The outcomes of mental health care for depression over time: A meta-regression analysis of response rates in usual care, 2024, Karyotaki

The outcomes of mental health care for depression over time: A meta-regression analysis of response rates in usual care
https://www.sciencedirect.com/science/article/pii/S0165032724007377

Highlights

• Response rates in usual mental health care have not improved over time.
  
  
• Only about one in five patients in usual care respond.
  
  
• Response rates are low in specialized mental health care, perinatal and general medical care.
  
  
• Response rates are somewhat higher in primary care.
  

Background
Over the past decades dozens of randomized trials have shown that psychological treatments are more effective than care-as-usual (CAU). It could be expected that these treatments are implemented in routine care and that the response rates in usual care improve over time. The aim of the current meta-analysis is to examine if response and remission rates in usual care have improved over time.

Results
We included 125 CAU control groups (8542 participants). The response rate for all CAU control groups was 0.22 (95 % CI: 0.19; 0.24) with high heterogeneity (I2 = 83; 95 % CI: 80; 85), with somewhat higher rates in primary care (0.27; 95 % CI: 0.23; 0.31). We found hardly any indications that the outcomes have improved over the years. The meta-regression analysis with publication year as predictor in the full dataset resulted in a coefficient of 0.1 (SE = 0.01; p = 0.0.35). A series of sensitivity analyses supported the main findings. Remission rates and pre-post effect sizes also did not significantly improve over time.

Conclusions
Response and remission rates in usual care are low, with the large majority of patients not responding or remitting, and the outcomes have probably not improved over time.

 

This caught my eye:

It is very important to examine what the exact reasons are for the stable response rates over time and how this rate can be improved. It makes little sense to conduct hundreds of randomized trials on psychological treatments, when they do not lead to better routine practice.​

In discussion elsewhere in the paper there is this:

There are two possible explanations for these findings. One is that the evidence-based interventions are implemented in routine care but are not as effective as RCTs suggest when applied in routine settings. This may be caused by differences between the context in which RCTs are conducted compared to routine care, or because patient characteristics in routine care differ from those in RCTs.
...
Another explanation for the absence of improved outcomes in routine care is that treatments that have been found to be effective in RCTs are not disseminated, or not in the right way.​

Which clearly neglects the most likely explanation, the giant wooly mammoth in the room: those randomized trials (not RCTs, mere randomized trials) are just not any good, and produced a wildly misleading evidence base. Which is one of the most common topic of discussion on this forum: clinical trials based on psychology are just awful and should basically be either fixed, which would likely invalidate most of the common beliefs, or shut down.

And it applies to, well, basically the whole thing:

It should be noted that not only psychological treatments can be held responsible for the lack of improvement of outcomes in routine care. Other general types of treatments like pharmacotherapy, exercise and bright light therapy, are also part of routine care and have apparently also not seem to be able to improve outcomes over the years.​

Which speaks to a much bigger problem of routine use of obviously ineffective care models, showing a blatant lack of adequate surveillance of real-life outcomes and odd systemic indifference to it.

Despite this, it's a common meme among MDs to call for medicine to be more biopsychosocial. Well, this is what biopsychosocial is: straight up fake. It's hard to understand this obsession with expanding something that has never been shown to work.