Wednesday, 19 January 2011

Care clusters: A race to the bottom?

Do you think you've got severe depression? Well you're wrong, you are in 'care cluster 4':
"This group is characterised by severe depression and/or anxiety and/or other increasing complexity of needs. They may experience disruption to function in everyday life and there is an increasing likelihood of significant risks.

Likely to include: F32 Depressive Episode (Non-Psychotic), F40 Phobic Anxiety Disorders, F41 Other Anxiety Disorders, F42 Obsessive-Compulsive Disorder, F43 Stress Reaction/Adjustment Disorder, F44 Dissociative Disorder, F45 Somatoform Disorder, F48 Other Neurotic Disorders, F50 Eating Disorder

Some may experience significant disruption in everyday functioning.

Some may experience moderate risk to self through self-harm or suicidal thoughts or behaviours.

Unlikely to improve without treatment and may deteriorate with long term impact on functioning."
So you're basically the same as someone with OCD or an eating disorder. Care clusters (see Figure 1 below) are the result of the Labour government's payment by results programme. You get allocated to a cluster partly based on 'clinical judgement' and partly automatically by a computer program using scores inputed by clinicians to answer 18 questions.*

Figure 1. The 21 Care Clusters
Unlike 'acute' medical trusts which are paid by 'activity' (e.g. how many operations they do) the mental health trusts will be paid by 'need', and that is defined basically by 'care cluster', which a different local tariff for each cluster. So the local mental health trust will get paid for 20 cluster 1 patients, 30 cluster 2 etc. It isn't entirely clear at this point how that will take into account that many patients only stay on a mental health team's books for a few weeks or months.

As with my local mental health trust care providers are now going to start deciding what services they are prepared to deliver for each care cluster based on that cluster's tariff (i.e. how much they'll get paid for treating that patient) and not by the current combination of supply and need. You are not paid by activity, that is by what care (e.g. therapy sessions or meetings) you deliver, but on what cluster someone comes under - therefore there is going to be a pressure to reduce the amount of care provided within each cluster to maximise profit. You also have to wonder whether patients that seem like they'll require more work, and thus cost, than others (e.g. personality disorders) will get taken on at all.

Locally we're developing care packages for each cluster, but you have to wonder how a cluster that includes major depression, OCD, and eating disorders can really have a generic package that is actually includes the appropriate evidence based treatments for those conditions. And where is the room for a bit of clinical judgement and addressing individual patient needs? 

This might not have happened if it was left to the current model where only a single mental health trust provides services to a given PCT but under the new government proposals for NHS commissioning by GPs they are subject to EU competition law and must commission services from 'any willing provider' based on price. So it seems likely there will be a race to the bottom, I may not want to deny you access to, say, psychological therapy, but if another provider is tendering for work with the GP consortium and they don't offer it they will be able to save money and come in under my quote.

This payment regime hasn't been introduced yet but the tools for implementing it are being put in place in 2010/11 ready to roll it out. Something worth looking out for.

* Bah, who says you can't quantify mental ill-health on a simple 72 point scale, and who says depression can't be nicely quantified on a 4 point scale e.g. 
"Question 7. Problems with depressed mood (current): 
0 No problem associated with depressed mood during the period rated. 
1 Gloomy; or minor changes in mood.
2 Mild but definite depression and distress (eg feelings of guilt; loss of self-esteem). 
3 Depression with inappropriate self-blame; preoccupied with feelings of guilt. 
4 Severe or very severe depression, with guilt or self-accusation."


Neuroskeptic said...

Hang on. "Some may experience significant disruption in everyday functioning."

Under DSM-IV criteria you must experience clinically significant impairment to get a diagnosis.

Yet for Cluster 4, it's optional. Don't tell me Clusters 1,2, and 3 are for people who experience no impairment in day-to-day functioning?

If so, what are they doing getting diagnosed at all?

pj said...

It is similar for ICD-10 (which I presume is the underlying basis) but what do you expect if you have a bureaucratic definition of mental illness?

I'm sure you're familiar with the way community mental health services work in the UK. Most patients are not seen by a psychiatrist but rather by a community psychiatric nurse who, in theory, isn't supposed to be diagnosing people with specific illnesses but rather addressing care 'needs'. This is just a further development of that, providing a semi-automated pseudo-diagnostic structure that circumvents the need for medical input to diagnosis.

Clusters 1-3 go from:
"Disorder unlikely to cause disruption to wider functioning."
"Disorder unlikely to cause disruption to wider function but some people will experience moderate problems."

So a question worth asking is why people with no dysfunction to functioning need to be so minutely subdivided when they're not really suitable for secondary care mental health services in the first place? Perhaps because payment is by cluster so there is money to be made wasting clinician time dealing with patients the GPs should be dealing with?

Maybe they're using the whole NICE/APA criteria for depression where all the labels are shifted so mild=none, moderate=mild, etc.

pj said...

I was reading this presentation from a consultant psychiatrist. He seems pretty upbeat but I noticed a few worrying things such as there was only 60% agreement between the clustering tool and clinician assessment and clinicians only got about 50% of the prototype vignettes into the 'right' cluster.

This tallies with my experience where the cluster tool doesn't even seem to get close to the cluster I'm thinking of when I've input some practice data (in one patient example which I designed to illustrate chronic severe psychosis it suggested dementia was most appropriate).

The examples of appropriate levels of care given in that presentation are pie in the sky stuff, which the author seems to acknowledge, but that is actively unhelpful to clinicians working on the front lines - being told by your own Trust that you should provide care X when they simultaneously prevent you from doing so.