eTable 1. Intake personality scales for groups defined by clinically significant symptoms of overt irritability during intake unipolar MDE
eTable 2. Association of key clinical correlates with overt irritability/anger after covarying for three categories of comorbid disorders
eTable 3. Bipolar diagnoses in first-degree relatives for groups defined by clinically significant symptoms of overt irritability during intake unipolar MDE
eTable 4. Results of preliminary analyses identifying significant correlates of psychomotor agitation during intake unipolar MDE
Judd LL, Schettler PJ, Coryell W, Akiskal HS, Fiedorowicz JG. Overt Irritability/Anger in Unipolar Major Depressive EpisodesPast and Current Characteristics and Implications for Long-term Course. JAMA Psychiatry. 2013;70(11):1171-1180. doi:10.1001/jamapsychiatry.2013.1957
Although symptoms of irritability or anger are not central to the diagnosis of unipolar major depressive episodes (MDEs), these symptoms have been found, in cross-sectional studies, to be highly prevalent and associated with increased comorbidity and depressive illness burden.
To determine the prevalence of overtly expressed irritability/anger and its effect on intake presentation and the long-term course of illness.
A prospective, naturalistic investigation of patients with unipolar MDEs, studied systematically at intake and during up to 31 years of follow-up.
Five US academic medical centers.
Patients entered the National Institute of Mental Health Collaborative Depression Study during an MDE in 1978, 1979, 1980, or 1981. Patients with unipolar MDE at intake (n = 536) were divided into those with and those without current comorbid overtly expressed irritability/anger.
In this observational, longitudinal study, patients received treatment that was recorded but not controlled.
Main Outcomes and Measures
Groups were compared on illness severity and chronicity, psychosocial impairment, quality of life, suicidal behavior, lifetime comorbid diagnoses, impulse control, and measures associated with bipolarity.
Overt irritability/anger was present in 292 of 536 participants with a unipolar MDE at study intake (54.5%). It was associated with significantly increased depressive severity, longer duration of the index MDE, poorer impulse control, a more chronic and severe long-term course of illness, higher rates of lifetime comorbid substance abuse and anxiety disorder, more antisocial personality disorders, greater psychosocial impairment before intake and during follow-up, reduced life satisfaction, and a higher rate of bipolar II disorder in relatives. No association was found with increased suicidal ideation or behavior. Results were not explained by comorbidity or other manic spectrum symptoms.
Conclusions and Relevance
This study extends results of cross-sectional investigations and indicates that irritability/anger during MDEs is a highly prevalent clinical marker of a more severe, chronic, and complex depressive illness. Findings have important implications for assessment and treatment.
Quiz Ref IDIrritability/anger has long been recognized as a symptom commonly seen during major depressive episodes (MDEs). This symptom was integral to the diagnosis of MDE until 1987, when it was dropped as a diagnostic criterion for adults but not for children and adolescents.1- 3
Cross-sectional studies have recently reported on the high prevalence of irritability/anger during unipolar MDEs, using different definitions and measures.4- 6 Benazzi and Akiskal4 found overtly expressed “irritable-hostile” depression in 37% of 254 patients who experienced a unipolar MDE. It was associated with younger age, younger age at onset of mood disorders, atypical depressive features, 3 or more concurrent hypomanic symptoms, and a family history of bipolar disorder. In 2307 participants who experienced a unipolar MDE in the Sequenced Treatment Alternatives to Relieve Depression study, Perlis and colleagues5 defined irritability/anger more broadly, based on subjective feelings of irritability during at least 50% of the time in the previous week using item 6 of the 30-item clinician-rated Inventory of Depressive Symptoms.7 An irritable mood occurred for 46% of participants with MDEs and was significantly linked to female sex, younger age, greater depressive severity, increased comorbid anxiety disorders, poorer subjective satisfaction and quality of life, a greater history of any prior suicide attempts, higher current suicidal ideation, and a higher rate of atypical depressive features. Using a US household survey of 9224 adults in the National Comorbidity Survey Replication, Fava and colleagues6 reported that 51% of 955 respondents with a nonbipolar MDE were “irritable, grouchy, or in a bad mood” most every day during the worst 2 weeks of their worst lifetime MDE. Feeling irritable/grouchy was associated with significantly younger age, earlier age at onset of mood disorders, and higher rates of comorbid anxiety and impulse-control disorders, as well as drug (but not alcohol) dependence, longer duration of MDEs, and higher prevalence of symptoms of fatigue and self-reproach during MDEs—but not with greater depressive severity.
The National Institute of Mental Health (NIMH) Collaborative Depression Study (CDS)8,9 provides a unique opportunity to examine the association of irritability/anger during the intake MDE with important clinical features occurring before study intake, at study intake, and during long-term follow-up (up to 31 years) in a large clinical cohort with systematically diagnosed unipolar MDEs.
The participants enrolled in the NIMH CDS8,9 at 5 academic medical centers (Boston, Massachusetts; Chicago, Illinois; Iowa City, Iowa; New York, New York; and St Louis, Missouri), from 1978 to 1981, while seeking treatment for a major affective episode. Diagnoses at study intake were made using Research Diagnostic Criteria (RDC)10 based on Schedule for Affective Disorders and Schizophrenia (SADS) interviews,11 as well as available medical and research records. Patients entering the CDS during a definite MDE with no prior history of bipolar disorder (type I or II) were selected for analysis. After excluding participants who showed any evidence of schizophrenia or schizoaffective disorder by the end of follow-up, a total of 536 participants with a unipolar MDE at study intake made up the analysis sample. Unlike our prior studies of the course of unipolar major depressive disorder (MDD),12- 15 we included 107 participants (20.0% of the sample) who developed their first lifetime mania or hypomania during follow-up, because conversion to bipolar disorder (type I or II) was an outcome of interest.
All patients in the CDS were required to be white (to test genetic hypotheses), speak English, have an IQ score of at least 70, and have no evidence of organic brain syndrome or terminal medical illness. Written informed consent was obtained at each of the 5 sites for participation in the research.
Quiz Ref IDThe SADS items assess overtly expressed irritability, anger, or annoyance during 2 time periods—the worst week of the study intake MDE and the week prior to the SADS interview. Instructions specify that symptoms of irritability in the presence of other manic/hypomanic symptoms were to be excluded. Suggested interview prompts and response options emphasize overt irritability/anger rather than mere feelings. Descriptors for severity ratings, specific to each SADS item, are as follows for overt irritability/anger: 1 = not at all, or only associated with other manic/hypomanic symptoms; 2 = slight (eg, occasional snappiness of doubtful clinical significance); 3 = mild (eg, somewhat argumentative or quick to express annoyance); 4 = moderate (eg, often shouts or loses temper); 5 = severe (eg, throws things, breaks windows, or is occasionally assaultive); and 6 = extreme (eg, repeatedly violent against things or people). In keeping with the SADS rating convention, we considered overt irritability/anger to be “clinically significant” if rated 3 or above on the severity scale, for the worst week, the past week, or both. Over half the sample had clinically significant overt irritability/anger during their study intake MDEs (292 of 536 participants [54.5%]) and are referred to hereafter as the “irritable group.” This group was compared with the “nonirritable group” (244 of 536 participants [44.5%]).
The CDS contains a remarkable set of data on patient characteristics at study intake and during follow-up. The study intake SADS interview provided information on MDE severity and suicidality (prior to and during study intake). A battery measuring 17 standardized personality characteristics with relevance to affective disorders was administered soon after study intake. The RDC diagnoses were made at study intake and updated at 2 and 5 years, and at subsequent 5-year intervals. For 59.3% of the analysis sample, affective diagnoses were obtained for 1 or more first-degree relatives (biological parents, siblings, or children) from lifetime RDC, based on lifetime SADS interviews. As described in detail elsewhere,12 trained professional raters interviewed patients every 6 months during the first 5 years of follow-up and yearly thereafter, using variations of the Longitudinal Interval Follow-up Evaluation (LIFE).16 Weekly Psychiatric Status Ratings in the LIFE forms are linked, as previously described,12,17 to diagnostic thresholds and were used to identify the start week and the end week of affective episodes during follow-up, as well as to create summary measures of illness chronicity and severity. Information on suicide attempts was obtained from LIFE interviews and from separate forms reporting suicidal behavior during follow-up. Because the wording and timeframe for ratings of psychosocial impairment changed after the period when the original LIFE form was used, LIFE variants provided monthly ratings of psychosocial impairment analyzed herein for all months of follow-up years 3 to 5 and the final month of years 6 to 31.14
The 536 patients entering the CDS during a unipolar MDE, with no lifetime schizophrenia or schizoaffective disorder by the end of follow-up, were systematically followed up for a mean (SD) of 16.2 (10.3) years, with a median of 17.0 years, and no significant difference between the analysis groups. A small portion of the sample (8.4%) participated in follow-up for less than 2 years, while 25.7% were followed for 2 to less than 10 years, 20.5% for 10 to less than 20 years, 40.3% for 20 to less than 30 years, and 5.0% for 30 or 31 years. With 45.3% of the sample followed up systematically for 20 years or more, the CDS provides a unique resource for examining the long-term course of unipolar MDD.
Statistical comparisons between participants with a unipolar MDE with overt clinically significant irritability/anger at study intake and those with a unipolar MDE without overt clinically significant irritability/anger at study intake were made by means of analysis of variance (t tests) for continuous variables if normally or nearly normally distributed, or by means of Wilcoxon rank sum tests if nonnormally distributed. The χ2 or Fisher exact tests were used for categorical variables, and Wilcoxon χ2 tests were used on survival function distributions for time to remission/recovery from the study intake MDE and time to subsequent relapse. Study intake personality measures were analyzed by analysis of variance, covarying for age and sex. Finally, we determined whether key outcome variables differed by study intake group after covarying for each of the 3 available RDC categories of current comorbidity (any substance abuse disorder, any anxiety disorder, and any RDC disorder).
A 2-tailed α level of .05 was used to determine statistical significance. Adjustments for multiple comparisons were not made for 2 reasons. First, to the extent that multiple dependent variables are correlated, the Bonferroni adjustment results in an increasing number of type II errors (ie, the failure to detect true differences).18,19 Subgroups of dependent variables in the study, such as those related to depressive severity, psychosocial function, or long-term course, are highly intercorrelated. Second, we believe that it is most useful if actual probability values are given for each dependent variable because the consistency of findings across related variables supports the overall findings about that area.20
The irritable group was significantly younger than the nonirritable group (P = .002), had a higher percentage of female participants (P = .02), had a lower percentage of participants with at least some college education (P = .03), and had an earlier age at onset of their first lifetime affective episodes (P = .03) (Table 1).
Half (49.3%) of the irritable group and 65.2% of the nonirritable group had a primary MDE at study intake (ie, MDE not preceded in time by another Axis I diagnosis) (P = .001).
The Global Assessment Scale measures of overall severity of the study intake MDE were similar for the irritable and nonirritable groups. However, the irritable group had significantly more severe depressive symptoms at study intake (P = .006) as measured by the CDS “extracted Hamilton” score (sum of all 17 items of the SADS measuring the severity of MDE diagnostic symptoms). The duration of the study intake MDE, from onset to remission/recovery, was also significantly longer for the irritable group (median duration, 91 weeks) than for the nonirritable group (median duration, 49 weeks) (P = .002).
We did not find significant differences on any SADS item assessing symptoms of DSM-IV-TR atypical depression (ie, hypersomnia, hyperphagia, mood reactivity, or leaden paralysis).
In a population study,6 irritability/anger during unipolar MDEs was associated with low impulse control. The CDS group with overt irritability/anger had significantly lower scores on the Ego Control scale (P = .009) of the revised Minnesota Multiphasic Personality Inventory21 and the Restraint scale (P = .01) of the Guilford-Zimmerman Temperament Survey22 (eTable 1 in Supplement). Compared with the nonirritable group, the irritable group also had a significantly higher percentage of participants who engaged in irresponsible, antisocial, or reckless behavior during their study intake MDEs (SADS item 417; 18.2% vs 10.2%; χ21 = 6.68; P = .01) or who had a lifetime diagnosis of antisocial personality disorder at study intake (4.8% vs 1.2%; P = .02, determined by use of the Fisher exact test).
Survival analysis showed a nonsignificant difference between the 2 groups in the length of time from the end of the intake MDE to the start of the subsequent affective episode (Table 2). Although the irritable group had a nonsignificantly shorter median length of time out of any RDC affective episode (144 vs 195 weeks) and a lower probability of remaining free of an MDE for at least a year, the very long first well interval for some participants in each group (up to 27 years) made the overall comparison of survival distribution functions nonsignificant (P = .31).
The irritable group had a more chronic and severe long-term course of illness, with a significantly higher mean percentage of follow-up weeks with affective symptoms (P = .004), in an affective episode (P = .003), or in a major affective episode (P = .03).
By current RDC and DSM-IV-TR definitions, major affective episode remission/recovery is reached when an individual either is fully asymptomatic (has no symptoms of the episode) or has only minimal affective symptoms below the threshold for minor depression/dysthymia or hypomania for 8 consecutive weeks. The irritable group had a significantly higher percentage of time with residual minimal affective symptoms between affective episodes (P = .03).
A significantly higher percentage of the irritable group had a lifetime diagnosis of substance abuse disorders (P < .001)—either alcoholism (P < .001) or drug use disorder (P = .02)—and of all anxiety-related disorders (from P = .001 to P = .04) except obsessive-compulsive disorder (P = .29) (Table 3). Rates of other specific RDC disorders were very low in both analysis groups, being significantly higher for the irritable group only with regard to antisocial personality disorder (P = .005).
Prior to study intake, the irritable group had significantly greater impairment in global functioning (P = .002) and social relationships (P = .03) including with spouse/partner (P < .001) (Table 4), plus a higher rate of poor occupational and/or school performance since 15 years of age (P < .001). During long-term follow-up, the irritable group had significantly worse impairment on the global rating of impairment (P = .001) and LIFE–Range of Impaired Functioning Tool composite score23 (P < .001) and in specific areas of work/occupational performance (P = .04), household tasks (P = .009), and relationship with one’s spouse/partner (P = .03).
At study intake, CDS participants rated their overall level of satisfaction (contentment, gratification, and feeling that their needs and desires have been fulfilled) for the best 6-month period during the prior 5 years. Mean scores for both groups were in the range reflecting “good—generally content with only mild or intermittent dissatisfaction.” However, the irritable group had significantly greater overall life dissatisfaction than the nonirritable group, with a mean (SD) score of 2.25 (0.89) vs 2.05 (0.94) (t533 = 2.43; P = .02).
During their long-term course, the irritable group had a significantly worse overall subjective quality of life than did the nonirritable group, with a mean (SD) score of 2.82 (0.74) vs 2.59 (0.82) (t455 = 3.22; P = .001).
The association of key clinical variables with overt irritability during the study intake MDE remained statistically significant after covarying for each of the 3 available RDC categories of current comorbidity: substance abuse, anxiety disorder, and any substance abuse or other RDC mental disorder (eTable 2 in Supplement). These results indicate that the major findings of our study are not fully explained by the higher rates of comorbid disorders in the irritable group.
There were no significant differences between the 2 groups in suicidal ideation or behavior at study intake or during follow-up, although the irritable group was slightly higher on all measures.
The irritable group showed some characteristics that have been associated with underlying bipolarity. They were significantly younger at study intake (P = .002; Table 1), had a significantly earlier age at onset of first lifetime depression (P = .03; Table 1), and had significantly greater emotional lability, as evidenced by lower mean scores on the Emotional Stability scale (P = .04; eTable 1 in Supplement).
There was no significant difference between the analysis groups in rates of switching to a bipolar diagnosis during long-term follow-up.
The irritable group had a significantly higher percentage of interviewed first-degree relatives with bipolar disorder (P = .02; eTable 3 in Supplement), owing entirely to a higher rate of type II bipolar disorder (P = .008).
The mean (SD) percentage of weeks with any prescribed antidepressant during the study intake MDE was significantly lower for the irritable group than for the nonirritable group: 61.6% (38.9%) vs 68.2% (37.6%) (P = .02, determined by use of the Wilcoxon rank sum test). However, the mean (SD) percentage of weeks receiving a therapeutic level of antidepressant was not significantly different between the irritable group and the nonirritable group: 27.2% (33.0%) vs 32.3% (36.0%) (P = .24, determined by use of the Wilcoxon rank sum test). During the recovery interval following the end of the study intake MDE, the 2 analysis groups did not have significantly different mean percentages of weeks with an antidepressant at any level (P = .29) or at a therapeutic level (P = .48).
We have previously reported on the clinical correlates of overt irritability/anger during bipolar MDEs.24 A second concurrent symptom, psychomotor agitation, was found to be associated with several significant clinical outcomes in the bipolar study, but only with significantly older age and worse depressive severity of study intake unipolar MDEs (eTable 4 in Supplement), leading us to focus the present study on overtly expressed irritability/anger.
To our knowledge, this is the first investigation to examine the association between irritability/anger during unipolar MDEs at study intake and the subsequent long-term course of illness. Quiz Ref IDWe found that 292 of 536 participants (54.5%) who experienced a unipolar MDE had symptoms of overtly expressed irritability/anger during their study intake MDE. Although these are not core symptoms for the diagnosis of an MDE, they are present to a clinically significant degree in more than half of the participants. Concurrent anger/irritability was associated with significantly increased depressive severity and a longer duration of the index MDE. We did not find features of atypical depression, as reported by Benazzi and Akiskal4 and Perlis et al,5 nor an increased history of suicide attempts or suicidal ideation as reported by Perlis et al.5 Although the irritable group did not have more MDEs prior to study intake or during follow-up, their study intake MDEs lasted longer, and their prospectively observed course of affective illness was significantly more chronic (more time symptomatic) and severe (more of follow-up in a major affective episode). In addition to a greater symptomatic burden of illness and psychosocial dysfunction, they also had significantly more comorbidities with anxiety and substance use disorders. Using CDS longitudinal data, we were able to extend the findings from prior cross-sectional studies and identify increased long-term chronicity, severity, psychosocial impairment, and comorbidity and lower life satisfaction associated with overt irritability/anger during the study intake MDE. We found no such long-term outcomes in relation to another common characteristic of unipolar MDEs: psychomotor agitation.
The present study provides limited evidence consistent with the hypothesis of underlying bipolarity25,26 in patients with MDEs accompanied by overtly expressed irritability/anger. Although we did not find a significantly greater risk of developing hypomania or mania during follow-up, our study participants were significantly younger, had an earlier age at onset of depressive illness, and had significantly more first-degree relatives with bipolar II disorder.
Quiz Ref IDThe current literature uses the term irritability in a broad-ranging and often imprecise manner. It has been used to represent inner feelings of annoyance, hostility, or inner psychic tension and as outwardly expressed anger toward people or objects ranging from displeasure or argumentativeness to rage.27 Despite considerable evidence of high prevalence and an association with important clinical features, and despite being a defining characteristic of manic episodes since 1952, the lack of specificity for the term irritability has been an impediment to effective diagnosis, research, and treatment.28 Within this variety of meanings, many of which are vaguely defined constructs, we are fortunate that the SADS included an item with a very explicit and precise meaning—overt expression of irritability, annoyance, or anger with clear behavioral descriptors for a 6-point rating of severity. Use of a single symptom item rather than a standardized scale to examine correlates of irritability/anger during MDEs is consistent with other large studies including the Sequenced Treatment Alternatives to Relieve Depression study5 and the National Comorbidity Survey Replication study.6 The SADS is unique in assessing overtly expressed anger/irritability not only during the past week but also during the week of the study intake MDE when depressive symptoms were at their most severe (worst week). We defined the irritable group based on their having clinically significant overt irritability/anger during either or both of those weeks.
Analysis based on external expression, rather than subjective feelings alone, is closely aligned with the meaning of irritability as measured by the Present State Examination,29 in which it is evidenced by outward expressions such as shouting at or quarreling with other people, throwing or breaking objects, or being physically assaultive to others. It is also similar to the meaning of the “Anger-Out” subscale of the State-Trait Anger Expression Inventory–2 of Spielberger.30 We found this very specific definition of irritability to have a significant association with important clinical characteristics related to long-term illness severity, chronicity, comorbidity, and psychosocial impairment. Those correlates may or may not be associated with other definitions of irritability.
Enrollment in the CDS took place from 1978 through 1981. The interim has seen a number of changes in the treatment of MDDs, changes that could potentially affect the course of illness. Furthermore, CDS patients received a diagnosis and were treated at 5 different academic medical centers in the United States. As was customary at that time, the majority (76%) were initially inpatients. This raises the possibility that the results reported herein may not generalize to nonwhite patients or to patients currently seen in private settings, who may have bipolar or schizoaffective MDE or lifetime schizophrenia, different levels of severity, different levels or types of comorbid disorders, or different symptoms or behaviors causing them to present for treatment.
Questions can be raised about the accuracy of follow-up data obtained at 6-month or yearly intervals. A high level of interrater agreement (an intraclass correlation coefficient of ≥0.80) has been demonstrated for assigning Psychiatric Status Ratings and using them to indicate the start and end of affective episodes.16 The accuracy of psychosocial impairment ratings is supported by their consistency across general measures of functioning (global ratings and Range of Impaired Functioning Tool23), as well as specific areas of work, household duties, and relationship with spouse/mate.
The CDS data on the percentage of weeks with any antidepressant treatment, or with antidepressants at a therapeutic level, does not support any conclusion about the effects of specific treatments, which is beyond the scope of our study. Unfortunately, the CDS contains no data on the presence of overt irritability/anger during MDEs that occurred later during follow-up. Thus, we cannot determine whether this is a persistent characteristic, which could help explain the correlation with a more severe and chronic long-term course of illness characterized by greater psychosocial impairment and poorer quality of life.
Quiz Ref IDThere is a strong confluence of scientific data from other investigators and the results of our study indicating that concurrent anger/irritability symptoms are important indicators of increased severity, chronicity, and complexity of unipolar major depression. We found that, despite its high prevalence, overt irritability had a low correlation (0.00 to 0.09) with criterion symptoms for MDE in our sample. We interpret those findings as indicating that overt irritability/anger should not be regarded as a core symptom of depression but, rather, as a possible indicator of a clinically important subtype of MDE that needs to be confirmed with further research. Our findings that certain personality characteristics (poor impulse control, high rejection sensitivity, and greater emotional lability), as well as measures of greater psychosocial impairment prior to entering the study and during long-term follow-up, suggest that subthreshold or threshold personality disorders, particularly borderline personality disorder, may predispose a person to outward expressions of irritability/anger. We cannot examine this hypothesis because the CDS did not include a diagnostic assessment of personality disorders other than the RDC diagnosis of antisocial personality disorder, which was significantly higher in the group with overt irritability/anger. Nor can we evaluate the possible etiological role of posttraumatic stress disorder, which was also not assessed.
The results of our study support the possibility that unipolar depression with overt irritability/anger may be a marker for a distinct subtype of unipolar MDD. Correlates such as poor anger management, poor impulse control, increased substance abuse, and greater long-term psychosocial dysfunction raise the question of whether MDEs with irritability/anger arise from a distinct biological substrate. Recent evidence indicates a possible relationship between anger attacks during MDEs and serotonergic dysfunction,31- 34 including its effects on specific brain regions,35 as well as subcortical vascular lesions,36 on particular patterns of cerebral blood flow,37 and on factors related to vascular morbidity.38,39 Katz et al31 have compiled an impressive body of evidence suggesting that symptoms of hostile irritability in severely depressed patients are particularly responsive to selective serotonin reuptake inhibitor antidepressants. Selective serotonin reuptake inhibitors were not yet available during the early years of the CDS. In the present study, we did find that antidepressants of any type were prescribed significantly less during study intake MDEs for the group with overt irritability/anger, despite the fact that this group had significantly more comorbidities, more severe episodes of depression, and greater levels of psychosocial impairment at study intake. Further research is needed to explore how combinations of biological mechanisms, personality vulnerabilities, poor social circumstances, or prior trauma may be involved in the pathogenesis of MDE with overt irritability/anger. Findings from such research could lead to the identification of a distinct subtype of unipolar MDD, along with specific diagnostic tools and more effective treatments.
Results from the present investigation have important clinical implications. First, it is important for clinicians and researchers to identify overt irritability/anger in patients who experience an MDE because such symptoms are a clinical marker for a significantly more complex, chronic, and severe form of MDD. Second, closer clinical monitoring of such patients is warranted. Third, the treatment plan should include specific strategies to address anger management issues, as well as the frequently associated problems of comorbid anxiety disorder, substance abuse disorder, poor impulse control, and psychosocial impairment when these are present. Such an MDD requires thoughtful adjunctive treatment and closer monitoring by the clinician to ameliorate these attendant problematic features.
Submitted for Publication: August 23, 2012; final revision received February 26, 2013; accepted February 27, 2013.
Corresponding Author: Lewis L. Judd, MD, Department of Psychiatry, University of California, San Diego, 9500 Gilman Dr, La Jolla, CA 92093 (email@example.com).
Published Online: September 11, 2013. doi:10.1001/jamapsychiatry.2013.1957.
Author Contributions: Dr Schettler had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Judd, Schettler, Coryell, Akiskal.
Acquisition of data: Judd, Schettler, Coryell.
Analysis and interpretation of data: Judd, Schettler, Akiskal, Fiedorowicz.
Drafting of the manuscript: Judd, Schettler.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Judd, Schettler.
Obtained funding: Coryell.
Administrative, technical, or material support: Judd, Fiedorowicz.
Study supervision: Akiskal.
Conflict of Interest Disclosures: Dr Akiskal has received compensation for presentations from the López-Ibor, Lilly, and Lundbeck foundations and from AstraZeneca, Bristol-Myers Squibb and Otsuka, GlaxoSmithKline, Merck, and Sanofi-Synthélabo. Dr Schettler has provided statistical consulting services to BrainCells Inc, Methylation Sciences Inc, Novartis BioVentures, and Clintara LLC. No other disclosures were reported.
Funding/Support: The initial funding source for the database used in this study was the NIMH.
Role of the Sponsors: This study has been approved for submission by the Publications Committee of the NIMH Collaborative Depression Study.
Additional Contributions: This study was conducted with the participation of the following investigators: M. B. Keller, MD (Chairperson, Providence, Rhode Island); W. Coryell, MD (co-chairperson, Iowa City, Iowa); T. I. Mueller, MD, and D. A. Solomon, MD (Providence, Rhode Island); J. Fawcett, MD, and W. A. Scheftner, MD (Chicago, Illinois); J. Haley (Iowa City, Iowa); J. Endicott, PhD, and J. Loth, MSW (New York, New York); and J. Rice, PhD, and T. Reich, MD (deceased) (St Louis, Missouri). Other contributors include H. S. Akiskal, MD, N. C. Andreasen, MD, PhD, P. J. Clayton, MD, J. Croughan, MD, R. M. A. Hirschfeld, MD, L. Judd, MD, M. M. Katz, PhD, P. W. Lavori, PhD, J. D. Maser, PhD, M. T. Shea, PhD, R. L. Spitzer, MD, and M. A. Young, PhD, as well as the following deceased contributors: G. L. Klerman, MD, E. Robins, MD, R. W. Shapiro, MD, G. Winokur, MD, and A. C. Leon, PhD.