Main findings
This study estimated the non-fatal burden of disease for subtypes of depression such as single episode and recurrent depression, graded by severity (mild, moderate, severe depression). All estimates were assessed from an individual or a population perspective. In addition, the disease burden estimates were adjusted for comorbidity.
We hypothesised that from an individual perspective, the disease burden of a single depressive episode would be equal to the disease burden of a recurrent depression. However, results show that a single episode is associated with greater non-fatal disease burden than recurrent depression. Apparently, DWs are equal (both 0.27) but the time spent in recurrent depression versus a single episode is shorter (107 vs 149 days respectively).
The other three hypotheses were confirmed; from a public health perspective, recurrent episodes are associated with greater disease burden (H2). Even though the mean time spent in a recurrent depression is shorter, the large number of people affected comes into play (n = 115 single episode and n = 246 recurrent episode). As expected, the burden of disease follows the gradient of symptom severity (H3). Finally, the burden of disease for each subtype of depression is lower after adjusting for comorbidity (H4).
Context and other studies
We need to place our findings in the wider context of the literature. Depressive disorder has been consistently identified as a leading cause of disability [
29,
30]. Currently, depression is the single leading cause of non-fatal YLD disease burden in high-income countries and it is projected to become the second leading cause of DALY disease burden (which also accounts for mortality) by 2020, second only to ischemic heart disease [
31]. More recent projections predict that depression might become the single leading cause of DALY disease burden in the high-income countries by the year 2030 [
29]. The Global Burden of Disease (GBD) studies [
32] conducted in 1990 and 2000 have quantified non-fatal health outcomes across a range of disorders at the global and regional level. The leading causes of YLDs were much the same in 2010 as they were in 1990, with depressive disorder contributing 8.1 % of total YLDs, ranking second after low back pain. Depressive disorder caused 63 million YLDs globally, but this figure was not disaggregated across the various types of depressive disorder. More recently, another GBD publication [
4] showed that depressive disorders accounted for most YLDs within the group of mental and substance use disorders (42.5 %, 95%CI: 33.3–51.7)).
Kruijshaar et al. [
7] studied the associations of severity and type of depression with functional impairment of the individual in a Dutch general population sample. Functional impairment was defined as limitations in physical, psychological and social functioning. Impairment was measured using the Short-Form-36 Health Survey (SF-36) [
33] and two other indicators of impairment were added to reflect some of the economic consequences of depression as well: the number of days spent in bed due to psychiatric, drug- or alcohol-related problems, and the time missed from work due to these problems. The study did not make a difference between disability at individual or population level. They concluded that higher severity classes were associated with more impairment. However, recurrent depression was found not to be associated with more impairment than single episode depression. In contrast, Vos et al. [
8] arrived at another conclusion. Their study focused on quantifying the burden of disease currently averted in people seeking care for major depression and the amount of disease burden that could be averted in these people under optimal episodic and maintenance treatment strategies. Results suggested that recurrent depression is the key driver of disease burden and that depression should be treated as a chronic episodic disorder in order to reduce this great burden of disease.
A study by Lokkerbol et al. [
9], aimed at estimating the non-fatal burden of disease at both individual and population level due to several mental disorders in a Dutch population sample, estimated an unadjusted DW of 0.35 for depression and an adjusted DW of 0.25. YLD/mln were estimated 9117 and 6524 respectively. These findings showed that comorbidity plays an important role in causing disability, analogue to our findings. Contrary to our findings however, Lokkerbol’s adjusted and unadjusted DWs and YLD/mln led to different rank orders. However, Lokkerbol et al. used the first NEMESIS data [
34] and as a consequence, their analyses were based on DSM-III-R disorders while ours were based on DSM-IV disorders. The fact that, from an individual perspective, the burden of disease for single depression decreased more than the burden of disease for recurrent depression after adjusting for comorbidity (0.017 versus 0.014 respectively) indicates that single depression is more often comorbid with other conditions.
Strengths and limitations
The strengths of this study include the population-based representative dataset on which the analyses are conducted and using valid instruments like the CIDI and the SDS.
Another strength is that the disability weights were derived from the general population. The most recent GBD study [
35] also accommodated data collection through population-based household surveys rather than from expert panels. This is important, because there are several ways of eliciting these valuations (e.g. from professionals in the medical field), but these are surrounded by controversy, and ultimately we need to understand how people value their own health [
36].
A third strength is that we could compute both unadjusted and adjusted estimates. Unadjusted QALY decrements and YLDs portray an accurate picture of the burden of disease in subtypes of people that are likely to have comorbid conditions—after all, in real life, we do not encounter people who have been adjusted for comorbidity. Therefore, unadjusted estimates may have value from a public health perspective. However, when the aim is to assess the burden of disease attributable to a specific type of disorder, then adjusted estimates are preferred, because adjusted estimates give information about the level of disability due to a single disorder. In our study, we explored both approaches and were thus able to shed light on the different conclusions that depend on the chosen perspective.
A final strength is the distinction between a clinical perspective on individuals and a public health perspective on populations. This distinction might be important to inform clinicians, decision makers and researchers in the health care sector in a conceptually clear and unambiguous way.
We acknowledge the following limitations of our study.
First, we used a 1-year timeframe for the analysis of disease burden. Therefore we might have missed the intermediate dynamics of incidence, remission and recurrence within that year. Perhaps more importantly, the longer-term dynamics of the epidemiology of depressive disorder beyond 1 year were missed in our study. To illustrate, Vos and colleagues [
8] estimated that in people with a history of depression the risk of a recurrence is 60 % within the 1st year after remission of the index episode, 70 % after 2 years, 80 % after 20 years and might be as high as 90 % on a lifetime basis. Despite this limitation, our approach is well-accepted and is used in the most recent WHO Global Burden of Disease (GBD) study [
5], which aims to estimate the burden of disease consistently across diseases, risk factors and regions. To estimate burden of disease the GBD study was based on point (current/past month) or past year prevalence estimates and excluded lifetime estimates as recall bias invalidated them as credible measures of disease burden.
Second, we focused on the non-fatal disease burden, ignoring excess mortality, but elsewhere we computed that the risk of premature death is a factor 1.65 higher in people with depression as compared to people without depression [
37]. Especially when taking the life-course perspective, mortality may have significant impacts on disease burden, and these impacts were missed due to our focus on ‘instantaneous’ non-fatal disease burden.
Third, people with severe conditions may have been unable to participate in this population-based survey, because they were hospitalised, and this is likely to have resulted in an under-estimation of the disease burden in the more severe forms of depression.
Fourth, we used the Brazier algorithm to calculate utilities and disability weights, but this algorithm was based on health state valuations in a sample of British people, while our sample was Dutch. This may have distorted our outcomes somewhat, although it is unlikely to change the overall results in a material way, as differences in Western Europe between national value sets, such as the set for the SF-6D, are small [
38]. While we feel that the people should be the ultimate judges of their own health, a panel of laypeople may be associated with limitations that are worth noting, such as lesser consistency, and the possibility that (healthy) laypeople have difficulties passing judgments on the severe conditions. This may have caused some errors in the estimation of the disability weights associated with the more severe disorders. In fact, regarding the more severe conditions, Brazier et al. [
22] pointed out that ‘inconsistent estimates (…) of the value of the poorest health states’ might be seen as a limitation of their method.
A fifth limitation is based on the fact that we relied on the DSM-IV diagnosis of major depression thus to the exclusion of minor depression which only became recognised as a disorder in the DSM-5. Nevertheless, it should be noted that minor depression also causes disability and is more common [
39]. This findings may have resulted in an under-estimation of population estimates of disease burden, which depends on the proportion of people affected.
The sixth limitation is concerned with the validity of asking someone who may be depressed about their quality of life. Their answers may be biased because of the negative perceptive biases so common in depression. Therefore, we expect a downward bias in the estimated QALYs of depressed people and conversely an upward bias in the estimation of YLD.
A seventh limitation is about the standard gamble technique. It is a well-tested technique and is consistent with QALY theory [
40]. However, in practice, appropriate adjustments for time preferences (i.e. discounting) for trade-offs made over a long period of time may be challenging to apply [
41]. This technique may result in a lack of sensitivity if the respondent does not perceive the temporary health state as sufficiently impaired to induce them to trade time from their life or to gamble with a probability of death. This ‘ceiling effect’ may occur despite the presence of disutility from the impaired health state [
42].
An eighth limitation is that data on somatic illnesses is based on self-report rather than diagnostic assessment, as well as the fact that we did not adjust for every illness, but only for 17 illnesses. This may have inflated DWs and YLDs estimates for depressive disorder.
A ninth limitation is that we did not adjust for severity of comorbidity, which might have provided a more realistic picture.
A final point: our analyses were performed on a sample that was diagnosed with depression during the past year, whereas the MOS SF-36 is designed to assess disability over the period of the past 4 weeks. The depressive episode may have occurred well before the last month as the average duration of an episode is 4–6 months. As a result, past month level of disability may not be truly representative for the disability that was actually experienced during the disorder. The individuals that are no longer suffering from depression may have therefore contributed to an underestimation of DWs, QALY decrements and YLDs. The use of a 12-month timeframe to estimate 1-month disability has the advantage that remitted disorders which continue to have residual adverse effects on disability are included. In sum, we captured the average effect of both acute and remitted episodes in the past 12 months on disability.
In this paragraph we mentioned a number of limitations. If these limitations led to bias overall, this bias will probably have led to an underestimation of the burden of disease but will probably not (or at least to a lesser degree) have impacted the ranking of the subtypes of depressive disorder.
Implications
Major depression is a leading cause of non-fatal disease burden, but it is not a homogeneous condition and assessing the disease burden for subtypes of depression and according to a gradient of symptom severity may help prioritise treatment allocation.
Overall, we saw that the disease burden differs from one subtype to another and that comorbidity influences the results. In addition, our study showed that the burden of depression poses a substantial challenge both from a clinical perspective (at individual level) as well as seen from a public health perspective (at population level). This justifies the prevailing dichotomy of medicine into clinical medicine (directed at individual patients) and public health (directed at collectives). Health care providers who focus on helping individual patients may conclude from our results that a single episode depression causes a greater burden of disease than a recurrent depression. However, the high prevalence of recurrent depression in the population does raise questions about how to best alleviate the disease burden stemming from recurrent depression when taking the public health perspective. The distinction between a clinical and a public health perspective may cause confusion in debates when not made explicit.
Our message is that individual and population perspectives are neither absolute nor independent concepts. Both perspectives serve different purposes and may require careful alignment when being used jointly. Such an alignment may result in the optimal balance between an individual approach directed, for example, at the episodic treatment of acute single episode depressions, in combination with a public health care approach with an emphasis on the longer-term preventive management of recurrences.