Validation of a Depression Screening Measure for Stroke Inpatients Bruce Rybarczyk, Ph.D., David R. Winemiller, Ph.D., Lawrence W. Lazarus, M.D., Allyson Haut, Ph.D., Carolyn Hartman, M.D. The Stroke Inpatient Depression Inventory (SID/) was developed and tested on 50 stroke inpatients on a rehabilitation unit. The format and content of the SIDI questions were intended to improve on the limited validity of the Center for Epidemiological Studies-Depression Scale (CES-D) and other measures of depression with stroke inpatients. The validity of the SIDI was supported by higher correlations with a psychiatrist's rating of depression level (r = 0.70), SADS-C score (r = 0.68), and Hamilton Rating Scale for Depression score (r = O. 78) compared with the CBS-D. Also, the SID] was more accurate than the CES-D in correct classification ofpatients with regard to diagnosis of depression (86% us. 76% correct). (American Journal of Geriatric Psychiatry 1996; 4:131-139) S tro ke is ranked among the 10 most common admitting diagnoses for acute care hospitals} and is the most common diagnosis for patients in rehabilitation hospttals.i A significant number of these stroke patients become depressed during the first few weeks after the stroke, with estimates ranging between 29% and 64%.3-8Yet the data suggest that the majority of older adults with depression secondary to any physical illness go undi- agnosed and untreated in both acute care and rehabilitation hospitals. 3,8- 10 Detection and treatment are particularly critical for stroke patients, given the finding that in-hospital levels of depression fredict physical rehabilitation outcome, activities of daily living (ADL) functioning 2 years later,11 and to-year mortality.12 Existing depression scales have significant limitations for use with stroke patients in the hospital setting.8,13 First, Received December 28, 1994; revised March 16, 1995; accepted May 10, 1995. From the Department of Psychology and Social Sciences, Rush-Presbyterian-St. Luke's Medical Center, Chicago, IL.Address correspondence to Dr, Rybarczyk, Dept. of Psychology and Social Sciences, Rush-Presbyterian-St. Luke's Medical Center, 1653 W. Congress Pkwy. Chicago, IL 60612. Copyright © 1996 American Association for Geriatric Psychiatry THE AMERICAN JOURNAL OF GERIATIUC PSYCHIATRY 131 Depression Screening and Stroke because of the sudden onset) acute physical impairments) and subsequent hospitalization) the questions contained in standard measures of depression are often not appropriate for recent stroke patients (e.g., "Have you dropped many ofyour activities and interests?"). Second) with the notable exception of the Geriatric Depression Scale (GDS) , 14 most depression inventories use a rating-scale answer format) usually requiring respondents to rate how frequently they experience a symptom. This type of answer format is often too complex and timeconsuming for patients who have either some level of cognitive impairment) language impairment, and/or fatigue secondary to the stroke. Pinally standard depression inventories often include somatic-type symptoms that are common to both the general older adult population and nondepressed stroke patients) in particular. For example) two studies have found that 4 of the 20 items of the Center for Epidemiological Studies-Depression scale (CES-D)15 correlate significantly with the presence of physical disease, regardless of mood status (e.g., UI could not get going. ") 16,17 Several self..report depression scales have been validated for stroke outpatients. 7,15 The most widely studied scale has been the CES-Dwhich, in general, has demonstrated good criterion validity, concurrent validity; and test-retest reliability with stroke outpatients.l' However) when the CES-Dwas tested with a sample of stroke inpatients, it lacked sufficient concurrent and criterion validity'' For example) the CES-Dcorrelation with psychiatric diagnosis was 0.95 with outpatients but only 0.53 with inpatients. Thus) the limitations of general depression questionnaires for stroke inpatients may diminish their validity with this population. Given the large number of stroke patients on inpatient medical or rehabilitation units) the large percentage who have 132 concomitant depression, and the problems with existing depression questionnaires, a screening test tailored for an inpatient population would address an important need. Accordingly; we developed a 30-item, yes/no-format depression scale, the Stroke Inpatient Depression Inventory (SIDI). Unlike general depression measures, SIDI items address issues that are unique to this population and refer explicitly to the time period since the patient had a stroke. To measure validity; we compared scores obtained on the SIDI with the results of an independent psychiatric evaluation of a group of stroke inpatients on a rehabilitation unit. The CES-D scale was also administered as a comparison measure. METHODS Subjects Over a period of 16 months) consecutive patients admitted to an inpatient rehabilitation service by two participating physiatrists were screened for inclusion in the study. Inclusion criteria were voluntary consent (consent rate was approximately 80%); a recent cerebrovascular accident (CVA) verified by history; physical exam, and CT scan or MRI; and a combined score of 25 or more on the abbreviated Neurobehavioral Cognitive Screening Exam (NCSE; therebyeliminating patients with significant dysphasia) . All 50 subjects included in the study were inpatients on the 66-bed rehabilitation unit of a 900-bed urban hospital. Subjects were an average of 71 ± 6.5 years old, with a mean of 11 ± 3.7 years of formal education. In all, 23 were married, 14 widowed, 10 divorced) 2 never married, and 1 separated; 56% of the subjects were black, and 44% were white; 56% of the CVAs were localized to the right hemisphere, 35% to the left hemisphere, and the remainder were either VOLUME 4 • NUMBER 2 • SPRING 1996 Rybarczyk et at. FIGURE 1. Items of the Stroke Inpatient Depression Inventory (SIDI) STROKE INPATIENT DEPRESSION INVENTORY INSTRUCTIONS: The foUowing questions address how you have adjusted since your stroke. You must answer either "yes" or "no" for each question, whichever is the best answer. ITEMS 1-15: SINCE YOUR STROKE .•• Y 1. have you found less enjoyment in life's simple pleasures that arc still available to rou (e.g., W, newspaper)' 2. have you been cheerful most of the time? 3. arc you having more than your usual troubles falling asleep at night? 4. do other people get on your nerves more easily? Y N 19. Are you worrying a great deal about how you're going to get by after you leave the hospital? 20. Do you feci helpless the way you are nowt 21. Compared to before your 5. have you of/ell felt like crying? 22. 6. have you been worried about your health much more than usual? 7. are you getting more upset over minor things? 23. 8. have you had your usual zest for life? 9. have you been able to give it your all when doing exercises? 10. have you returned to feeling like your "old self' again? 11. have you been feeling empty much of the time? 25. 12. do you find yourself feeling downhearted and blue? N 18. Do you talk as much as usual during visits from family and friends? 24. 26. stroke, arc you in as good a mood when you get up in the morning? Compared to before your stroke, arc you laughing and smiling much less? Have you been discouraged by the recovery process? Even though you have some disability from your stroke, do you still find life exciting? Do you feel worthless the way you are now? Do you blame yourself for the stroke? 27. Have you been bored most of the time since being in the hospital? 28. Do you feel that most people are better off than you arc? 29. If you stayed in the condition you are in now, do you feel like life would be worth Iivin~? 30. Do you feel satisfiedwith the lifeyou have lived thus fur? 13. has your interest in food dropped off?, 14. have you had thoughts of wanting to die? 15. have you felt full of enerJW? 16. Currently, are you feeling hopeful about your future? 17. During your hospital stay, have you been in good spirits most of the time? SCORING:One point for each Uno" answer for items 2, 8-10, 15-18, 21, 22, 24, 29, and 30. One point for each "yes" answer for all other items. TOTAL SCORE: TliE AMERICAN JOURNAL OF GERIATRIC PSYCHIATRY 133 Depression Screening and Stroke mixed or subcortical; 72% of the subjects reported no previous eVA, 8 subjects had one previous eVA, and 3 subjects had three or more. The average length of time between the recent CVA and the first day of participation in the study was 23 ±20 days. Four patients reported a previous history of depression. In each case, the depression was not present at the time of the stroke. Procedure Upon admission to the rehabilitation unit, patients were screened for inclusion. Qualified patients who provided consent were scheduled for testing. All testing was conducted by staff psychologists, psychology residents, or psychology graduate students trained in administration of the study instruments. Patients first completed the abbreviated NCSE. If they met criteria on this measure, they were then administered both the CES-D and SIDI on 2 consecutive days. Order of testing was counterbalanced, with initial order randomly selected. One of two geriatric psychiatrists, blind to the CES-D and SIDI results, then conducted a standard TABLE 1. clinical interview and administered the Hamilton Rating Scale for Depression (Ham-D) and the Schedule for Affective Disorders and Schizophrenia-Change Form (SADS-C) interview rating scales within several days of the administration of the self-report scales (48 patients within 4 days; 2 patients within 6 days; mean = 2.6 days). The psychiatrist, using the information obtained in the interview and the rating scales, categorized the patients according to level of depression. To facilitate comparison with other poststroke depression studies, three categories based on DSM-III criteria were adopted from previous studies: 5-8 1) no depression; 2) minor depression (at least three of nine dysthymia symptoms, without the time criteria); and 3) major depression (five or more symptoms, with at least one being depressed mood or anhedonia). Instruments Stroke Inpatient Depression Inventory (SID/). The 30 items of the SIDI were selected from a pool of approximately 50 Comparison of criterion validity, concurrent validity, and reliability coefficients for the SID] and CES-D SIDI Criterion validity Psychiatrist rating of no/minor/major depression Concurrent validity Ham .. D SADS-C Total Score SADS-C Global Rating Scale Reliability 24-hour test-retest Cronbach's alpha CES-D 0.70* 0.53* 0.78** 0.68** -0.68** 0.66** 0.64** -0.61** 0.92** 0.90 0.89** 0.87 Note: SIDI = Stroke Inpatient Depression Inventory; CES..D = Center for Epidemiological Studies Depression Scale; Ham..D = Hamilton Rating Scale for Depression; SADS-C = Schedule for Affective Disorders and Schizophrenia-Change Form. *Spearman rho: P < 0.0001. **Pcarson r: P < 0.0001. 134 VOLUME 4 • NUMBER 2 • SPRING 1996 Rybarczyk et al. items written by a group offour psychologists and a psychiatrist (Figure 1). Items were selected to match the major symptom areas covered in other self-report depression inventories. To accommodate this particular population, the selected items refer to activities that can be carried out in the hospital, feelings and attitudes about physical impairment, and unique ways used by older adults to describe their dysphoria (e.g., not feeling like "my old self"). Nine items (#5, 7, 12, 15-17, 24, 25, 28) are modified questions from the GDS. 14 The maximum score is 30, with higher scores indicating more depressive symptoms. reliability, concurrent validity, and predictive validity in both nonmedical 15,19 and medical populations. 7 ,16,17 Internal consistency, as measured by Cronbach's alpha, ranged from 0.84 to 0.90 in separate samples. 19 Hamilton Rating Scale for Depression (Ham-D). The Ham-D is a widely used 18-, 21-, or 24..item interview measure of depression originally designed to assess the severity of depression in patients already having a diagnosable depressive disorder. 20 A comprehensive review of the Ham-D indicated that it has good criterion validity, predictive validity, and reliability.21 A study with stroke outpatients found that the Ham-D,was stronglycorrelated with the CES-D (r = 0.74), the GOS (r = 0.77), and the Zung self-report scales (r = 0.70).18 The 24-item version was used in the present study. Centerfor Epidemiologic Studies-Depression Scale (CBS-D). The CES-D is a 20item scale designed to assess the frequency of depressive symptomatol0f>: in samples from community surveys.' Respondents indicate the frequency of each symptom during the past 7 days on a 4-point scale: rarely or none of the time (less than 1 day); some or little of the time (1-2 days); occasionally or a moderate amount of time (3-4 days); and most or all of the time (5-7 days). To facilitate recall of the four possible answers, patients were provided with a set of 511 x 8 cards with the responses printed in large type. The CES-D has demonstrated good Schedule of Affective Disorders and Schizophrenia-Change Form (SADS-C). The SADS-C is a 45-item, interview-based rating scale developed for standardizing the DSM-III diagnosis of depresslon.v' The SADS-C has a seven-point response format and includes a single-item Global Rating Scale (GRS) that was a precursor to Axis V of DSM-III-R. It has been widely used in depression research and has dem- t1 TABLE 2. Agreement of SIDI and CES-D measures with psychlatrlst's diagnosis of depression/no depression, % Specificity Sensitivity Positive Predictive Value Correctly Classified SIDI cutoff = 1,.1 94 71 86 86 18 S6 100 6S 6S 39 64 60 76 CES·D cutoff= 17b cutoff = 21c cutoff = 263 82 44 6S a Optimal cutoff score (l.e., highest percentage correctly classified). b Cutoff score recommended in community studies of depression. 24 C Cutoff score recommended in previous poststroke study.6 THE AMERICAN JOURNAL OF GERIATRIC PSYCHIATRY 13S Depression Screening and Stroke onstrated good concurrent validity and interrater reliability.22 A single total score obtained by summing all depression items and the GRS were used in the present study. Abbreviated Neurobebaoioral Cognitive Screening Exam (NCSB). The NCSE was developed as an expanded mental status test that contains 10 subtests.v' It was intended for patients with neurological disorders and has demonstrated good reliability and predictive validity with this population. Five subtests (Orientation, Attention, Comprehension, Memory, and Social Judgment) were selected for an abbreviated version of the NCSE. These subtests represented cognitive functions that were judged to be essential for answering a self-report questionnaire. tered by the psychiatrist. The SIDI scores correlated highly with Ham-D scores (r = 0.78), SADS-C total score (r = 0.68), and SADS-C Global Rating Scale (r = - 0.68). Corresponding CES-D correlations were lower, at r = 0.66, 0.64, and - 0.61, respectively (see Table 1). The SIDI showed a correlation of 0.77 with the CES-D. Reliability was assessed by the correlation between identical tests administered 24 hours apart and the Cronbach alpha measure of internal consistency Test-retest reliability was 0.92 for the SIDI and 0.89 for the CES-D. Cronbach's alpha (equivalent to the average correlation of items with each other) was 0.90 for the SIDI and 0.87 for the CES-D (see Table I)'. To assess the accuracy of the SIDI and CES-D as screening measures, a standard meth odo1ogy61718 was used to measure the level of agreement of the tests with the psychiatrist's diagnosis. For the purpose of this analysis, minor depression and major depression were combined into a single diagnosis of depressive disorder. Optimal cutoff scores were determined by the highest percentage of patients being correctly classified as depressed or not depressed. The optimal cutoff score of 17 for the SIDI resulted in the correct classification of 86% of the patients. The optimal cutoff of 26 for the CES-D resulted in the correct classification of 76% of the patients. Sensitivity (percentage of subjects diagnosed as depressed who had a score at or above the cutoff), specificity (percentage of subjects diagnosed as not depressed who had a score below the cutoff), and positive predictive value (percentage of subjects with scores at or above cutoff who received a depression diagnosis) were calculated for the optimal cutoff scores as well as for two additional recommended cutoff scores for the CES_D6 24 (see Table 2). f RESULTS The breakdown ofdiagnoses according to psychiatric classification was as follows: 33 patients with no depression (66%); 8 patients with minor depression (16%); and 9 patients with major depression (18%). When mild and major depression were combined into one category, the overall rate of depression was 34%. This rate is in the lower end of the range reported by other studies. 3-8 However, it should be noted that the present study excluded dysphasic patients, a factor that may account for the lower-than-average rate of depression. The standard used to assess criterion validity was the psychiatrist's diagnosis of level of depression (Le., no depression/minor/major depression). The SIDI correlated 0.70 with depression level, whereas the CES-D correlated 0.53 with depression level. Concurrent validity was assessed by the correlation between the two scales, the SIDI and the CES-D, and the rating scales that had been adminis136 f t VOLUME 4 • NUMBER 2 • SPRING 1996 Rybarczyk et al. DISCUSSION The data presented in this paper indicate that the SIDI is a valid and reliable screening instrument for poststroke depression. In fact, the SIDI scored higher than the widely used CES-D on all measures of criterion validity, concurrent validity, and reliability. For example, the SIDl correlated 0.70 with the criterion of the psychiatrist's diagnosis oflevel ofdepression, compared with 0.53 for the CES-D. Also, when minor and major depression were combined into one category to allow for sensitivity and specificity calculations, the SIDI proved to be a more accurate screening instrument than the CES-D (86% vs. 76% correctly classified). The SIDI outperformed the CES-D by the largest margin on specificity, which is a measure of false positives (94% vs. 82%). These preliminary findings indicate that the SIDI has advantages over the CES-D as both a measure for quantifying level of depression and as a measure for screening for the presence or absence of depressive disorder. A primary reason for developing and testing the SIDI was the low level ofvalidity found for the CES-D when it was applied to 66 stroke inpatients enrolled in the Stroke Data Bank study'' Interestingly; the low correlation of the CES-D with psychiatrist's diagnosis of depression in this study (r = 0.53) was identical to the correlation reported in the Stroke Data Base study The studies used comparable methods for psychiatric evaluation and the same three categories of diagnosis. However, the CES·D appears to have a much higher level of validity for stroke outpatients who are no longer in the acute stages of recovery The Stroke Data Base sample, which followed up patients after discharge from the hospital over a 2-year period, showed a much higher correlation between patients' CES-D scores and psychiatric diagnosis (0.95).6 A simiTHE AMERICAN JOURNAL OF GERIATRIC PSYCHIATRY lar study found a moderately high correlation (0.73) between CES-D score and psychiatric diagnosis among outpatients. I 8 Although a CES·D cutoff score of 16 has been suggested for use in the general population, our analysis found that the optimal cutoff for the present sample of stroke patients was 26. This was even higher than the optimal cutoff of 21 obtained in the Stroke Data Base study 6 This difference is probably accounted for by the fact that the Stroke Data Base study primarily involved stroke outpatients. Taken together, these studies suggest that the period of acute care and rehabilitation immediately after a stroke is likely to involve a greater number of false-positive answers to questions involving sleep disturbance, fatigue, and other somatic symptoms. Also, during the initial poststroke period some of the hallmarksymptoms of depression may be present as part of the normal grief response without the presence of a full depression. For example, one study using the Zung scale found that crying and sadness items had a very low correlation with total depression score among stroke patients in an inpatient rehabilitation program. 25 Supplementary analyses of our data did not support previous findings linking the presence of poststroke depression with gender,3,7 age,7 history of depression, 7 or hemisphere of lesion. 26 In our sample, depressed patients did not differ from nondepressed patients in terms of gender, age, race, education, cognitive status, prior history of strokes, history of depression, or hemisphere of lesion. The link with hemisphere of the lesion, in particular, has been a controversial issue in the literature.8 ,26 However, by excluding individuals with significant dysphasia, we may have biased our sample against finding a higher risk for depression in patients with left-hemisphere lesions. Also, we were unable to obtain a sufficient number of completed measures 137 Depression Screening and Stroke of physical functioning at the time of assessment to investigate the previously reported link between depression and lower functional status." A strength of the SIDI) and also a limitation, is the reliance on items asking the patient to compare current status to the period before the stroke or hospitalization. This strategy could result in a patient not answering in the affirmative for depressive symptoms that existed before the stroke. In effect, this limits the SIDI to the detection of depression secondary to stroke or hospitalization. One advantage of an instrument like the SIDI, with a YeslNo format) is that it is easier to administer and therefore less time-consuming than open-ended questions. The development of user-friendly screening instruments would contribute to improving on the current low rate of recognition and treatment observed for cases of poststroke depression.' The undertreatment of poststroke depression is unfortunate, given the research showing that sttmulanr" and antidepressant medtcarionst'' are effective with this population. Moreover, the issue of early detection and treatment has become even more significant with the introduction of new antidepressant medications (e.g., SSRIs) with fewer side effects than tricyclics and improved efficacy in cases of less severe depressions (e.g., minor depression) dysthymia). Although well-controlled studies have not been conducted) it is reasonable to assume that amelioration of depression in stroke patients facilitates the rehabilitation process and improves overall functioning. In summary; this initial study indicates that the SIDI has good validity and appears to have some advantages over the CES-Dfor both clinical and research purposes. A study replicating the present findings is needed to further support the validity of the SlOt The modest sample size of50 subjects and our reliance on the clinical judgment of a single psychiatrist for the validity criterion were limitations of the present study that could be addressed in a replication stud}'. Combining the current study with a replication study would also allow for an item analysis and the eventual elimination ofany items that diminish the validity of the scale. Finally; by substituting the word "hospitalization" for "stroke" throughout the test (e.g., "since your hospltalization"), the SIDI could be tested with hospitalized older adults with any medical diagnosis. Most of the problems outlined above regarding assessment of poststroke depression (e.g., overlap of medical symptoms with depression symptoms) are problems inherent in the assessment of geriatric inpatients with any medical illness. Rates of depression among geriatric medical inpatients) in general) are nearly as high as those found among stroke inpatients (i.e., 34%-45%).9,10 Therefore) a valid and reliable screening measure for depression in medically ill elderly patients would also have considerable clinical utility References 1. U.S. Department of Health and Human Services: National Hospital Discharge Survey: Annual Summary, 1991. (PHS) 93·1775, Washington, . DC, 1991 2. Haut AE.Rybarczyk B, Nicholasj], et al: Negative bias against older women among rehabilitation professionals. Paper presented at the annual meeting of the Society for Behavioral Medicine, San Diego. CA, 1994 138 3. Bacher Y, Korner-Bitensky N, Mayo Nt et al: A longitudinal study of depression among stroke patients participating in a rehabilitation program. Canadian Journal of Rehabilitation 1990; 4:27-37 4. Sinyor 0, Amato P, Kaloupek DG, et al: Poststroke depression: relationships to functional impairment, coping strategies, and rehabilltation outcome. Stroke 1986; 17:1102-1107 VOLUME 4 • NUMBER 2 • SPIUNG 1996 Rybarczyk et al. 5. Shinar D, Gross CR, Price TR, et al: Screening for depression in stroke patients: the reliability and validity of the Center for Epidemiologic Studies Depression Scale. Stroke 1986; 17:241-245 6. Parikh RM, Eden DT, Price TR, et al: The sensitivity and specificity of the Center for Epidemiologic Studies Depression Scale in screening for poststroke depression. Int J Psychiatry Med 1988; 18:169-181 7. Morris PLP, Robinson RG, Rapheal B, et al: The relationship between risk factors for affective disorder and poststroke depression in hospitalized stroke patients. Aust N ZJ Psychiatry 1992; 26:208-217 8. Gordon WA, Hibbard MR, Egelko S, et al: Issues in the diagnosis of poststroke depression. Rehabilitation Psychology 1991; 36:71-87 9. Rapp SR, Parisi SA, Walsh DA, et a1: Detecting depression in elderly medical patients. J Consult Clin Psychol1988; 56:509-513 10. Lichtenberg PA, Gibbons TA, Nanna M, et al: Physician detection of depression in medically ill elderly. Clinical Gerontologist 1993; 13:81-90 11. Parikh RM, Robinson RG, Lipsey JR, et al: The impact of poststroke depression on recovery in activities of daily living over a 2-year follow-up. Arch Neurol 1990; 47:785-789 tl2.Morris PLP, Robinson RG, Andrzewski P, et al: Association of depression with 10-year poststroke mortality. AmJ Psychiatry; 150:124-129 13. Rusin MJ: Stroke rehabilitation: a geropsychological perspective. Arch Phys Med Rehabil 1990i 71:914-922 14.Yesavage JA, Brink TL, Rose TL, et al: Development of a geriatric depression screening scale: a preliminary report. J Psychiatr Res 1983; 17:37-49 15. Radloff 15: The CES-D Scale: a self-report depression scale for research in the general population. Applied Psychological Measurement 1977; 1:384-401 16.Callahan LF, Kaplan MR, Pincus T: The Beck Depression Inventory, Center for Epidemlologi- cal Studies Depression Scale (CES-D),and General Well-Being Schedule Depression Subscale in rheumatoid arthritis. Arthritis Care and Research 1991; 4:3-11 17. Turk DC, Okifujl A: Detecting depression in chronic pain patients: adequacy of self-reports. Behav Res Ther 1994; 32:9-16 18.Agrell B, Dehlln 0: Comparison of six depression rating scales in geriatric stroke patients. Stroke 1989; 20:1190-1194 19. Plutchik R, Conte HR: Self-report scales for the measurement of depression. Psychiatric Annals 1989; 19:367-371 20. Hamilton M: A rating scale for depression. J Neurol Neurosurg Psychiatry 1960; 12:56-62 21.HedlungJL, Vieweg BW: The Hamilton Rating Scale for Depression: a comprehensive review. J Oper Psychiatry 1979; 10:149-162 22. EndicottJ, Spitzer RL:A diagnostic interview for affective disorders and schizophrenia. Arch Gen Psychiatry 1978; 35:837-844 23. Kiernan RJ, Mueller J, Langston)W, et al: The Neurobehavioral Cognitive Status Examination: a brief but differentiated approach to cognitive assessment. Ann Intern Med 1987; 107:481-485 24. Van Natta CT: Current medication use and symptoms of depression in a general population. Am J Psychiatr 1978; 135:1036-1039 25. Rybarczyk BD, Martelli M: Zung depression items and subscale analysis for stroke patients enrolled in a rehabilitation program. Paper presented at the annual meeting of the Gerontological Society of America, Washington, DC, 1987 26. Robinson, RG, Szetela B: Mood changes following left hemisphere brain injury. Ann Neural 1981; 9:447-453 27. Lingam VR, Lazarus LW, Groves L, et al: Methylphenidate in treating poststroke depression. J Clin Psychiatry 1988; 49: 151-153 28. Lipsey JR, Robinson RG, Pearlson GD, et al: Nortriptyline treatment of poststroke depression. Lancet 1984; 1 :297-300 THE AMERICAN JOURNAL OF GERIATIUC PSYCHIATRY 139