Influence of Premorbid Psychopathology and Lesion Location on Affective and Behavioral Disorders After Ischemic Stroke

We prospectively studied a population of patients with ischemic stroke who were admitted to the Neurology Unit in the Hospital Virgen del Puerto. They were evaluated upon admission and again at 4, 12, and 26 weeks after their strokes. The recruitment period was from May 2007 to December 2008.

Patients admitted to the hospital with an acute cerebral infarct (CI) confirmed by neuroimaging techniques, and who had a responsible caregiver, qualified for the study. The exclusion criteria were the following: knowledge or suspicion of previous dementia or cognitive decline, as assessed by clinical record or by means of the Jorm's Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE);¹¹ cerebral hemorrhage or another suspected etiology of brain injury (tumor or others); transient ischemic attack; persistent coma or severe alteration of consciousness 4 weeks after the stroke; and death or appearance of a new brain lesion before 4 weeks had passed (stroke recurrence, cerebral hemorrhage, or others).

We used the following instruments for patient evaluation: 1) A clinical record form for clinical, demographic, and social variables: age, gender, years of formal education, documented previous clinical diagnosis of hypertension, diabetes, or hypercholesterolemia; current smoking habit or hazardous/harmful alcohol consumption (defined as current intake of >280 g per week in men and >168g per week in women, according to the recommended criteria of the World Health Organization¹²); and documented before the diagnosis of depression, anxiety disorder, or psychosis; 2) The Spanish versions of the NPI,¹³ the 17-item Hamilton Rating Scale for Depression (Ham-D),¹⁴ and the Hamilton Rating Scale for Anxiety (Ham-A)¹⁵ were used for psychopathological assessment. The Ham-D has been identified as a reliable instrument for the diagnosis of poststroke depression in previous studies.^7,16 The NPI is composed of 12 subscales that measure different behavioral symptoms (delusions, hallucination, agitation, dysphoria, anxiety, euphoria, apathy, disinhibition, irritability, aberrant motor behavior, nighttime behavioral disturbances, and eating abnormalities). Each subscale rates the frequency and severity of the behavior as well as the caregiver's resulting distress. The frequency rating multiplied by the severity rating gives a composite score for each behavior. A global score for the NPI is generated by summing up the total scores of the individual subscales.⁹ 3) The Spanish versions of the Folstein's Mini-Mental State Exam (MMSE),¹⁷ the short form of the IQCODE,¹⁸ the Frontal Assessment Battery (FAB),¹⁹ and the Clock-Drawing Test²⁰ for cognitive evaluation; 4) the Modified Rankin Scale,²¹ Barthel Index,²² Lawton Index,²³ and the Canadian Neurological Scale²⁴ for functional and neurological assessment.

All patients were evaluated for inclusion within 48 hours after occurrence of the stroke. After checking inclusion and exclusion criteria, we obtained informed consent from the patient and/or caregiver. The clinical record form, NPI, IQCODE, Barthel Index, and Lawton Index were then administered to the caregiver. Patients with data from the IQCODE that suggested previous cognitive decline or dementia were excluded in order to avoid the inclusion of patients with previous psychopathology associated with neurodegenerative disease, which could bias the study. For those patients who qualified to participate in the study, all of the psychopathological, cognitive, and functional scales described in the instruments section were applied at 4, 12, and 26 weeks after the event.

CI locations were clinically diagnosed according to the Oxfordshire Community Stroke Project criteria²⁵ and were confirmed by computerized tomography or brain MRI. For the purpose of this study, CIs were classified as right hemispheric, left hemispheric, or brainstem/cerebellum lesions, distinguishing whether or not they were lacunar infarcts.

Statistical analyses were performed with the SPSS program (Version 11.0). We used descriptive statistics to analyze frequency distributions and summarize data. The Student's t-test was used for comparison of means, and Pearson's coefficient was used to examine correlations. Multivariate analysis was also performed, with a backward, stepwise linear-regression analysis. A p value of<0.05 was regarded as statistically significant.

Eighty-nine patients fulfilled the inclusion criteria and met none of the exclusion criteria. Table 1 details their baseline demographic data, medical and psychiatric antecedents, and clinical characteristics.

The number and percentage of patients who presented at least one PS, defined as a score >0 in any NPI subscale, at baseline (before stroke) and 4, 12, and 26 weeks after the acute event, are represented in Table 2.

Eighty-three patients were able to accomplish an assessment with Hamilton's scales (Ham-D and Ham-A). Among the other six patients, five had a severe aphasic disorder, and the remaining patient refused to perform the test; 38% of patients showed detectable depression on the Ham-D (defined as a score >7). Moderate-to-severe depression (defined as a score >14) was seen in 19.9% of patients, whereas anxiety was detected with the Ham-A (defined as a score >5) in 33.7% of patients. Moderate-to-severe anxiety (defined as a score >14) was detected in 6% of patients. Pearson′s correlation coefficient between the NPI Dysphoria subscale and Ham-D scores at Week 4 was 0.837 (p<0.01), and the correlation coefficient between the NPI Anxiety subscale and Ham-A scores was 0.763 (p<0.01).

A comparison of the mean scores in the NPI-4W according to the presence or absence of vascular risk factors (HTA, diabetes, hypercholesterolemia, smoking), alcohol consumption, and psychiatric antecedents (depression, anxiety, or psychotic disorders) did not reveal any statistically significant differences. There were also no statistically significant differences on any of the NPI subscales regarding stroke location or size (lacunar versus nonlacunar infarcts).

Fifty-five patients completed the study up to Week 26 (Table 3, Figure 1). In this population, the mean NPI Global score and the mean scores for the Caregiver's Distress, Dysphoria, and Apathy scales were significantly increased (p<0.05) at Weeks 4, 12, and 26, as compared with the baseline NPI. However, the mean NPI Global score and Anxiety score significantly decreased between Weeks 4 and 12. There were no statistically significant differences in the mean NPI Global score or in any NPI subscale between Weeks 12 and 26.

Table 4 shows the NPI-4W subscales that correlated significantly with cognitive and functional variables at Week 4. The Dysphoria score correlated negatively with more cognitive (Clock-Drawing Test, MMSE, and FAB) and functional (Lawton Index and Rankin Scale) scales. On the other hand, functional variables were negatively correlated with more psychopathological variables, including Caregiver's Distress.

In the group of patients who completed the follow-up (N=55), apathy was found to be significantly correlated with the Lawton Index (p<0.001) and Rankin score (p<0.05) at Week 12. At Week 26, the Lawton Index was significantly correlated with Dysphoria (p<0.05) and Apathy (p<0.001) scores, but there were no other correlations between psychopathological variables and the rest of the cognitive and functional scales.

Predictive factors for the occurrence of early and long-term PSs after stroke were determined by means of a backward, stepwise linear-regression analysis, which produced explanatory models for some of the psychopathological variables. The NPI Global score and NPI subscales scores at 4 (NPI-4W) and 26 weeks (NPI-26W) after stroke were introduced as dependent variables. The following variables, which could be assessed at the moment of admission, were included as independent variables: baseline NPI Global score and baseline NPI subscales (reflecting previous psychopathological status), psychiatric antecedents (depression, anxiety, or psychotic disorder), demographic characteristics (age, gender, and years of formal education), medical antecedents (vascular risk factors, smoking, and alcohol consumption) and stroke location (right hemispheric, left hemispheric, and brainstem/cerebellum). Qualitative variables were recodified as dummy variables to be introduced in the linear-regression models.

The following NPI-4W subscales produced valid regression models: Agitation, Dysphoria, Apathy, Disinhibition, Irritability, Motor Disturbance, Sleep Disturbance, and Eating Disturbance (Table 5). Predictive factors for PSs at 4 weeks were younger age, male gender, right-hemispheric location of the CI, and baseline NPI scores. Younger age was a predictive factor for Agitation and Disinhibition, whereas male gender was a predictive factor for Agitation and Irritability. Right-hemispheric location of the CI predicted the development of Dysphoria, Apathy, Disinhibition, and Sleep Disturbance, although its coefficient value was only significant for Sleep Disturbance.

The related baseline NPI subscale score was a predictive factor for the behavior studied at 4 weeks after stroke in every model except in Disinhibition and Sleep Disturbance (e.g., a high baseline NPI score for Dysphoria was a risk factor for having a high score for Dysphoria in the NPI-4W). Apathy at Week 4 was also predicted by a higher score in other baseline NPI subscales (Agitation and Dysphoria). Moreover, a history of hazardous/harmful alcohol consumption was a risk factor for Eating Disturbance, whereas antecedents of depression predicted the occurrence of Sleep Disturbance.

At 26 weeks after stroke, we found valid regression models for Agitation, Dysphoria, Anxiety, Apathy, Irritability, and Sleep Disturbance (Table 5). Antecedent depression, right-hemispheric location, and scores in the baseline NPI subscales of Agitation, Anxiety, Irritability, and Dysphoria emerged as predictive variables in these models. Right-hemispheric location persisted as a predictive factor for Dysphoria and Apathy, with a significant coefficient value for Dysphoria (p=0.03). Previous history of depression was a risk factor for Dysphoria and Agitation on the NPI-26W, but it made Irritability less probable.

The percentage of variance explained was higher at 26 weeks than at 4 weeks in the models for Agitation, Dysphoria, Apathy, and Irritability, with more than 50% of the variance explained by the models of agitation and dysphoria.