Gender Difference in the Prevalence of Clinical Depression: The Role Played by Depression Associated With Somatic Symptoms

The National Comorbidity Survey (7) is a nationwide sample (N=8,098) of the U.S. population, ages 15–54 years. While the data were weighted to compensate for variations in the probabilities of selection, the preliminary analyses presented here are based on unweighted data.

In the National Comorbidity Survey, DSM-III-R diagnoses were based on a modified version of the Composite International Diagnostic Interview (7). Respondents were categorized here as exhibiting somatic depression if they met criteria for major depression (without use of any hierarchies of exclusion based on subjects exhibiting other diagnoses) and also reported somatic symptoms in all three of the following categories: 1) sleep disturbance (trouble falling asleep or staying asleep or early morning wakening or hypersomnia), 2) fatigue, and 3) appetite disturbance (lost or increased appetite). Respondents were categorized as exhibiting pure depression if they met criteria for major depression but did not report symptoms in these three categories. Several exploratory analyses were performed. Respondents were categorized as exhibiting pure depression in one analysis if they exhibited symptoms in none of the three somatic categories, in another analysis if they exhibited symptoms in fewer than two categories, and in a third analysis if they exhibited symptoms in fewer than all three of the somatic categories. Separate analyses were also done by using lifetime and 6-month criteria for depression. The primary analyses were three-by-two chi-square analyses comparing the prevalence of no depression, somatic depression, and pure depression exhibited by male and female respondents.

Respondents in the two depressive subgroups were compared on the prevalence of any anxiety disorder (agoraphobia, generalized anxiety disorder, panic disorder, simple phobia, or social phobia) and on the prevalence of body aches, a common somatic symptom that is not included in current criteria for depression. Finally, because research has reported the development of a gender difference in the prevalence of depression during adolescence (8, 9), the age at onset of depression was compared for respondents exhibiting somatic depression and pure depression.

When 6-month depressive criteria were used and pure depression was defined as the presence of symptoms in fewer than all three of the somatic categories, female subjects exhibited twice the prevalence of somatic depression as male subjects (7.6% versus 3.6%) (χ²=64.19, df=2, p<0.00001) but a prevalence of pure depression very similar to that of male subjects (3.0% versus 2.3%). In an analysis that was identical except for the use of lifetime depressive criteria, female subjects again exhibited twice the prevalence of somatic depression as male subjects (15.2% versus 7.5%) (χ²=123.90, df=2, p<0.00001) but a prevalence of pure depression similar to that of male subjects (6.9% versus 6.0%). The gender difference in the prevalence of pure depression remained very low when the definition of pure depression was the presence of symptoms in fewer than two somatic categories (female subjects: 1.5%, male subjects: 1.3%) or symptoms in none of the somatic categories (female subjects: 0.1%, male subjects: 0.1%). (It is rare for people to meet current criteria for depression if they exhibit none of the somatic symptoms.) Because analyses using any of the criteria were so similar, the ancillary analyses reported here maximized cell sizes by using the lifetime depressive criteria and the definition of pure depression as the presence of symptoms in fewer than all three somatic categories. Results of analyses using other criteria were similar but did not always meet statistical requirements for minimal acceptable cell sizes.

Among both male and female subjects, respondents with somatic depression were more likely than those with pure depression to exhibit an anxiety disorder. Of male respondents, 55.0% with somatic depression exhibited one of the anxiety disorders, compared to 36.0% with pure depression (χ²=17.72, df=1, p=0.00003). Among female respondents, 59.8% with somatic depression exhibited an anxiety disorder, compared to 37.7% with pure depression (χ²=38.78, df=1, p<0.00001).

Among male subjects, there were no significant differences among respondents exhibiting somatic depression and those exhibiting pure depression in the prevalence of body aches (18.8% versus 13.9%) (χ²=1.88, df=1) and the age at onset of depression (mean=23.9 versus 24.5 years) (t=0.62, df=517). In fascinating contrast, among female subjects, those with somatic depression reported a higher prevalence of body aches than those with pure depression (29.0% versus 15.4%) (χ²=19.18, df=1, p=0.00001) and a significantly earlier age at onset of depression (mean=23.1 versus 25.0 years) (t=2.73, df=935, p<0.01). This is due in part to the greater likelihood of onset in early adolescence (ages 11–13) among female respondents exhibiting somatic depression than among those exhibiting pure depression (9.9% versus 5.5%) (χ²=4.53, df=1, p=0.02).

These analyses of research interview data from a nationwide sample are congruent with previous self-report studies and analyses of gender differences in individual symptoms in the ECA and the collaborative studies. They indicate that the gender difference in depression may derive primarily from a difference in depression associated with fatigue and appetite and sleep disturbance (as well as anxiety and possibly other somatic symptoms such as body aches) but not from depression that is not greatly associated with these symptoms. Further analysis of the National Comorbidity Survey, ECA, and collaborative study data exploring differences between somatic depression and pure depression would be useful. Given that even people with pure depression usually exhibit some somatic symptoms, much work is needed to clarify the criteria used in defining both somatic depression and pure depression. The data reported here and elsewhere would seem to justify additional research into anxious somatic depression.

Received Jan. 20, 1998; revisions received April 27 and July 30, 1998; accepted Aug. 25, 1998. From the City College of New York. Address reprint requests to Dr. Silverstein, Department of Psychology, City College of New York, New York, NY 10031; [email protected] (e-mail). Analyses reported in this article are based on the public-use data set of the National Comorbidity Survey sponsored by NIMH and administered by the Survey Research Center of the University of Michigan. Ramkumari Baliram provided assistance in the downloading of the data.