Effective Tools to Predict Depression in Acute and Subacute Phase of Ischemic Stroke

Stroke was the third leading cause of mortality in Japan in 2017 (1). In 2019, those over age 65 accounted for 28.4% of the total population in Japan, which ranked first among 201 countries (2, 3).

In large cohorts exceeding 25,000 people, poststroke depression has been found to occur in approximately 30% of survivors during the first year following stroke (4). Depression following stroke has been shown in numerous studies to impair both cognitive and motor recovery as well as increase mortality for as long as 10 years following acute stroke (5–10).

Although several studies have found that antidepressant medications can prevent the development of depression (11, 12), many patients are reluctant to take antidepressants before they are depressed. Thus, methods to predict the risk of developing depression after acute stroke are needed. However, very few studies have tried to identify reliable methods for predicting the development of depression after the acute stroke period.

The present study was designed to examine frequency and risks of major depression after ischemic stroke. We tested the hypothesis that subsyndromal depressive symptoms during the acute and subacute stroke periods would predict the development of major depression during the chronic 1-year period following stroke. The aim of this study, therefore, was to assess the reliability of standardized instruments during the acute and subacute stroke periods to predict patients’ likelihood of developing major depression during the first year chronic stage following stroke.

Methods

Participants

Study subjects were enrolled and followed between July 2014 and March 2019 from inpatients at the Department of Neurology, Tokai University Hospital, with ischemic stroke aged 20–85 years within 6 weeks of onset. Patients were excluded if they had impairment of consciousness, history of dementia, another form of brain injury or aphasia with severe comprehension deficits, or cognitive impairment due to stroke that impaired their ability to understand informed consent or mental status questions. In addition, patients were excluded if they met DSM-IV criteria for alcohol or substance abuse or dependence within the 24 weeks prior to enrollment or if they met DSM-IV criteria for schizophrenia or other psychotic disorders. The patients were examined at initial evaluation (i.e., within 6 weeks of stroke onset) and 3, 6, 9, and 12 months after the ischemic stroke. Depression, cognition, and quality of life were assessed at each of these visits. To predict onset of major depression during the chronic stage of the first year after stroke, patients with major depression in the acute and subacute stroke periods (i.e., within 6 weeks of stroke onset) were excluded from the analysis.

The Institutional Review Board for Clinical Research of the Tokai University School of Medicine approved this study. Written consent was obtained from all enrollees.

Neurological Evaluations

At least two neurologists diagnosed an ischemic stroke based on a complete physical and neurological examination, and MRI brain scans. The severity of the stroke was assessed using the National Institutes of Health Stroke Scale (13).

Assessment for Depression

The patients were examined using the Structured Clinical Interview for DSM-IV (SCID) (14, 15) mood and anxiety disorder section. The diagnosis of major depression developing for the first time after stroke was based on symptoms elicited by the SCID and DSM-IV criteria.

The severity of depressive symptoms was objectively assessed using the Japanese version of the Montgomery-Åsberg Depression Rating Scale (MADRS); it comprises 10 items, each of which is scored on a scale that ranges from 0 to 6 (16, 17). The MADRS has been used widely in clinical double-blind studies to assess the efficacy of various antidepressant therapies in Western countries (18).

The Patient Health Questionnaire (PHQ-9) Japanese version consists of questions about the nine symptom criteria upon which the DSM-IV major depression diagnosis is based (19, 20). The PHQ-9 is a self-administered questionnaire that asks patients to rate their symptoms of depression from 0 to 3 (0=not at all; 3=nearly every day) over the past 2 weeks. Total scores from 1 to 4 are categorized as minimal, from 5 to 9 mild, from 10 to 14 moderate, from 15 to 19 moderately severe, and from 20 to 27 severe (19). The PHQ-9 is widely used throughout the world to screen for depressive disorder and monitor depressive symptoms. The original version of the PHQ-9 has been translated into multiple languages and validated in numerous studies globally (20–22).

Cognitive Assessment

Global cognitive impairment was assessed at all evaluations using the Japanese version of the Mini-Mental State Examination (23, 24). Scores may range from 0 to 30, with lower scores indicating greater impairment. In addition, the Frontal Assessment Battery, developed as a short bedside cognitive and behavioral battery to assess frontal lobe functions, was also administered (25, 26). The Frontal Assessment Battery consists of six items, and the score on each item ranges from 0 to 3. A higher score indicates better frontal lobe performance.

Assessment for Quality of Life

The 36-item Short Form Survey (SF–36) (27) is one of the most widely used generic health survey instruments. The Japanese version of the SF–36 (28–30), which has been demonstrated to address issues of validity and comparability with the English version, was used to evaluate health-related quality of life. The SF–36 health questionnaire generates three scores: a physical component summary, a mental component summary, and a role/social component summary (31). All scores were transformed to fit a norm-based score of 50 and standard deviation of 10, with higher scores indicating better quality of life.

Statistical Analysis

This study used a prospective design. Continuous variables were presented as means and standard deviation according to the diagnosis of depression. Categorical variables were summarized by descriptive statistical methods. Continuous variables were compared using the Mann-Whitney U test, whereas categorical variables were compared using the Fisher’s exact test. For each mental, cognitive, or physical characteristic at baseline, a simple logistic regression model was applied to identify factors that predict the onset of depression (32). For each identified factor, the predictive performance of discrimination was evaluated by calculating area under the receiver operating characteristic curve (AUROC). The AUROC was tested to determine whether it was >0.5. The predictive performance of calibration was evaluated by using the Hosmer-Lemeshow test. Threshold value for each predictive factor was determined considering the values of sensitivity and specificity. In order to evaluate the influence of missing values for each predictive factor, multiple imputation was applied (33). All statistical tests were performed with two-sided significance level of 0.05 using SAS 9.4 (SAS Institute, Cary, N.C.).

Results

Participants

A total of 152 patients met inclusion criteria; of these, 99 refused informed consent. In addition, four withdrew their previously granted consent. To predict patients in the acute and subacute stroke periods who would develop major depression during the chronic period of the first year following stroke, one patient who had a diagnosis of major depression in the acute and subacute phase was excluded from the analysis. The remaining 48 patients were included in the analysis. The number of patients with major depression during the chronic first year after stroke onset was five (10.4%). The mean time from stroke to onset of major depression was 7.2 months. Demographic and clinical characteristics of depressed and nondepressed patients are presented in Table 1. Psychiatric history included previous psychiatric outpatient treatment or use of a psychotropic medication.

Depressive State, Cognitive and Physical Function, and Quality of Life in a Subacute Stroke Phase

The mental, cognitive, and physical characteristics of patients at baseline are summarized in Table 2. The patients who developed major depression during the first year had significantly higher scores at baseline on the MADRS and PHQ-9.

Predictive Tools for Depression

The predictive performance of discrimination and calibration for the identified two factors are shown in Table 3. MADRS and PHQ-9 in the subacute stroke phase had good classification ability to discriminate those who developed depression during the first year because the AUROCs were 0.83 and 0.88 (i.e., significantly greater than 0.5 compared with random classification). In addition, the Hosmer-Lemeshow test was nonsignificant.

Furthermore, the threshold values were determined to make sensitivity plus specificity maximum. The threshold value of MADRS was 11, and its specificity and sensitivity were 0.881 (37/42) and 0.800 (4/5), respectively. The positive predictive value (PPV) and the negative predictive value (NPV) were 0.444 (4/9) and 0.974 (37/38), respectively. The threshold value of PHQ-9 was 9, and its specificity and sensitivity were 0.921 (35/38) and 0.800 (4/5), respectively. The PPV and NPV were 0.571 (4/7) and 0.972 (35/36), respectively. There was one missing MADRS score and five missing PHQ-9 scores in the nondepressed group. Similar results were obtained by the multiple imputation approach, suggesting little influence of missing values. The Pearson correlation coefficient between MADRS and PHQ-9 was 0.64 (p<0.001).

Discussion

To our knowledge, this is the first study to demonstrate that both the PHQ-9 and MADRS in the acute and subacute phase of ischemic stroke (i.e., within 6 weeks poststroke) were able to predict major depression during the chronic first-year period with an accuracy >80%.

Previous studies have demonstrated the efficacy of antidepressants to prevent the development of depression during the first year after stroke (11) and to decrease the mortality rate for depressed patient up to 5 years after stroke (34). Therefore, one might logically ask why the acute and subacute prediction of major depression during the chronic poststroke period is useful in the clinical setting. Based on our prior randomized studies of antidepressants to treat poststroke depression, we found that eight out of 30 patients randomized to medication refused to take medications due to fear of side effects of medications and/or because they did not feel depressed at that time (35). Thus, a significant number of patients with stroke are resistant to taking medications. The current study, however, demonstrated a clear method to identify stroke patients at high risk for developing major depression in the chronic period after stroke. Clinicians, therefore, may monitor the patients who have increased risk of major depression in the chronic stroke phase.

The only other study that we are aware of to examine risk of future poststroke depression was done by Fuentes et al. (36). This study found that higher scores on the Hamilton Depression Rating Scale (HAM-D) was associated with a three times greater risk of poststroke depression at the 3-month follow-up. However, the HAM-D, similar to the MADRS, is based on a clinical interview that requires a trained clinician.

Among combined studies of 2,769 patients hospitalized for subacute stroke or rehabilitation, 21.6% were diagnosed with major depression (37). Furthermore, approximately 15% of patients who were not depressed at initial assessment developed depression during the first year following stroke (4). In Japan, Kaji et al. (38) examined incident depression among 100 patients at 2 to 5 weeks after onset of stroke using the Mini-International Neuropsychiatric Interview (MINI); they found that the prevalence of major depression was 5%. In addition, a multiple-center study examined incident depression among 373 patients in a subacute-chronic phase of stroke (1 month to 1 year after the onset of stroke) using the MINI, and found that the prevalence of major depression was 9.9% (39). In the current study, the prevalence of patients with major depression in the chronic phase of stroke (7 weeks to 1 year after onset of stroke) was 10.4%. In Japan, the prevalence of poststroke chronic onset major depression may be lower than that in the United States and European countries. Therefore, the PPV and NPV should be carefully interpreted for a different population of interest because the PPV and NPV are highly dependent on the prevalence. For the Japanese stroke patient population, we can interpret that the risk for depression increases from 10% to 57% if the score of PHQ-9 is ≥9. For a different population where the prevalence is higher, the PPV or NPV should be interpreted differently.

The present study had the following limitations that should be acknowledged. First, the number of cases was rather small because only one institution was involved. Second, almost two thirds of the patients who met inclusion criteria declined to sign a consent form and therefore could not be included. The reason why so many patients refused to participate is unclear, but most of the patients in this study lived around the Tokai University Hospital; they might have been afraid that neighbors might discover information about their stroke or depression. In any event, additional patients need to be studied to assess the accuracy of the cut-off scores derived from this study as well as the utility of each of these depression scales.

In conclusion, a PHQ-9 score ≥9 in the acute and subacute phase of stroke was associated with an increased risk of developing major depression during the chronic period of the first year following stroke. The self-administered PHQ-9 may be a beneficial tool to help clinicians to identify patients who are at increased risk of developing depression during the first year after stroke.