Letters
to the Editors
1077
References 1. Snedcor GW, Cochran WC. Statistical Methods. Iowa State University Press, Ames, Iowa, 1989. 2. Vineis I’, Schulte PA, Vogt RF Jr. T ec h mea I variability in laboratory data. In: Molecular Epidemiology, Principles and Practice (Schulte PA, Perera FP, eds.). Academic Press, San Diego, California, 1993. PI1 SO895-4356(96)00052-2
Time Trends in Risk of Depression Simon et al. [l], using data from 15 international primary care sites, raise questions as to whether the rates of major depression are increasing in recent birth cohorts. Since their data were collected 515 years later than those in earlier reports of a birth cohort effect [2-41, these data provide a unique opportunity to study trends in early-onset (~20 years) depression in countries around the world. When we examined the rates of early-onset major depressive disorder (MDD) by birth cohort shown in Fig. 3 of their article, we were able to discern some very interesting patterns that the authors appear to have overlooked. We found that these rates do not keep increasing during each decade of this century, but rather show specific patterns by site. For example, in both Western Europe and the United States there is a three- to fourfold increase in rates between the 1937-1946 birth cohort and the 1947-1956 birth cohort, but the rates then remain constant between the I947- I956 birth cohort and 1957-1966 birth cohort. There is then a twofold increase between the 1957-1966 cohort and the 1967-1974 cohort. Although there is undoubtedly some bias associated with the recall of these reported rates due to the retrospective nature of this study, we think that it is unlikely that these specific patterns are solely a consequence of methodologic effects. Similar effects are also observed in Greece and Turkey. In Chile and Brazil, on the other hand, the rates remain constant between the 1927-1957 cohorts and then increase dramatically (sixfold) in the 1967-1974 cohort. In China, Japan, and India the rates appear to be increasing with each successive cohort for the three youngest birth cohorts, while in Nigeria only the youngest cohort reported rates of early-onset MDD. (If the hypothesis that all of these patterns are due to methodologic effects were correct, that would imply that patterns of recall must be very different among these sites. Simon et al. have hypothesized that if all of the psychiatric disorders studied show the same cohort pattern it is more likely that this pattern is a result of methodologic artifact, not a real phenomenon. In the same vein, we hypothesized that if the cohort patterns we observe for early-onset MDD represent an artifact of reporting, we should observe similar patterns for panic disorder and agoraphobia. However, their Fig. 4 shows that in contrast to the rates of earlyonset MDD, where rates in the youngest cohort were higher than in the cohort immediately preceding it at every site (a finding that was also seen when all of the sites were combined in Fig. 1 of their article), the rates of panic and agoraphobia in the youngest cohort do not follow this pattern. A careful examination of Fig. 4 of their article shows that the rates of panic in the 1967-1974 cohort are approximately equal to the rates in the 1957-1966 cohort, while the rates of agoraphobia in the 1967-1974 cohort are, in fact, less than the rates in the 1957-1966 cohort. An alternative approach to investigating whether or not this phenomenon is an artifact of methodologic bias is to determine if cohort
effects exist for outcomes that are known to be highly correlated with depression but are not as vulnerable to forgetting and/or “telescoping” as is major depression. One such outcome is the rate of suicide. The specific cohort patterns observed in this set of data for the United States and Western Europe are consistent with the findings of McIntosh [5] who showed that in the United States, the “Boom” generation (born between 1943 and 1960) and the young group of Americans called the 13ers (the 13th generation of Americans; born between 1961 and 1981) were at greater risk for suicide than earlier generations at the same chronological age, with 13ers higher than boomers for the ages they have attained. Other outcomes known to be highly associated with depression are rates of suicide attempts and rates of divorce/separation. We have found that, in the ECA data, cohort effects for rates of suicide attempts and rates of divorce/separation exist and are similar in pattern to that of major depression. Murphy [6] has examined the relationship between age at time of interview and recent and/or current prevalence ofdepression at different time periods among entirely different groups of subjects selected to represent those time periods in the United States. Murphy’s findings suggest that the peak prevalence appears to have moved downward from older to younger people as time has moved forward from 1952 to 1992. As noted by Murphy, this trend (which is not subject to recall bias) is consistent with a genuine birth cohort effect. Thus, although we do not question that there is a certain amount of recall bias in the rates of MDD reported by cohort resulting in some underestimation of lifetime prevalence in older cohorts, we believe that these data support the view taken by Fombonne [7]increasing rates of major depression observed in adolescents and young adults over time is a genuine phenomenon and not primarily a methodologic artifact.
College
of Physicians
PRIYA J. WKKRAMARATNE MYRNA M. WEISSMAN Department of Psychiatry and Surgeons of Columbia University New York, New York
References 1. Simon GE, Vonkorff M, Ustun B, Cater R, Gureje 0, Sartorius N. Is the lifetime risk of depression actually increasing? J Clin Epidemiol 1995; 48: 1109-1118. 2. Wickramaratne PJ, Weissman MM, Leaf PJ, Holford TR. Age, period and cohort effects on the risk of major depression: Results from five United States communities. J Clin Epidemiol 1989; 42: 333-343. 3. Cross Narional Collaborative Group. The changing rate of major depression: Cross national comparisons. J Am Med Assoc 1992; 268: 30983105. Klerman CL, Weissman MM. increasing rates of depression. J Am Med Assoc 1989; 261: 2229-2235. McIntosh JL. Generational analyses of suicide: Bahy hoomers and 13ers. Suicide Life Threatening Behav 1994; 24: 334-341. Murphy JM. The Stirling County study: Then and now. Int Rev Psychiat 1994; 6: 329-348. Fombonne E. Depressive disorders: Time trends and possible explanatory mechanisms. In: Psychosocial Disorders in Young People: Time Trends and Their Causes (Rutter M, Smith DJ, eds.). John Wiley & Sons, Chichester, 1995, pp. 545-615. PI1 SO895-4356(96)00049-Z
RESPONSE We thank Drs. Wickramaratne and Weissman for their to our manuscript. Their comments raise two questions:
attention