the prevalence of and characteristics associated with antidepressant use at stroke discharge. Data included patients admitted with an acute stroke or transient ischemic attack (TIA) from October 1, 2019 – October 31, 2020. Per registry protocols,^19,20 the presence of acute stroke was confirmed on neuroimaging with CT or MRI by neurologists. Only the first event was included in the analysis for patients with more than 1 stroke event during the study period. We excluded patients <18 years of age, patients with symptoms representing as stroke-mimics, and those who left hospital care against medical advice or died before discharge. Lastly, we also excluded patients with missing data for any variables of interest. The study protocol was reviewed and approved by the institutional review boards. The need for informed consent was waived due to the retrospective nature of the study.

Outcome - antidepressant prescription

The primary outcome of interest was the receipt of antidepressant prescriptions at discharge from acute stroke or TIA admission. Antidepressants of interest included selective serotonin reuptake inhibitors (SSRIs; citalopram, escitalopram, fluoxetine, sertraline, paroxetine, vortioxetine), serotonin and norepinephrine reuptake inhibitors (SNRIs; duloxetine, venlafaxine), tricyclic or tetracyclic antidepressants (TCA; amitriptyline, doxepin, nortriptyline, mirtazapine), and other atypical (bupropion, trazodone) antidepressants. Data on antidepressant prescriptions were collected from the medication list in the electronic health records (EHR) with the assistance of the institutional administrative data warehouse. For quality assurance purposes, all EHR data were manually reviewed. Data on antidepressant use upon admission was also collected from the EHR to differentiate patients initiating antidepressant use at discharge as a secondary outcome from patients with prevalent antidepressant users. Additionally, we also assessed SSRI prescription prevalence and SSRI initiation at discharge as secondary outcomes.

Factors potentially associated with antidepressant use

Based on clinical significance and evidence from prior studies, we were interested in the relationship between demographic variables (age, sex, and race), health behaviors (tobacco and alcohol use), clinical comorbidities (hypertension, diabetes, dyslipidemia, atrial fibrillation, coronary artery disease, congestive heart failure, prior stroke/TIA, depression, anxiety, other mood disorders), and stroke prognosis variables. We utilized the institutional data warehouse to obtain the data from EHR and manually reviewed all EHR data for quality assurance. We created a continuous variable representing the number of clinical comorbidities in each patient to assess the impact of multimorbidity on receiving antidepressant prescriptions. Prognostic stroke-specific variables included stroke severity measured using the National Institutes of Health Stroke Scale (NIHSS) and discharge disposition (home, inpatient rehabilitation facility, other). We assessed demographics, health behaviors, and clinical comorbidities upon admission, while we used data throughout the hospital stay to identify other clinical and prognostic variables.

Statistical analysis

We described patient characteristics overall and by antidepressant prescription at discharge. We used chi-square and t-tests for categorical and continuous variables, respectively, to evaluate differences in characteristics by antidepressant prescription at discharge. We used logistic regression to identify the individual variables associated with receipt of antidepressant prescriptions at discharge. We used multivariable logistic regression that included variables of interest into 1 model, to identify the independent variables associated with receipt of antidepressants at discharge.

We performed a secondary analysis excluding patients with antidepressant prescriptions at stroke admission to assess if factors associated with receipt of antidepressants at discharge differed between prevalent and new users. Given the increasing interest in the role of SSRIs in stroke recovery and rehabilitation, we limited to only prevalent and new SSRI prescriptions at discharge for an additional secondary analysis. Finally, given that TIA is often referred to as a “warning stroke” rather than a type of stroke,²¹ we performed a sensitivity analysis by excluding patients with TIA from our regression models to assess whether the results differed for patients with stroke events. We used SAS v.9.4 (Cary, NC, USA) for all analyses.

Results

During the study period, we identified 1183 stroke admissions from 1156 unique patients (Figure 1). After exclusions, our final analytic sample included 803 patients with acute stroke or TIA, out of which 220 (27.4%) patients received a prescription for at least 1 antidepressant medication, including 31 (14.1%) who received prescriptions for more than 1 antidepressant at discharge. The primary antidepressant prescribed was an SSRI (60%), followed by other (23.2%), SNRI (18.6%), and TCA (12.7%).

Figure 1.

Patient selection flowchart.

The average age in our cohort was 62.7 ± 14.8 years, with higher proportions of participants being male (51.7%), White (61.8%), and presenting with ischemic strokes (75.8%) (Table 1). Patients receiving antidepressants at discharge were more likely to be female (56.8% vs 45.1%), have a higher average number of clinical comorbidities (3.1 ± 1.8 vs 2.5 ± 1.5), have a history of depression (17.7% vs 3.1%) and anxiety (13.2% vs 2.9%), have a higher median NIHSS score (5.0 vs 4.0), have a longer length of stay (4.0 vs 3.0), and be discharged to a rehabilitation facility (30.9% vs 23.5%) (Table 1).

Table 1.

Patient characteristics by antidepressant prescription status upon discharge from acute stroke hospitalization.

		Antidepressant prescription on discharge
	Overall N = 803	Yes N = 220	No N = 583	P-value
Age, mean ± SD	62.7 ± 14.8	61.9 ± 14.6	63.0 ± 14.9	0.3546
Sex, N (%)				0.0031
Male	415 (51.7)	95 (43.2)	320 (54.9)
Female	388 (48.3)	125 (56.8)	263 (45.1)
Race, N (%)				0.1867
White	496 (61.8)	144 (65.5)	352 (60.4)
Black	307 (38.2)	76 (34.6)	231 (39.6)
Tobacco use, N (%)				0.2676
Current	218 (27.2)	66 (30.0)	152 (26.1)
Former	216 (26.9)	63 (28.6)	153 (26.2)
Never	369 (45.7)	91 (41.4)	278 (47.7)
Alcohol use, N (%)				0.2189
Current	265 (33.0)	68 (30.9)	197 (33.8)
Former	95 (11.8)	33 (15.0)	62 (10.6)
Never	443 (55.2)	119 (54.1)	324 (55.6)
Clinical comorbidities, N (%)
Number of comorbidities, mean ± SD	2.7 ± 1.6	3.1 ± 1.8	2.5 ± 1.5	<0.0001
Hypertension	596 (74.2)	170 (77.3)	426 (73.1)	0.2247
Diabetes	232 (28.9)	74 (33.6)	158 (27.1)	0.0684
Dyslipidemia	307 (38.2)	93 (42.3)	214 (36.7)	0.1477
Atrial fibrillation	137 (17.1)	36 (16.4)	101 (17.3)	0.7469
Congestive heart failure	81 (10.1)	21 (9.6)	60 (10.3)	0.7542
History of stroke/TIA	225 (28.0)	71 (32.3)	154 (26.4)	0.0993
Coronary artery disease	120 (14.9)	39 (17.7)	81 (13.9)	0.1742
Mental health variables, N (%)
History of depression	57 (7.1)	39 (17.7)	18 (3.1)	<0.0001
Anxiety	46 (5.7)	29 (13.2)	17 (2.9)	<0.0001
Antidepressant use on admission	456 (56.8)	121 (55.0)	335 (57.5)	0.5300
Stroke admission
Admission NIHSS, median (IQR)	4.0 (1.0, 11.0)	5.0 (2.0, 12.5)	4.0 (1.0, 10.0)	0.0166
Stroke type, N (%)				0.9108
Ischemic	609 (75.8)	165 (75.0)	444 (76.2)
Hemorrhagic	110 (13.7)	32 (14.6)	78 (13.4)
TIA/Other	84 (10.5)	23 (10.5)	61 (10.5)
Length of stay, median (IQR)	2.0 (1.0, 4.0)	4.0 (2.0, 9.0)	3.0 (2.0, 7.0)	0.0193
Discharge disposition, N (%)				0.0371
Home	484 (60.3)	117 (53.2)	367 (63.0)
Rehabilitation facility	205 (25.5)	68 (30.9)	137 (23.5)
Other	114 (14.2)	35 (15.9)	79 (13.6)

IQR, interquartile range; NIHSS, National Institutes of Health stroke scale; SD, standard deviation; TIA, transient ischemic attack.

Crude models indicated significant associations for predictors like sex, number of clinical comorbidities, history of depression, and discharge disposition (Table 2). Compared to males, females had 60% higher odds (95% CI = 1.17, 2.19) of anti-depressant prescriptions at discharge. We also observed that with every unit increase in the number of clinical comorbidities, the odds of receiving antidepressant prescriptions increased by 24% (95% CI = 1.13, 1.36). A history of depression was associated with a 6.76-fold odds of antidepressant prescription at discharge (95% CI = 3.78, 12.12), and being discharged to a rehabilitation facility had 56% higher odds (95% CI = 1.09, 2.23) of antidepressant prescription at discharge. In the multivariable model, being female (OR = 1.53; 95% CI = 1.09, 2.16), number of clinical comorbidities (OR = 1.16; 95% CI = 1.03, 1.32), having a history of depression (OR = 4.91; 95% CI = 2.59, 9.31), and discharge to rehabilitation facility (OR = 1.52; 95% CI = 1.01, 2.30) remained significantly associated with antidepressant prescription at discharge (Table 2). Race was significantly associated with antidepressant prescription in the multivariable model, with Black patients having 33% lower odds of receiving an antidepressant prescription at discharge (95% CI = 0.47, 0.96) compared to White patients.

Table 2.

Factors associated with prevalence and initiation of antidepressant prescriptions at discharge.

	Prevalent antidepressant prescriptions		Initiation of antidepressant
	Crude	Multivariable	Multivariable
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Age	1.00 (0.99, 1.01)	0.99 (0.98, 1.01)	0.98 (0.97, 1.00)
Sex
Male	Referent	Referent	Referent
Female	1.60 (1.17, 2.19)	1.53 (1.09, 2.16)	0.91 (0.55, 1.51)
Race
White	Referent	Referent	Referent
Black	0.80 (0.58, 1.11)	0.67 (0.47, 0.96)	0.85 (0.51, 1.40)
Tobacco
Never	Referent	Referent	Referent
Current	1.33 (0.91, 1.93)	1.38 (0.92, 2.10)	1.21 (0.67, 2.19)
Former	1.26 (0.86, 1.83)	1.23 (0.81, 1.85)	1.25 (0.68, 2.30)
Alcohol
Never	Referent	Referent	Referent
Current	0.94 (0.67, 1.33)	0.99 (0.68, 1.46)	0.87 (0.50, 1.52)
Former	1.45 (0.90, 2.32)	1.56 (0.93, 2.63)	1.47 (0.71, 3.05)
Number of comorbidities	1.24 (1.13, 1.36)	1.16 (1.03, 1.32)	1.03 (0.85, 1.24)
Prior stroke/TIA	1.33 (0.95, 1.86)	0.89 (0.56, 1.35)	0.84 (0.44, 1.61)
History of depression	6.76 (3.78, 12.12)	4.91 (2.59, 9.31)	2.34 (0.79, 6.93)
Admission NIHSS	1.02 (1.00, 1.04)	1.02 (0.99, 1.04)	1.02 (0.98, 1.06)
Discharge disposition
Home	Referent	Referent	Referent
Rehabilitation facility	1.56 (1.09, 2.23)	1.52 (1.01, 2.30)	2.92 (1.62, 5.28)
Other	1.39 (0.89, 2.18)	1.39 (0.80, 2.42)	2.59 (1.17, 5.76)

CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; TIA, transient ischemic attack.

Multivariable model included age, sex, tobacco, alcohol, number of clinical comorbidities, prior stroke/TIA, history of depression, admission NIHSS, discharge disposition. Bold typeface indicates statistical significance.

To assess whether the factors associated with receipt of antidepressant use at discharge were similar in the new antidepressant users, we excluded patients with antidepressant use upon admission (N = 163; 20.3%). Among the 640 patients who were antidepressant-naive at admission, 82 (12.8%) initiated antidepressants at discharge, the majority of which were SSRI prescriptions (65.9%). The multivariable logistic regression model showed that discharge disposition was the only independent factor significantly associated with antidepressant initiation, with patients discharged to an inpatient rehabilitation facility having 2.9 times higher odds (95% CI = 1.62, 5.28) (Table 2).

We also limited to SSRI prescriptions as a secondary outcome with 132 (16.4%) patients out of 803 receiving an SSRI prescription at discharge. Race, history of depression, and discharge disposition were independently associated with SSRI use. (Table 3). We also evaluated new SSRI prescriptions in patients who were not on any antidepressants before stroke admission (N = 640) and observed that 54 (8.4%) patients received new SSRI prescriptions at stroke discharge. Similar to the primary analysis, patients discharged to an inpatient rehabilitation facility had 2.7-fold higher odds of receiving SSRI prescriptions (95% CI = 1.38, 5.40) compared to patients discharged home. However, the results from the multivariable logistic regression model for SSRI initiation differed from the results for the initiation of any antidepressant, revealing race and stroke severity were independently associated with SSRI initiation.

Table 3.

Factors associated with prevalence and initiation of SSRI prescriptions at discharge.

	Prevalent SSRI aOR (95%CI)	Initiation of SSRI aOR (95% CI)
Total N	132/803	54/640
Age	0.98 (0.97, 0.99)	0.97 (0.95, 0.99)
Female	1.28 (0.85, 1.94)	0.66 (0.36, 1.19)
Race
White	Referent	Referent
Black	0.49 (0.31, 0.76)	0.53 (0.28, 0.97)
Tobacco
Never	Referent	Referent
Current	1.08 (0.66, 1.77)	0.99 (0.51, 1.93)
Former	1.21 (0.74, 1.96)	0.84 (0.41, 1.74)
Alcohol
Never	Referent	Referent
Current	0.92 (0.58, 1.46)	0.85 (0.45, 1.62)
Former	1.26 (0.68, 2.36)	1.24 (0.53, 2.93)
Number of comorbidities	1.15 (0.99, 1.33)	1.07 (0.86, 1.33)
Prior stroke/TIA	0.71 (0.43, 1.18)	0.71 (0.33, 1.51)
History of depression	3.10 (1.57, 5.93)	2.75 (0.98, 7.71)
Admission NIHSS	1.04 (1.01, 1.07)	1.04 (1.01, 1.08)
Discharge disposition
Home	Referent	Referent
Rehabilitation facility	1.73 (1.06, 2.81)	2.73 (1.38, 5.40)
Other	1.89 (1.01, 3.56)	2.44 (0.98, 6.07)

aOR, adjusted odds ratio; CI, confidence interval; NIHSS, National Institutes of Health stroke scale; TIA, transient ischemic attack. Bold typeface indicates statistical significance.

Finally, we performed a sensitivity analysis by excluding patients with TIA (Table 4) and similar to the main analysis, we found sex (OR = 1.53; 95% CI = 1.06, 2.19), race (OR = 0.66; 95% CI = 0.45, 0.96), history of depression (OR = 4.44; 95% CI = 2.26, 8.72), and discharge disposition (OR = 1.64; 95% CI = 1.07, 2.50) to be independent factors significantly associated with antidepressant use (Table 4).

Table 4.

Identifying independent factors associated with prevalence and initiation of antidepressant use and initiation at discharge, after excluding patients with TIA (N = 719).

	Prevalent antidepressant use	Antidepressant initiation
	OR (95% CI)	OR (95% CI)
Age	0.99 (0.98, 1.01)	0.98 (0.96, 1.00)
Sex
Male	Referent	Referent
Female	1.53 (1.06, 2.19)	0.83 (0.49, 1.41)
Race
White	Referent	Referent
Black	0.66 (0.45, 0.96)	0.82 (0.49, 1.38)
Tobacco
Never	Referent	Referent
Current	1.33 (0.87, 2.05)	1.07 (0.58, 1.97)
Former	1.19 (0.77, 1.84)	1.08 (0.57, 2.03)
Alcohol
Never	Referent	Referent
Current	1.04 (0.70, 1.54)	0.86 (0.49, 1.52)
Former	1.54 (0.88, 2.68)	1.40 (0.66, 2.99)
Number of comorbidities	1.14 (0.99, 1.30)	1.07 (0.88, 1.30)
Prior stroke/TIA	0.92 (0.59, 1.44)	0.77 (0.38, 1.52)
History of depression	4.44 (2.26, 8.72)	2.19 (0.72, 6.63)
Admission NIHSS	1.01 (0.99, 1.04)	1.02 (0.98, 1.06)
Discharge disposition
Home	Referent	Referent
Rehabilitation facility	1.64 (1.07, 2.50)	2.86 (1.06, 5.46)
Other	1.47 (0.84, 2.58)	2.86 (1.57, 5.21)

CI, confidence interval; NIHSS, National Institutes of Health Stroke Scale; OR, odds ratio; TIA, transient ischemic attack. Bold typeface indicates statistical significance.

Discussion

In this cross-sectional study of patients hospitalized with stroke or TIA at a tertiary health care center in the Southeastern US, we found sex, race, number of clinical comorbidities, history of depression, and discharge disposition to be independently associated with receiving an antidepressant prescription at acute stroke discharge. Specifically, antidepressant use was higher in females, those with more comorbidities, those with a history of depression, and those discharged to rehabilitation facilities.

We observed a higher prevalence of antidepressant prescriptions in our cohort compared to the average US population (19%) 60 years and older,²² but comparable to prior studies assessing antidepressant use in patients following stroke (32% – 75%).^15,16 The higher odds of antidepressant prescriptions at discharge in females vs males and Black vs White patients in this study are comparable to the trends in antidepressant use by sex and race in US adults 60 years and older.²² Prior studies have reported similar trends of antidepressant use by sex and race among patients with post-stroke depression.^23,24 These observed trends are likely more representative of the prescription patterns in the larger population outside of the patients who experience strokes. These trends also highlight possible differences in receiving mental health screenings and subsequent treatment by sex and race. Female patients are more likely to receive a depression diagnosis than males in the overall population²⁵ as well as in patients after stroke.²⁶ Depression symptomatology also differs by sex with female patients more likely to report somatic symptoms and male patients more likely to present with psychomotor changes after stroke,²⁷ which may also contribute to differences in the prevalence of depression and antidepressant use after stroke. Similarly, patients of racial or ethnic minorities are less likely to seek mental health care and to be diagnosed with post-stroke depression compared to White patients.^24,28 Other potential mechanisms for these differences may include discriminatory care due to provider bias and structural barriers such as insurance which inhibit access to mental health care in minoritized groups,²⁹ which are not well-studied in the current stroke literature and could be avenues to explore in future research.

We found that a higher number of comorbidities was associated with higher odds of receiving antidepressant prescriptions. Studies have shown that multimorbidity is independently associated with a 2 to 3 times higher risk of depression,³⁰ which would be consistent with our findings. However, our findings contradict a previous study showing that in a cohort of community-dwelling patients with stroke from the Medical Expenditure Panel Survey, compared to patients with no chronic disease conditions, those with ≥3 conditions were 79% less likely (95% CI = 0.08, 0.52) to be prescribed with antidepressant medications with co-occurring depression.¹⁶ These discrepancies could be attributed to the differences in study setting (eg, cohort vs admissions from 1 hospital). Our study assessed the prevalence of antidepressant use in the acute setting, whereas the previous study was conducted among patients with stroke, a concurrent depression diagnosis, and had a follow-up of up to 1 year.¹⁶

Patients discharged to an inpatient rehabilitation facility had both higher odds of the prevalence and initiation of antidepressant prescriptions including SSRIs compared to those discharged home. A similar association was observed in the Get With The Guidelines-Stroke cohort wherein patients initiating antidepressant use were more likely to be discharged to an inpatient rehabilitation facility than home (34% vs 31%, P < 0.001).¹⁷ There has been an increasing interest in utilizing antidepressants for stroke recovery,^13,31 which coincides with the management goals for patients discharged to inpatient rehabilitation facilities. Particularly, SSRI use has been linked to improved post-stroke functional outcomes, with a recent meta-analysis of placebo-controlled RCTs indicating that SSRI-treated patients demonstrated significant improvement in motor and cognitive outcomes.¹³ Although RCTs in the past decade have produced conflicting evidence,^31–34 this possible link between SSRI use and stroke recovery may have contributed to the higher odds of initiating antidepressants among patients discharged to inpatient stroke rehabilitation facilities compared to those discharged at home. The positive relationship between higher admission NIHSS and the higher odds of receiving SSRI prescriptions has been observed in prior studies³⁵ and may also be explained by the perceived benefits of antidepressant use in improving stroke outcomes. These results indicate that the off-label use of antidepressants may have impacted prescription behaviors in this population and highlight the need for structured guidelines in the acute post-stroke care settings for mental health care.

Studies have previously evaluated factors associated with antidepressant use in later post-stroke periods, and have shown older age (65+ years),¹⁶ white collar occupation,³⁵ lower education levels (high school or less),¹⁶ current smoking,^16,35 poor perceived mental status,¹⁶ overweight or obesity,¹⁶ moderate stroke severity (NIHSS between 7–14),³⁵ and mild to severe limb paresis³⁵ to be independent predictors of initiating an antidepressant at 1-year post-stroke. Given our data source, we were unable to evaluate factors like occupation and education, and we did not observe independent associations with age, smoking, or history of depression.

Limitations

The results of this study should be interpreted cautiously. Being a single-center study, our findings have limited generalizability, nationally and globally. However, the Southeastern US, also known as the “Stroke Belt”, contributes significantly to the stroke burden and stroke-related mortality in the US with higher proportions of Black Americans, rural residents, and a higher prevalence of stroke risk factors compared to other regions.³⁶ These factors make it an important population for the assessment of stroke and post-stroke outcomes. We also chose a complete case analysis, thus excluding 74 patients with missing covariate data, potentially underestimating the outcome and the exposures. However, given the small proportion of excluded cases (<10%), we expect the impact of this bias to be relatively small. Additionally, our assessment of factors associated with the initiation of antidepressants at discharge was conducted in a small sample with a low event rate. This lack of power may have resulted in discrepancies in the results between the prevalence of antidepressant prescriptions and initiation of antidepressants at discharge. It is also important to acknowledge that the associations observed in our multivariable model for each independent variable are conditional upon the inclusion of other independent variables in the model.³⁷ Given that information on factors of interest came from EHR data, we were limited in the factors evaluated in the analyses and were unable to evaluate factors like socioeconomic variables, education level, some lifestyle factors, and provider prescribing behaviors that could be related to the receipt of antidepressants. Also, we relied on ICD codes to identify a history of depression in our cohort, which has limited sensitivity likely due to under-coding,³⁸ and can underestimate the prevalence. We also lacked a direct measure of depressive symptoms during the hospital stay, preventing us from assessing poststroke depression severity. Due to these reasons, prescriptions were likely reliant on physician discretion. Lastly, prognostic outcomes after stroke admission were not available for many of the patients receiving their follow-up care outside of our hospital system. Therefore, it is difficult to determine how antidepressant medication use may have impacted their health beyond their initial encounter. Despite these limitations, our study had many strengths, including the use of a cohort of patients with stroke events confirmed on neuroimaging, which is the gold standard for identifying strokes for research. Additionally, our use of detailed EHR review for an accurate assessment of antidepressant prescriptions and their clinical predictors demonstrates rigor toward ensuring the accuracy of the data used in our study.

Conclusion

This cross-sectional study assessed the prevalence and initiation of antidepressant prescriptions at discharge from acute stroke. We observed a higher prevalence of antidepressant prescriptions in our study compared to the average US population. Prescriptions were more frequent among women, White adults, and those discharged to an inpatient rehabilitation facility. The inconsistencies in access to care by sex and race may indicate a need to further examine socioeconomic factors, provider-level characteristics, clinical decision-making, and structural barriers that shape these patterns. Additionally, assessing the optimal timing and frequency of depression screening is also required to inform future guidelines that may promote optimal and equitable access to post-stroke mental health care.