An 82-Year-Old Woman With Mood Changes Following a Stroke

DR AUDET: Mrs K is an 82-year-old woman who experienced changes in her mood and behavior 2 months after a thrombotic stroke. She has lived in a long-term care facility for the past 2 years. She is retired, has separated from her husband, and until this recent event, enjoyed participating in frequent visits from her children. Her insurance is through Medicare.

Mrs K has cerebrovascular disease characterized by mild multi-infarct dementia. Until her recent stroke, she had some memory loss, but no restrictions in her usual activities of daily living.

The acute right internal capsular stroke resulted in a left-sided ataxia-hemiparesis syndrome and left her with moderate to severe restrictions in her ability to use both her left arm and leg. She experienced significant deterioration in her activities of daily living. She now needs to use a wheelchair and has not been able to resume her social activities. Her acute hospital stay was uncomplicated, and she returned to her long-term care facility 1 month after her initial presentation.

Three months after her stroke, Mrs K continued to have episodic changes in her mood and behavior. On occasion, complaining of tiredness, she declined to participate in physical and occupational therapy sessions. Mrs K often lost her temper and regretted the outbursts afterwards, stating that she was not herself anymore. She expressed fear that medications might be used to poison her and was suspicious of people around her. She expressed the wish to die during her sleep.

Mrs K's appetite has been poor, and she has lost 5 pounds. She has not complained of any problems with sleeping. She admits to episodes of sadness and hopelessness. She is resigned that she must accept her body as it is now, with all its limitations. Her daughter has also noticed changes in Mrs K's affect, but notes that her mother brightens up when she and her son visit.

Mrs K's past medical history includes hypertension, type 2 diabetes mellitus, duodenal ulcers, coronary artery disease, and a pacemaker for sick sinus syndrome. There is no family history of psychiatric disease, and Mrs K has no history of depression. She is a nonsmoker and has never abused alcohol. Her medications include digoxin, 0.125 mg/d; glyburide, 15 mg/d; atenolol, 50 mg/d; atorvastatin calcium, 10 mg/d; and nizatidine, 150 mg/d. Her treating physician initially started Mrs K on fluoxetine, but this was changed to sertraline, 25 mg/d, because of its shorter half-life and the need to use a hospital formulary medication.

On examination, Mrs K is a pleasant, cooperative woman whose speech reveals mild dysarthria. She is oriented to person, but not place or time—although she comments that she is aware of this memory problem. Physical examination reveals a left facial droop, decreased strength in the left upper and lower extremities (2-3/5), normal sensation, decreased vibration sense bilaterally, loss of position sense, and a left Babinski sign. Her thought process is logical and goal directed, without loosening of association, flight of ideas, or tangentiality. She denies hallucinations. Her cognitive function is diminished—she can spell "world" backward correctly, but with notable effort. She recalls 3 out of 3 objects immediately, but none at 5 minutes. The computed tomographic scan of her brain reveals an old frontal lobe stroke and cerebral atrophy of the right ventricle. Her electroencephalogram findings are normal.

Well right now, I don't do too much. I don't do much walking anymore. I used to love to walk, but I don't anymore. I feel sadness at times, but not too often. I sort of buried that part of me. I really don't think of it. As my life goes on now, I'll do what I can, and that's it. I don't think of [being sad] much. I don't want to think of it because then I would be sad all the time. Because if I did I would cry and cry and cry. I feel there's no reason for me to do that because it's not going to help. So, I go along with my life as best as I can.

She's had a couple of strokes before. And they left no residual—maybe a little left-sided facial weakness. But this stroke with the paralysis was the most substantial part of her life change. As far as losing hope, I think that has a lot to do with her mood. I think she has every right to lose hope. In the beginning, she was down a lot, really depressed. As days and months go by, as she's gaining more acceptance of existing like this, she's not always doom-and-gloom. Usually when I come by, I try to bring her out of that.

I saw Mrs K recently for a visit subsequent to my initial evaluation. Her affect was considerably more depressed than when I had seen her the first time. She was tearful. Her outlook was very grim. She was feeling hopeless about her recovery. She was less talkative than at the time of the first meeting, and really did not offer much spontaneously. She admitted to some passive suicidal ideation—she said that she had hoped at times that she would die in her sleep. Staff reported that she was continuing to resist physical rehabilitation efforts. Her unwillingness to participate in physical therapy is striking, not only because of her previous level of physical activity but also because she is known to be a very independent woman.

I'd like to ask Dr Robinson what his experience has been with time of onset of poststroke depression. Do patients often have other symptoms that may be different from the recognized diagnostic criteria for major depressive episode? How do lesions from the cerebrovascular accident complicate the diagnostic approach?

What is known about the prevalence of depression after stroke? What are the pathophysiologic mechanisms for this connection? What is the differential diagnosis for mood changes after a stroke—depression vs other poststroke phenomena? How should one evaluate patients to ascertain the correct diagnosis? Are there data to support cognitive and/or psychological interventions for this phenomenon? Do medications help with poststroke depression? Which do you recommend, and why? What are your recommendations for Mrs K?

DR ROBINSON: The annual incidence of stroke in the United States is approximately 500,000 cases per year.¹ Numerous investigators in almost all continents of the world have studied the prevalence of depressive disorder among the 75% to 80% of patients who survive their stroke.^{1 - 8} The major finding from these studies is that the prevalence of poststroke depression is relatively consistent across cultures, but depends on whether patients are examined in a hospital setting or surveyed in the community (Figure 1). In hospitalized patients with acute stroke, the mean prevalence of major depression is 22% and for minor depression, 17%. In a random sample of patients in the community examined at about the same time following stroke as hospitalized patients, the mean prevalence of major and minor depression was 13% and 10%, respectively. Based on the poststroke survival rate, approximately 150,000 new cases of poststroke depression occur each year in the United States. The scientific evidence available on poststroke depression is stongest as to data on prevalence of disease. Many of the other data are fraught with methodological problems such as comparing patients who were studied at different times poststroke or not using structured interviews and diagnostic criteria. In addition, many important issues, including the prevalence and clinical correlates of disorders such as apathy, anosognosia, or even anxiety disorder, have been examined in detail by only 1 or 2 investigators. Thus, the field is still emerging and many of the findings discussed here need to be replicated in other patient populations.

The pathophysiological mechanisms of poststroke depression are not known, but a large number of studies have examined the causative role of lesion location. Mrs K developed depression about 3 to 4 months following a right hemisphere stroke. Our group first reported in 1984 that, during the acute poststroke period, patients with left anterior lesions (left frontal cortex or left basal ganglia lesions) had the highest frequency of depression, and the severity of depression correlated directly with proximity of the anterior border of the lesion to the frontal pole.^{9 - 10} However, these findings have generated controversy in the literature on poststroke depression, with some studies reporting the same phenomenon^{2 - 3 ,11} and others finding no difference between right and left frontal injury in the frequency of poststroke depression^{7 ,12 - 14} (Figure 2).

A recent longitudinal study of poststroke depression and lesion location reported that the most likely explanation for these conflicting findings is interstudy differences in the time patients were examined following stroke.¹⁵ The association of left frontal and left basal ganglia lesions with major depression has been found only during the acute poststroke period (ie, 2 weeks to 2 months). Although 2 studies done between 30 and 60 days did not find an association with left hemisphere lesions (Figure 2), the great majority of studies were done at 3 or more months following stroke when the frequency of depression following right hemisphere stroke was no different than that following left-hemisphere stroke.^{1 ,15}

Mrs K's lesion location and time of onset of depression are consistent with this time-dependent phenomenon. The stroke occurred in early June 1999, and 2 months later an examination did not provide evidence for a major depression. Although she began receiving 10 mg of fluoxetine in late August, mostly because of her increasing feelings of suspicion, several days later a psychiatrist examined her and did not feel that she had clinical depression. By the fourth month after the stroke, however, the physician felt that Mrs K had a clinically significant depressive disorder and switched her medication from fluoxetine to sertraline.

Another interesting aspect of the pathophysiology of depression in this case is that Mrs K had more than 1 stroke. A computed tomographic scan revealed the presence of a prior right frontal lesion and clinical evidence of an acute right internal capsule lesion. Our group conducted a study of patients with multiple lesions to determine whether the existence of a second or third lesion would influence the effect of the acute stroke on mood and found that subsequent strokes bring out symptoms associated with an earlier stroke.¹⁶ Thus, among patients with old lesions of the left frontal cortex or left basal ganglia, a new lesion in any location was associated with the acute onset depressive disorder regardless of whether the previous lesions were associated with depression. For Mrs K, the onset of depression 3 to 4 months after her stroke suggests that a new lesion of the internal capsule may have brought on symptoms associated with the prior lesion located in the right frontal cortex. The right frontal and right parietal cortices are right hemisphere sites where stroke lesions are most strongly associated with depression.¹

Left frontal and left basal ganglia lesions may be associated with depression in the acute stroke period, and right frontal and right parietal lesions may be associated with depression in the subacute period because of lateralized biochemical or neurophysiological changes occurring in response to brain infarction. Laboratory studies in rats and receptor imaging in humans using positron emission tomography have shown that right hemisphere lesions produce more profound decreases in norepinephrine and serotonin than left hemisphere lesions.^{17 - 18} These right hemisphere changes lead to compensatory up-regulation of serotonin receptors^{17 - 18} in parietal and temporal cortex and perhaps delay the onset of depression. On the other hand, biogenic amine depletion is milder following a left frontal lesion but is uncompensated and leads to acute onset of depression.

In addition to lesion location, other factors associated with increased risk of poststroke depression include prior personal history, family history of major depression,⁵ poor social support,¹ greater impairment in activities of daily living⁴ (but not stroke severity¹ ), high neuroticism score as measured by the Eysenck Personality Inventory,¹⁹ and 1 or more negative life events during the 6 months prior to stroke.²⁰ Morris et al²⁰ showed that the risk of poststroke depression increased linearly from 20% in patients with no risk factors to 65% if family history, neuroticism, and negative life events were all present. This finding suggests that the effect of risk factors may be additive.

This brief discussion of the causes of poststroke depression would not be complete without a description of the consequences (burden) of depression. Studies by both our group,²¹ based on our database from Baltimore, and Morris et al,²² based on an Australian study, demonstrated that, when all factors influencing physical recovery (eg, lesion size and hours of physical therapy) are taken into account, depression has an independent effect inhibiting recovery in activities of daily living over the first 2 years following stroke. In addition, both Morris et al²³ and our study²⁴ demonstrated that at 15-month or 10-year follow-up, patients with poststroke major or minor depression were between 3.5 and 8 times more likely to die than nondepressed patients even when other factors such as associated illnesses and prior stroke were taken into account. Thus, depression represents a very significant burden on both recovery and actual survival following stroke. These findings suggest that treatment of depression should be initiated as soon as it is recognized. There is preliminary evidence that treatment with amphetamine may enhance physical recovery in nondepressed patients.²⁵

The differential diagnosis of poststroke depression includes syndromes that may mimic some symptoms of depression (Table 1). The condition perhaps most likely to mimic the symptom of depressive disorder is pathological crying. In this condition, patients cry frequently, in excess of their underlying feelings of sadness or depression. Approximately half of the patients with pathological crying in our study also had major or minor depressive disorders,³⁰ but the prevalence of depression was lower in other studies.^{31 - 32} Pathological crying may be differentiated from poststroke depressive symptoms in that patients are aware that the emotional display is uncontrollable and in excess of their underlying feelings of sadness. In addition, some patients with pathological crying will not display symptoms of depressive disorder such as feelings of hopelessness, suicidal thoughts, or sleep disturbance. Although Mrs K was noted to be crying at times, she did not display tearfulness during the interview, and she did not complain of uncontrollable crying.

Another disorder that should be considered in the differential diagnosis in Mrs K's case is apathy. As with pathological crying, apathy occurs in patients with or without depression. In our study,³³ approximately half the patients with apathy had an associated depressive disorder while the other half reported apathy alone. The diagnosis is based on the existence of apathetic symptoms such as loss of interest, loss of motivation, unconcern and indifference, and an inability to get started on things. Although Mrs K was not specifically examined for apathy, she did not appear apathetic during the interview. She did, however, withdraw from physical therapy because it was "too hard" and her energy level was low.

Another diagnostic issue to consider is whether the depression is major or minor (subsyndromal).³⁴ The Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DMS-IV)³⁵ diagnosis of minor depression is a subsyndromal form of major depression in which the presence of depressed mood or loss of interest and pleasure must be present but there are less than 5 symptoms of major depression. During the interview, Mrs K was noted by her treating physician to have sadness, loss of interest, loss of energy, decreased appetite, hopelessness, psychomotor retardation, passive death wishes, and social withdrawal. If these symptoms were present for 2 weeks or more, she would have met the DSM-IV criteria for "depression due to stroke with a major depressive-like episode."

Another differential diagnosis to consider is anxiety disorder, characterized by worry, restlessness, irritability, tension, sleep disturbance, low energy, and poor concentration. Both our studies^{1 ,6} and that of Astrom³⁶ have found that more than half of the patients with anxiety disorders have comorbid depression. The anxiety and depression are thought to be comorbid because the course of anxiety and depression are different, and they each have different clinical and pathological correlates.¹ For example, anxiety is associated with prior history of alcohol abuse and injury to the right hemisphere, while depression is associated with left frontal and left basal ganglia lesions. Mrs K had some symptoms but did not report anxiety.

The final diagnostic consideration is whether this could be a catastrophic reaction, characterized by the sudden onset of tears, aggressive behavior, anger, refusal, anxiety, and shouting usually precipitated when the patient is asked to perform a task. It has been suggested that many patients diagnosed as having major depression actually have catastrophic reaction.¹² Our group examined patients with acute stroke for catastrophic reaction and found that the majority, but not all, of patients who display this reaction had major depression.^{1 ,37} Catastrophic reaction was associated with right or left basal ganglia lesions. In addition, major depression, but not catastrophic reaction, was associated with frontal lobe lesions. Although Mrs K had tearful episodes, she did not display, during the interview or by history, any episodes of sudden emotional display.

Another clinical issue worthy of comment in this patient is the development of paranoia. Although it is unusual for patients to develop delusions or hallucinations following a stroke, a significant number of patients reported in the literature developed these psychotic symptoms. Mrs K expressed fears that people were trying to poison her and told her treating physician that she felt her roommate might try to kill her. Our group conducted a study of patients with delusions, including paranoid ideas like those Mrs K expressed, or hallucinations following stroke and compared them with controls matched for lesion location (without delusions or hallucinations) and with general patients with acute stroke.^{1 ,38} Delusions and hallucinations were significantly associated with older age, family history of psychiatric disorder, right hemisphere lesions (particularly of the temporal parietal occipital junction), subcortical atrophy as evidenced by enlarged ventricular-to-brain ratios in the frontal horns and third ventricle, and the existence of a seizure disorder. Mrs K fits part of this profile in that she is older (aged 82 years) and had a right hemisphere lesion. Furthermore, examination of her computed tomographic scan revealed cerebral atrophy with enlargement of the anterior horn of her right lateral ventricle. Mrs K did not, however, have a history of poststroke seizure disorder. Delusions may begin any time from a few hours to several months following stroke and are usually short-lived, lasting only a few months.^{1 ,38} The improvement in Mrs K's symptoms at about 4 months poststroke is consistent with this time course. Finally, although some patients are more refractory to treatment than patients without stroke, psychotic symptoms following stroke will usually respond to treatment with antipsychotic medications.^{1 ,38}

There has been a debate in the literature concerning the usefulness of traditional depressive symptoms for the diagnosis of depression in patients with acute medical illnesses such as stroke.³⁹ Symptoms such as loss of appetite, loss of energy, difficulty with concentration, and sleep disturbance may all be attributable to acute medical illness rather than depressive disorder. We have extensively examined this issue in patients with acute stroke and during the first 2 years following stroke.^{1 ,40 - 41} The frequency of depressive symptoms in patients who acknowledge feelings of sadness was compared with the frequency in patients who deny feelings of sadness. If sleep disturbance, for example, was produced by the hospital routine or the physical illness, the frequency of sleep disturbance would be expected to be as common among patients without depressed mood as it is among patients with depressed mood. Among 205 patients with acute stroke, we found that during acute hospitalization, patients with depressed mood had a mean (SD) frequency of 3.6 (2.1) vegetative symptoms such as sleep disturbance, appetite disturbance, loss of energy, loss of libido, or diurnal mood variation, while the nondepressed patients had a mean (SD) frequency of 1.0 (1.1) (P = .001). This means that vegetative symptoms are not rampant in patients with an acute illness like stroke. With the exception of weight loss and early morning awakening, the frequency of all individual vegetative and psychological symptoms of depression were significantly more common in patients with a depressed mood compared with patients without a depressed mood. Furthermore, when we accounted for this frequency of vegetative symptoms among the nondepressed patients by requiring this group to have 1 additional symptom of depression to be classified as having depression, the frequency of major depression in the overall population of 205 patients decreased by only 1%, from 22% to 21%. Furthermore, the rate of depression declined only 2% if 2 additional symptoms were required. Thus, the diagnosis of major and minor depression in patients with stroke can be made using the standard DSM-IV criteria for major and minor depression regardless of the fact that some symptoms of depression might be attributable to acute physical illness.

Although cognitive behavioral and other psychological treatments have been proposed in the treatment of poststroke depression,^{42 - 43} no controlled clinical trials have been done to demonstrate the efficacy of these treatments in either major or minor depression. In some patients, impairments in language production or comprehension may interfere with the use of traditional psychological methods or even the assessment of depression. However, uncontrolled studies of group therapy⁴³ and cognitive therapy⁴² have been conducted. In 1 of these studies, of 19 patients given cognitive therapy, 4 responded, 6 had some benefit, and 9 failed to respond.^{41 - 42}

At least 4 double-blind, placebo-controlled clinical trials of drug therapy in poststroke depression have been reported in the literature. The first study, reported by our group in 1984, demonstrated the superiority of nortriptyline (n = 14) vs placebo (n = 20) in the treatment of poststroke depression.^{1 ,26} The efficacy of citalopram (20 mg given to patients younger than age 66 years, 10 mg to those aged 66 years or older) (n = 33) over placebo (n = 33) in the treatment of poststroke depression for 6 weeks was reported in 1994.²⁷ In addition, trazodone, given to patients with a positive dexamethasone suppression test result (n = 7, trazodone; n = 9, placebo), was shown to be effective in improving recovery in activities of daily living for at least 6 weeks following a stroke.²⁹ The most recent drug treatment study was a double-blind, randomized trial conducted by our group that compared fluoxetine hydrochloride (n = 23) with nortriptyline (n = 16) and placebo (n = 17).²⁸ Patients were enrolled if they had a diagnosis of either major or minor poststroke depression and had no contraindication to the use of fluoxetine or nortriptyline such as intracerebral hemorrhage for fluoxetine or cardiac conduction abnormality for nortriptyline. Patients treated with fluoxetine received 10-mg dosages for the first 3 weeks, 20 mg for weeks 4 through 6, 30 mg for weeks 7 through 9, and 40 mg for the last 3 weeks. Patients treated with nortriptyline received 25 mg for the first week, 50 mg for weeks 2 and 3, 75 mg for weeks 3 through 6, and 100 mg for weeks 6 through 12. Intent-to-treat analysis demonstrated significant time-by-treatment interaction with patients treated with nortriptyline showing a significantly greater decline in Hamilton depression scores than patients treated with either placebo or fluoxetine at 12 weeks. There were no differences between fluoxetine and placebo. Nortriptyline had a significantly higher rate of response (10/16, 62%) than either fluoxetine (2/23, 9%) or placebo (4/17, 24%) (Fisher exact test, P = .001). In addition, fluoxetine treatment was associated with a mean weight loss of 15.1 lbs or 8% of initial body weight from beginning to end of treatment, which was not seen with nortriptyline or placebo. Among 12 patients treated with fluoxetine, 10 lost 10 lbs or more while only 2 of 13 nortriptyline and 1 of 11 placebo patients lost this amount of weight (Fisher exact test, P = .004). Based on these data, if there are no contraindications to nortriptyline such as heart block, cardiac arrhythmia, or narrow angle glaucoma, Mrs K should be treated with nortriptyline for her poststroke depression. The course of her paranoid ideas should be followed closely to see if they resolve without treatment. Doses of nortriptyline should be increased slowly and blood levels monitored to achieve serum concentrations between 190 and 570 nmol/L. If there are contraindications to the use of nortriptyline, citalopram (I use 20 mg for patients younger than age 66 years, 10 mg for those aged 66 years or older) would be the next choice.

In conclusion, Mrs K's poststroke psychiatric disorders include cognitive impairment, paranoid ideation (perhaps delusions), and probably poststroke major depression. Her cognitive function may improve somewhat with successful treatment of depression, and her depression would be expected to improve following several weeks of treatment with antidepressant medication. Her paranoid ideas will probably slowly dissipate with or without antipsychotic medication. She is, however, vulnerable to the development of similar symptoms should another stroke occur. If she is able to tolerate nortriptyline, I would treat her with this medication, slowly titrate her dose, and continue the therapy for at least 1 year.

A PHYSICIAN: Could you comment on the anatomical and pathophysiological models that might explain the phenomenon of poststroke depression?

DR ROBINSON: We do not know a lot about the pathophysiology of depressive disorders, let alone poststroke depression. Most of the work has been done in the context of primary depressive disorders not associated with brain lesions or other secondary depressions such as Parkinson disease. The consensus about the anatomical basis for depression is that some of the cortical and subcortical neuronal loops described by Alexander et al⁴⁴ may play an important role in the development of depression. A number of structures, such as the frontal cortex, basal ganglia, and thalamus, have been implicated. Brain lesions that interfere with the function of these structures and therefore interrupt the integrity of these loops have consistently been associated with depressive symptoms.^{45 - 46}

A PHYSICIAN: Psychiatrists usually tell internists to combine drugs with psychotherapy in the management of people who are depressed. Are any data available on cognitive approaches to depression after stroke?

DR ROBINSON: Unfortunately, no controlled treatment trial data are available on this issue. It seems reasonable to me that patients who don't have significant language impairment should be candidates for cognitive or interpersonal therapy. These structured psychotherapies should be studied in clinical trials. No studies have yet investigated whether such interventions are as effective or, in certain kinds of patients, perhaps more effective than pharmacotherapy.

A PHYSICIAN: In cases of patients with dementia—and for Mrs K allegedly multi-infarct dementia—should clinicians worry about the anticholinergic effects of nortriptyline?

DR ROBINSON: A number of medications have anticholinergic effects, and clinicians taking care of geriatric patients should be aware of their potential to produce adverse effects such as urinary retention and delirium. Yet, clinicians must balance this with the benefits of medications.We found that patients treated with nortriptyline whose depression resolved showed significantly greater cognitive improvement compared with patients treated with placebo or nortriptyline whose depression did not respond (R.G.R., unpublished data, 2000). Thus, depression has a significantly greater effect on cognition than anticholinergic medications. Clinicians should monitor patients closely to see whether cognitive function improves as depression is treated. Anticholinergic effects to monitor are severe dry mouth, difficulty urinating, and constipation. Usually these anticholinergic effects do not necessitate stopping the nortriptyline and can be managed by lowering the dosage (50 mg or 75 mg of nortriptyline may produce serum concentrations between 190 and 570 nmol/L in the elderly) or using symptomatic treatments (eg, laxatives).

A PHYSICIAN: If 40% of patients with stroke suffer from depression, should we approach this complication through prevention, by giving nortriptyline to everyone who has a stroke?

DR ROBINSON: In our study, we treated nondepressed patients to see whether we could either prevent them from developing a depressive disorder or improve their recovery. Our hypothesis was that antidepressants might foster recovery independent of their effect on depression as well as preventing the development of depression. We treated patients with stroke for 12 weeks and followed up for 2 years. The longitudinal data have yet to be analyzed, but, in the future, we should have information on long-term outcomes of stroke patients treated with antidepressants before they develop depression.

A PHYSICIAN: I am delighted that you have done a study in which patients were treated for 12 weeks. In the study you mentioned of nortriptyline vs fluoxetine vs placebo, there was a decrease in scores at 6 weeks, with residual depressive symptoms. At 12 weeks, the nortriptyline patients clearly got better. This finding implies that it takes older people longer to develop a full therapeutic response to antidepressants. One important clinical implication is that physicians need to treat older patients for longer and not quit too early. Six weeks is often a bare minimum.

DR ROBINSON: I agree that a 12-week treatment trial in the elderly is preferable to a shorter 6-week trial. It has been demonstrated that continued treatment with fluoxetine beyond the usual treatment period without an increase in antidepressant dose produces response in previous nonresponders to the same extent as increased dosage.⁴⁷ This suggests that extended treatment trials such as the one used in our recent study are essential to assess accurate treatment response rates.