Reply to Predicting Suicide Attempt Risk: Logistic Regression Requires Large Sample Sizes

Alison M. Gilbert, PhD; Raphael J. Braga, MD; Katherine E. Burdick, PhD

J Clin Psychiatry 2011;72(12):1698

Article Abstract

Letter to the Editor

Because this piece does not have an abstract, we have provided for your benefit the first 3 sentences of the full text.

We thank Dr Bagley for his comments on our article. We agree that a larger sample size would have been optimal; however, given that the predictors included in the models were not altogether independent of one another (eg, the cognitive measures have substantial intercorrelations), the number of truly independent factors tested was—to some degree—reduced. Nonetheless, as we acknowledged in the article, there is always a possibility that a study with negative findings is underpowered to detect a true effect.

See original letter by Bagley, related article by Gilbert, et al, and correction for article by Gilbert, et al.

Dr Gilbert and Colleagues Reply

To the Editor: We thank Dr Bagley for his comments on our article. We agree that a larger sample size would have been optimal; however, given that the predictors included in the models were not altogether independent of one another (eg, the cognitive measures have substantial intercorrelations), the number of truly independent factors tested was—to some degree—reduced. Nonetheless, as we acknowledged in the article,1 there is always a possibility that a study with negative findings is underpowered to detect a true effect. We did attempt to evaluate the degree to which our study was powered to detect effect sizes previously reported in the literature and noted this in the article. As with most studies, especially those reporting negative results, replication will be critical.

In response to Dr Bagley’s question about Table 3 and odds ratios (ORs) reported in the article, he is correct in pointing out the error: instead of reporting the raising of e to the power of Î², we reported Î² twice. The correct OR values for the respective predictors are as follows: age, 0.98; sex, 0.48; race, 0.95; mood type, 3.11; psychosis, 0.04; trauma history, 0.28; Barratt Impulsiveness Scale-version II total score, 0.99; Aggression Questionnaire, 0.99; Reason for Living Inventory total score, 0.98; 17-item Hamilton Rating Scale for Depression, 0.90; Clinician-Administered Rating Scale for Mania, 1.02; Wide Range Achievement Test reading subscale, 0.93. We apologize for this error.

Reference

1. Gilbert AM, Garno JL, Braga RJ, et al. Clinical and cognitive correlates of suicide attempts in bipolar disorder: is suicide predictable? J Clin Psychiatry. 2011;72(8):1027-1033. PubMed doi:10.4088/JCP.10m06410

Alison M. Gilbert, PhD

[email protected]

Raphael J. Braga, MD

Katherine E. Burdick, PhD

Author affiliations: Division of Psychiatry Research, Zucker Hillside Hospital, Glen Oaks, New York. Dr Burdick is now at the Department of Psychiatry and Neuroscience, Mount Sinai School of Medicine, New York, New York. Potential conflicts of interest: The authors of this response have no conflicts of interest to disclose. Funding/support: Financial support for the work referenced in this letter included grants from the National Institute of Mental Health (NIMH) to Dr Burdick (K23MH077807; R03MH079995).

doi:10.4088/JCP.11lr07393a

Download Free PDF

Save Save