The 2021 – 22 IAPT Annual Report delivered an unexpected surprise. Set against other therapy modalities, and its own past performance, the recovery rate for CBT has deteriorated significantly in the past year. Here, we showcase the data and speculate on what might lie behind this drop in performance.


Estimated reading time: 6 minutes


I have to confess I don’t get much joy from my yearly peer under the hood of the IAPT Annual Report. I also worry that in giving you the headlines, the result is starting to sound like a rehashed annual whinge-fest.

This year, I’m therefore delighted to be able to offer you something rather different. Not a stunning improvement in IAPT’s overall KPI data. Not much has really changed there. No, what really stands out for me this time is the very sizeable drop in the overall 2021 – 22 recovery rate for CBT, compared both to the previous year and to other bona fide therapies.

Why would this be the case? What’s happened? Many factors might be implicated, and the truth is I really don’t know. I doubt anyone else does either, assuming they’ve noticed. But speculating is fun, isn’t it, so we offer one or two options further on. Why not join us and offer your own in the comment section?

My, how you’ve grown….

IAPT has come a long way since its early days, when you could have any therapy you wanted as long as it was CBT. Now, the 2021 -22 IAPT Annual Report lists outcome and other KPI data for no less than fifteen different interventions.

Of these, I’d argue that five would broadly meet the criteria for recognition as a bona fide psychological therapy. These are Counselling for Depression (CfD), Cognitive Behavioural Therapy (CBT), Interpersonal Therapy (IPT) and Brief Psychodynamic Psychotherapy (BPP). I’d also included Couples Therapy for Depression, though it’s an intervention I know relatively little about.

One of the datasheets in the report (clip below) contains outcome percentages by category (recovery, reliable recovery, improvement, no reliable change and reliable deterioration) for each.

The remarkable drop in the recovery rates of CBT

From these data I’ve extracted below the recovery rates, and the number of clients on which they are based, for each of the four individual-focused psychological therapies listed above. First, take a look at the 2020 – 21 recovery rates for each. Across the four, there’s not much difference, right? IPT is a little lower than the others, but its numbers are small enough to raise issues of statistical significance.

Now, look at the recovery rates for 2021 – 22.

Whereas the recovery rates for CfD, IPT and BPP are within a whisker of each other, that of CBT has dropped by a full four and a half percentage points in a year. It is now six to seven percent below the others.

What’s behind the drop in performance?

What’s going on? All other things being equal, we would expect the outcomes of bona fide therapies to be broadly equivalent in their outcomes. That was the case in 2020 – 21, but not the following year. Factors might include:

  1. A change in the characteristics of the workforce delivering the interventions
  2. A change in the characteristics of the interventions being offered
  3. Insufficient sessions provided for the therapy to be optimally effective
  4. A preference effect, whereby clients receiving their preferred therapy achieve superior outcomes

I’d be inclined to discount the first option. It’s hard to see how the characteristics of the workforce would have changed so significantly as to impact recovery rates so dramatically in the course of one year.

I certainly wouldn’t discount option two. I’m not an industry insider, but I have the sense that within IAPT, CBT may sometimes be more of an organising term that a distinct therapeutic modality. Any significant shifts in the way those interventions are delivered may impact on outcomes.

So to option three. Are clients getting a sufficient number of sessions for therapy to show an effect? Comparing the mean number of treatment appointments for each therapy across the two years shows a modest increase in the mean for each. So not much has changed there. Unless some combination of options one and two are also at play, it’s hard to see how the number of sessions alone accounts for the reduction in the recovery rate of CBT.

Option four. Here the question is, are clients who receive CBT less likely to recover because they are not receiving their preferred treatment? We know that client preferences have a positive effect on treatment outcome, so might this be at play here?

This would be a reasonable speculation, and if we knew what clients’ therapy preferences were, we could test this. Even if that were possible, however, we have the inescapable fact that not only is CBT’s recovery rate lower than other therapies for 21 – 22, it’s also lower than its own in the previous year.

Please, if you have alternative answers to this conundrum, we’d love to hear them.

Behind the data; the people

We should always remember that behind these data are people. People in need who came seeking help. Some were helped, but using the headline indicator of recovery as a benchmark, many were not.

It seems bizarre, especially given the IAPT model of using sessional measures and being able to track progress through therapy, that the data for so many clients are failing to show demonstrable impact. But being able to track progress isn’t the same as actually doing it.

It leads me to conclude that either the model of therapy intervention isn’t sufficiently flexible to adapt to progress feedback, or worse, that no-one’s paying much attention to it. Either way, for CBT at least, and the clients receiving it, the situation looks like it’s getting worse.

You can add your thoughts below. ↓↓

Just like you we thrive on feedback.

Please leave your thoughts on what you’ve read in the comments section below.


Share with your networks
Posted by:Barry McInnes

19 replies on “Is CBT losing its shine?

  1. Hi Barry, really interesting article – thanks for sharing the data. Unfortunately am not an industry insider to be able to explain this drop. One question aside though, am wondering why EMDR is not included amongst the 5 bona fide therapies you list? Given that more and more robust research is emerging https://www.frontiersin.org/research-topics/19948/present-and-future-of-emdr-in-clinical-psychology-and-psychotherapy—volume-ii
    and increasingly more queries from customers and clients asking if we provide it?

    1. Hi Elspeth and thanks for your comment. Briefly, I didn’t include EMDR for two reasons. First, thought I’m prepared to be wrong on this, I don’t think EMDR is generally factored into studies of therapy equivalence. so, (second point) I looked at the recovery rate for 21 – 22 for EMDR and it’s 19.3% on a count of 1,164. Doesn’t really surprise me though – the recovery rates are based on using GAD 7 and PHQ 9, and not really sure they’re the best measure of change for a trauma oriented measure? Hope that helps! Barry

      1. Unpopular view perhaps, but EMDR smells like pseudo science to me riding high off the back of a growing fad. Growing amount of evidence isn’t growing quality of evidence. Is it worth considering that patience preference for the brand effects skews perceptions of efficacy? Essentially, people might just be getting exhausted by a decade of cbt fanfare and are gradually seeing it as less valuable.

        1. Rick – many thanks for you comments. I can’t speak for anyone but myself when it comes to exhaustion with the CBT fanfare, but I think it’s hard to deny the way in which other modalities were discounted in the set up of IAPT and its early days. All this despite the overwhelming evidence for broad equivalence as far as therapy model effects are concerned.

      2. Thanks Barry – I agree about GAD7 / PHQ9 not being most effective measures of change in trauma: I use EMDR a lot in my practice and generally use the IES-R, which gives a much clearer picture.

  2. As an IAPT worker I can’t see why recovery rates for CBT specifically have reduced by such a large percentage. If I were to guess could it be waiting times for CBT are significantly longer than the other modalities?
    Could it also be starting symptom severity is higher for CBT clients and therefore they are less likely to recover?
    Also you raised the cohort of workers changing as a possible factor, then discounted it. It could be a factor, IAPT has a recruitment crisis as it’s system can feel depleting and lead to burnout. There is a high staff turnover, meaning they are losing more experienced therapists each year. Could the pandemic have accelerated this churn? I don’t know, lots of questions here.

    1. Hi Jen and thanks for your thoughts. You raise some interesting possibilities, and maybe I did discount workforce factors prematurely. You have me thinking! It’s worrying though, isn’t it? Has anyone that’s in a position to investigate at a top level, and/or make a difference, actually noticed? I can only hope so……. 🙂

  3. Please read William M Epstein’s three key books relating to psychotherapy.

    The Illusion of Psychotherapy
    Psychotherapy as Religion
    Psychotherapy and the social clinic in the united states, soothing fictions.

    Across these books he takes apart the key psychotherapy research including the foundational studies – what he demonstrates is there is NO evidence that any psychotherapy is effective and they can be harmful. If you can’t afford those books try The Therapy Industry by Paul Maloney and all of David Smails works.

    The entire system is a house of cards kept together by self and vested interest.

    1. Righto. Would this be the same William M Epstein who in July 2022 published the paper Corrective Abortion and Crime? In which there’s a proposal for the “corrective abortion” of individuals anywhere from birth to the age of 18, on the basis that it’s concluded that:

      1) the parents, particularly the mother, would have been better off if the living organism had been aborted as a fetus and that future contributions by the living organism will not compensate them for harms created during the years since birth;
      2) the society, judged by the actions of the living organism, would have gained from a timely abortion and that its unfavorable profit/loss ratio will probably not be corrected; and
      3) that the living organism itself would have been better off never to have achieved its current viability
      The mother is first in line to decide. Of course, no man would want to sully their hands with such a decision.

      https://www.researchgate.net/publication/362080097_Corrective_Abortion_and_Crime

      I think I’ll skip his other works thanks.

    2. I have no idea about this author either way, but I am wondering how he accounts for therapeutic successes. Perhaps such studies are the shadow side of, say, the hype surrounding cbt.
      I do think that those who publish books which deny observable reality should examine their own critical thinking skills, and confirmation biases.

  4. Hi. I’ve just come across this excellent site. My hunch is that clients are increasingly suffering from the kinds of real-life problems and adversities that CBT is just not equipped to deal with. No amount of tweaking your negative cognitions will put food on the table or money in your meter. And if you can’t afford to go out – to a cafe, a gym etc – then most forms of ‘behavioural activation’ will be beyond your reach. Other therapies aren’t always good at acknowledging real life contexts either, but they are not quite as committed to gaslighting as CBT.

    1. Hi Petronella and thanks for your comments.
      I agree with so much of what you say – I’d even say that if we’re seeking to alleviate the individual misery caused by wider societal problems then we’re in danger of colluding with a broken system. Let’s not forget that part of IAPT’s founding ideas was to get people off benefits and into work. We often talk about the GAD-7 and PHQ-9, and forget the Work and Social Adjustment Scale. If that’s not value laden then what is? If IAPT is the answer, then what is the question?

  5. Of course that is the purpose …. Why else pour so much money into a project that patently isn’t producing the predicted outcomes? Two ironies strike me: 1. If IAPT actually lived up to its own much-vaunted ‘evidence-based’ approach, it would close itself down (I understand this is what happened to an IAPT-style project in Sweden which was found to be making people WORSE overall) and 2. The IAPT workforce is so stressed by the ruthless target-driven culture that they themselves are developing anxiety and depression in large numbers. They probably know better than to refer themselves to IAPT as clients, though.

  6. Hi. I’ve just come across your blog in the course of searching for IAPT recovery data by therapy type. I found it a few months ago (and like you, found that counselling was performing better and more consistently than CBT). Next time I looked, the dashboard had a notice saying something like ‘errors had been found and the dashboard is temporarily unavailable’. Now I can only find data organised by region, and not by therapy type. Can you shed any light on this?

    1. Hi Rinda and thanks for getting in touch. Yes, you’re right, the search parameters seem to have been changed and that makes the data we can get at from the dashboard much more restricted. It’s very annoying.

      There’s an enquiry page for NHS Digital here https://digital.nhs.uk/about-nhs-digital/contact-us though frankly I doubt you’ll get much joy there. You might also access the data files from Resources section here https://digital.nhs.uk/data-and-information/publications/statistical/psychological-therapies-annual-reports-on-the-use-of-iapt-services/annual-report-2021-22, but they are very hard work as they’re huge .csv files in the main.

      The notice “Therapy-Based Outcomes Interactive Dashboard 2021-22: This dashboard has been temporarily removed due to an issue identified in this publication. We are working to resolve this and will republish the dashboard as soon as possible.” isn’t encouraging and looks like it’s been there for a while.

      I’m sorry I can’t be more help – but good luck nonetheless!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.