- Email: [email protected]
Cincinnati criteria for mixed mania and suicidality in patients with acute mania
Cincinnati Criteria for Mixed Mania and Suicidality in Patients With Acute Mania Tetsuya Sato, Ronald Bottlender, Akira Tanabe, and Hans-Ju¨rgen Mo¨ller The association between suicidality and diagnoses of mixed mania, as defined using both DSM-IV and Cincinnati criteria, was studied in 576 consecutive manic inpatients. Of the whole sample, 51 (8.9%) had suicidal ideation and 13 (2.3%) attempted suicide during the index episode. Suicidality was significantly more frequent in patients with a diagnosis of mixed mania, whether the diagnosis was made by DSM-IV or Cincinnati criteria. A multiple logistic regression analysis revealed that an additive combination of a diagnosis of mixed mania, the depression severity, and the Global Assessment of Functioning (GAF) score was significant in predicting suicidal ideation, when using the DSM-IV criteria. A diagnosis of mixed mania alone was significant in a similar analysis, when using the Cincinnati criteria. The adjusted odds ratio for a diagnosis of mixed mania to having suicidality was much higher when using the latter criteria (4.0 v 14.0). A subsequent logistic regression analysis indicated that the Cincinnati mixed mania alone, rather than an additive combination of the DSM-IV mixed mania and
the depression severity, achieved the most appropriate prediction of suicidal ideation in the sample. These findings did not differ, even when suicidality was defined as having a suicide attempt during the index episode. Our finding that suicidality was more strongly associated with Cincinnati mixed mania than with DSM-IV mixed mania is probably due to that suicidal patients who do not meet DSM-IV criteria for mixed mania are classified into mixed mania, or/and that the depressive syndrome, related to suicidality, is more appropriately assessed among manic patients, when using the Cincinnati criteria. There was no evidence that marital status, employment, a lifetime history of alcohol or substance abuse, or a history of suicide attempts before the index episode was significantly associated with suicidality in the sample. Manic patients with suicidality may have a greater severity of residual depressive symptoms at discharge. © 2004 Elsevier Inc. All rights reserved.
S
ideation in patients simultaneously meeting DSMIII-R criteria for both manic and major depressive episodes than in pure manic patients. Interestingly, Strakowski et al.3 provided evidence that the severity of current depressive symptoms, rather than the presence of a full depressive syndrome per se, is associated with suicidal ideation among manic patients, suggesting that narrow definitions of mixed mania may not play a definite role in predicting suicidality in these patients. DSM-III-R and DSM-IV adopt a narrow definition of mixed mania. However, there are several studies indicating that broader definitions of mixed states result in a more meaningful subtyping of mixed mania in terms of demographic variables,7-11 natural course,7-9,11,12 and specific treatment responses.13,14 The association between a broader definition of mixed mania and suicidality has not been well studied. One study reported that suicidal ideation is not infrequent in manic patients, even when full major depression does not accompany mania.5,6 However, it is unclear whether or not a broader definition of mixed mania plays a more definite role in predicting suicidality in manic patients, as compared with a narrow definition of mixed mania. McElroy et al.4,11 proposed one of well-known broad definitions of mixed mania, the Cincinnati
UICIDE RISK in bipolar patients is generally high. An exhaustive review by Goodwin and Jamison1 showed that 18.9% of deaths in bipolar patients were due to suicide. Although previous studies on suicidality in bipolar patients tended to be restricted to depressive episodes, several recent studies have emphasized that mixed states (or mixed mania), as defined as a simultaneous admixture of both depressive and manic syndromes, may also mediate the high suicide risk in bipolar patients.2-6 Using a narrow definition of mixed mania that requires patients to simultaneously meet both full manic and depressive syndromes, Dilsaver et al.2 reported that 54.5% of patients with mixed mania had suicidal ideation, in contrast to 2.0% of patients with pure mania. Strakowski et al.3 also found a significantly higher frequency of suicidal From the Psychiatrische Klinik und Poliklinik, Ludwig-Maximilians-Universita¨t, Mu¨nchen, Germany. T.S. is supported by the Alexander von Humboldt Foundation. Address reprint requests to Tetsuya Sato, M.D., Ph.D., Psychiatrische Klinik und Poliklinik, Ludwig-MaximiliansUniversita¨t Mu¨nchen, Nussbaumstrasse 7, D-80336 Munich, Germany. © 2004 Elsevier Inc. All rights reserved. 0010-440X/04/4501-0015$30.00/0 doi:10.1016/S0010-440X(03)00145-7 62
Comprehensive Psychiatry, Vol. 45, No. 1 ( January/February), 2004: pp 62-69
SUICIDALITY AND MIXED MANIA
criteria, which require the presence of three or more symptoms of major depression during a full manic episode. The present study examined the association between mixed mania and suicidality in 576 hospitalized patients with acute mania. For diagnosing mixed mania, both Cincinnati and DSM-IV criteria were used. METHOD
Patients All patients who were hospitalized at the Psychiatric Hospital of the Ludwig-Maximilian University in Munich, Germany for any affective disorder during the period of 1980 to 1997 were considered as subjects in this study. Routine clinical diagnoses were made according to the International Classification of Diseases system (ICD-9 or ICD-10),15,16 but a broad range of 196 psychiatric and related somatic symptoms were, as part of the routine documentation at the hospital, systematically evaluated for all patients at both admission and discharge by using a standardized instrument (the AMDP system; see below), which allowed for precise diagnoses of DSM-IV manic episode, nonmixed and mixed, and mixed mania according to the Cincinnati criteria.4,9 The following inclusion criteria were used: (1) patients be diagnosed using the ICD system as currently having a manic or mixed episode (ICD-9 diagnosis of 296.0, 296.2, or 296.4; or ICD-10 diagnosis of F30 F31.0, F31.1, F31.2, F31.6, or F38.00); (2) patients be younger than 70 years; and (3) patients who met DSM-IV criteria for manic episode, nonmixed or mixed. Finally, 576 consecutive inpatients with DSM-IV manic episode, nonmixed or mixed, were included for the following analyses. All patients gave informed consent to be assessed using several instruments described below. Of the subjects, 292 (51%) were women, and 284 (49%) were men. Their mean age was 38.6 (SD 12.6) years. Their mean age at onset of first affective episode was 28.9 (SD 10.7) years. Fiftyeight patients (10.1%: 38 women and 20 men) met criteria for DSM-IV manic episode, mixed, while 88 patients (15.3%: 55 women and 33 men) met criteria for mixed mania according to Cincinnati criteria. The frequency of mixed mania as defined using DSM-IV or Cincinnati criteria was constant during the whole study period when the frequency was compared between three entry periods, 1980 to 1985, 1986 to 1990, and 1991 to 1997. Both diagnoses were significantly correlated in the whole sample ( ⫽ 0.67, P ⬍ .0001). The patients were treated with medications as clinically appropriate during the hospital stay. Before admission, the majority of the patients had received various medications including mood stabilizers (58%) and/or antipsychotics (36%). There was no significant association between presence of these medications and diagnoses of mixed mania.
Clinical Assessments The Association for Methodology and Documentation in Psychiatry (AMDP) system17 was used to assess psychiatric symptoms. The AMDP system is a comprehensive rating instrument, developed on the basis of German traditional descriptive psychopathology on functional psychoses; it is commonly used in most psychiatric institutes in German-speaking coun-
63
tries. Each psychiatric symptom of the AMDP system is scored from 0 (absent) to 3 (severe) with defined anchor statements by using a semistructured interview method. Several studies indicated moderate to high inter-rater agreements for most included symptoms.18,19 Rater-training sessions are performed regularly in our hospital to establish and maintain high inter-rater reliability of the instrument. Inter-rater reliability of all AMDP items in our hospital, calculated based on joint interviews by multiple raters of over 50 diagnostically diverse patients, ranged from 0.65 to 0.92 (analysis of variance intraclass correlation). Based on the AMDP ratings, several summary scales for depressive, manic, and paranoid-hallucinatory syndromes (the AMDP scores for depressive, manic, and paranoid-hallucinatory syndromes), which have been validated in large psychiatric samples,20-22 can be calculated by summing up the scores on 13 items (rumination, loss of feeling, loss of vitality, depressed mood, hopelessness, feeling of inadequacy, feeling of guilt, inhibition of drive, worse in the morning, interrupted sleep, shortened sleep, early waking, decreased appetite) for depressive syndrome, seven items (flight of idea, euphoria, exaggerated self-esteem, increased drive, motor restlessness, logorrhea, excessive social contact) for manic syndrome, and 13 items (delusional mood, delusional perceptions, sudden delusional ideas, delusional ideas, systematized delusions, delusional dynamics, delusions of reference, delusions of persecution, verbal hallucinations, bodily hallucinations, depersonalization, thought withdrawal, other feelings of alien influence) for paranoid-hallucinatory syndrome. The Global Assessment of Functioning (GAF) score was also assessed for all subjects at both admission and discharge. All subjects gave written informed consent to be assessed by using the instruments. Well-trained psychiatrists administered these two instruments. The presence of suicidality was defined in this study by using the highest score on the AMDP suicide item at admission. The item scores the severity of suicidality as follows: 0 (absent); 1 (mild): frequent thoughts that she/he would be better off dead; 2 (moderate): frequent thoughts about committing suicide or mental rehearsal of suicide; 3 (severe): suicide attempt during the index episode. The present study defines suicidality as a score equal to or higher than 1, since this definition appears to be equivalent to that used in previous similar studies.2,3
Statistical Analyses Clinical and demographic variables were compared between patients with and without suicidality by using chi-square test for categorical variables and Mann-Whitney tests for continuous variables. The frequencies of mixed mania as defined by the DSM-IV and Cincinnati criteria were also compared between the two groups. The nonparametric tests were used for all examined continuous variables, since the distributions of these variables were not normally distributed. In each of these univariate analyses, the significance level was adjusted at a P ⬍ .0025 (.05 divided by 20 tests) level by using Bonferroni’s method. An adjusted odds ratio for patients with mixed mania to having suicidal ideation was computed by using multiple logistic regression analysis. This analysis was performed for mixed mania as defined by the DSM-IV and Cincinnati criteria separately. Several other variables, which were, in the above univariate analyses, associated with suicidal ideation at a P ⬍ .10 level, were entered into the logistic analyses. A second logistic
64
SATO ET AL
Table 1. Demographic and Clinical Characteristics of Nonsuicidal and Suicidal Manic Patients
Gender, female, n (%) Age, years (SD) Marital status Single, n (%) Married or living together, n (%) Divorced, separated or bereaved, n (%) Unemployed, n (%) Mother tongue German, n (%) Non-German, n (%) Age of onset, years (SD) GAF score, baseline Alcohol or substance abuse, lifetime, n (%) Alcohol or substance abuse, index episode, n (%) Suicide attempts, before index episode, n (%) Suicide attempts, index episode, n (%) AMDP depressive syndrome mean score at admission (SD) AMDP manic syndrome mean score at admission (SD) AMDP paranoid-hallucinatory syndrome mean score at admission (SD) DSM-IV mixed mania, n (%) Mixed mania by the Cincinnati criteria, n (%) Duration of index admission, days, mean (SD) GAF score at discharge, mean (SD) AMDP depressive syndrome mean score at discharge (SD) AMDP manic syndrome mean score at discharge (SD) AMDP paranoid-hallucinatory syndrome mean score at discharge (SD)
Nonsuicidal Patients (n ⫽ 525)
Suicidal Patients (n ⫽ 51)
257 (49) 38.7 (12.5)
35 (69) 37.1 (13.3)
230 (44) 223 (42) 72 (14) 24 (5)
22 (43) 22 (43) 7 (14) 5 (10)
468 (89) 57 (11) 28.9 (10.6) 36.3 (13.1) 103 (20) 61 (12) 87 (17) 0 (0) 4.5 (3.3) 11.5 (4.0) 3.5 (4.9) 36 (7) 54 (10) 50.7 (34.1) 66.8 (13.7) 0.6 (1.2) 1.4 (2.3) 0.1 (0.5)
46 (90) 5 (10) 28.2 (11.6) 32.4 (11.1) 9 (18) 7 (14) 10 (20) 13 (25) 8.4 (6.0) 9.2 (4.5) 5.3 (5.6) 22 (43) 34 (67) 49.8 (32.0) 64.7 (12.5) 1.0 (1.4) 1.3 (2.2) 0.3 (0.9)
Group-Differences Statistics
7.20
P
.01
⫺1.96
Not tested ⫺4.85 ⫺3.62 ⫺2.48 .01 67.57 114.16 ⫺1.68 ⫺3.23
.09
⫺2.23
.03
NOTE. Chi-square values with df ⫽ 1 were used for categorical variables, and z values were used for continuous variables. Marital status was analyzed by using a chi-square value with df ⫽ 2. The significance level recommended is P ⬍ .0025 (.05 divided by 20 tests).
regression analysis, with the presence of suicidal ideation being used as the dependent variable, was then performed, in which the status of mixed mania as diagnosed by DSM-IV, the severity of depressive syndrome as measured by the AMDP score for depressive syndrome, and the status of mixed mania as diagnosed by the Cincinnati criteria were sequentially entered as independent variables in this order. An SPSS software (release 7.0)23 was used for statistical analyses. All statistical statements in the present study were two-tailed.
RESULTS
Of the whole sample, 51 patients (8.9%; 35 women and 16 men) had suicidal ideation at admission. Of these patients, 13 patients (2.6%; nine women and four men) attempted suicide during the index episode. The frequency of patients with suicide ideation was fairly constant during the whole study period when the frequency was compared between three entry periods, 1980 to 1985, 1986 to 1990, and 1991 to 1997. The presence of suicide ideation was not significantly associated with medications of mood stabilizers or antipsychotics before patients’ admission.
Table 1 shows demographic and clinical variables in patients with and without suicidal ideation. After Bonferroni’s correction, the severity of depressive and manic syndromes at admission, the status of mixed mania as diagnosed by both DSM-IV and Cincinnati criteria, and the severity of depressive syndrome at discharge significantly differentiated the two groups. The AMDP scores for depressive and manic syndromes were significantly higher and lower in patients with suicidality than in patients without suicidality, respectively. Mixed mania as diagnosed by DSM-IV was significantly frequent in patients with suicidality (43%) than in patients without suicidality (7%). The frequency of mixed mania as diagnosed by the Cincinnati criteria was also significantly higher in patients with suicidal ideation (67%) than in patients without suicidal ideation (10%), but the computed chi-square value for the Cincinnati criteria was much greater than that for DSM-IV criteria in these analyses. Although there was no evidence
SUICIDALITY AND MIXED MANIA
65
Table 2. Adjusted Odds Ratio for Differently Defined Mixed Mania in Predicting Suicidality, as Computed by Multiple Logistic Regression Analyses Diagnostic Criteria for Mixed Mania Entered Into Multiple Logistic Regression Analyses Mix Mania by DSM-IV Model Improvement, 2 (df ⫽ 7) 66.50, P ⫽ .0000 AIC ⫽ 293.0
Mixed Mania by Cincinnati Criteria Model Improvement, 2 (df ⫽ 7) 92.01, P ⫽ .0000 AIC ⫽ 269.5
Odds Ratio (95% CI)
Odds Ratio (95% CI)
1.79 (0.91-3.52) 1.66 (1.04-2.67)* 0.91 (0.60-1.39) 0.93 (0.81-1.07) 1.07 (0.90-1.28) 1.43 (1.06-1.92)* 3.98 (1.60-9.95)†
1.94 (0.94-4.01) 1.48 (0.93-2.35) 1.01 (0.66-1.54) 0.93 (0.81-1.07) 1.14 (0.95-1.38) 1.28 (0.94-1.73) 13.86 (5.84-33.74)†
Gender AMDP depressive syndrome, baselinea AMDP manic syndrome, baselinea Interaction of depressive and manic syndromes AMDP paranoid-hallucinatory syndromea GAF score, baselineb Diagnosis of mixed states a
Odds ratios and their 95% confidence intervals (CI) were calculated for a 3-point higher score. Odds ratios and their 95% confidence intervals were calculated for a 10-point lower score. Significant odds ratios at *P ⬍ .05 and †P ⬍ .0001, respectively. Model improvements were calculated from the full model.
b
that patients with suicidality required a longer hospitalization, these patients had a significantly more severe residual depressive syndrome at discharge. Table 2 demonstrates adjusted odds ratios for the status of mixed mania. The variables that were associated with suicidality at a P ⬍ .05 in Table 1, such as gender, the severity of depressive, manic, and paranoid-hallucinatory syndromes, and the GAF score at admission, were entered as independent variables into the analyses, together with status of mixed mania. Interaction of manic and depressive syndromes was also entered, since a combination of these psychiatric syndromes has been suggested to be specifically associated with the presence of suicidality among manic patients.1 When the DSM-IV criteria were used for diagnosing mixed mania, the model improvement was highly significant (2 ⫽ 66.5, df ⫽ 7, P ⬍ .0001), where the depression severity, GAF score, and the status of mixed mania were significant in predicting suicidality. Subjects having a 3-point higher score on the AMDP depressive syndrome and a 10-point lower GAF score were 1.7 times and 1.4 times more likely to have suicidality, respectively. The adjusted odds ratio for DSM-IV mixed mania was 4.0. When using the Cincinnati criteria for diagnosing mixed mania, the model improvement was much higher (2 ⫽ 92.0, df ⫽ 7, P ⬍ .0001). In this model, only the diagnosis of mixed mania by the Cincinnati criteria was significant in predicting suicidality. A large increase in the adjusted odds ratio for status of mixed mania was observed
(14.0 from 4.0), suggesting that the Cincinnati criteria, rather than DSM-IV criteria, provided a more appropriate subtyping of mixed mania in terms of predicting suicidality. This suggestion was confirmed by the final logistic regression analysis, where the DSM-IV definition of mixed mania, the depression severity, and the Cincinnati definition of mixed mania were entered into the analysis sequentially. The first step including the DSM-IV mixed mania alone produced a significant model improvement from the null model (2 ⫽ 44.1, df ⫽ 1, P ⬍ .0001). Akaike Information Criterion (AIC),24 which assesses the goodness of fit of the model, was 304.5. In this model, the DSM-IV diagnosis of mixed mania was significant in predicting suicidality (Wald statistic ⫽ 49.6, df ⫽ 1, P ⬍ .0001). The second step, the AIC of which was 299.6, improved the model fit (2 ⫽ 7.0, df ⫽ 1, P ⫽ .008). In this model, both DSM-IV mixed mania (Wald statistic ⫽ 13.0, df ⫽ 1, P ⫽ .0003) and the depression severity (Wald statistic ⫽ 6.7, df ⫽ 1, P ⫽ .0096) were significant, indicating that a more appropriate prediction of suicidality was achieved by using both DSM-IV mixed mania and the depression severity than by using DSM-IV mixed mania alone. The third step further improved the model fit (AIC ⫽ 256.6; 2 ⫽ 36.0, df ⫽ 1, P ⬍ .0001), where only mixed mania as diagnosed by the Cincinnati criteria was significant (Wald statistic ⫽ 40.2, df ⫽ 1, P ⬍ .0001). The significance for DSM-IV mixed mania (Wald statistic ⫽ 0.8, df ⫽ 1, P ⫽ .38) and the depression
66
severity (Wald statistic ⫽ 3.35, df ⫽ 1, P⫽ .067) disappeared at this step, indicating that mixed mania as diagnosed by the Cincinnati criteria alone, rather than an additive combination of DSM-IV mixed mania and the depression severity, achieved the most appropriate prediction of suicidality. Similar analyses were performed with the presence of a suicide attempt during the index episode being used as the dependent variable, to know whether or not the results provided by the above analyses are applicable to a more severe form of suicidality. Univariate comparisons between patients with and without a suicide attempt produced significant differences in three variables at a P ⬍ .05 level, the depression severity, and mixed mania as defined by the DSM-IV and Cincinnati criteria. The depression severity (8.8 [SD 6.4] v 4.8 [SD 3.7] for patients with and without a suicide attempt, respectively; z ⫽ ⫺2.9, P ⫽ .002), and the frequencies of mixed mania as defined by the DSM-IV (38% [n ⫽ 5] v 9% [n ⫽ 53] for patients with and without a suicide attempt, respectively; 2 with continuity correction ⫽ 11.8, df ⫽ 1, P ⫽ .003) and Cincinnati criteria (62% [n ⫽ 8] v 14% [n ⫽ 80] for patients with and without a suicide attempt, respectively; 2 with continuity correction ⫽ 22.0, df ⫽ 1, P ⬍ .0001) were higher in patients with a suicide attempt than in patients without a suicide attempt. DSM-IV mixed mania, the depression severity, and mixed mania as defined by the Cincinnati criteria were then entered sequentially in this order into a logistic regression analysis, with the presence of a suicide attempt being used as the dependent variable. The first step produced a significant model improvement (2 ⫽ 7.6, df ⫽ 1, P ⫽ .006) with an AIC value of 120.7. DSM-IV mixed mania was significant in this model (Wald statistic ⫽ 9.3, df ⫽ 1, P ⫽ .002). The second step significantly improved the model fit (2 ⫽ 7.9, df ⫽ 1, P ⫽ .005; AIC ⫽ 114.8). In this model, only the depression severity was significant (Wald statistic ⫽ 7.5, df ⫽ 1, P ⫽ .006), and the significance of DSM-IV mixed mania disappeared (Wald statistic ⫽ .03, df ⫽ 1, P ⫽ .87). The third step further improved the model fit (2 ⫽ 9.6, df ⫽ 1, P ⫽ .002, AIC ⫽ 107.2), where only mixed mania as defined by the Cincinnati criteria was significant (Wald statistic ⫽ 12.9, df ⫽ 1, P ⫽ .0002).
SATO ET AL
DISCUSSION
As shown by other similar studies,2,3,5,6 the present study indicates that suicidality is not infrequent among patients with acute mania. Of 576 consecutive inpatients meeting DSM-IV criteria for manic episode, 51 (8.9%) had suicidal ideation, 13 of whom attempted suicide during the index episode. The frequency of suicidal ideation in our study is lower than that in the Dilsaver et al. study2 (26.8% of 93 manic patients), but is similar to that in the Strakowski et al. study3 (14.2% of 91 manic patients). A highly significant association between suicidal ideation and the status of DSM-IV mixed mania was found in our study. Twenty-two (37.9%) of 58 patients with a diagnosis of DSM-IV mixed mania had suicidal ideation, while 36 (6.9%) of 518 pure manic patients had suicidal ideation: these frequencies are placed within a range of those provided by the previous two studies (26% to 54.5% for patients with DSM-IV mixed mania, and 2% to 7% for patients with pure mania).2,3 The present study showed that a more severe form of suicidality, as defined as the presence of a suicide attempt during the index episode, was also significantly associated with the status of DSM-IV mixed mania. The frequency of a suicide attempt during the index episode was 8.6% (n ⫽ 5) and 1.5% (n ⫽ 8) for patients with and without the diagnosis, respectively, and the difference was significant (P ⫽ .003). However, a diagnosis of mixed mania by using the Cincinnati criteria was more strongly associated with suicidality, whether suicidality was defined as the presence of suicidal ideation or a suicide attempt. Of 88 patients with Cincinnati mixed mania, 38.6% (n ⫽ 34) had suicidal ideation and 9.1% (n ⫽ 8) attempted suicide during the index episode. Of 488 patients with pure mania, 3.5% (n ⫽ 17) had suicidal ideation and 1.0% (n ⫽ 5) attempted suicide during the index episode. The frequency of suicidal ideation among patients with Cincinnati mixed mania in our study was slightly lower than that in the study by Goldberg et al.5 Interestingly, the chi-square value, which estimates the association between suicidality and mixed mania, was much higher when using the Cincinnati criteria than when using the DSM-IV criteria. Multiple logistic regression analyses showed that the adjusted odds ratio for Cincinnati mixed mania to
SUICIDALITY AND MIXED MANIA
having suicidal ideation was much higher than that for DSM-IV mixed mania. A subsequent logistic regression analysis, where variables were entered sequentially, provided direct evidence that a diagnosis of mixed mania by using the Cincinnati criteria alone was the most appropriate predictor of suicidality, compared to DSM-IV mixed mania, the depression severity, or an additive combination of these two variables: this finding was not different, whether suicidality was defined as the presence of suicidal idea or a suicide attempt during the index episode. These results indicate that the Cincinnati criteria, rather than the DSM-IV criteria, achieve a more appropriate classification of mixed mania in terms of predicting suicidality. Several explanations can be noted to understand why the Cincinnati criteria achieve a better prediction of suicidality among manic patients than do the DSM-IV criteria. First, the Cincinnati criteria provide a broad definition of mixed mania, which requires the presence of three or more symptoms of major depression during a full manic episode; this may help patients who are suicidal but do not reach a strict definition of mixed mania be classified into mixed mania. This explanation is supported by our findings that a higher frequency of mixed mania was observed when using the Cincinnati criteria, and that eight suicidal patients (16% of all suicidal patients in the present study) who did not meet the DSM-IV criteria were classified into Cincinnati mixed mania (see Table 1). Second, the nine depressive symptoms listed in the Cincinnati criteria do not include several symptoms (agitation, appetite loss, insomnia, difficulty in concentration) that are included in the DSM-IV criteria, since these symptoms may also appear in pure mania.4 This deletion from the Cincinnati criteria may lead, among patients having a full manic syndrome, to a more appropriate assessment of the depressive syndrome related to suicidality, thereby resulting in a more accurate prediction of suicidality in these patients. One may question whether or not a smaller number of depressive symptoms required to diagnose Cincinnati mixed mania may cause an increase in the chance probability to being suicidal among mixed manic patients. However, this does not appear to be true. To attempt to ensure this, we calculated the theoretical chance probability to being suicidal in patients diagnosed as having mixed mania. The number of symptoms required for di-
67
agnosing mixed mania is different between the Cincinnati (three or more) and DSM-IV criteria (five or more, in which at least one symptom should be depressed mood, or loss of pleasure or interest). Calculations that assume an equivalent appearance rate of each included criterion were made for mixed mania as defined by using the DSM-IV and Cincinnati criteria separately. Of all combinations (N ⫽ 227) of criteria producing a DSM-IV diagnosis of mixed mania, 62% (n ⫽ 141) include the item “suicidality.” Of all combinations (N ⫽ 466) of criteria producing a Cincinnati diagnosis of mixed mania, 53% (n ⫽ 247) include the item “suicidality,” indicating that theoretically to say, the chance probability to being suicidal among patients with a diagnosis of mixed mania might be even higher when using the DSM-IV criteria. The Cincinnati criteria provide one of wellknown broad definitions of mixed mania. Based on evidence obtained by studies on demographic variables,7-11 natural course,7-9,11,12 and specific treatment responses,13,14 many researchers are often putting forward the argument that narrow definitions of mixed mania such as the DSM-IV criteria, which require a simultaneous overlap of full depressive and manic syndromes, are too strict. Our study, along with other studies attempting to validate mixed mania,7-14 suggests that a broad definition of mixed mania is more appropriate than a narrow definition in terms of predicting suicidality in manic patients. Our finding that manic patients with suicidality at admission were likely to have depressive residual symptoms at discharge suggests that manic patients with suicidality have differential natural history from manic patients without suicidality. Our finding is consistent with one recent study reporting that patients with mixed mania were more likely than other manic patients to experience a cycling into depression from a first episode of mania.25 However, the medication during the index manic episode was not controlled in either of these studies. Further study is needed to clarify whether our finding regarding the association between suicidality at admission and residual depressive symptoms reflect differential medications during the index manic episode or natural history of bipolar disorder. Some possible limitations may be noted in in-
68
SATO ET AL
terpreting the results of the present study: suicidal ideation was not assessed blindly to other psychopathological variables. However, data for most patients in this study were collected before the clinical finding that acute manic patients may frequently be suicidal became well known. Most included patients (88% [n ⫽ 507] of the total subjects) were hospitalized during the period of 1980 to 1994, while the first report on the frequency of suicidality in manic patients was published in September 1994.2 This suggests that the bias, possibly caused by nonblindness of assessing suicidal ideation, has not in a large amount influenced the results of this study. Another limitation is a relatively low rate of alcohol or substance abuse in our sample. The present study found no evidence of an association between suicidality and alcohol or substance abuse, which may have been due to a relative low rate of substance or alcohol
abuse in our sample. Goldberg et al.5 reported that dysphoric manic patients with substance abuse are more likely to have an elevated risk for suicidality. The present study evidenced that suicidality is not infrequent among manic patients, and that the diagnosis of mixed mania by using the Cincinnati criteria alone is, among the variables studied, the best predictor of the presence of suicidality in these patients. However, it is still unknown to what extent such a relatively high frequency of suicidal tendencies in patients with mixed states explains the high rate of suicidal deaths (18.9%) in bipolar patients.1 Further study with a prospective naturalistic design is required to explore whether or not mixed states are really associated with a high rate of suicidal deaths. Such a study might provide a meaningful knowledge for developing a more effective strategy for preventing suicides in bipolar patients.
REFERENCES 1. Goodwin FK, Jamison KR. Manic-Depressive Illness. New York, NY: Oxford University Press, 1990. 2. Dilsaver SC, Chen YW, Swann AC, Shoaib AM, Krajewski KJ. Suicidality in patients with pure and depressive mania. Am J Psychiatry 1994;151:1312-1315. 3. Strakowski SM, McElroy SL, Keck PE, West SA. Suicidality among patients with mixed and manic bipolar disorder. Am J Psychiatry 1996;53:674-676. 4. McElroy SL, Keck PE, Pope HG, Hudson JI, Faedda GL, Swann AC. Clinical and research implications of the diagnosis of dysphoric or mixed mania or hypomania. Am J Psychiatry 1992;149:1633-1644. 5. Goldberg JF, Garno JL, Portera L, Leon AC, Kocsis JH, Whiteside JE. Correlates of suicidal ideation in dysphoric mania. J Affect Disord 1999;56:75-81. 6. Goldberg JF, Garno JL, Leon AC, Kocsis JH, Portera L. Association of recurrent suicidal ideation with nonremission from acute mixed mania. Am J Psychiatry 1998;155:1753-1755. 7. Akiskal HS, Bourgeois ML, Angst J, Post R, Moller H, Hirschfeld R. Re-evaluating the prevalence of and diagnostic composition within the broad clinical spectrum of bipolar disorders. J Affect Disord 2000;59:S5-S30. 8. Perugi G, Akiskal HS, Micheli C, Musetti L, Paiano A, Quilici C, et al. Clinical subtypes of bipolar mixed states: validating a broader European definition in 143 cases. J Affect Disord 1997;43:169-180. 9. Akiskal HS, Hantouche EG, Bourgeois ML, Azorin J-M, Sechter D, Allilaire J-F, et al. Gender, temperament, and the clinical picture in dysphoric mixed mania: findings from a French national study (EPIMAN). J Affect Disord 1998;50:175186. 10. Arnold LM, McElroy SL, Keck PE Jr. The role of gender in mixed mania. Compr Psychiatry 2000;41:83-87. 11. McElroy SL, Strakowski SM, Keck PE, Tugrul KL, West
SA, Lonczak HS. Differences and similarities in mixed and pure mania. Compr Psychiatry 1995;36:187-194. 12. Bauer MS, Whybrow PC, Gyulai L, Gonnel J, Yeh H-S. Testing definitions of dysphoric mania and hypomania: prevalence, clinical characteristics and inter-episode stability. J Affect Disord 1994;32:201-211. 13. Greil W, Kleindienst N, Erazo N, Muller-Oerlinghausen B. Differential response to lithium and carbamazepine in the prophylaxis of bipolar disorder. J Clin Psychopharmacol 1998; 18:455-60. 14. Swann AC, Bowden CL, Morris D, Calabrese JR, Petty F, Small J, et al. Depression during mania: treatment response to lithium or divalproex. Arch Gen Psychiatry 1997;54:37-42. 15. World Health Organization. The ICD-9 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva, Switzerland: World Health Organization, 1977. 16. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva, Switzerland: World Health Organization, 1992. 17. Association for Methodology and Documentation in Psychiatry (Arbeitsgemeinschaft fu¨ r Methodik und Diagnostik in der Psychiatrie). Das AMDP-System: Manual zur Dokumentation Psychiatrischer Befunde. Ed. 5. Berlin, Germany: Springer, 1981. 18. Kuny S, von Luckner N, Baenninger R, Bauer P, Eichenberger G, Woggon B. Interrater reliability of AMDP and AMP symptoms. In: Bobon D, Baumann U, Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:143-160. 19. Renfordt E, Busch H, von Cranach M, Gulbinat W, Tegeler M. Particular aspects of the interrater reliability of the AMDP psychopathology scale. In: Bobon D, Baumann U,
SUICIDALITY AND MIXED MANIA
Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:125-142. 20. Baumann U, Pietzcker A, Woggon B. Syndroms and scales in the AMP-systems. In: Bobon D, Baumann U, Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:74-87. 21. Pietzcker A, Gebhardt R, Strauss A, Sto¨ ckel M, Langer C, Freudenthal K. The syndrome scales in the AMDP-system. In: Bobon D, Baumann U, Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:88-99.
69
22. Angst J, Stassen HH, Woggon B. Effect of neuroleptics on positive and negative symptoms and the deficit state. Psychopharmacol Berl 1989;99:S41-S46. 23. SPSS. SPSS for Windows Release 7.0. Chicago, IL: SPSS, 1995. 24. Akaike H. Factor analysis and AIC. Psychometrika 1987;52:317-332. 25. Zarate CA Jr, Tohen M, Fletcher K. Cycling into depression from a first episode of mania: a case-comparison study. Am J Psychiatry 2001;158:1524-1526.
the depression severity, achieved the most appropriate prediction of suicidal ideation in the sample. These findings did not differ, even when suicidality was defined as having a suicide attempt during the index episode. Our finding that suicidality was more strongly associated with Cincinnati mixed mania than with DSM-IV mixed mania is probably due to that suicidal patients who do not meet DSM-IV criteria for mixed mania are classified into mixed mania, or/and that the depressive syndrome, related to suicidality, is more appropriately assessed among manic patients, when using the Cincinnati criteria. There was no evidence that marital status, employment, a lifetime history of alcohol or substance abuse, or a history of suicide attempts before the index episode was significantly associated with suicidality in the sample. Manic patients with suicidality may have a greater severity of residual depressive symptoms at discharge. © 2004 Elsevier Inc. All rights reserved.
S
ideation in patients simultaneously meeting DSMIII-R criteria for both manic and major depressive episodes than in pure manic patients. Interestingly, Strakowski et al.3 provided evidence that the severity of current depressive symptoms, rather than the presence of a full depressive syndrome per se, is associated with suicidal ideation among manic patients, suggesting that narrow definitions of mixed mania may not play a definite role in predicting suicidality in these patients. DSM-III-R and DSM-IV adopt a narrow definition of mixed mania. However, there are several studies indicating that broader definitions of mixed states result in a more meaningful subtyping of mixed mania in terms of demographic variables,7-11 natural course,7-9,11,12 and specific treatment responses.13,14 The association between a broader definition of mixed mania and suicidality has not been well studied. One study reported that suicidal ideation is not infrequent in manic patients, even when full major depression does not accompany mania.5,6 However, it is unclear whether or not a broader definition of mixed mania plays a more definite role in predicting suicidality in manic patients, as compared with a narrow definition of mixed mania. McElroy et al.4,11 proposed one of well-known broad definitions of mixed mania, the Cincinnati
UICIDE RISK in bipolar patients is generally high. An exhaustive review by Goodwin and Jamison1 showed that 18.9% of deaths in bipolar patients were due to suicide. Although previous studies on suicidality in bipolar patients tended to be restricted to depressive episodes, several recent studies have emphasized that mixed states (or mixed mania), as defined as a simultaneous admixture of both depressive and manic syndromes, may also mediate the high suicide risk in bipolar patients.2-6 Using a narrow definition of mixed mania that requires patients to simultaneously meet both full manic and depressive syndromes, Dilsaver et al.2 reported that 54.5% of patients with mixed mania had suicidal ideation, in contrast to 2.0% of patients with pure mania. Strakowski et al.3 also found a significantly higher frequency of suicidal From the Psychiatrische Klinik und Poliklinik, Ludwig-Maximilians-Universita¨t, Mu¨nchen, Germany. T.S. is supported by the Alexander von Humboldt Foundation. Address reprint requests to Tetsuya Sato, M.D., Ph.D., Psychiatrische Klinik und Poliklinik, Ludwig-MaximiliansUniversita¨t Mu¨nchen, Nussbaumstrasse 7, D-80336 Munich, Germany. © 2004 Elsevier Inc. All rights reserved. 0010-440X/04/4501-0015$30.00/0 doi:10.1016/S0010-440X(03)00145-7 62
Comprehensive Psychiatry, Vol. 45, No. 1 ( January/February), 2004: pp 62-69
SUICIDALITY AND MIXED MANIA
criteria, which require the presence of three or more symptoms of major depression during a full manic episode. The present study examined the association between mixed mania and suicidality in 576 hospitalized patients with acute mania. For diagnosing mixed mania, both Cincinnati and DSM-IV criteria were used. METHOD
Patients All patients who were hospitalized at the Psychiatric Hospital of the Ludwig-Maximilian University in Munich, Germany for any affective disorder during the period of 1980 to 1997 were considered as subjects in this study. Routine clinical diagnoses were made according to the International Classification of Diseases system (ICD-9 or ICD-10),15,16 but a broad range of 196 psychiatric and related somatic symptoms were, as part of the routine documentation at the hospital, systematically evaluated for all patients at both admission and discharge by using a standardized instrument (the AMDP system; see below), which allowed for precise diagnoses of DSM-IV manic episode, nonmixed and mixed, and mixed mania according to the Cincinnati criteria.4,9 The following inclusion criteria were used: (1) patients be diagnosed using the ICD system as currently having a manic or mixed episode (ICD-9 diagnosis of 296.0, 296.2, or 296.4; or ICD-10 diagnosis of F30 F31.0, F31.1, F31.2, F31.6, or F38.00); (2) patients be younger than 70 years; and (3) patients who met DSM-IV criteria for manic episode, nonmixed or mixed. Finally, 576 consecutive inpatients with DSM-IV manic episode, nonmixed or mixed, were included for the following analyses. All patients gave informed consent to be assessed using several instruments described below. Of the subjects, 292 (51%) were women, and 284 (49%) were men. Their mean age was 38.6 (SD 12.6) years. Their mean age at onset of first affective episode was 28.9 (SD 10.7) years. Fiftyeight patients (10.1%: 38 women and 20 men) met criteria for DSM-IV manic episode, mixed, while 88 patients (15.3%: 55 women and 33 men) met criteria for mixed mania according to Cincinnati criteria. The frequency of mixed mania as defined using DSM-IV or Cincinnati criteria was constant during the whole study period when the frequency was compared between three entry periods, 1980 to 1985, 1986 to 1990, and 1991 to 1997. Both diagnoses were significantly correlated in the whole sample ( ⫽ 0.67, P ⬍ .0001). The patients were treated with medications as clinically appropriate during the hospital stay. Before admission, the majority of the patients had received various medications including mood stabilizers (58%) and/or antipsychotics (36%). There was no significant association between presence of these medications and diagnoses of mixed mania.
Clinical Assessments The Association for Methodology and Documentation in Psychiatry (AMDP) system17 was used to assess psychiatric symptoms. The AMDP system is a comprehensive rating instrument, developed on the basis of German traditional descriptive psychopathology on functional psychoses; it is commonly used in most psychiatric institutes in German-speaking coun-
63
tries. Each psychiatric symptom of the AMDP system is scored from 0 (absent) to 3 (severe) with defined anchor statements by using a semistructured interview method. Several studies indicated moderate to high inter-rater agreements for most included symptoms.18,19 Rater-training sessions are performed regularly in our hospital to establish and maintain high inter-rater reliability of the instrument. Inter-rater reliability of all AMDP items in our hospital, calculated based on joint interviews by multiple raters of over 50 diagnostically diverse patients, ranged from 0.65 to 0.92 (analysis of variance intraclass correlation). Based on the AMDP ratings, several summary scales for depressive, manic, and paranoid-hallucinatory syndromes (the AMDP scores for depressive, manic, and paranoid-hallucinatory syndromes), which have been validated in large psychiatric samples,20-22 can be calculated by summing up the scores on 13 items (rumination, loss of feeling, loss of vitality, depressed mood, hopelessness, feeling of inadequacy, feeling of guilt, inhibition of drive, worse in the morning, interrupted sleep, shortened sleep, early waking, decreased appetite) for depressive syndrome, seven items (flight of idea, euphoria, exaggerated self-esteem, increased drive, motor restlessness, logorrhea, excessive social contact) for manic syndrome, and 13 items (delusional mood, delusional perceptions, sudden delusional ideas, delusional ideas, systematized delusions, delusional dynamics, delusions of reference, delusions of persecution, verbal hallucinations, bodily hallucinations, depersonalization, thought withdrawal, other feelings of alien influence) for paranoid-hallucinatory syndrome. The Global Assessment of Functioning (GAF) score was also assessed for all subjects at both admission and discharge. All subjects gave written informed consent to be assessed by using the instruments. Well-trained psychiatrists administered these two instruments. The presence of suicidality was defined in this study by using the highest score on the AMDP suicide item at admission. The item scores the severity of suicidality as follows: 0 (absent); 1 (mild): frequent thoughts that she/he would be better off dead; 2 (moderate): frequent thoughts about committing suicide or mental rehearsal of suicide; 3 (severe): suicide attempt during the index episode. The present study defines suicidality as a score equal to or higher than 1, since this definition appears to be equivalent to that used in previous similar studies.2,3
Statistical Analyses Clinical and demographic variables were compared between patients with and without suicidality by using chi-square test for categorical variables and Mann-Whitney tests for continuous variables. The frequencies of mixed mania as defined by the DSM-IV and Cincinnati criteria were also compared between the two groups. The nonparametric tests were used for all examined continuous variables, since the distributions of these variables were not normally distributed. In each of these univariate analyses, the significance level was adjusted at a P ⬍ .0025 (.05 divided by 20 tests) level by using Bonferroni’s method. An adjusted odds ratio for patients with mixed mania to having suicidal ideation was computed by using multiple logistic regression analysis. This analysis was performed for mixed mania as defined by the DSM-IV and Cincinnati criteria separately. Several other variables, which were, in the above univariate analyses, associated with suicidal ideation at a P ⬍ .10 level, were entered into the logistic analyses. A second logistic
64
SATO ET AL
Table 1. Demographic and Clinical Characteristics of Nonsuicidal and Suicidal Manic Patients
Gender, female, n (%) Age, years (SD) Marital status Single, n (%) Married or living together, n (%) Divorced, separated or bereaved, n (%) Unemployed, n (%) Mother tongue German, n (%) Non-German, n (%) Age of onset, years (SD) GAF score, baseline Alcohol or substance abuse, lifetime, n (%) Alcohol or substance abuse, index episode, n (%) Suicide attempts, before index episode, n (%) Suicide attempts, index episode, n (%) AMDP depressive syndrome mean score at admission (SD) AMDP manic syndrome mean score at admission (SD) AMDP paranoid-hallucinatory syndrome mean score at admission (SD) DSM-IV mixed mania, n (%) Mixed mania by the Cincinnati criteria, n (%) Duration of index admission, days, mean (SD) GAF score at discharge, mean (SD) AMDP depressive syndrome mean score at discharge (SD) AMDP manic syndrome mean score at discharge (SD) AMDP paranoid-hallucinatory syndrome mean score at discharge (SD)
Nonsuicidal Patients (n ⫽ 525)
Suicidal Patients (n ⫽ 51)
257 (49) 38.7 (12.5)
35 (69) 37.1 (13.3)
230 (44) 223 (42) 72 (14) 24 (5)
22 (43) 22 (43) 7 (14) 5 (10)
468 (89) 57 (11) 28.9 (10.6) 36.3 (13.1) 103 (20) 61 (12) 87 (17) 0 (0) 4.5 (3.3) 11.5 (4.0) 3.5 (4.9) 36 (7) 54 (10) 50.7 (34.1) 66.8 (13.7) 0.6 (1.2) 1.4 (2.3) 0.1 (0.5)
46 (90) 5 (10) 28.2 (11.6) 32.4 (11.1) 9 (18) 7 (14) 10 (20) 13 (25) 8.4 (6.0) 9.2 (4.5) 5.3 (5.6) 22 (43) 34 (67) 49.8 (32.0) 64.7 (12.5) 1.0 (1.4) 1.3 (2.2) 0.3 (0.9)
Group-Differences Statistics
7.20
P
.01
⫺1.96
Not tested ⫺4.85 ⫺3.62 ⫺2.48 .01 67.57 114.16 ⫺1.68 ⫺3.23
.09
⫺2.23
.03
NOTE. Chi-square values with df ⫽ 1 were used for categorical variables, and z values were used for continuous variables. Marital status was analyzed by using a chi-square value with df ⫽ 2. The significance level recommended is P ⬍ .0025 (.05 divided by 20 tests).
regression analysis, with the presence of suicidal ideation being used as the dependent variable, was then performed, in which the status of mixed mania as diagnosed by DSM-IV, the severity of depressive syndrome as measured by the AMDP score for depressive syndrome, and the status of mixed mania as diagnosed by the Cincinnati criteria were sequentially entered as independent variables in this order. An SPSS software (release 7.0)23 was used for statistical analyses. All statistical statements in the present study were two-tailed.
RESULTS
Of the whole sample, 51 patients (8.9%; 35 women and 16 men) had suicidal ideation at admission. Of these patients, 13 patients (2.6%; nine women and four men) attempted suicide during the index episode. The frequency of patients with suicide ideation was fairly constant during the whole study period when the frequency was compared between three entry periods, 1980 to 1985, 1986 to 1990, and 1991 to 1997. The presence of suicide ideation was not significantly associated with medications of mood stabilizers or antipsychotics before patients’ admission.
Table 1 shows demographic and clinical variables in patients with and without suicidal ideation. After Bonferroni’s correction, the severity of depressive and manic syndromes at admission, the status of mixed mania as diagnosed by both DSM-IV and Cincinnati criteria, and the severity of depressive syndrome at discharge significantly differentiated the two groups. The AMDP scores for depressive and manic syndromes were significantly higher and lower in patients with suicidality than in patients without suicidality, respectively. Mixed mania as diagnosed by DSM-IV was significantly frequent in patients with suicidality (43%) than in patients without suicidality (7%). The frequency of mixed mania as diagnosed by the Cincinnati criteria was also significantly higher in patients with suicidal ideation (67%) than in patients without suicidal ideation (10%), but the computed chi-square value for the Cincinnati criteria was much greater than that for DSM-IV criteria in these analyses. Although there was no evidence
SUICIDALITY AND MIXED MANIA
65
Table 2. Adjusted Odds Ratio for Differently Defined Mixed Mania in Predicting Suicidality, as Computed by Multiple Logistic Regression Analyses Diagnostic Criteria for Mixed Mania Entered Into Multiple Logistic Regression Analyses Mix Mania by DSM-IV Model Improvement, 2 (df ⫽ 7) 66.50, P ⫽ .0000 AIC ⫽ 293.0
Mixed Mania by Cincinnati Criteria Model Improvement, 2 (df ⫽ 7) 92.01, P ⫽ .0000 AIC ⫽ 269.5
Odds Ratio (95% CI)
Odds Ratio (95% CI)
1.79 (0.91-3.52) 1.66 (1.04-2.67)* 0.91 (0.60-1.39) 0.93 (0.81-1.07) 1.07 (0.90-1.28) 1.43 (1.06-1.92)* 3.98 (1.60-9.95)†
1.94 (0.94-4.01) 1.48 (0.93-2.35) 1.01 (0.66-1.54) 0.93 (0.81-1.07) 1.14 (0.95-1.38) 1.28 (0.94-1.73) 13.86 (5.84-33.74)†
Gender AMDP depressive syndrome, baselinea AMDP manic syndrome, baselinea Interaction of depressive and manic syndromes AMDP paranoid-hallucinatory syndromea GAF score, baselineb Diagnosis of mixed states a
Odds ratios and their 95% confidence intervals (CI) were calculated for a 3-point higher score. Odds ratios and their 95% confidence intervals were calculated for a 10-point lower score. Significant odds ratios at *P ⬍ .05 and †P ⬍ .0001, respectively. Model improvements were calculated from the full model.
b
that patients with suicidality required a longer hospitalization, these patients had a significantly more severe residual depressive syndrome at discharge. Table 2 demonstrates adjusted odds ratios for the status of mixed mania. The variables that were associated with suicidality at a P ⬍ .05 in Table 1, such as gender, the severity of depressive, manic, and paranoid-hallucinatory syndromes, and the GAF score at admission, were entered as independent variables into the analyses, together with status of mixed mania. Interaction of manic and depressive syndromes was also entered, since a combination of these psychiatric syndromes has been suggested to be specifically associated with the presence of suicidality among manic patients.1 When the DSM-IV criteria were used for diagnosing mixed mania, the model improvement was highly significant (2 ⫽ 66.5, df ⫽ 7, P ⬍ .0001), where the depression severity, GAF score, and the status of mixed mania were significant in predicting suicidality. Subjects having a 3-point higher score on the AMDP depressive syndrome and a 10-point lower GAF score were 1.7 times and 1.4 times more likely to have suicidality, respectively. The adjusted odds ratio for DSM-IV mixed mania was 4.0. When using the Cincinnati criteria for diagnosing mixed mania, the model improvement was much higher (2 ⫽ 92.0, df ⫽ 7, P ⬍ .0001). In this model, only the diagnosis of mixed mania by the Cincinnati criteria was significant in predicting suicidality. A large increase in the adjusted odds ratio for status of mixed mania was observed
(14.0 from 4.0), suggesting that the Cincinnati criteria, rather than DSM-IV criteria, provided a more appropriate subtyping of mixed mania in terms of predicting suicidality. This suggestion was confirmed by the final logistic regression analysis, where the DSM-IV definition of mixed mania, the depression severity, and the Cincinnati definition of mixed mania were entered into the analysis sequentially. The first step including the DSM-IV mixed mania alone produced a significant model improvement from the null model (2 ⫽ 44.1, df ⫽ 1, P ⬍ .0001). Akaike Information Criterion (AIC),24 which assesses the goodness of fit of the model, was 304.5. In this model, the DSM-IV diagnosis of mixed mania was significant in predicting suicidality (Wald statistic ⫽ 49.6, df ⫽ 1, P ⬍ .0001). The second step, the AIC of which was 299.6, improved the model fit (2 ⫽ 7.0, df ⫽ 1, P ⫽ .008). In this model, both DSM-IV mixed mania (Wald statistic ⫽ 13.0, df ⫽ 1, P ⫽ .0003) and the depression severity (Wald statistic ⫽ 6.7, df ⫽ 1, P ⫽ .0096) were significant, indicating that a more appropriate prediction of suicidality was achieved by using both DSM-IV mixed mania and the depression severity than by using DSM-IV mixed mania alone. The third step further improved the model fit (AIC ⫽ 256.6; 2 ⫽ 36.0, df ⫽ 1, P ⬍ .0001), where only mixed mania as diagnosed by the Cincinnati criteria was significant (Wald statistic ⫽ 40.2, df ⫽ 1, P ⬍ .0001). The significance for DSM-IV mixed mania (Wald statistic ⫽ 0.8, df ⫽ 1, P ⫽ .38) and the depression
66
severity (Wald statistic ⫽ 3.35, df ⫽ 1, P⫽ .067) disappeared at this step, indicating that mixed mania as diagnosed by the Cincinnati criteria alone, rather than an additive combination of DSM-IV mixed mania and the depression severity, achieved the most appropriate prediction of suicidality. Similar analyses were performed with the presence of a suicide attempt during the index episode being used as the dependent variable, to know whether or not the results provided by the above analyses are applicable to a more severe form of suicidality. Univariate comparisons between patients with and without a suicide attempt produced significant differences in three variables at a P ⬍ .05 level, the depression severity, and mixed mania as defined by the DSM-IV and Cincinnati criteria. The depression severity (8.8 [SD 6.4] v 4.8 [SD 3.7] for patients with and without a suicide attempt, respectively; z ⫽ ⫺2.9, P ⫽ .002), and the frequencies of mixed mania as defined by the DSM-IV (38% [n ⫽ 5] v 9% [n ⫽ 53] for patients with and without a suicide attempt, respectively; 2 with continuity correction ⫽ 11.8, df ⫽ 1, P ⫽ .003) and Cincinnati criteria (62% [n ⫽ 8] v 14% [n ⫽ 80] for patients with and without a suicide attempt, respectively; 2 with continuity correction ⫽ 22.0, df ⫽ 1, P ⬍ .0001) were higher in patients with a suicide attempt than in patients without a suicide attempt. DSM-IV mixed mania, the depression severity, and mixed mania as defined by the Cincinnati criteria were then entered sequentially in this order into a logistic regression analysis, with the presence of a suicide attempt being used as the dependent variable. The first step produced a significant model improvement (2 ⫽ 7.6, df ⫽ 1, P ⫽ .006) with an AIC value of 120.7. DSM-IV mixed mania was significant in this model (Wald statistic ⫽ 9.3, df ⫽ 1, P ⫽ .002). The second step significantly improved the model fit (2 ⫽ 7.9, df ⫽ 1, P ⫽ .005; AIC ⫽ 114.8). In this model, only the depression severity was significant (Wald statistic ⫽ 7.5, df ⫽ 1, P ⫽ .006), and the significance of DSM-IV mixed mania disappeared (Wald statistic ⫽ .03, df ⫽ 1, P ⫽ .87). The third step further improved the model fit (2 ⫽ 9.6, df ⫽ 1, P ⫽ .002, AIC ⫽ 107.2), where only mixed mania as defined by the Cincinnati criteria was significant (Wald statistic ⫽ 12.9, df ⫽ 1, P ⫽ .0002).
SATO ET AL
DISCUSSION
As shown by other similar studies,2,3,5,6 the present study indicates that suicidality is not infrequent among patients with acute mania. Of 576 consecutive inpatients meeting DSM-IV criteria for manic episode, 51 (8.9%) had suicidal ideation, 13 of whom attempted suicide during the index episode. The frequency of suicidal ideation in our study is lower than that in the Dilsaver et al. study2 (26.8% of 93 manic patients), but is similar to that in the Strakowski et al. study3 (14.2% of 91 manic patients). A highly significant association between suicidal ideation and the status of DSM-IV mixed mania was found in our study. Twenty-two (37.9%) of 58 patients with a diagnosis of DSM-IV mixed mania had suicidal ideation, while 36 (6.9%) of 518 pure manic patients had suicidal ideation: these frequencies are placed within a range of those provided by the previous two studies (26% to 54.5% for patients with DSM-IV mixed mania, and 2% to 7% for patients with pure mania).2,3 The present study showed that a more severe form of suicidality, as defined as the presence of a suicide attempt during the index episode, was also significantly associated with the status of DSM-IV mixed mania. The frequency of a suicide attempt during the index episode was 8.6% (n ⫽ 5) and 1.5% (n ⫽ 8) for patients with and without the diagnosis, respectively, and the difference was significant (P ⫽ .003). However, a diagnosis of mixed mania by using the Cincinnati criteria was more strongly associated with suicidality, whether suicidality was defined as the presence of suicidal ideation or a suicide attempt. Of 88 patients with Cincinnati mixed mania, 38.6% (n ⫽ 34) had suicidal ideation and 9.1% (n ⫽ 8) attempted suicide during the index episode. Of 488 patients with pure mania, 3.5% (n ⫽ 17) had suicidal ideation and 1.0% (n ⫽ 5) attempted suicide during the index episode. The frequency of suicidal ideation among patients with Cincinnati mixed mania in our study was slightly lower than that in the study by Goldberg et al.5 Interestingly, the chi-square value, which estimates the association between suicidality and mixed mania, was much higher when using the Cincinnati criteria than when using the DSM-IV criteria. Multiple logistic regression analyses showed that the adjusted odds ratio for Cincinnati mixed mania to
SUICIDALITY AND MIXED MANIA
having suicidal ideation was much higher than that for DSM-IV mixed mania. A subsequent logistic regression analysis, where variables were entered sequentially, provided direct evidence that a diagnosis of mixed mania by using the Cincinnati criteria alone was the most appropriate predictor of suicidality, compared to DSM-IV mixed mania, the depression severity, or an additive combination of these two variables: this finding was not different, whether suicidality was defined as the presence of suicidal idea or a suicide attempt during the index episode. These results indicate that the Cincinnati criteria, rather than the DSM-IV criteria, achieve a more appropriate classification of mixed mania in terms of predicting suicidality. Several explanations can be noted to understand why the Cincinnati criteria achieve a better prediction of suicidality among manic patients than do the DSM-IV criteria. First, the Cincinnati criteria provide a broad definition of mixed mania, which requires the presence of three or more symptoms of major depression during a full manic episode; this may help patients who are suicidal but do not reach a strict definition of mixed mania be classified into mixed mania. This explanation is supported by our findings that a higher frequency of mixed mania was observed when using the Cincinnati criteria, and that eight suicidal patients (16% of all suicidal patients in the present study) who did not meet the DSM-IV criteria were classified into Cincinnati mixed mania (see Table 1). Second, the nine depressive symptoms listed in the Cincinnati criteria do not include several symptoms (agitation, appetite loss, insomnia, difficulty in concentration) that are included in the DSM-IV criteria, since these symptoms may also appear in pure mania.4 This deletion from the Cincinnati criteria may lead, among patients having a full manic syndrome, to a more appropriate assessment of the depressive syndrome related to suicidality, thereby resulting in a more accurate prediction of suicidality in these patients. One may question whether or not a smaller number of depressive symptoms required to diagnose Cincinnati mixed mania may cause an increase in the chance probability to being suicidal among mixed manic patients. However, this does not appear to be true. To attempt to ensure this, we calculated the theoretical chance probability to being suicidal in patients diagnosed as having mixed mania. The number of symptoms required for di-
67
agnosing mixed mania is different between the Cincinnati (three or more) and DSM-IV criteria (five or more, in which at least one symptom should be depressed mood, or loss of pleasure or interest). Calculations that assume an equivalent appearance rate of each included criterion were made for mixed mania as defined by using the DSM-IV and Cincinnati criteria separately. Of all combinations (N ⫽ 227) of criteria producing a DSM-IV diagnosis of mixed mania, 62% (n ⫽ 141) include the item “suicidality.” Of all combinations (N ⫽ 466) of criteria producing a Cincinnati diagnosis of mixed mania, 53% (n ⫽ 247) include the item “suicidality,” indicating that theoretically to say, the chance probability to being suicidal among patients with a diagnosis of mixed mania might be even higher when using the DSM-IV criteria. The Cincinnati criteria provide one of wellknown broad definitions of mixed mania. Based on evidence obtained by studies on demographic variables,7-11 natural course,7-9,11,12 and specific treatment responses,13,14 many researchers are often putting forward the argument that narrow definitions of mixed mania such as the DSM-IV criteria, which require a simultaneous overlap of full depressive and manic syndromes, are too strict. Our study, along with other studies attempting to validate mixed mania,7-14 suggests that a broad definition of mixed mania is more appropriate than a narrow definition in terms of predicting suicidality in manic patients. Our finding that manic patients with suicidality at admission were likely to have depressive residual symptoms at discharge suggests that manic patients with suicidality have differential natural history from manic patients without suicidality. Our finding is consistent with one recent study reporting that patients with mixed mania were more likely than other manic patients to experience a cycling into depression from a first episode of mania.25 However, the medication during the index manic episode was not controlled in either of these studies. Further study is needed to clarify whether our finding regarding the association between suicidality at admission and residual depressive symptoms reflect differential medications during the index manic episode or natural history of bipolar disorder. Some possible limitations may be noted in in-
68
SATO ET AL
terpreting the results of the present study: suicidal ideation was not assessed blindly to other psychopathological variables. However, data for most patients in this study were collected before the clinical finding that acute manic patients may frequently be suicidal became well known. Most included patients (88% [n ⫽ 507] of the total subjects) were hospitalized during the period of 1980 to 1994, while the first report on the frequency of suicidality in manic patients was published in September 1994.2 This suggests that the bias, possibly caused by nonblindness of assessing suicidal ideation, has not in a large amount influenced the results of this study. Another limitation is a relatively low rate of alcohol or substance abuse in our sample. The present study found no evidence of an association between suicidality and alcohol or substance abuse, which may have been due to a relative low rate of substance or alcohol
abuse in our sample. Goldberg et al.5 reported that dysphoric manic patients with substance abuse are more likely to have an elevated risk for suicidality. The present study evidenced that suicidality is not infrequent among manic patients, and that the diagnosis of mixed mania by using the Cincinnati criteria alone is, among the variables studied, the best predictor of the presence of suicidality in these patients. However, it is still unknown to what extent such a relatively high frequency of suicidal tendencies in patients with mixed states explains the high rate of suicidal deaths (18.9%) in bipolar patients.1 Further study with a prospective naturalistic design is required to explore whether or not mixed states are really associated with a high rate of suicidal deaths. Such a study might provide a meaningful knowledge for developing a more effective strategy for preventing suicides in bipolar patients.
REFERENCES 1. Goodwin FK, Jamison KR. Manic-Depressive Illness. New York, NY: Oxford University Press, 1990. 2. Dilsaver SC, Chen YW, Swann AC, Shoaib AM, Krajewski KJ. Suicidality in patients with pure and depressive mania. Am J Psychiatry 1994;151:1312-1315. 3. Strakowski SM, McElroy SL, Keck PE, West SA. Suicidality among patients with mixed and manic bipolar disorder. Am J Psychiatry 1996;53:674-676. 4. McElroy SL, Keck PE, Pope HG, Hudson JI, Faedda GL, Swann AC. Clinical and research implications of the diagnosis of dysphoric or mixed mania or hypomania. Am J Psychiatry 1992;149:1633-1644. 5. Goldberg JF, Garno JL, Portera L, Leon AC, Kocsis JH, Whiteside JE. Correlates of suicidal ideation in dysphoric mania. J Affect Disord 1999;56:75-81. 6. Goldberg JF, Garno JL, Leon AC, Kocsis JH, Portera L. Association of recurrent suicidal ideation with nonremission from acute mixed mania. Am J Psychiatry 1998;155:1753-1755. 7. Akiskal HS, Bourgeois ML, Angst J, Post R, Moller H, Hirschfeld R. Re-evaluating the prevalence of and diagnostic composition within the broad clinical spectrum of bipolar disorders. J Affect Disord 2000;59:S5-S30. 8. Perugi G, Akiskal HS, Micheli C, Musetti L, Paiano A, Quilici C, et al. Clinical subtypes of bipolar mixed states: validating a broader European definition in 143 cases. J Affect Disord 1997;43:169-180. 9. Akiskal HS, Hantouche EG, Bourgeois ML, Azorin J-M, Sechter D, Allilaire J-F, et al. Gender, temperament, and the clinical picture in dysphoric mixed mania: findings from a French national study (EPIMAN). J Affect Disord 1998;50:175186. 10. Arnold LM, McElroy SL, Keck PE Jr. The role of gender in mixed mania. Compr Psychiatry 2000;41:83-87. 11. McElroy SL, Strakowski SM, Keck PE, Tugrul KL, West
SA, Lonczak HS. Differences and similarities in mixed and pure mania. Compr Psychiatry 1995;36:187-194. 12. Bauer MS, Whybrow PC, Gyulai L, Gonnel J, Yeh H-S. Testing definitions of dysphoric mania and hypomania: prevalence, clinical characteristics and inter-episode stability. J Affect Disord 1994;32:201-211. 13. Greil W, Kleindienst N, Erazo N, Muller-Oerlinghausen B. Differential response to lithium and carbamazepine in the prophylaxis of bipolar disorder. J Clin Psychopharmacol 1998; 18:455-60. 14. Swann AC, Bowden CL, Morris D, Calabrese JR, Petty F, Small J, et al. Depression during mania: treatment response to lithium or divalproex. Arch Gen Psychiatry 1997;54:37-42. 15. World Health Organization. The ICD-9 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva, Switzerland: World Health Organization, 1977. 16. World Health Organization. The ICD-10 Classification of Mental and Behavioural Disorders: Clinical Descriptions and Diagnostic Guidelines. Geneva, Switzerland: World Health Organization, 1992. 17. Association for Methodology and Documentation in Psychiatry (Arbeitsgemeinschaft fu¨ r Methodik und Diagnostik in der Psychiatrie). Das AMDP-System: Manual zur Dokumentation Psychiatrischer Befunde. Ed. 5. Berlin, Germany: Springer, 1981. 18. Kuny S, von Luckner N, Baenninger R, Bauer P, Eichenberger G, Woggon B. Interrater reliability of AMDP and AMP symptoms. In: Bobon D, Baumann U, Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:143-160. 19. Renfordt E, Busch H, von Cranach M, Gulbinat W, Tegeler M. Particular aspects of the interrater reliability of the AMDP psychopathology scale. In: Bobon D, Baumann U,
SUICIDALITY AND MIXED MANIA
Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:125-142. 20. Baumann U, Pietzcker A, Woggon B. Syndroms and scales in the AMP-systems. In: Bobon D, Baumann U, Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:74-87. 21. Pietzcker A, Gebhardt R, Strauss A, Sto¨ ckel M, Langer C, Freudenthal K. The syndrome scales in the AMDP-system. In: Bobon D, Baumann U, Angst J, Helmchen H, Hippius H (eds). AMDP-System in Pharmacopsychiatry. Basel, Sweden: Karger, 1983:88-99.
69
22. Angst J, Stassen HH, Woggon B. Effect of neuroleptics on positive and negative symptoms and the deficit state. Psychopharmacol Berl 1989;99:S41-S46. 23. SPSS. SPSS for Windows Release 7.0. Chicago, IL: SPSS, 1995. 24. Akaike H. Factor analysis and AIC. Psychometrika 1987;52:317-332. 25. Zarate CA Jr, Tohen M, Fletcher K. Cycling into depression from a first episode of mania: a case-comparison study. Am J Psychiatry 2001;158:1524-1526.