Depression, anxiety and their comorbidity in the Swedish general population: point prevalence and the effect on health-related quality of life

Introduction

Depression and anxiety disorders are major world-wide health problems that affect a substantial number of individuals every year (Ebmeier, Donaghey & Steele, 2006; Kessler, Merikangas & Wang, 2007). In the US National Comorbidity Survey Replication (NCS-R), lifetime prevalence of mood and anxiety disorders were 20.8% and 28.8%, respectively (Kessler, Merikangas & Wang, 2007). Twelve-month prevalence estimates of mood and anxiety range from 6.6% to 11.9% and 5.6% to 18.1% across surveys from Europe, Australia and the US (Baumeister & Härter, 2007). In Sweden in 1957, the point prevalence of depression was estimated to be 4.7%, based on data from the total population (n = 2612) of Lundby, a small rural area in southern Sweden (Rorsman et al., 1990). Using the national Swedish Twin Registry, lifetime prevalence for depression was estimated to be 13.2% among men and 25.1% among women (Kendler et al., 2006). In the Lundby study, lifetime prevalence for depression was 27% among men and 45% among women, when participants were followed from 1957 up to 1972 (Rorsman et al., 1990). Importantly, the Lundby study did not use DSM criteria for major depression, which makes comparisons to prevalence rates from other countries complicated (Rorsman et al., 1990). To our knowledge, there exist no up-to-date point estimates of DSM-IV depression from the Swedish general population.

For anxiety disorders, there is a 12-month prevalence study from the Swedish general population regarding panic disorder (2.2%; Carlbring et al., 2002) and a point prevalence study on social phobia (15.6%; Furmark et al., 1999). In the NCS-R, the 12-month prevalence for these disorders were 2.7% and 6.8%, respectively (Kessler et al., 2005). From Sweden, there exist point prevalence data on generalized anxiety disorder (GAD) from primary care (3.6%; Allgulander & Nilsson, 2003; Munk-Jørgensen et al., 2006). This can be compared to NCS-R, where 12-month prevalence of GAD was 3.1% (Kessler et al., 2005). However, the point prevalence of GAD in the general Swedish population seems unknown.

Comorbidity between mood and anxiety disorders is known to be common. For example, among participants in the NCS-R that had a diagnosis of major depression, 57.5% also met criteria for at least one anxiety disorder (Kessler, Merikangas & Wang, 2007). Other epidemiological data suggests that 59.0% of individuals with GAD fulfill criteria for major depression (Carter et al., 2001). This suggests that comorbidity between depression and anxiety disorders is the rule rather than the exception. Comorbidity has consistently been associated with a poorer prognosis and greater demands for professional help (Albert et al., 2008; Schoevers et al., 2005). In addition, comorbidity between depression and anxiety seems strongly associated both with role impairment and higher symptom severity (Kessler et al., 2003). Generally, psychiatric comorbidity is known to affect various aspects of health-related quality of life (Carpentier et al., 2009; Saarni et al., 2007; Sherbourne et al., 2010).

The aim of the present study was to investigate the point prevalence of depression, GAD, anxiety disorders in general, and comorbidity in a representative Swedish sample. We also investigated the effect of depression, anxiety and comorbidity on health-related quality of life.

Materials & Methods

This study was carried out in accordance with the STROBE initiative for reporting epidemiological data (Von Elm et al., 2007). Approval for the study was obtained from the Institutional Review Board at the Department of Behavioural Sciences and Learning, Psychology Section, Linköping University, Sweden. By responding to the postal survey, participants gave their consent to take part in the study.

Results

Demographic data are illustrated in Table 1. There were some gender differences in this data, as women (56.5% had post-secondary education) were more educated than men (45.3% had post-secondary education). Women also had more experience of treatments for worry and for psychiatric problems in general. Moreover, the rate of experience of treatment for any psychiatric disorder was 52.2% (95% CI: 45.3–59.1; n = 107) among individuals with clinically significant depression or anxiety (n = 205), and 64.9% (95% CI: 56.7–73.1; n = 85) among those with major depression or GAD (n = 131).

Comorbidity

As seen in Table 2, the rate of comorbidity was 8.3% (95% CI: 6.8–9.9) and 3.1% (95% CI: 2.1–4.1), when using the two different definitions. Put differently, among those with clinically significant depression or anxiety, 48.3% (100/207) had comorbid depression and anxiety, while the same figure was 28.2% (37/131) among those with major depression or GAD. This is illustrated in further detail in Fig. 1.

Patients with comorbid clinically significant depression and anxiety had higher symptom severity, compared to those with a single diagnosis as measured by the PHQ-9 (14.91 compared to 8.21) and the GAD-7 (13.00 compared to 8.55). The same was true for those with comorbid major depression and GAD (PHQ-9 score 18.81, GAD-7 score 14.17), compared to those with either of the diagnoses (PHQ-9 score 10.55, GAD-7 score 9.99). All the differences were statistically significant (all t’s > 5.13, all p’s < .001).

Discussion

This study suggest that, at any given time point, around 17.2% of the Swedish general population are experiencing clinically significant depression (10.8%) or anxiety (14.7%) that likely affect their daily lives. Around 5.2% experienced a current major depressive episode and 8.8% had generalized anxiety disorder. Half of those with either clinically significant depression or anxiety also had a comorbid disorder. Comorbidity among disorders was associated with higher symptom severity and lower health-related quality of life.

Between half and two thirds of participants with a disorder had treatment experiences. This is in line with what was found in the US National Comorbidity Survey Replication (NCS-R), where only 51.6% (95% CI: 46.1–57.2) of 12-month cases received health care treatment for depression, and where only 21.7% (95% CI: 18.1–25.2) were adequately treated (Kessler et al., 2003). Furthermore, there are data from a community survey in Finland, where the use of health services for mental health was investigated (Hämäläinen et al., 2008). Among individuals with depression, anxiety disorders, or both, only 34%, 36%, and 59% used health services, respectively. In summary, data from the present study are in line with treatment consumption studies from other countries, and indicate that depression and anxiety disorders are undertreated conditions also in the Swedish general population.

This study estimated the point prevalence of major depression in Sweden to be 5.2%. This figure is in line with the 12 month prevalence rates from the US (6.7%, 95% CI: 6.1–7.3; Kessler et al., 2005, Australia (6.3%, 95% CI: 5.7–6.9; Andrews, Henderson & Hall, 2001) and the Netherlands (5.8%, 95% CI: 5.2–6.4; Bijl, Ravelli & Van Zessen, 1998). The figure 5.2% is also close to the original estimate of 4.7%, from the 1957 population in the Lundby study (Rorsman et al., 1990). When using a stricter definition (only 5% risk of detecting false positives), 4.4% of individuals were identified as having major depression. This and other prevalence estimates provide further indications that 5.2% is indeed a valid estimate of current point prevalence of major depression in Sweden, and therefore complements epidemiological research from the Lundby study (Rorsman et al., 1990) and the national Swedish Twin Registry (Kendler et al., 2006).

A cut-off of PHQ-9 ≥ 10 was used to identify clinically significant depression, and was found among 10.8% of individuals in this study. This rate is similar or somewhat higher compared to previous estimates using the same definition in the US (8.6%; Kroenke et al., 2009) and in Canada (8.4%; Patten & Schopflocher, 2009). Our result on clinically significant depression seems to mirror estimates of ‘any mood disorder’, e.g., from the NCS-R (Kessler et al., 2005) where this was found among 9.5% (95% CI: 8.7–10.3).

We found the point prevalence of clinically significant anxiety in Sweden to be 14.7%, using the definition of GAD-7 ≥ 8. This seems similar or somewhat higher than previous estimates from Germany (12.1%; Löwe et al., 2008). The mean GAD-7 score in the present study (3.59, 95% CI: 3.36–3.81) was higher than the average score from the German population study (2.95, 95% CI: 2.85–3.04) (Löwe et al., 2008). The rate of clinically significant anxiety found in this study can be compared with that of 18.1% for any anxiety disorder from the NCS-R (Kessler et al., 2005). Importantly, the GAD-7 is known only to identify GAD, social phobia, panic disorder and PTSD (Kroenke et al., 2007). The NCS-R estimate did also include specific phobia, which was found in 8.7% of US 12 month cases (Kessler et al., 2005). In conclusion, our estimates of prevalence of general anxiety seem comparable to other epidemiological studies.

Our estimate of GAD prevalence, 8.8%, using the definition of GAD-Q-IV ≥ 5.7 (Newman, Zuellig & Kachin, 2002), seem significantly higher than in other epidemiological studies from e.g., US (3.1%; Kessler et al., 2005) and Australia (2.6%; Andrews, Henderson & Hall, 2001). The stricter definition of GAD-Q-IV ≥ 9.0 (95% specificity) gave an estimate of 4.7%, that indicates that the true prevalence in Sweden is probably at least that high. Importantly, as noted by Kessler and colleagues (Kessler, Walters & Wittchen, 2004) in a review of the epidemiology of GAD, there are uncertainties about the basic epidemiological characteristics of GAD that could lead to underestimates of the true prevalence of GAD in the general population. There are known complicating factors due to changing DSM criteria over time, and the fact that DSM definitions of GAD seem to exclude a substantial amount of individuals with chronic worry, tension and nervousness (Kessler, Walters & Wittchen, 2004). Kessler, Walters & Wittchen (2004) argued that the true current prevalence of GAD in the community could actually be as high as 5 to 8%. Our study has provided indications that the prevalence of GAD in the Swedish general population is probably within this interval.

The EQ-5D index value obtained in this study (0.84) was identical to that obtained in a study that validates the EQ-5D in the Swedish general population (Burström, Johannesson & Diderichsen, 2001). Absence of reported confidence intervals in Burström, Johannesson & Diderichsen (2001) prohibits a more detailed comparison, but when comparing HRQoL data from this study (Table 3) to Swedish normative data, no dimension differs more than 5.4%. This gives indications that HRQoL data collected in this study are overall similar to previous collected data from Sweden.

There are methodological limitations that need to be considered. In this study, we used self-reporting, rather than a diagnostic interview, such as the SCID (First et al., 1997) or the CIDI (Robins et al., 1988). While the measures used in this study are thoroughly validated (Kroenke, Spitzer & Williams, 2001; Newman, Zuellig & Kachin, 2002; Spitzer et al., 2006), also for use in the general population (Kroenke et al., 2009; Löwe et al., 2008) we still consider this as a limitation. A further limitation is that with the data collected on educational level it was impossible to interpret whether a respondent had a certain education ongoing or completed. This makes the collected rate of 52% with post-secondary education hard to interpret. According to national statistics from Statistics Sweden (Statistics Sweden, 2010), this figure is 35.5%. This indicates that our data on educational level may not be valid. A final important limitation of this study is the response rate of 44.3%, as for example compared to Kessler et al. (2005) who had 70.9% response rate. Among community surveys using the PHQ-9, our response rate is low in comparison to 64.5% (Martin et al., 2006), but higher than in Pirkis et al. (2009) who reported 28.6%. This could have biased the prevalence estimates in this study. Importantly, there are indications that individuals with mental illness might be more reluctant than others to participate in mental health surveys (Allgulander, 1989; Eaton et al., 1992). Based on this, prevalence estimates of depression and anxiety disorders in this study would be underestimated, if biased. Future research with a similar design to this study should make use of established strategies to increase the response rate, for example, by using monetary incentives, using pre-notification and by having a shorter questionnaire (Edwards et al., 2007).

Some implications of this study are discussed as follows. The results from this study complements and corroborates earlier findings from Sweden and other countries by providing up-to-date prevalence rates. The fact that 17.2% of the population suffer from a clinically significant depression or anxiety condition calls for further efforts regarding prevention and treatment. This includes both novel and more established intervention alternatives. The use of the Internet is one way of enabling evidence-based interventions to reach larger proportions of the general population (Muñoz, 2010). Internet-based psychological treatments seem to be a promising and feasible treatment alternative that may increase availability of evidence-based treatments for psychiatric disorders such as depression and anxiety (Andersson, 2009; Hedman, Ljótsson & Lindefors, 2012). Future challenges include establishing means of delivering such treatment in regular health-care (Hedman et al., 2013), but also on national level with preserved efficacy (Calear et al., 2009).

A further implication regards the significant amount of comorbidity between depression and anxiety disorders identified in this study (see Fig. 1). The large overlap calls for alternatives to disorder-focused assessments and treatments that address transdiagnostic factors, such as emotion regulation deficits (Aldao, Nolen-Hoeksema & Schweizer, 2010; Mennin, McLaughlin & Flanagan, 2009). This could potentially enable assessment that is grounded in experimental research and that provide an opportunity for treatment alternatives that target these processes (e.g., Farchione et al., 2012; Johansson et al., 2012; McEvoy, Nathan & Norton, 2009). Hence, transdiagnostic treatment approaches seem as an important step in addressing psychiatric comorbidity.

Finally, there are indications that positive mental health likely is more than absence of psychiatric symptoms (Huppert & Whittington, 2003). For example, the absence of positive emotions have been found to be a better predictor of mortality than the presence of psychological symptoms (Huppert & Whittington, 2003). Also, there are indications that disability and lack of social roles are important determinants of psychological symptoms, but have less influence on positive well-being. Similarly, paid employment has been found to be an important determinant of positive well-being but it seem to have little influence on psychiatric symptoms (Huppert & Whittington, 2003). Therefore, future epidemiological research that aims to understand the whole picture of human health seems to require assessments that go beyond mere symptoms (Cloninger, 2006; Huppert & Whittington, 2003).

Conclusions

The findings reported here provide updated prevalence rates from the Swedish general population. Depression, anxiety and their comorbidity are undertreated conditions that are associated with lower quality of life. This study adds to existing evidence that these conditions are clearly major health problems in Sweden, that need further efforts regarding preventive and treatment interventions.