To the best of our knowledge, this is the first population-based study of prognosis of anxiety and depression caseness and their comorbidity over this long a time-period, and assessed at more than two points in time. For the three-year associations, we took advantage of all available data, thus, from baseline to 3-YFU and from 3-YFU to 6-YFU. The results showed that when being an anxiety caseness, it was most common, and about equally common, to recover as it was to remain with anxiety. It was less common to develop comorbidity, and least common to recover from the anxiety and instead develop depression. Being a depression caseness most commonly resulted in recovery, and less common to either recover from depression and instead develop anxiety, develop also anxiety, or remain with depression. However, this calls for caution considering the small number of cases with depression at baseline (n = 33) and 3-YFU (n = 31). Being a comorbidity caseness, it was most common, and about equally common, to recover as to remain with comorbidity. It was less common to recover from depression, but remain with anxiety, and least common to recover from anxiety, but remain with depression. Being neither an anxiety nor depression caseness, it was most common to remain so after three years, less common to develop anxiety, even less common to develop comorbidity, and least common to develop depression.
Overall, the results from the three-year associations show that it was most common to recover when being an anxiety, depression or comorbidity caseness, and when not being a caseness it was most common to remain so. Apart from this, it was rather common to remain in the same caseness condition after three years, and this was particularly so for comorbidity caseness. Anxiety and depression caseness differed, such that in comorbidity it was more likely to recover from depression than from anxiety. In addition, being no caseness was more likely to develop into anxiety than into depression caseness.
Regarding the prognoses across six years, most of them were simply concordant with the three-year associations, whereas the most common additional pattern was for the affective caseness to continue from 3-YFU to 6-YFU. Notably, when being no caseness at baseline, the second most common prognosis, after continued no caseness at 3-YFU, was to develop anxiety at 6-YFU.
Some of the progressions were considerably more common than other progressions. Thus, close to half of the 64 possible six-year prognoses were followed by only two or fewer of the participants. In contrast, taken together, 80.5% of the participants followed either the most common or the second most common six-year prognosis for the four caseness conditions. This implies good representativeness in these prognoses for the general population.
Although different types of samples, designs and methods have been applied, the present results show similarities with prior work. Comorbidity between anxiety and depression has been reported to have a poorer recovery prognosis than depression alone [8, 12, 13], which the present recovery results for three-year associations support (36.8% for comorbidity vs 59.4% for depression). Furthermore, in accordance with earlier studies [29,30,31], relapse in depression was more common in comorbid anxiety (9 cases) than without comorbid anxiety (1 case). Global prevalence and incidence rates for depression are quite similar, whereas the prevalence rate for anxiety is much higher than the incidence rate, indicating that anxiety is a more persistent condition than depression [1]. This may partly be explained by anxiety being a trait [32], thereby easily evoked by environmental triggers due to dysregulated cognition and affect. There is also genetic and epidemiological evidence linking trait anxiety with stress-related vulnerability to depression [32].
The present results on recovery across three years are also in line with this difference (59.4% for depression vs 46.0% for anxiety). A meta-analysis also showed that anxiety predicts depression better than vice versa, although the effect size was very small and unlikely to be clinically meaningful [16]. The present data support this description based on the proportions of participants who develop comorbidity three years after anxiety compared to after depression (13.7% develop comorbidity after anxiety vs 12.5% after depression). However, the likelihood of developing depression and anxiety without comorbidity exhibits the opposite pattern, but the small sample size for depression (n = 64) calls for caution.
There are numerous theories regarding the very close relationship between anxiety and depression [9, 11]. For example, Gold [33] argues that depression and anxiety can cause a vicious circle due to increased activity in amygdala. According to other theories, avoidance behavior as a result of anxiety causes decreased positive affect, leading to depression. In the opposite direction, social withdrawal as a result of depression causes fear and anxiety linked to social situations [11]. In line with this, another Swedish population-based survey showed that about half of the participants who met the criterion for depression caseness (“clinically significant depression”) also met the criterion for anxiety caseness (“clinically significant anxiety”) [4], whereas the present study showed as many as three out of four participants fulfilling this description.
The present findings suggest that in the general adult population, about 54% of those with a high probability of being a clinical case with anxiety, about 40% of those with depression, and about 63% with comorbidity of anxiety and depression will not recover within a three-year period. The results further show that the second most common six-year prognoses in these subpopulations are to remain in these conditions also after six years. An implication of these results is need of further efforts in prevention of long-term affective health problems. Among those with anxiety and depression comorbidity at baseline, about 37% had this condition also after six years, and 22% in between these endpoints as well. This makes this comorbidity a particular public health problem since it is strongly associated with severe somatic symptomatology, very poor quality of life, and increased risk for suicide [3, 4, 7, 8, 14, 15].
The present study intended to describe relatively long time-period changes in common affective health conditions in a general adult population, in order to provide further understanding relevant for healthcare planning and disease prevention. Thus, the present focus was the associations between these conditions, not risk factors for the progression of each of them per se. Common ones being identified for affective conditions are treatment history, demographics (e.g., age, sex and marital status) and socioeconomics [29, 34, 35].
As part of guidelines for prognosis studies [36] the present study had a prospective design and used a large sample. The prospective design limits the risk of recall bias, and being population-based is a particular advantage regarding mental health issues, since individuals with psychiatric disorders are often reluctant to seek treatment, especially men and the elderly [4, 6]. The study also had a population-based sample stratified for age and sex, recruited from a population with an age and sex distribution that is very similar to that of Sweden in general [20].
A number of limitations with the present study should be considered. First, the relatively low response rate of 40.0% at baseline has implications for the representativeness due to potential selection bias, in particular among men aged 18–29 years, for whom the participation rate at baseline was 17.3% (although higher at 3-YFU, 54.7%, and 6-YFU, 72.7%). The response rate increased, however, to 73.4% at 3-YFU, and to 82.5% at 6-YFU for the total sample. This may be a concern from a clinical perspective, as young adults tend to respond poorer to psychological therapy than do older adults [37, 38]. Second, the relatively long time interval (three years) between assessments may be seen as a limitation, insomuch as several changes in the respondents’ conditions might have occurred in the interval. It should be noted that the items referred only to the past week. It is possible that the participant may have recovered from or developed anxiety and/or depression, resulting in an inaccurate prognosis. Future studies may use shorter time interval between assessments, and perhaps also assess the condition over a longer period than one week, as was done in the present study. Third, although statistical analyses were not conducted regarding the most common six-year prognoses of caseness, certain caution should be taken when interpreting the results due to the relatively small sample sizes for depression only (n = 33) and comorbidity (n = 98). Fourth, and finally, the use of questionnaire instruments to assess anxiety and depression brings both advantages and disadvantages. As much as they are validated, they inform on high probability of being a clinical case, even if the respondent has not actually received a diagnosis from a physician. This reduces under-reporting due to not (yet) having sought help or other administrative lacunae related to health journal reporting and communication. On the other hand, it is open to erroneous self-report, rather than explicitly meeting diagnostic criteria based on a diagnostic interview. However, such error is arguably unlikely to be systematic in this kind of population survey that is independent from clinical practice.