Health effects of depression: keeping economists on track

We know that depression is bad for a person’s health-related quality of life; but how bad?

That’s an important question because when it comes to resource allocation decisions, the value of treatments for depression have to be compared with treatments for a whole host of other diseases. The NHS operates within a budget after all. This is why NICE calls in the economists.

NICE’s favourite type of evaluation in health technology assessments is something called cost-utility analysis. Here utility refers to a person’s health-related quality of life, which is measured on a 0-1 scale and summed up over time to generate quality-adjusted life years (QALYs). NICE usually want these data (along with cost data) to be plugged into a model that extrapolates effects beyond what a trial is normally able to capture. This means that the choice of parameters for the model (such as the health impact of a disease) plays a crucial role in determining whether or not a treatment is recommended by NICE.

When economists build models, they generally turn to the literature to find their parameters. A new systematic review and meta-analysis of utility values for adults with unipolar depression could therefore have important implications for future NICE decisions relating to depression. It suggests that current approaches adopted by economists could lead to dodgy results, and it isn’t entirely the fault of the economists.

Methods

The authors conducted a systematic review, searching all the usual databases: EMBASE, MEDLINE, PsycINFO. The search included disease-related terms like ‘unipolar depression’ and ‘major depressive disorder’, as well as utility-related terms like ‘utility score’ and ‘EQ-5D’ (a widely used measure).

The authors subsequently carried out a meta-analysis to pool reported utility values. Papers were included in the meta-analysis if the reported utility values were categorised by disease severity in the form mild/moderate/severe. The authors divided the papers into two groups; those eliciting utility values directly, and those eliciting them indirectly. Put simply, direct valuations ask respondents how they would value depression-related health states, while indirect valuations apply weights to health states based on values elicited from the public. Whether we should use public or patient values is an ongoing debate.

Papers that didn’t classify disease severity were summarised in the review, but excluded from the meta-analysis.

Results

The search identified 420 references. Ultimately, 35 papers were included in the final review and 6 were included in the meta-analysis. Mean pooled utility values are shown in the table below, categorised by disease severity and elicitation method.

	Mild depression	Moderate depression	Severe depression
Direct elicitation	0.69	0.52	0.27
Indirect elicitation	0.56	0.45	0.25

For health state utility values, a score of 1 represents ‘full health’, while a score of 0 is a health state of equivalent value to being dead.

Discussion

Utility values like this do not communicate the nature or true health burden of depression. However, it’s clear that they reflect the substantial negative impact of depression at all levels of severity. It’s also clear that the mild/moderate/severe categories can be used to differentiate between the size of effect on health-related quality of life amongst patients. The pooled values for mild depression are consistent with values observed in various chronic physical health problems, while severe depression is far worse.

The review identifies a lot of variation between the studies, but economists tend to just pick one utility value from the literature that they think is most appropriate.

The primary purpose of this study is to provide economists with more robust figures with which to parameterise their models. However, there are also implications for the wider research community.

The review finds that the majority of studies do not classify depression based on whether it is mild, moderate or severe. While such an approach may not hold value in some primary research, the review highlights its potential value for future secondary research. This study lost a lot of data in the meta-analysis due to the non-categorisation of disease severity. Wherever possible, primary research should report severity of depression. Furthermore, practitioners should seek to reach consensus on the best classifications of severity to use.

The review also highlights the potential for huge differences in cost-effectiveness results based on whether direct or indirect elicitation methods are used. NICE’s preferred elicitation method is to use indirect valuation, and in particular the EQ-5D. One implication of this study is that while using indirect methods would attach a greater value to complete remission, they would attach a lesser value than direct methods to reductions in severity. This suggests that public valuation may underestimate the value of some treatments. The debate over which method to use takes place almost exclusively amongst economists, yet practitioners and mental health researchers are better informed about the nature of depression. Their engagement is crucial if the best approach is to be adopted.