INT’L. J. PSYCHIATRY IN MEDICINE, Vol. 22(2)183-195,1992
A BRIEF DEPRESSION SCALE FOR USE IN THE MEDICALLY ILL*
HAROLD G. KOENIG, MD, MHS
DAN G. BLAZER, MD, PHD
HARVEY J. COHEN, MD
KEITH G. MEADOR, MD, MPH RON WESTLUND, MS
Duke University Medical Center Geriatric Research Education and Clinical Center VA Medical Center Durham, North Carolina
Duke University Medical Center Durham, North Carolina
ABSTRACT
Objective: Using items from two existing depression scales, we have sought to develop a brief self-rated instrument for detecting major depressive disorder (M.D.D.) in medically ill, hospitalized patients. Method: Forty-two items from the Geriatric Depression Scale (G.D.S.) and Carroll Depression Scale were administered to 559 men under age 40 or over age 70 consecutively admitted to the hospital. Eighty-two M.D.D.’s were diagnosed in this group by structured psychiatric interview. After eliminating 12 items confounded by medical illness, 11items were selected using regression analysis, correlation with the total score, and factor analysis. The 11-item scale includes an assessment of the five DSM-111-R criteria for M.D.D. which are least confounded by medical illness (mood, suicidal intent, guilt or worthlessness, concentration, and psychomotor agitation). The scale was then tested in 78 medical inpatients who were later assessed for M.D.D. using a structured psychiatric interview. Results: Ten out of twelve M.D.D.’s were identified (83% sensitivity) and depression excluded in 51 of 66 non-depressed subjects (77% specificity) (compared with 82% sensitivity and 76% specificity for the 30-item G.D.S.). Scores on the 11-item scale were also correlated with the G.D.S. (.92), the Zung Depression Scale (.58), and the C.E.S.-D (.67). *Funding provided by Mellon Foundation, Sandoz Pharmaceutical Corporation, Center for the Study of Aging and Human Development, Duke University Medical Center (grant #AGoo371), and Clinical Research Center for the Study of Depression in the Elderly, NIMH (grant # MH40159). 183 0 1992, Baywood Publishing Co., Inc.
doi: 10.2190/M1F5-F40P-C4KD-YPA3 http://baywood.com
184 / KOENIG ETAL.
Conclusion: The 11-item scale is a practical tool for clinicians who screen patients for depression and for investigators who need a brief measure of depression in studies involving medical inpatients. (Intl J. Psychiatry in Medicine 22:183-195, 1992)
Key Words: major depression, medical inpatients, elderly, geriatrics, screening, self-rated depression scale, test characteristics, hospital, institutionalized
BACKGROUND
Depression is the most common reversible psychiatric disorder in medical inpatients. The prevalence of depressive symptoms and aisorders in this setting surpasses that in the community by a factor of at least 10-fold [l, 21. Depressive disorders may adversely affect survival [3-71, length of hospital stay [6], compliance with therapy [8,9], ability to care for self, and quality of life. Despite this, the detection rate of depressive disorders in medical inpatients has been notably poor [lo, 111. When made aware of this possibility of depression through feedback from depression scores, however, physicians are more likely to pay attention to this mental health problem [12]. Interest has centered on the possible use of self-rated depression scales to improve case-finding in this setting. Lack of time is the number one reason given by physicians for their failure to screen for depression using these scales [13]. A search, then, has begun for a brief, valid screening measure that is acceptable to medical inpatients who are often quite ill, have a limited attention span, and lack the physical energy to plod through an extended questionnaire. Unfortunately, most depression scales today such as the Beck, Zung, or CES-D have a relatively complex set of response options that may be confusing to sick or elderly patients and are unwieldy for either self or assisted administration. Even the Geriatric Depression Scale (GDS), originally designed and validated in healthy elders [14, 151, has a number of items confounded by symptoms of physical illness and may be burdensome for some frail patients because of its length. Thus, there is need for a brief, easily administered and understood depression scale that is capable of detecting serious clinical depression in medical inpatients and differentiatingthis group from patients with symptoms due to physical illness. In two recent epidemiological studies of depression in medically ill hospitalized patients [2, 161, the GDS and the Brief Carroll Depression Scale (BCDRS) [17] were administered to 559 inpatients, who also underwent a structured psychiatric evaluation to determine the presence or absence of major depressive disorder. From the 42 items on the combined scales, we have identified 11 items which most strongly predict major depression and which are least likely to be confounded by physical illness. These 11 items were then amalgamated into a scale and validated against a standard psychiatric interview in a separate set of patients.
DEPRESSION SCALE / 185
DEVELOPMENTOFTHE SCALE Between January 1987 and January 1989, two epidemiological studies of depression in medical inpatients took place at the Durham VA Medical Center [2, 161. During this time, 559 men under age forty (22%) or over age seventy (78%) admitted to medical or neurological services were evaluated for depression with two self-rated depression scales, the GDS and BCDRS, and by structured psychiatric interview. Because the original design of the study was to examine age differences in rate of depression between younger and older patients, those age forty to sixty-four were excluded. The methods, sociodemographic and health characteristics of these patients have already been described [2, 171. Eligibility required that patients not be admitted to intensive care settings or have severe medical illness or communication problems precluding psychological testing. Participants were required to score 15 or higher on the Mini-Mental State Exam (MMSE) [18]. There were eighty-two diagnoses of major depression made in patients who also received the self-rated depression scales. Diagnoses were made by a psychiatrist, blinded to results of the self-rated scales, who reviewed data collected during a forty-five to ninety minute Structured psychiatric interview using the affective disorders section of the NIMH Diagnostic Interview Schedule [19], Hamilton Depression Rating Scale [20], and other mental and physical health data. In making the diagnosis of major depressive disorder, the psychiatrist excluded symptoms judged to be clearly secondary to medical illness.
Removal of Confounding Items The 30 items of the GDS and 12 items of the BCDRS were examined by three of the authors (HJC, DGB, HGK) for questions which might be confounded by physical illness. Each examiner was asked to choose ten to fifteen of the 42 items most likely confounded by health; each was blinded of the others’ choices. If agreement was obtained by at least two of the three examiners on an item, then it was removed from the list. There were fourteen items excluded by each of the examiners: six were unanimous exclusions and six were agreed upon by two examiners. Hence, 12 questions were removed from the list (Table 1) leaving 30 items for further evaluation (Table 1).These 30 items represented those questions least confounded by the physical health of the respondent (Table 2).
Selection of Final Questions The responses of the 559 patients to these 30 items were then subjected to three methods of analysis: 1)entry into a logistic regression model with major depression as the dependent variable, 2) correlation (Pearson) between each of the items and the total 30-item scale score, and 3) principle components analysis. A backwards stepwise method of variable elimination was used in the logistic regression analysis; in other words, items which contributed insignificantly
186 / KOENIG ET AL.
Table 1. Excluded Items Based on Probable Confounding by Physical Illness 1. Have you dropped many of your activities and interests? 2. Do you prefer to stay home, rather than going out and doing new things? 3. Is it hard for you to get started on new projects? 4. Do you feel full of energy? 5. Do you have trouble concentrating? 6. Are you basically satisfied with your life? 7. I am losing weight. 8. I have dropped many of my activities and interests. 9. My sleep is restless and disturbed. 10. I still enjoy my meals as much as usual. 11. I get hardly anything done lately. 12. I am exhausted much of the time. Note: First six items are from the GDS, last six are from the BCDRS.
(p > .05) to the likelihood of major depression (controlling for other items) were successively dropped from the model. All subjects responding to at least 75 percent of all items on the original scales were included in the model. With all 30 items included in the model, 36 percent of the variance in the likelihood of major depression was explained. The final model contained five variables that explained 34 percent of the variance (94% of the total) (Table 3). In other words, with knowledge of the patient’s responses to these five items only, the chances of their having a major depression could be predicted almost as well (94% as well) as if responses to all 30 items were known. Next, each item of the 30-item scale was correlated with the total score obtained by summing responses to all 30 items in the scale. This enabled us to see which items were most important in determining the score achieved by patients on this scale-and give us further data about which questions might be dropped from the scale without losing information. Table 4 gives the 10 strongest correlations. Finally, a principle components analysis was performed using a varimax rotation technique. This statistical technique identifies groups of items in the scale that correlate highly with one another and thus appear to be measuring specific components of depression. A depression scale may tap several aspects of depression-anxiety, hopelessness, activity level, concentration, and so forth. This analysis picks out how many such components of depression are being measured and then which items in the scale best measure each component or factor. Using this technique, seven distinct factors were identified with 2 to 11items per factor (Table 5). The final 11 items were chosen as follows. First, all five significant predictors of major depression in the regression model were included (#2, 9, 11, 13, 30).
DEPRESSION SCALE 1 187
Table 2. Items Remaining After Excluding Questions Most Confounded by Physical Illness 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.
Do you feel that your life is empty? Do you often get bored? Are you hopeful about the future? Are you bothered by thoughts you can’t get out of your head? Are you in good spirits most of the time? Are you afraid that something bad is going to happen to you? Do you feel happy most of the time? Do you often feel helpless? Do you often get restless and fidgety? Do you frequently worry about the future? Do you feel you have more problems with memory than most? Do you think it is wonderful to be alive now? Do you often feel downhearted and blue? Do you feel pretty worthless the way you are now? Do you worry a lot about the past? Do you find life very exciting? Do you feel your situation is hopeless? Do you think that most people are better off than you are? Do you frequently get upset over little things? Do you frequently feel like crying? Do you enjoy getting up in the morning? Do you prefer to avoid social gatherings? Is it easy for you to make decisions? Is your mind as clear as it used to be? It must be obvious that I am disturbed and agitated. I can concentrate easily when reading the papers. I am miserable or often feel like crying. I often wish Iwere dead. Ifeel worthless and ashamed about myself. Ifeel in good spirits.
Next included were three of the seven questions having the highest correlation with the 30-item total score (#7,8,29) and not already chosen from the regression analysis (#2,13,30); the remaining item (#5) was not included because its content overlapped heavily with another item already chosen from the regression model (#30). The final three questions for the 11-item scale were chosen on the basis of their clinical and diagnostic value in the detection of depression in this population. Because of concern over the risk of suicide in depressed patients, an item
188 / KOENIG ET AL.
Table 3. Logistic Regression Model of the Five Strongest Predictors of Major Depressive Disorder Beta (SE)
F Score
p Value
0.31 (0.05)
39.7
A BRIEF DEPRESSION SCALE FOR USE IN THE MEDICALLY ILL*
HAROLD G. KOENIG, MD, MHS
DAN G. BLAZER, MD, PHD
HARVEY J. COHEN, MD
KEITH G. MEADOR, MD, MPH RON WESTLUND, MS
Duke University Medical Center Geriatric Research Education and Clinical Center VA Medical Center Durham, North Carolina
Duke University Medical Center Durham, North Carolina
ABSTRACT
Objective: Using items from two existing depression scales, we have sought to develop a brief self-rated instrument for detecting major depressive disorder (M.D.D.) in medically ill, hospitalized patients. Method: Forty-two items from the Geriatric Depression Scale (G.D.S.) and Carroll Depression Scale were administered to 559 men under age 40 or over age 70 consecutively admitted to the hospital. Eighty-two M.D.D.’s were diagnosed in this group by structured psychiatric interview. After eliminating 12 items confounded by medical illness, 11items were selected using regression analysis, correlation with the total score, and factor analysis. The 11-item scale includes an assessment of the five DSM-111-R criteria for M.D.D. which are least confounded by medical illness (mood, suicidal intent, guilt or worthlessness, concentration, and psychomotor agitation). The scale was then tested in 78 medical inpatients who were later assessed for M.D.D. using a structured psychiatric interview. Results: Ten out of twelve M.D.D.’s were identified (83% sensitivity) and depression excluded in 51 of 66 non-depressed subjects (77% specificity) (compared with 82% sensitivity and 76% specificity for the 30-item G.D.S.). Scores on the 11-item scale were also correlated with the G.D.S. (.92), the Zung Depression Scale (.58), and the C.E.S.-D (.67). *Funding provided by Mellon Foundation, Sandoz Pharmaceutical Corporation, Center for the Study of Aging and Human Development, Duke University Medical Center (grant #AGoo371), and Clinical Research Center for the Study of Depression in the Elderly, NIMH (grant # MH40159). 183 0 1992, Baywood Publishing Co., Inc.
doi: 10.2190/M1F5-F40P-C4KD-YPA3 http://baywood.com
184 / KOENIG ETAL.
Conclusion: The 11-item scale is a practical tool for clinicians who screen patients for depression and for investigators who need a brief measure of depression in studies involving medical inpatients. (Intl J. Psychiatry in Medicine 22:183-195, 1992)
Key Words: major depression, medical inpatients, elderly, geriatrics, screening, self-rated depression scale, test characteristics, hospital, institutionalized
BACKGROUND
Depression is the most common reversible psychiatric disorder in medical inpatients. The prevalence of depressive symptoms and aisorders in this setting surpasses that in the community by a factor of at least 10-fold [l, 21. Depressive disorders may adversely affect survival [3-71, length of hospital stay [6], compliance with therapy [8,9], ability to care for self, and quality of life. Despite this, the detection rate of depressive disorders in medical inpatients has been notably poor [lo, 111. When made aware of this possibility of depression through feedback from depression scores, however, physicians are more likely to pay attention to this mental health problem [12]. Interest has centered on the possible use of self-rated depression scales to improve case-finding in this setting. Lack of time is the number one reason given by physicians for their failure to screen for depression using these scales [13]. A search, then, has begun for a brief, valid screening measure that is acceptable to medical inpatients who are often quite ill, have a limited attention span, and lack the physical energy to plod through an extended questionnaire. Unfortunately, most depression scales today such as the Beck, Zung, or CES-D have a relatively complex set of response options that may be confusing to sick or elderly patients and are unwieldy for either self or assisted administration. Even the Geriatric Depression Scale (GDS), originally designed and validated in healthy elders [14, 151, has a number of items confounded by symptoms of physical illness and may be burdensome for some frail patients because of its length. Thus, there is need for a brief, easily administered and understood depression scale that is capable of detecting serious clinical depression in medical inpatients and differentiatingthis group from patients with symptoms due to physical illness. In two recent epidemiological studies of depression in medically ill hospitalized patients [2, 161, the GDS and the Brief Carroll Depression Scale (BCDRS) [17] were administered to 559 inpatients, who also underwent a structured psychiatric evaluation to determine the presence or absence of major depressive disorder. From the 42 items on the combined scales, we have identified 11 items which most strongly predict major depression and which are least likely to be confounded by physical illness. These 11 items were then amalgamated into a scale and validated against a standard psychiatric interview in a separate set of patients.
DEPRESSION SCALE / 185
DEVELOPMENTOFTHE SCALE Between January 1987 and January 1989, two epidemiological studies of depression in medical inpatients took place at the Durham VA Medical Center [2, 161. During this time, 559 men under age forty (22%) or over age seventy (78%) admitted to medical or neurological services were evaluated for depression with two self-rated depression scales, the GDS and BCDRS, and by structured psychiatric interview. Because the original design of the study was to examine age differences in rate of depression between younger and older patients, those age forty to sixty-four were excluded. The methods, sociodemographic and health characteristics of these patients have already been described [2, 171. Eligibility required that patients not be admitted to intensive care settings or have severe medical illness or communication problems precluding psychological testing. Participants were required to score 15 or higher on the Mini-Mental State Exam (MMSE) [18]. There were eighty-two diagnoses of major depression made in patients who also received the self-rated depression scales. Diagnoses were made by a psychiatrist, blinded to results of the self-rated scales, who reviewed data collected during a forty-five to ninety minute Structured psychiatric interview using the affective disorders section of the NIMH Diagnostic Interview Schedule [19], Hamilton Depression Rating Scale [20], and other mental and physical health data. In making the diagnosis of major depressive disorder, the psychiatrist excluded symptoms judged to be clearly secondary to medical illness.
Removal of Confounding Items The 30 items of the GDS and 12 items of the BCDRS were examined by three of the authors (HJC, DGB, HGK) for questions which might be confounded by physical illness. Each examiner was asked to choose ten to fifteen of the 42 items most likely confounded by health; each was blinded of the others’ choices. If agreement was obtained by at least two of the three examiners on an item, then it was removed from the list. There were fourteen items excluded by each of the examiners: six were unanimous exclusions and six were agreed upon by two examiners. Hence, 12 questions were removed from the list (Table 1) leaving 30 items for further evaluation (Table 1).These 30 items represented those questions least confounded by the physical health of the respondent (Table 2).
Selection of Final Questions The responses of the 559 patients to these 30 items were then subjected to three methods of analysis: 1)entry into a logistic regression model with major depression as the dependent variable, 2) correlation (Pearson) between each of the items and the total 30-item scale score, and 3) principle components analysis. A backwards stepwise method of variable elimination was used in the logistic regression analysis; in other words, items which contributed insignificantly
186 / KOENIG ET AL.
Table 1. Excluded Items Based on Probable Confounding by Physical Illness 1. Have you dropped many of your activities and interests? 2. Do you prefer to stay home, rather than going out and doing new things? 3. Is it hard for you to get started on new projects? 4. Do you feel full of energy? 5. Do you have trouble concentrating? 6. Are you basically satisfied with your life? 7. I am losing weight. 8. I have dropped many of my activities and interests. 9. My sleep is restless and disturbed. 10. I still enjoy my meals as much as usual. 11. I get hardly anything done lately. 12. I am exhausted much of the time. Note: First six items are from the GDS, last six are from the BCDRS.
(p > .05) to the likelihood of major depression (controlling for other items) were successively dropped from the model. All subjects responding to at least 75 percent of all items on the original scales were included in the model. With all 30 items included in the model, 36 percent of the variance in the likelihood of major depression was explained. The final model contained five variables that explained 34 percent of the variance (94% of the total) (Table 3). In other words, with knowledge of the patient’s responses to these five items only, the chances of their having a major depression could be predicted almost as well (94% as well) as if responses to all 30 items were known. Next, each item of the 30-item scale was correlated with the total score obtained by summing responses to all 30 items in the scale. This enabled us to see which items were most important in determining the score achieved by patients on this scale-and give us further data about which questions might be dropped from the scale without losing information. Table 4 gives the 10 strongest correlations. Finally, a principle components analysis was performed using a varimax rotation technique. This statistical technique identifies groups of items in the scale that correlate highly with one another and thus appear to be measuring specific components of depression. A depression scale may tap several aspects of depression-anxiety, hopelessness, activity level, concentration, and so forth. This analysis picks out how many such components of depression are being measured and then which items in the scale best measure each component or factor. Using this technique, seven distinct factors were identified with 2 to 11items per factor (Table 5). The final 11 items were chosen as follows. First, all five significant predictors of major depression in the regression model were included (#2, 9, 11, 13, 30).
DEPRESSION SCALE 1 187
Table 2. Items Remaining After Excluding Questions Most Confounded by Physical Illness 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30.
Do you feel that your life is empty? Do you often get bored? Are you hopeful about the future? Are you bothered by thoughts you can’t get out of your head? Are you in good spirits most of the time? Are you afraid that something bad is going to happen to you? Do you feel happy most of the time? Do you often feel helpless? Do you often get restless and fidgety? Do you frequently worry about the future? Do you feel you have more problems with memory than most? Do you think it is wonderful to be alive now? Do you often feel downhearted and blue? Do you feel pretty worthless the way you are now? Do you worry a lot about the past? Do you find life very exciting? Do you feel your situation is hopeless? Do you think that most people are better off than you are? Do you frequently get upset over little things? Do you frequently feel like crying? Do you enjoy getting up in the morning? Do you prefer to avoid social gatherings? Is it easy for you to make decisions? Is your mind as clear as it used to be? It must be obvious that I am disturbed and agitated. I can concentrate easily when reading the papers. I am miserable or often feel like crying. I often wish Iwere dead. Ifeel worthless and ashamed about myself. Ifeel in good spirits.
Next included were three of the seven questions having the highest correlation with the 30-item total score (#7,8,29) and not already chosen from the regression analysis (#2,13,30); the remaining item (#5) was not included because its content overlapped heavily with another item already chosen from the regression model (#30). The final three questions for the 11-item scale were chosen on the basis of their clinical and diagnostic value in the detection of depression in this population. Because of concern over the risk of suicide in depressed patients, an item
188 / KOENIG ET AL.
Table 3. Logistic Regression Model of the Five Strongest Predictors of Major Depressive Disorder Beta (SE)
F Score
p Value
0.31 (0.05)
39.7
