| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
LETTERS TO THE EDITOR |
Department of Psychiatry and Behavioral Sciences; Box 3119; Duke University Medical Center; Durham, NC 7710
Ford suggests that it cannot be concluded that exercise and sertraline were superior to placebo because the conventional 
.05 significance level was not achieved (p = .057). We recognize that some readers may discount our findings for this reason. However, the claim that excluding "early responders" from the analysis introduced instability into the analysis (in which the p value improved to .022) is unlikely given the relatively small number of patients (n = 14) who were removed from the primary analysis. We did not intend to suggest that sertraline and exercise were "equivalent," as such terminology has a special technical meaning. Equivalence trials typically aim to determine whether one treatment (e.g., exercise) is therapeutically similar to another, usually an existing treatment (e.g., sertraline) (1). We purposefully did not use the term "equivalent" because the study was not designed for such a purpose and our trial also lacked the statistical power to detect group differences in the active treatments. Thase (2) has suggested that at least 300 patients per arm are needed to compare two active anti-depressant treatments, but few randomized trials have ever enrolled this many patients. To our knowledge, our study is the largest exercise trial of patients with major depressive disorder (MDD) conducted to date, and is the only placebo-controlled exercise study of which we are aware. We stated that the two interventions appeared "comparable" based upon the observation that 47% of patients receiving sertraline were considered "remitted" compared to 45% of patients in supervised exercise and 40% in home-based exercise. Clearly a larger trial would be necessary to demonstrate equivalence or noninferiority.
In his letter, Malt suggests that an important limitation of our study was that many patients with MDD enrolled in our trial had relatively mild symptoms. The average HAM-D for our sample was 17, with 78 patients (39%) obtaining HAM-D scores 
18. We defined "moderately/severely" depressed as 18 on the HAM-D for conditional randomization purposes to ensure that MDD severity would be similar across groups; to this end, our strategy was successful insofar as there were no baseline group differences in MDD severity. We also noted that there were no differential effects of the treatments as a function of MDD severity at the conclusion of the 16-week trial. We observed that even patients with HAM-D scores >20 could exercise and achieve similar clinical benefits to those individuals with less severe MDD. We agree that patients with scores >20 may be more typical of patients enrolled in pharmaceutical trials, and may be more common in outpatient psychiatric settings, but such patients are not necessarily more representative of outpatients seen in general medicine or family practice settings. Our participants were patient volunteers, and as Malt correctly notes the majority were at least willing to consider exercise as a potential treatment. However, to be enrolled in our trial, patients also had to be willing to potentially receive medication or a placebo pill, which may also have affected patients’ willingness to volunteer for our study.
Although Malt is critical of our methodology, we don’t believe that the diagnostic process that we employed was a significant shortcoming of our study. We used a standardized psychiatric interview to diagnose depression and to rate the severity of symptoms. In addition to the Structured Clinical Interview for Depression (SCID) (3) and Hamilton Rating scale for Depression (HAM-D) (4), clinical interviewers administered all sections of the SCID pertaining to the following disorders: Mood Disorders (including Major Depressive Disorder, Bipolar I and II Disorders, Mood Disorder Due to a General Medical Condition, Substance Induced Mood Disorder, and Mood Disorder NOS), Psychotic Symptoms, Substance Use Disorders, and Anxiety Disorders. Interviews were audiotape recorded for quality control. In order to evaluate inter-rater reliability, 20 randomly selected interviews were independently rated by two clinicians. The presence of MDD was confirmed in all cases, and the Shrout-Fleiss reliability coefficient was 0.98, suggesting that our diagnostic assessments and clinical ratings of depression severity were highly reliable.
Patients with a primary psychiatric diagnosis other than MDD were excluded from the trial. Patients were excluded if they met diagnostic criteria for bipolar disorder, dementia, delirium, current obsessive compulsive disorder, schizophrenia, schizoaffective, or other psychotic disorder, psychotic features including any delusions or hallucinations during the current depressive episode, current alcohol or other substance abuse disorder, and patients who posed an acute suicide (e.g., score >2 on the suicide item on the HAM-D) or homicide risk. Patients with comorbid generalized anxiety disorder or panic disorder were permitted to participate in the trial, provided that MDD was their primary psychiatric diagnosis. In fact, 56 patients (28%) were considered to have a concurrent anxiety disorder.
Malt also noted that our sample was overweight, which is consistent with epidemiological studies that have reported obesity to be associated with a 25% increase in risk of mood disorders (5). The body mass index (BMI) of our sample was 30 ± 7.3 kg/m2; in contrast, the epidemiological study by McIntyre and colleagues (6), to which Malt referred, consisted of 36, 984 Canadians aged 15 years and older in which the BMI for the sample was 25.96 kg/m2. This difference can be explained by the fact that the Canadian sample was younger, included a smaller proportion of women (54% vs 76%), and unlike our current intervention trial, did not exclude patients who were physically active or who were on anti-depressant medications. At Malt’s suggestion, we examined the possibility that obesity might moderate the relationship between treatment and depression status. We found that the BMI by treatment group interaction term in the fully adjusted model predicting the primary endpoint was not significant (p = .60). Thus, it is unlikely that the BMI of our sample limits the generalizability of our findings. Furthermore, because there were no baseline treatment group differences in BMI, we don’t consider body weight to be a threat to the internal or external validity of our study. We agree with Malt that more studies are needed to fully understand how exercise can be used effectively in the treatment of mood disorders.
DOI:10.1097/01.psy.0000311332.34779.56
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
