Imagine for a minute that there was a ticking time bomb that needed defusing. It’s just like in the movies, a jumble of wires, some sticks of dynamite, and a rather obvious digital countdown on the side of it. It’s ticking ominously closer to zero. If you defuse it in the wrong way, the bomb goes off.
Now imagine, that you have an expert who knows how to safely defuse this bomb. The expert does this regularly, many times a day, and knows what to do like she knows the back of her hand.
So my question to you is this. If you were locked in the room with this bomb, how would you want it to be defused?
This is where the expert walks into the room, swiftly snips a few wires (usually the red one in the films) and safely defuses the bomb.
The expert employs a skilled technician. The technician’s skills lie elsewhere in maintaining electronic circuits. This technician has very little experience of specific types of bomb but knows something about bombs in general. The expert has sent the technician detailed instructions about this type of bomb and how to defuse it. The instructions are quite long and detailed, and take a large amount of time to read and understand. The technician also has many other sets of instructions that she needs to read. The expert guides the technician by shouting instructions to her, and the technician sends pictures of her progress in defusing the bomb back to the expert. The expert then uses these pictures to decide on the next step. She then shouts more instructions back to the technician. This process is repeated until either the bomb is safely defused, or everything goes BOOM!
Oh, and the technician has five other bombs that are similar but different to defuse at the same time. Each separate bomb needing different instructions from different experts simultaneously.
At this point, I’m guessing that if you were relying on these bombs being safely defused, you’d want option A.
It should be quite obvious to everyone that option B has a much greater risk of error. It should also be obvious that the risk comes not from the technician’s level of competence, but from the system that she is working in.
Yet, stunningly, in the NHS, we choose the equivalent of option B to manage some of our highest risk prescribing. It’s called “shared care” and like option B above, it has a large amount of risk due to the way it is structured. I shall endeavour to explain why.
Instead of a bomb, imagine a high-risk medication. This is a medication that needs regular dose changes and monitoring for serious side effects. Instead of going BOOM when things go wrong, a patient is potentially harmed. Perhaps they even die.
“Shared care” is where the dosing of a high-risk medication is decided by the hospital, yet it is prescribed by the GP. It is done for reasons of convenience to the hospital and the patients. For medications that need regular monitoring, the monitoring may be organised by the hospital, or it may be organised by the GP. As the GP is the one who is prescribing the medication, the ultimate responsibility for any errors lies with her.
“Shared care” is meant to be done by agreement. In practice, it isn’t. The agreements are long, unwieldy, and badly written. Sometimes they don’t exist at all. If they do, the hospitals rarely bother sending them. GPs don’t have the time to read them thoroughly. Neither are they able to internalise the knowledge because it's only a relatively small part of the GP’s job.
With high-risk medications, monitoring, dosing, and prescribing should all be done by the same organisation. Ideally, they should all be done by the same person. This person should be doing this often enough that they are unconsciously competent at it. Going back to the bomb metaphor, this is equivalent to option A, where the expert comes in and defuses the bomb. This is inherently more safe.
In theory, there should be no problem with GPs doing this - if they know what they're doing, and they're adequately resourced, and they do it often enough to be competent at it…
Arguments for “shared care”
There are three main arguments that are used to perpetuate “shared care.” It will not surprise you to hear that I reject all three.
The first is that it is convenient for our patients. I disagree. The dosing, monitoring, and prescribing of high-risk medications from secondary care could be made as simple as it is from primary care. There are many methods of safe secure electronic communication that can make direct communication between the patient and secondary care very convenient.
The second is one of cost. Again I disagree. Shared care reduces costs, by saving on secondary care appointments. However I think this is a false saving and in fact, what it does is transfer the costs from secondary care to primary care. It also ignores the cost of iatrogenic harm through medication errors. Also, volume effects would make flow efficiencies much easier to achieve in secondary care than primary care. These flow efficiencies would then lead to significant savings. I will elaborate on this later on in this blog.
The third is that by prescribing, the GP will be aware of all the medications that the patient is on. This will lead to less medication interaction errors when the GP prescribes other medications for their patient. This is true. However, with modern medical systems, it is quite possible to record that a patient is being prescribed high-risk medication elsewhere without having to prescribe it yourself. This reduces this risk. At the same time, the secondary clinic is quite capable of getting an up to date medication history from the patient or surgery if it chose to. The system would be unwieldy at first, but it would undoubtedly be possible to streamline it. This argument is also fallacious. Secondary care clinics already start patients on high-risk medications. This, if anything, is even riskier than continuation prescribing.
Be that as it may, there is still a really big safety issue with “shared care”.
If you have one person whose sole responsibility is to monitor a high-risk medication, they are likely to have an ordered system for that to happen.
However, every other drug that you add in that needs monitoring increases complexity in the system.
So, what in theory would be safe for one single drug becomes progressively less safe for each additional drug that you add in. This is a classic values mismatch. Secondary care only sees the value to them in handing over prescribing of a single drug. They don’t see the risk involved in handing over the monitoring. They also don't see all the other dangerous drugs that primary care is being asked to monitor as well.
Then add in fatigued GPs with limited resources and time, and it's a recipe for disaster.
Incidentally, these types of values mismatches combined with micromanagement are a very common cause for unsafe systems in the NHS. More on that in a future blog.
K.I.S.S - Keep It Simple Stupid
In system design, every step in a process is an opportunity for the system to fail. We all know this from the game that we played as children. Nowadays it’s known as the “telephone game”. You whisper a message in the ear of one child. That child passes the message to the next one down the line. They do this by whispering it to each other. By the time the message reaches the end of the line, it has changed beyond recognition. This occurs because there are many steps in this process.
Therefore good system design simplifies a process by removing any unnecessary steps. There are some steps in a process that are particularly high risk. These are known as “hand-offs”. These are when a process has to pass from one organisation to a different organisation. We’ve all had the experience when one company swears that they’ve sent something and the other says they’ve never received it. Safe system design involves keeping these “hand-offs” to an absolute minimum.
In the case of high-risk medications, this involves making sure that dosing, monitoring and prescribing are all done in the same place. This minimises “hand-offs”. This also reduces the number of other steps between each stage. This makes the process inherently safer. For the vast majority of high-risk medications, because the competency in dosing will lie with secondary care, so should the responsibility for monitoring and prescribing.
To use the bomb metaphor, this means the expert is defusing the bomb directly, rather than relying on shouting messages and getting picture messages of progress back.
I said, Keep It Simple Stupid
The second principle of safe system design is not to have unnecessary cross-over of process streams, as this is an area of risk too. Cross-over is when you ask someone to do multiple jobs simultaneously within a period of time. As we’ve all experienced, being interrupted when doing one job, often means that we forget something. This means that rather than putting multiple workloads onto a single organisation, it is inherently safer if each separate clinic does it for their own drugs and monitoring.
To go back to our bomb metaphor, this is the equivalent of getting someone to defuse a single type of bomb, rather than asking them to defuse many types of bomb simultaneously.
Reaching proficiency and the opportunity cost of streamlining
Finally, there are volume effects. If you have many patients on the same thing, it will be easier for you to become competent at managing them. It takes effective practice to become safe and efficient at any process. If you only do a couple a month, it will take much more time to become effective at it. In fact, if it’s really complicated, you may never reach the threshold for effective practice if you’re only doing it infrequently . The threshold is the level of repetition at which you start to learn new things about the task that you are doing. This is also known as unconscious competence.
The other volume effect is that to safely and effectively streamline any process takes time, resources and effort. If you’re only doing something for a relatively short amount of time out of your day, there is very little incentive to streamline the process. After all, what’s the point in spending many hours working on improving something that only takes a few minutes out of your day? However, if a single organisation is spending a large amount of time doing a process, Pareto’s law says that they’re going to get much more benefit out of streamlining it. It’s also going to be easier for them to streamline because they will be able to collect better quality metrics in a quicker time for their improvement process.
These volume effects are arguments that to improve safety, these high-risk medications are best managed in one place. The logical choice for this one place is the secondary care clinic that is managing the condition that the medications are being prescribed for.
Indeed, this is probably why we’re getting so much resistance from our colleagues in secondary care when we pass this work back to them. It’s a lot of work, and up until now, they’ve had no need to streamline it. Instead, they’ve been able to hide it by dispersing it across many organisations. This also means that secondary care has little incentive to improve the process.
To return to the bomb metaphor, this means making sure that the expert has enough experience of defusing a certain type of bomb. This means that they will get better at it more quickly. It also means that they are more likely to discover quicker and safer ways of defusing the bomb.
Cost of risk being born by primary care
Of course one of the problems is that we cannot make a quantified assessment of these safety issues as the NHS doesn’t systematically collect data about iatrogenic harm to patients. Neither has there been a real world randomised controlled trial looking at the impact that share care has on patient safety.
Another effect of this lack of data about the risk of harm to patients is that GP surgeries and CCGs cannot possibly price in risk when commissioning these services. This means that unless their LES is particularly generous (unlikely!), it is likely that GP surgeries are supplying this service at less than cost.
Part of the reason that GP indemnity fees are going through the roof is that there has been this untested and uncosted transfer of risk from secondary to primary care.
I shall write more on the effects of risk on costs in a future post.
“Shared care” is like option B in the bomb disposal metaphor above. It is inherently unsafe, even when implemented as specified. The fact that it is rarely implemented as specified only adds to the risk involved.
- - It is unsafe because it creates unnecessary separation between expertise and action.
- - It is unsafe because it adds unnecessary complexity and “hand-offs” to the system.
- - It is unsafe because it takes no account of the competing workload pressures on the GP and the effects that they have on safety.
- - It is unsafe because it takes no account of the asymmetry of complexity between handing over one high-risk area and having to deal with multiple high-risk areas because they’ve been handed over to you.
- - It is unsafe because it takes no account of the beneficial effects of repetition on proficiency.
- - It is unsafe because it limits the potential for flow efficiency improvements to be analysed and made.
This is before we start to account for all the second order effects of shared care on General Practice. These include, but are not limited to, resources diverted from patient care because of the complexity and administration, resources to deal with the increased risk of complaints, the effects of increased stress on clinician performance…
The list goes on.
If we are to take the Berwick Report seriously, we have to start to prioritise patient safety above all else.
This author believes all the reasons given above present an overwhelming argument for halting “shared care” in its current form. Don Berwick states that a “culture change” is needed if the NHS is to survive. I believe that this is one of the first steps towards that culture change that we should be taking. Until we have reliable data about the safety of “shared care”, we should not be participating in it.
Furthermore, if General Practice and the NHS is going to survive in a recognisable form, we have to stop prioritising patient convenience over safety.
I will expand on how the NHS needs to change its approach to systems improvement in a future post...
Please let us know what you think in the comments
As ever, if this makes sense to you, please share this as widely as possible. If you do print it off, we’d appreciate it if you’d also print off links to our website, facebook and twitter accounts.
We at itamus hope that you find these blog posts interesting and useful. In return, we’d appreciate your help in raising our profile.
Please follow us on twitter: @itamuscom
Please “like” and more importantly “share” our facebook page and posts:
Please have a look at our prototype web apps, and give us some feedback: http://www.itamus.com/
Colin & Richard