Current anxiety (19.4%) and mood disorders (18.5%) were prevalent in residents of the selected Georgia counties. Overall, these findings agree with published U.S. national-level surveys. The 2001 to 2003 NCS-R, the most comprehensive national study, used the Composite International Diagnostic Interview (CIDI) [29
] and reported 18.2% of U.S. adults had an anxiety disorder and 19.5% a mood disorder [32
]. Occurrence of major depressive disorder in the selected Georgia counties was similar to NCS-R national estimates (7.3% and 6.7%, respectively) [32
]. Most of those with a mental illness detected in our study (19%) suffered only one, 7.4% had two psychiatric conditions, 2% had three, and 1.4% four or more, which also mirrors NCS-R reported findings; 14.4% of people with a psychiatric disorder had only one, 5.9% had two, and 5.9% three or more [32
]. However both order and magnitude of specific anxiety disorder prevalence in our Georgia population sample differed markedly from 2001 to 2003 NCS-R national estimates. In Georgia, PTSD (6.6%) and GAD (5.8%) were the most common anxiety disorders followed by specific phobia (4.0%) and social phobia (1.2%); while, specific phobia (8.7%), social phobia (6.8%), PTSD (3.6%) and GAD were the most common anxiety disorders reported from NCS-R (2.7%) [32
]. This difference in our findings from Georgia and NCS-R highlights that national surveys may not reflect category-specific occurrence of mental illness in individual states. For example, a recent report indicates depression and psychological distress are more common in the southeastern states than the nation as a whole [33
]. However, it is also possible that this variation in prevalence estimates may be due to methodological differences between our study and the NCS-R, such as using a state-specific sample rather than a national sample and the SCID rather than CIDI as a diagnostic tool.
As we hypothesized, both current anxiety and current mood disorders were significantly more common among people identified by telephone interview as unwell (functional somatic syndrome
or other unwell
). Fifty three percent of functional somatic syndrome
participants had a current anxiety disorder and 32% a current mood disorder; among those classified as well
only 18.9% had an anxiety disorder and 3.7% a mood disorder. The increased prevalence of anxiety and mood disorders in people with a functional somatic syndrome is important for primary care providers, who should consider additional psychiatric screening or referral of individuals presenting with somatoform symptoms. This association also has implications for psychiatric nosology and development of DSM-V [19
]. As noted in a recent review by Lieb et al. [34
], there is a remarkable lack of data evaluating occurrence of anxiety and mood disorders among people with a functional somatic syndrome.
Research surveys for mental illness should consider screening for functional somatic syndromes
to help resolve questions of nosology and surveillance systems intended for public health use should also measure functional somatic syndromes
as part of screening for anxiety and mood disorders.
In our Georgia study sample, anxiety and mood disorders varied considerably according to socio-demographic factors among metropolitan, urban and rural samples. Our study found no significant difference among the three geographic areas in the prevalence of having any current mood disorder, whereas previous U.S. studies have reported a slight but significantly higher prevalence of depression in rural areas than metropolitan area [35
]. Future work should investigate whether there may be unique aspects of mental health treatment or access to treatment in urban and rural Georgia that are associated with this non-significant effect. On the contrary, for current anxiety disorders, our study results showed that there was a significantly higher prevalence in urban and rural areas than that in the metropolitan area. Logistic regression modeling of PTSD and GAD (the most common anxiety disorders) found both significantly associated with female sex and education less than high school. PTSD was also significantly associated with rural and urban residence but residence and ethnicity were not associated with GAD. This finding might indicate more limited availability of and referral to treatment opportunities for those living in rural and urban locations who have encountered traumatic events. A recent study has shown that veterans with PTSD had significantly fewer treatment visits when living in rural areas than those living in bigger cities [37
]. Another study observed that referrals after a natural disaster were markedly higher in large cities (Atlanta among them) compared to urban locations [38
]. Logistic regression modeling of moderate to severe depression showed less than high school education as the most significant factor and no significant association with sex or urbanicity. These findings add to the evidence base that should be considered when designing mental illness control programs. For both anxiety and mood disorders, messaging and education must consider educational attainment. In Georgia, programs for PTSD and GAD should also consider outreach to gynecologists, who often serve as women’s primary care providers. Finally, in Georgia the association of urban and rural residence with PTSD should be considered when developing clinical services and messaging for these segments of the population.
To our knowledge, CDC’s Behavioral Risk Factor Surveillance System (BRFSS) comprises the only population survey of depression in Georgia against which to compare our findings. In 2006 and 2008 BRFSS conducted random digit dial telephone interviews with adults and estimated that 9.0% were depressed during the last 2 weeks, as reflected by a PHQ-8 score >10 [33
], which is similar to our finding that 9.8% of the selected Georgia counties’ study population suffered moderate to severe depression during the last month. BRFSS has also estimated that 3.4% of Georgia adults had major depression, which is quite different from our estimated 7.3% prevalence of major depressive disorder in the selected Georgia counties. This most likely occurred because the PHQ-8 screens for depression in the last two weeks and does not identify DSM-IV major depressive disorder as stringently measured by means of the SCID (at least two weeks during the last month). BRFSS has not published results concerning factors associated with depression in Georgia against which to compare our present findings (e.g., associations with age, sex, race/ethnicity, metropolitan/urban/rural residence, education, or household income). Unfortunately, BRFSS does not measure other categories of mental illnesses, such as anxiety disorders, which our survey found to be as common and disabling as depression and which also reflected different associations with socio-demographic variables than did depression.
Strengths and limitations
One of the strengths of this study is the use of a large sample identified from metropolitan, urban, and rural Georgia populations through phone interviews and clinical evaluations. In addition, this study used the SCID as a diagnostic tool for psychiatric disorders evaluated in the paper.
There are two major limitations of our study. Most importantly, we used Fulton and DeKalb counties to represent metropolitan Georgia, Macon and Warner Robins represented urban Georgia, and counties surrounding them represented rural Georgia. We chose this approach for logistical reasons. Fulton and DeKalb counties are the heart of Atlanta and meet US Department of Agriculture metropolitan population criteria [39
]. However, 14 additional counties comprise the Atlanta metropolitan area and were not sampled. Macon (Bibb County) and Warner Robins are considered urban [39
] but are not necessarily similar to other Georgia urban populations (e.g., Savannah, Athens and Rome). Due to cross-sectional data, we could not infer causal relationships between the psychiatric diagnoses and socio-demographic variables. The study also suffered from the non-coverage for institutionalized and non-English speaking populations. Finally, although our rural study counties fulfill USDA criteria for rural [39
], their populations may differ from those of rural counties in the north Georgia mountains or southern coastal plain. Further analyses of the publically available BRFSS database may provide information concerning the effects of such possible differences on occurrence of depression.
Second, selection for clinical evaluation was not random, but was based on wellness status detected by the telephone interview. The source study was enriched in fatiguing illness and potentially identified high prevalence of undiagnosed psychiatric conditions. We attempted to adjust for clinic selection bias by weighting, which has attendant methodological limitations. Finally, reflecting sample size, some standard errors are large.