Two-Year Outcome of Psychotic Depression in Late Life

The study group was derived from subjects consecutively referred to the Geriatric Affective Disorders Program of The Toronto Hospital. Most referrals to this program are from family physicians or other primary care services. Men and women aged 60 years or older who met DSM-III-R criteria for unipolar major depression and had a score of 16 or more on the 17-item Hamilton Depression Rating Scale (16) after at least 7 days without psychotropic medication were eligible for the study. Exclusion criteria were any other concurrent axis I diagnosis; a lifetime history of DSM-III-R-defined schizophrenia, schizoaffective disorder, paranoid disorder, or dementia; any neurological disorder affecting the central nervous system; acute, uncontrolled medical illness at the time of recruitment, although patients with stable chronic medical conditions could be included; or any condition that contraindicated the use of any of the study medications. Patients who had delusions or hallucinations or both (as defined by the Structured Clinical Interview for DSM-III-R [SCID] [17]) at index assessment were diagnosed as having a psychotic depression. After a complete description of the study to the patients, written informed consent was obtained.

Before treatment, patients underwent a psychiatric interview that employed the SCID, a physical assessment, routine blood work, and ECG. The Mini-Mental State (18) was used to quantify cognitive function; however, a diagnosis of dementia, which was used to exclude patients from the study, was based on a comprehensive history and cognitive evaluation. In particular, no patient had cognitive impairment before the index episode of depression. Concurrent chronic medical problems were rated on the physical illness rating scale of Burvill et al. (19).

Treatment consisted of two phases. The purpose of the first phase was to achieve remission of the index episode of depression. Patients with nonpsychotic depression were treated with nortriptyline, which was increased to a dose of 75 mg/day by the end of the first week. Plasma nortriptyline levels were determined at the end of the second week of treatment, and, if necessary, the dose was adjusted to achieve levels of 150–500 nmol/liter. Patients remained on a regimen of nortriptyline for 6 weeks from the time of its initiation. If patients failed to respond (response was defined as a Hamilton depression scale score of 10 or lower) by the end of the sixth week of treatment, they were then treated with 2 weeks of adjunctive lithium carbonate; a 12-hour postdose lithium plasma level of 0.5–1.0 mmol/liter was achieved by the end of the first week of augmentation. The recommended treatment for patients with psychotic depression was ECT, which was administered with a brief-pulse, constant-current device (MECTA SR-1). Patients were initially treated with fixed high-dose (375–500 millicoulombs), nondominant unilateral ECT three times a week. If there was no improvement (on the basis of the judgment of the study psychiatrist) in depressive symptoms after five treatments, electrode position was changed to bilateral. In the psychotic group, response was defined as a Hamilton depression score of 10 or lower and the absence of delusions and hallucinations, at which point ECT was discontinued. However, if, by the end of 12 treatments, patients had not responded and had failed to show ongoing improvement between treatments 10 and 12, ECT was discontinued at that point. If ECT-induced cognitive impairment severely interfered with the patient's functioning, the frequency of treatments was reduced to twice per week. Psychotically depressed patients who declined ECT were treated with 6 weeks of nortriptyline and perphenazine, followed, if necessitated by nonresponse, by 2 weeks of lithium augmentation. The dosing and duration of nortriptyline and adjunctive lithium were the same as those for patients without psychosis. Perphenazine was started at a dose of 4 mg/day and could be increased by 4–8-mg increments, up to every 4 days, depending on clinical response and side effects.

Patients whose depression remitted with treatment (the start of remission was defined as the point of response followed by at least 2 consecutive weeks of Hamilton depression scores of 10 or lower and, in the case of psychotic subjects, the absence of delusions and hallucinations) entered the 2-year follow-up phase. It should be noted that maintenance treatment was not restricted to patients with a prior history of depression. This decision was based on evidence that patients who experience a first episode of depression after the age of 60 years are also at high risk of recurrence of depression (15). Patients continued to take the same dose of medication that led to response. Nortriptyline and lithium were continued throughout the study. However, in order to minimize the risk of tardive dyskinesia, perphenazine was withdrawn (by tapering the dose over a 4-week period) 16 weeks after the start of remission. For patients who responded to ECT, nortriptyline was titrated up to a therapeutic dose within 2 weeks after ECT was stopped. Patients who responded to ECT were not prescribed prophylactic antipsychotic medication or lithium. During this 2-year phase of the study, patients were followed on a monthly basis for clinical assessment of their mental state, medication side effects, and compliance with treatment. Plasma nortriptyline and, when applicable, lithium levels were determined on a periodic basis and were maintained in the range of 150–500 nmol/liter and 0.5–1.0 mmol/liter, respectively.

The outcome events of primary interest were relapse and recurrence of depression. Relapse was defined as the reemergence of depressive symptoms within 16 weeks of remission of the index episode of depression. Recurrence was defined as a new episode of depression occurring after at least 16 consecutive weeks without a relapse. Relapse and recurrence were diagnosed when the patient met symptomatic criteria for DSM-III-R major depression for at least 1 week and had a Hamilton depression score of 16 or above. Reemergence of psychotic symptoms was not required for the diagnosis of relapse or recurrence. If a patient required a change of treatment as a result of a relapse or recurrence, the change was made only after it was determined that she or he met the study's criteria for these events.

Comparison of characteristics between groups was performed by t tests for continuous variables and chi-square analysis (or Fisher's exact test, when appropriate) for categorical variables. Long-term outcome was analyzed in two ways. First, the frequency of relapse and recurrence was compared by chi-square analysis or Fisher's exact test. Stepwise logistic regression analysis was then used to examine whether psychosis had an impact on the frequency of relapse or recurrence independent of other clinical variables. Second, survival methods (the Cox proportional hazards model and the Kaplan-Meier product limit method) were used to estimate the probability of a patient remaining well without relapse or recurrence during the 2-year follow-up. For all analyses, statistical significance was set at p≤0.05 (two-tailed tests).

Of the 283 patients initially evaluated, 126 (25 psychotic and 101 nonpsychotic) met entry criteria and entered the first phase of treatment. Of the psychotic group, 17 patients underwent ECT, and 15 (88.2%) responded (eight to unilateral ECT alone and seven to unilateral crossed over to bilateral ECT; mean number of treatments=9.7, SD=3.4). Eight psychotic patients opted for pharmacotherapy, and four (50%) responded (two to nortriptyline and perphenazine and two to these medications plus lithium). Thus, 19 (76.0%) of 25 psychotic patients responded to treatment of the index episode. Sixty-eight (67.3%) of the nonpsychotic group responded to treatment (61 to nortriptyline alone and seven to nortriptyline and lithium). In addition to having Hamilton depression scores of 10 or below, all responders had a 50% or greater reduction in baseline Hamilton depression scores. Psychotic and nonpsychotic groups did not significantly differ in the rate of attrition during treatment of the index episode (4.0% and 17.8%, respectively) (p=0.12, Fisher's exact test).

Table 1 describes the demographic and clinical characteristics of the 87 patients who entered the 2-year follow-up phase. Patients with psychotic depression had a higher mean Hamilton depression score at index assessment (p<0.001) and were more likely to have had a past history of depression (p=0.03), more depressive episodes (p=0.001), and previous treatment with ECT (p<0.001, Fisher's exact test). Patients without psychotic features had a longer depressive episode at index assessment (p=0.03). At the time of index assessment all 19 psychotic subjects had delusions, and two (10.5%) of them also had hallucinations.

At the time of starting follow-up, psychotic and nonpsychotic groups, respectively, had mean nortriptyline doses of 69.7 mg/day (SD=21.4) and 70.1 mg/day (SD=20.1). The groups did not significantly differ on mean plasma nortriptyline levels at this stage of treatment (mean=336.1 nmol/liter, SD=116.4, versus mean=325.9, SD=113.5), at 12-month follow-up (mean=321.9, SD=96.7, versus mean=316.4, SD=98.6), at 24-month follow-up (mean=317.2, SD=80.0, versus mean=319.7, SD=106.8), or at the point of relapse or recurrence (mean=341.4, SD=87.5, versus mean=313.5, SD=119.6). The four psychotic subjects who responded to combined medications were taking a mean dose of 20.0 mg/day (SD=8.9) of perphenazine at the start of follow-up.

Table 2 indicates the number and percent of patients in each group who had a relapse or recurrence of depression (for brevity, these individuals will be collectively referred to as relapsers) or who left the study by the end of 2 years of follow-up. No incident cases of dementia were detected during the follow-up period. The psychotic group was more likely to have had either a relapse or a recurrence (p=0.005, Fisher's exact test; odds ratio=5.2, 95% confidence interval=1.7–16.0) or a recurrence only (p=0.01, Fisher's exact test; odds ratio=5.1, 95% confidence interval=1.5–17.2) than was the nonpsychotic group. Psychotic and nonpsychotic groups did not significantly differ in the proportion of relapsers who were aged 75 years and older (55.6% and 40.0% of relapsers, respectively). Since psychotic and nonpsychotic groups differed on baseline Hamilton depression scores, the number of previous episodes of depression, and duration of the index episode, a stepwise logistic regression analysis was performed across the cohort to assess whether patient group predicted the frequency of relapse or recurrence independent of these variables. Psychosis grouping was the only variable significantly associated with this outcome (χ²=8.31, df=1, p=0.004).

Within the psychotic group, eight (53.3%) of 15 subjects who responded to ECT became ill (one relapse, seven recurrences), compared to one (25.0%) of the four patients who responded to medication (one relapse during treatment with nortriptyline and perphenazine) (p=0.58, Fisher's exact test). Rates of relapse/recurrence following ECT have been found to be increased in patients who were previously unresponsive to adequate pharmacotherapy (20). Therefore, we examined whether ECT-relapsers and nonrelapsers differed in the treatments they had received for the index episode before entering the study. None of these individuals had been treated with combined antidepressant-antipsychotic medication or ECT; two (25.0%) of eight relapsers and two (28.6%) of seven nonrelapsers had received 4 weeks or more of a tricyclic antidepressant at 75 mg/day or more.

Figure 1 presents the Kaplan-Meier survival curves for psychotic and nonpsychotic groups. The cumulative probability of remaining well without relapse or recurrence was 0.488 for the psychotic group and 0.821 for the nonpsychotic group (p=0.001). The mean survival times for psychotic and nonpsychotic groups were 68.9 weeks (SE=8.7) (95% confidence interval=51.9–85.9) and 95.4 weeks (SE=2.9) (95% confidence interval=89.8–100.9), respectively.

A Cox proportional hazards analysis was performed to examine the effect of covariates on outcome as defined by time to first relapse or recurrence. In addition to psychosis grouping, the following demographic and clinical variables were entered into the model: age, age at onset of the first episode of depression, sex, number of previous episodes, duration of the index episode at the time of assessment, time to respond to treatment of the index episode, Hamilton depression score at index assessment, the sum of severity and disability scores on the chronic physical illness rating scale at index assessment, and life events (as determined by the Bedford Life Events and Difficulties Schedule modified for the elderly [21]) in the 6 months before finishing the study. Following stepwise backward elimination, psychosis grouping was the only variable that was significantly related to outcome (χ²=8.91, df=1, p=0.003).