A prospective longitudinal study suggests that retrospective surveys,
though fairly reliable for short-term snapshots, may be off the mark in
estimating lifetime prevalence rates of common mental disorders in the general
Large-scale surveys conducted in the community, such as the National
Comorbidity Survey (NCS) and NCS-Replication (NCS-R) in the United States and
the New Zealand Mental Health Survey (NZMHS), have estimated that one-fourth
to one-third of the general population have suffered an anxiety disorder in
their lifetime, and 17 percent to 19 percent have had symptoms that met the
criteria for major depression. However, a prospective study published online
September 1 in Psychological Medicine indicates that about half (49.5
percent) of adults up to age 32 have experienced at least one episode of
anxiety disorder, and 41 percent have experienced at least one episode of
depression at some point in their
Prospective Study Reveals
Higher Lifetime Prevalence
The study authors, led by Terrie Moffitt, Ph.D., the Knut Schmitt-Nielsen
Professor of Psychology and Neuroscience at Duke University, analyzed data
from the prospective, longitudinal Dunedin Multidisciplinary Health and
Development Study. More than 1,000 people born in Dunedin, New Zealand, from
April 1972 through March 1973 were registered at birth to participate in the
study and were periodically assessed for health status as they aged.
The participants were assessed for mental disorders at ages 18, 21, 26, and
32 by mental health professionals using DSM-III-R (before 1994) or
DSM-IV (after 1994) criteria. The study had an excellent follow-up
rate, as 96 percent of the study participants who were alive at the last time
point (32 years) were assessed.
The Dunedin study highlights the key divergence between prospective and
retrospective epidemiological data on mental illnesses: Prospective,
longitudinal data reveal a much higher long-term prevalence than retrospective
data that rely on respondents' recall.
Knowing the true prevalence of mental disorders is important for estimating
the economic burden, setting public-health policy, allocating resources, and
changing public perception and stigma about mental illness, the authors
Why is there such a large discrepancy between researchers' estimation of
cumulative lifetime prevalence? "Recall failure, the fundamental
difference between prospective versus retrospective measurement," the
authors wrote, "is the likely explanation.
The explanation is supported by the similarity between the Dunedin study
and previous retrospective data in short-term prevalence of mental disorders.
The past-year prevalence averaged 23 percent for any anxiety disorder, 17
percent for depression, and 13 percent for alcohol dependence across all
assessments during the Dunedin study. These rates approximate the past-year
prevalence found in the NCS, NCS-R, and NZMHS, which also implies that the
assessment method and validity of the Dunedin study were not significantly
different from those of the previous surveys.
That retrospective surveys are vulnerable to recall failure is a known
phenomenon in mental health. Previous studies have shown that the proportion
of patients who remember they have been diagnosed with depression decreases
progressively over time. Moffitt and coauthors cited a body of research that
illustrates the unreliability of retrospective mental health epidemiological
data, which are largely based on recall of symptoms. For example, among
patients who had been hospitalized for depression, half could no longer recall
enough symptoms to meet the diagnostic criteria for depression 25 years later
to be identified as a lifetime case. Ten percent of people who had been
diagnosed with depression could not be identified as lifetime cases in a
reinterview after only three years.
Another problem with counting lifetime prevalence of mental disorders was
symptom changes over time. Unlike other chronic diseases that persist or
progress once diagnosed—such as hypertension or diabetes—common
mental disorders often fluctuate in severity and symptoms among some but not
all patients. Moffitt and coauthors found that many young adults in the
Dunedin cohort did not experience chronic or recurrent disorders, even though
their symptoms at one point were severe enough to meet the diagnostic
criteria. More than half of those with a lifetime prevalence of anxiety,
depression, alcohol dependence, or cannabis dependence were diagnosed with the
disorder at only one of the four assessments.
"Epidemiologists agree that prospective studies are more reliable
than retrospective studies," Darrel Regier, M.D., M.P.H., executive
director of the American Psychiatric Institute for Research and Education
(APIRE) and director of research at APA, told Psychiatric News. To
understand what really happens over time, longitudinal studies capture a lot
of data that are missed by cross-sectional studies, he noted.
The higher-than-expected cumulative prevalence of mental disorders adds
fuel to a long-running argument that diagnostic criteria are too loose and
that too many people are unnecessarily diagnosed with mental disorders. A high
prevalence by itself, however, does not invalidate a diagnosis. For example,
almost every person who has normal blood pressure at 55 will develop
hypertension at some point in the rest of his or her life, according to the
Framingham Heart Study.
The authors of the Dunedin study believe that their data have probably
still underestimated the lifetime prevalence of these mental disorders,
because the study cohort was followed only to age 32. They stressed that their
findings are "uninformative (and agnostic) about the validity of
diagnoses of depression, anxiety, and substance dependence as defined by
DSM-IV," which is "a separate debate." Meanwhile,
the debate about the validity of DSM diagnostic criteria continues
within and outside of psychiatry.
Regier, who is the vice chair of the DSM-V task force, hopes the
revised diagnostic manual will put some of the controversies to rest.
Regier and William Narrow, M.D., M.P.H., associate director of research at
APA and APIRE, both said that the DSM-IV criteria have validity
issues inherited from DSM-III, particularly in the community patient
"DSM-III criteria were developed without epidemiological
data, and... were based mostly on inpatient data with some outpatient
evidence," said Regier. In other words, these diagnostic criteria worked
well in describing patients who are already in psychiatric care. These
criteria were necessary and extremely useful at the time to begin to describe
and classify mental disorders systematically and build a framework for
epidemiological studies. However, in describing the complex presentations of
mental disorders outside psychiatric hospitals and epidemiological research,
the DSM criteria have encountered difficulties. The criteria are not
always helpful in setting appropriate thresholds for those individuals in the
community, as captured by this and other epidemiological studies, who
apparently meet the criteria but have no need for treatment because of mild
severity or a self-limiting episode, Regier and Narrow pointed out.
"The public and media assume that everyone who meets criteria for a
DSM diagnosis in a community survey should automatically get
treatment, but that is not true," said Narrow. He gave an example: If a
person with a phobia to an object (e.g., snakes) can avoid the object all his
life without functional impairment, he may not need treatment for the
Thus, Regier said, an important goal for DSM-V is to revise the
diagnostic criteria substantially, particularly by integrating dimensional
severity criteria and symptomatic information across current criteria
boundaries, which would make them more useful for clinicians to better
identify patients who have underlying psychopathology and to provide a
measurement-based approach to treatment.
Rather than expanding the diagnoses of mental disorders as some people
fear, the DSM-V task force is aiming to make the diagnoses clearer
and more precise and to help epidemiologists set better thresholds for
clinically significant mental disorders in community settings, said Regier and
Narrow, and these changes will be informative for identifying patients in need
An abstract of "How Common Are Common Mental Disorders?
Evidence That Lifetime Prevalence Rates Are Doubled by Prospective Versus
Retrospective Ascertainment" is posted at<journals.cambridge.org/action/displayAbstract?fromPage=online&aid=6121524&fulltextType=RA&fileId=S0033291709991036>.▪