DEPRESSION AND ANXIETY 31:72–76 (2014) Research Article RECURRENCE OF MAJOR DEPRESSIVE EPISODES IS STRONGLY DEPENDENT ON THE NUMBER OF PREVIOUS EPISODES Andrew Bulloch, Ph.D., 1,2,3∗ Jeanne Williams, M.Sc., 1 Dina Lavorato, M.Sc., 1 and Scott Patten, M.D., Ph.D. 1,2,3 Background: A history of past major depressive episodes (MDEs) is known to be a risk factor for future MDEs. Additional information about the relation- ship between past and future episodes would be useful in clinical practice, it is therefore important to fully understand the epidemiology of major depression. We asked whether the number of previous MDEs is related to the probability of recurrence in the general population. Methods: Data were used from the Canadian National Population Health Survey (NPHS) that was repeated every 2 years from 1994/1995 to 2009/2010 (i.e., nine cycles). Prior year depression was assessed with the Composite International Diagnostic Interview Short Form (CIDI-SF). We estimated the total number of MDEs in individuals over the first eight cycles and examined recurrence in the ninth cycle. These analyses employed a generalized linear model (identity link) where recurrence in cycle 9 was the outcome and the predictor variables were age, gender, and the number of MDEs in the first eight cycles. Results: The risk for recurrence of depression in cycle 9 was found to progressively increase with the number of prior episodes, reaching a value of greater than 46% when the number of prior episodes was five to eight. Independent of this association, the risk of recurrence was greater in younger people and women, but the strength of association of these variables was much weaker for past episodes. Conclusions: MDE recurrence strongly depended on the number of preceding episodes. Those at highest risk of recurrence can be easily identified by their number of past episodes. Depression and Anxiety 31:72–76, 2014. C  2013 Wiley Periodicals, Inc. Key words: major depressive disorder; health survey; episode count; epidemio- logical studies; longitudinal studies 1 Department of Community Health Sciences, University of Calgary, Calgary, Canada 2 Department of Psychiatry, University of Calgary, Calgary, Canada 3 Mathison Mental Health Center for Research and Education, Hotchkiss Brain Institute, Calgary, Canada Contract grant sponsor: Clinical Research Unit of the Hotchkiss Brain Institute of the University of Calgary. ∗ Correspondence to: Andrew Bulloch, Ph.D., Department of Com- munity Health Sciences, 4D67 TRW4, University of Calgary, 3280 Hospital Dr NW, Calgary, Canada T2N 4Z6. E-mail: bul- loch@ucalgary.ca INTRODUCTION Major depressive disorder (MDD) is a chronic con- dition typified by recurrent major depressive episodes (MDEs). Not all those with a first MDE will experience recurrence, but those with recurrent episodes are likely to experience a more serious first episode and poorer response to treatment for recurrent episodes. [1] Ideally, algorithm-based tools would be beneficial for progno- sis of MDD and to guide its treatment. Such tools are Received for publication 23 May 2013; Revised 30 July 2013; Ac- cepted 2 August 2013 DOI 10.1002/da.22173 Published online 26 August 2013 in Wiley Online Library (wileyonlinelibrary.com). C  2013 Wiley Periodicals, Inc.