A Prospective 12-Year Study of Subsyndromal and Syndromal Depressive Symptoms in Unipolar Major Depressive Disorders FREE

LONGITUDINAL STUDIES of unipolar major depressive disorder (MDD) by Angst et al^{1 - 2} and Keller et al^{3 - 5} have described relapse and chronicity of major depressive episode (MDE) during the course of MDD. Detailed characterization of the disease between MDEs has rarely been the focus of systematic investigation. Some information on clinical course has been gleaned indirectly from cross-sectional studies of MDD in which alternative depressive subtypes have been described, eg, dysthymic disorder,⁶ minor depressive disorder (MinD),⁷ double depression,⁸ recurrent brief depression,⁹ and subsyndromal or subthreshold depressive symptoms (SSDs).^{10 - 16}

Study of depressive subtypes suggests that the clinical course of MDD may be more pleomorphic and varied than previously thought. This has stimulated debate about whether MDD is a single clinical disease characterized by a continuum of different subtypes or levels of symptoms vs the alternative proposition that unipolar disorders are a cluster of distinct clinical entities, each with different clinical characteristics and biological substrates.

After a lifetime investigating mood disorders, Winokur¹⁶ proposed that unipolar depression is clinically homogeneous but etiologically heterogeneous. Support for this position has been provided recently by various research groups.^{17 - 21} For example, we found a high rate of change in depressive subtype diagnoses for epidemiologic catchment area respondents with one diagnosis at wave 1 and another symptom level or depressive subtype 1 year later at wave 2, suggesting a dynamic, changeable clinical course of MDD even during short follow-up.¹⁷

To provide a clear test of the single clinical illness hypothesis of unipolar depression, we conducted weekly analyses of depressive symptom severity in a large cohort of patients with MDD evaluated prospectively for up to 12 years in the National Institute of Mental Health Collaborative Depression Study (CDS).^{22 - 23} We anticipated that if the hypothesis was correct, patients with MDD would present over time with multiple levels of depressive symptoms, and symptom levels would change relatively often.

The analysis sample consisted of all 431 patients with unipolar MDD who entered the CDS^{22 - 23} during an MDE at 1 of 5 tertiary care centers from 1978 to 1981. Patients were diagnosed using Research Diagnostic Criteria²⁴ based on the Schedule for Affective Disorders and Schizophrenia.²⁵ No patient had evidence of bipolar disorder (mania, hypomania, cyclothymic personality), schizoaffective disorder, or schizophrenia prior to intake or during follow-up. Subjects were white, spoke English, had an IQ score of at least 70, and had no evidence of organic mental disorder or terminal medical illness. Informed consent was obtained.

To evaluate the influence of mood disorder history on the course of illness, the MDD cohort was divided into 3 mutually exclusive groups: (1) patients in their first lifetime MDE in the absence of ongoing dysthymia (n=122, 28% of the sample), (2) patients with 1 or more prior MDEs in the absence of ongoing dysthymia (n=205, 47% of the sample), and (3) patients with double depression (MDE superimposed on ongoing dysthysmic disorder) (n=104, 24% of the sample). The demographic and clinical characteristics of the total MDD sample and the 3 subgroups are summarized in Table 1.

Trained raters interviewed patients every 6 months for the first 5 years and yearly thereafter using the Longitudinal Interval Follow-up Evaluation (LIFE).²⁶ Patient interviews were the primary information source for the LIFE, with chronological memory prompts used to obtain accurate information on changes in weekly symptom severity for all mood or other mental disorders. Interviews were supplemented through the first 5 years by detailed review of clinical, medical, and research records. Weekly depressive symptoms were rated using the LIFE Psychiatric Status Rating (PSR) scales,²⁶ as shown in Table 2. All CDS raters undergo rigorous training, resulting in very high intraclass correlation coefficients (ICCs) for rating changes in depressive symptoms (ICC=0.92), recovery from an episode (ICC=0.95), and subsequent appearance of depressive symptoms (ICC=0.88).²⁶

Raters also estimated, on a 5-point scale, information reliability (accuracy) for each LIFE form.²⁶ Of the 5776 LIFE forms available (containing 195825 weekly PSR ratings), 79.8% were rated "very good" or "good" and 18.2% "fair". Individual LIFE forms with "poor" or "very poor" ratings (2%) were omitted from the analyses. No patients were excluded from the analyses due to missing or unreliable data. Only 36 (8.4%) of 431 of patients with MDD had any missing or unreliable data before the end of their follow-up; for most, this represented 6-month or 1-year blocks of time. Because of the frequent changes in weekly PSR status, we felt it was not appropriate to impute these missing data and they were omitted from the analyses.

Follow-up for the 431 patients with MDD averaged 454.4 weeks (SD, 216.1 weeks), or 8.7 years (median, 572.0 weeks; range, 2-624 weeks). The distribution is skewed, since 195 (45.2%) of 431 patients had the full 12 years (624 weeks) of weekly PSR data, 272 patients (63.1%) had 10 years or more, 339 (78.7%) had 5 years or more, and only 46 (10.7%) had less than 2 years.

Weekly depressive symptoms were first rated using the PSR-MDD, PSR-MinD, and DSM-III scales (Table 2).²⁷ Next, follow-up weeks were assigned to 1 of 4 mutually exclusive depressive symptom severity levels (Table 3), independent of whether the patient was in an RDC episode. The 4 depressive symptom levels represent states of illness activity anchored to the diagnostic threshold for commonly observed depressive subtypes or asymptomatic status and constitute a continuum of severity: level 4 (most severe), at the threshold for MDD; level 3 (moderately severe), at the threshold for MinD; level 2 (mildly severe), SSDs (≥1 symptom below the diagnostic threshold for MDD or MinD); and level 1 (least severe), asymptomatic (complete absence of depressive symptoms and return to usual self).

Composite antidepressant treatment scores were calculated for the first 8 weeks of intake MDD episodes (Table 1) and for each week of follow-up.⁵ Mean percentages of weeks patients received any antidepressant medication (composite antidepressant score ≥1) were compared for each level of depressive symptom severity.

Total weeks at the 4 depressive symptom severity levels were summed and expressed as a percentage of the total number of weeks of available data for each patient. Primary analysis variables were percentages of follow-up time spent at the 4 depressive symptom severity levels. Percentages were also collapsed into binary categories and analyzed in terms of none vs any time at each level and at all possible combinations of the 4 levels. The number of symptom level changes per year and total changes during follow-up were also analyzed.

A 2-way analysis of variance was conducted to evaluate time at the 4 symptom levels and changes in symptom levels as a function of intake mood disorder history, site, and the interaction of these variables. Tukey-Kramer²⁷ post hoc t tests were used to compare group means, controlling for the type I error rate for pairs of subgroups. Arcsine transformation was performed to create uniform units of measurement and consistent variance over the entire range of values, and χ² tests or analyses of variance were used to analyze group or site differences in demographic and clinical descriptors. Analyses were performed using the SAS statistical analysis software package.²⁸ Two-tailed α=.05 was used for all significance tests.

As summarized in Table 1, 61% of patients in the overall CDS MDD cohort were women, 52% were married, the mean intake age was 40.3 years, the mean age at first depressive episode was 29.4 years, 71.7% had experienced at least 1 prior depressive episode, the median number of lifetime depressive episodes was 2, and 33.5% had experienced episode onset before age 21 years. Among the groups divided by mood disorder history at intake, the double depression subgroup had significantly earlier onset and had the most patients (52.9%) with onset before age 21 years.

As shown in Table 4, patients with MDD spent 41.5% of follow-up weeks asymptomatic, 26.7% of weeks with MinD symptoms, 16.5% of weeks with SSDs, and 15.3% of weeks with symptoms at the MDD level. Patients with their first lifetime MDE were symptom-free in 53.8% of weeks, significantly more often than the other 2 mood disorder history subgroups. There were no significant differences in the percentage of weeks spent at the SSD or MDD symptom levels. Patients in the double depression subgroup spent significantly more time with MinD symptoms (34.6%) than patients in the other 2 subgroups. The age of onset was significantly different between the groups but did not account for differences in time at the 4 symptom levels.

Patients with unipolar MDD manifested symptoms during 59% of follow-up weeks, and 27% never had an asymptomatic week during follow-up. To allow sufficient recovery time from the intake MDE, additional analyses were conducted on 385 patients with at least 2 years of follow-up. Table 5 shows that 22.6% of these patients were never free of depressive symptoms at any time, with rates higher for the prior episode (26.4%) and the double depression (25.3%) subgroups than for the first episode subgroup (14.3%).

Table 5 shows that 59.0% of patients with MDD with at least 2 years of follow-up spent weeks at each of the 4 levels of depressive symptom severity, 28.8% at 3 levels, and 11.2% at 2 levels; only 4 patients (1.0%) spent their entire follow-up at 1 symptom level. Patients changed symptom levels an average of 1.8 (SD, 1.4) times per year, a total of 17.1 (SD, 15) changes during follow-up. Patients in the first episode subgroup had significantly fewer total changes (13.1 [SD, 12.3]) than patients in the prior episode (18.8 [SD, 16]) or double depression (18.4 [SD, 15]) subgroups.

During 25989 person-weeks at the MDD level, patients received any antidepressant treatment (composite antidepressant score ≥1) in 60.8% of weeks. During 49388 person-weeks at the MinD level, antidepressants were administered in 54.4% of weeks, and during the 32108 person-weeks at the SSD level, antidepressants were administered in 51.3% of weeks. During 88266 person-weeks asymptomatic, antidepressants were administered in 27.8% of weeks.

The National Institute of Mental Health CDS is a 5-site, prospective, naturalistic longitudinal study of patients with affective illness that began during 1978 to 1981 and is ongoing. There were site differences in some demographic and clinical characteristics as a result of the CDS sampling strategy to obtain a diverse unipolar MDD cohort; however, site differences did not account for differences in symptomatic course. For purposes of reference, symptom levels can be associated with symptomatic thresholds of common depressive subtypes or comparable DSM-IV categories and asymptomatic status. Although interrater agreement was high (ICC ≥ 0.88), there may have been some degree of error in assigning PSR levels. One would expect any such error to attenuate systematic group differences, which proved to be quite robust, providing support for the reliability of symptom severity rating. In addition, since patients entered the CDS during an MDE, the amount of follow-up time available for each patient is positively correlated with the percentage of time asymptomatic (r=+0.21) and negatively correlated with time at the MinD and MDD symptom levels (r=−0.13 and−0.20, respectively). Although statistically significant, this accounts for no more than 4.4% of the variance of the percentage of time at the 4 depressive symptom severity levels and has relatively little explanatory power for our findings. The interpretability of CDS treatment data is limited, since treatment was not controlled in this study. The treatment data reflect standards of care in the regions surrounding the 5 academic centers beginning in 1978; these standards have changed over time as new antidepressants (etc) and psychotherapeutic techniques have been introduced into clinical practice.

Weekly analyses of unipolar MDD revealed prolonged chronic symptoms in the course of illness, with only 41.5% of weeks asymptomatic during an average of 8.7 years. Twenty-three percent of patients with at least 2 years of follow-up were never free of depressive symptoms for even 1 week. These new findings add the dimension of prolonged lifetime symptomatic chronicity to the grim prognosis of frequent MDE relapse/recurrence that characterizes unipolar MDD.^{1 - 5 ,29 - 30} It appears that the disease model of unipolar MDD is more analogous in terms of lifetime chronicity to hypertension than to lobar pneumonia or other acute episodic diseases.

Patients in their first lifetime MDE had significantly more benign courses of illness, experiencing their first depressive episodes later; fewer had onset of their first episode before age 21 years, and they were asymptomatic during more than half of the follow-up weeks. Patients with a history of MDEs, in contrast, had a significantly earlier onset of illness and had more chronic symptoms during follow-up. Patients with double depression had the most chronic symptomatic courses and earliest age of first episode onset; more than 50% had onset of their first episode before age 21 years. Once chronic episodes of dysthymia appeared during the course of MDD, they persisted and recurred. These group differences were consistent across all CDS sites and were not accounted for by differences in age of onset, affirming the strength of these findings.

In the total cohort, symptoms at the MinD level accounted for 26.7% of follow-up weeks, compared with 16.7% at the SSD level and 15.3% at the MDD level. Although the percentage of weeks at the MDD level was the lowest, it is striking that more than 15% of follow-up weeks were spent at the most severe level of depressive symptoms. Unipolar MDD has classically been defined by episodes of major depression; however, symptomatic analyses indicate that the weekly course of unipolar MDD primarily involves MinD and SSD symptoms (43.2% of follow-up weeks); these symptoms were approximately 3 times more common during the course of illness than MDD symptoms.

In this unipolar MDD cohort, SSDs were residual to MDEs and were as common as MDD symptoms during the longitudinal course of unipolar depression. Compared with no symptoms, SSDs are associated with small but significant decreases in psychosocial function^{10 - 15 ,17 ,31} ; thus, we conclude that residual SSDs constitute an active state of illness in unipolar MDD, indicating that patients have not recovered completely from their prior MDEs. There is a need for systematic investigation of the clinical significance and treatment responses of MinD and SSDs, which combined are the most common clinically active states in unipolar major depressive disorders.

The vast majority of patients (88%) spent some follow-up weeks at 3 or 4 different levels of depressive symptom severity, and they changed symptom levels almost twice a year. This is consistent with data we¹⁷ and others^{19 - 21} have reported indicating that different levels of depressive symptom severity are very commonly observed in the same patients during their course of illness. Weekly symptomatic analysis of patients with unipolar MDD during up to 12 years of follow-up supports the proposal of Kendler and Gardner,³² based on twin studies, that unipolar depression manifests as a continuum of depressive symptoms that vary in severity and duration. We believe a growing confluence of scientific evidence^{16 - 17 ,19 - 21 ,32} supports the hypothesis that unipolar MDD is a clinically homogeneous illness in which major, minor, and subsyndromal depressive symptoms commonly alternate as different manifestations and levels of illness activity.

Clinicians evaluating patients with MDD should extend their thresholds for detecting clinically relevant illness to include all symptomatically active clinical states of unipolar depression. Researchers and public health experts should recognize that, although MDD is the most severe state of illness, the traditional, relatively exclusive focus on the MDD level of symptoms represents only the tip of the iceberg in this common, chronic, and disabling disease.

Accepted for publication March 24, 1998.

Funds for this study were provided in part by the Mental Health Clinical Research Grants PHSMH30914 and PHSMH49671from the National Institute of Mental Health, Rockville, Md, and by the Roehr Fund of the University of California, San Diego.

This manuscript has been reviewed by the Publications Committee of the Collaborative Depression Study and has its endorsement.

Additional contributors to this study include: Philip W. Lavori, PhD, M. Tracie Shea, PhD, Jan Fawcett, MD, William A. Scheftner, MD, James Haley, Jo Ellen Loth, MSW, Theodore Reich, MD, Nancy C. Andreasen, MD, PhD, Paula J. Clayton, MD, Jack Croughan, MD, Gerald L. Klerman, MD (deceased), Robert M. A. Hirschfeld, MD, Martin M. Katz, PhD, Timothy Mueller, MD, Eli Robins, MD, Robert W. Shapiro, MD, Robert L. Spitzer, MD, George Winokur, MD (deceased), and Michael A. Young, PhD.

Reprints: Lewis L. Judd, MD, Department of Psychiatry, University of California, San Diego, 9500 Gilman Dr, La Jolla, CA 92093-0603.