It is generally accepted that unipolar major depression tends to run a chronic or recurrent course (1–3). Although clinical (1–3) and biological (4, 5) predictors of episode recurrence have been identified, the mechanism remains poorly understood.

Thyroid hormones have been extensively studied in major depressive illness. It is well established that thyroxine (T₄) levels are higher in treatment responders (6) and decrease with antidepressant response (7) but are not associated with degree of treatment resistance. Studies of thyroid hormones in major depression have largely been confined to acute episodes and not to the long-term course, although it has been shown that perturbations of the thyroid axis may influence the course of illness in bipolar disorder (8). Therefore, we studied the relationship between basal thyroid hormone levels and recurrence of illness in a group of patients with unipolar depression.

Seventy-five outpatients (25 men and 50 women) whose mean age was 37.9 years (SD=11.3) gave written and oral consent to be included in a database of subjects treated under usual clinical conditions. All had a diagnosis of nonpsychotic unipolar major depressive disorder, determined by using the Schedule for Affective Disorders and Schizophrenia—Lifetime Version (SADS-L) (9), a structured interview that uses Research Diagnostic Criteria (RDC) (10). All subjects had a minimum score of 16 on the 17-item Hamilton Rating Scale for Depression (11) at study entry. At that time, blood was drawn for measurement of T₄ and triiodothyronine (T₃) by standard radioimmunoassays and for measurement of thyrotropin (TSH) by ultrasensitive assay.

Patients were treated by clinician choice and were evaluated in the clinic at 1–3-month intervals. Response to treatment was defined as the absence of an RDC-defined major depressive episode and a score of 1 or 2 on a 5-point clinical global improvement scale (1=remission, 2=marked improvement, 3=moderate improvement, 4=minimal improvement, 5=no response). Recurrence was defined as follows: 1) stable response for a minimum of 2 months, 2) occurrence of RDC-defined major depressive episodes determined by the SADS-L, and 3) need for further treatment intervention and/or hospitalization.

Thyroid hormone levels were collected at the beginning of the index episode just before antidepressant treatment was given and were recorded separate from course of illness data. The study period extended over a 10-year period; patients were followed as long as possible or until recurrence. Data were analyzed by using Cox regression survival analysis, with each thyroid hormone (T₄, T₃, and TSH), comorbid anxiety, sex, age, duration of current episode, duration of illness, and number of previous episodes entered as covariates. A stepwise method was used with an inclusion value of p<0.05 and a removal value of p>0.10.

The survival curve for the study group (including positions of "censored" cases; i.e., patients who had not suffered a recurrence by the end of the study period) is shown in F1. The significant positive predictors of recurrence were presence of comorbid anxiety (Wald χ²=21.6, df=1, p<0.0001), higher numbers of previous episodes (Wald χ²=7.0, df=1, p<0.01), and increased durations of current episodes (Wald χ²=5.3, df=1, p<0.03). Furthermore, T₃ was significantly related to recurrence (Wald χ²=43, df=1, p<0.04) but T₄ was not. Specifically, each 0.10-μg increase in T₃ level was associated with a 22% decrease in risk of recurrence.

In addition to well-described predictors of recurrence in unipolar depression, such as number of previous episodes, duration of current episode, and comorbidity (1–3), we found that serum T₃ levels but not T₄ levels were inversely related to time to recurrence of a depressive episode. Our data are of interest because T₃ is more commonly used to augment antidepressant response (reviewed in reference 7), suggesting that elevation of T₃ levels in the circulation and presumably in the brain may have a beneficial effect on depression.

Our study is limited by several methodological problems, including the relatively small number of subjects and the lack of precise measures of depressive symptoms. Moreover, the studies to date of T₃ augmentation (reviewed in reference 7) have been of very short duration. On the basis of the findings from our study, it would be of interest to see whether long-term T₃ treatment may have a beneficial effect on not only acute episodes but also the risk of future recurrence of depression.

Received Nov. 16, 1999; revision received March 15, 2000; accepted May 10, 2000. From the Mood Disorders Program and Department of Psychiatry, Faculty of Health Sciences, McMaster University. Address reprint requests to Dr. Joffe, McMaster University, 1200 Main St. West, Rm. 2E1, Hamilton, Ont., L8N 3Z5, Canada; joffe@fhs.csu.mcmaster.ca (e-mail).