Thursday, 27 June 2013

Time to stop wasting data: The Clinical Practice Research Datalink

Let's stop wasting data

I was speaking at an event last night for general dental practitioners who are involved with, or interested in joining, studies in dental practice. One of my fellow speakers was a very engaging speaker called John Parkinson, who is director of the The Clinical Practice Research Datalink (CPRD) about which I will speak a little later.

Diagnostic codes

A little while ago I blogged about how useful it would be if dental electronic health records were to incorporate diagnostic codes to allow tracking of patient outcomes over time. The potential to observe the success or otherwise of patient management could be huge. Not only could we observe outcomes to routine treatment but potentially also conduct trials in practice using the codes to track what problems patients arrive with, any interventions we randomise them to, and the outcome.

So, for example, one of the audience last night asked the question: in a time-restricted environment is it better to provide oral hygiene advice or do a quick scale and polish, presumably for a patient with gingivitis rather than destructive periodontal disease. Well, using the SNOMED codes that I blogged about at the first encounter a dentist would record the following from a drop down box in the patient's record:

Code 66383009 Gingivitis (disorder)

Then the clinician records how it was managed. This could be as part of a randomised controlled trial - half patients are randomly allocated to scale and polish, half to oral hygiene instruction:

Code 234696006 Scale and polish teeth (procedure)


Code 58707002 Oral hygiene education (procedure)

Then, at the next visit the dentist, hygienist or therapist records whether gingivitis is present using the same code:

Code 66383009 Gingivitis (disorder)

If this data - that we probably record for thousands of patients every day anyway - were collected in this organised way then there would be a massive amount of potentially really useful data that we could use to answer the question asked by our colleague last night.

Clinical Practice Research Datalink

This is where the The Clinical Practice Research Datalink comes in. John Parkinson already churns the numbers for general medical practices using data from the last 25 years. What they can now do is to identify potential participants in research based on their medical conditions and create a pop-up that asks the clinician to ask the patient if they would be happy to take part in a trial. If the patient consents then they are randomised automatically to one treatment or another.

So imagine this as a dentist participating in a trial to find out whether you should give oral hygiene advice or a scale and polish in your short appointment: 
  • a patient attends for a dental examination. You diagnose gingivitis and click the appropriate code (as above). 
  • A pop-up screen asks you to ask the patient if they would like to participate in the trial. They consent and you click the appropriate button.
  • The pop-up then tells you that for this patient you should give oral hygiene advice. So that's what you do.
  • Then you see the patient again in 6 weeks / 6 months / 12 months and without even having to think about the trial you just record what you see. If there's gingivitis then you record it. If there isn't you don't.
  • The number crunchers process the data and in a year or two we have a massive data set that allows us with some confidence to say whether one or the other is more effective when done alone.

So why aren't we doing this?

This effortless utilisation of information for the betterment of our patients is only likely to happen if we can get dental electronic health record systems to adopt the coding that enables us to track the problems patients present with, the interventions that we use, and the outcomes some time afterwards. John Parkinson wants CPRD to facilitate this - he just needs access to the data we have on our computers. And whilst this has generally been an NHS programme because GPs are almost all in it, it doesn't mean private colleagues without NHS contracts couldn't also be joining the network. 

The CPRD could also link our records to general health records. So potentially we could research whether, say, giving dietary advice in the dental clinic results in wider benefits to the patient - like reduction in obesity-related diseases, for example. And we could also track the outcomes of treatment completed in, for example, dental hospitals once they have been discharged back into the care of a general practitioner so allowing us to understand better the long term impact of hospital dental care.

As a dental profession I think we really need to join our medical colleagues and engage with this great opportunity and we need to start by persuading the dental systems to play ball. I don't know who the best people are to help with this...but perhaps you do.

Sunday, 9 June 2013

Respecting tacit knowledge in knowledge transformation

I used the term knowledge transformation in the title of this blog instead of the common terms knowledge translation or knowledge implementation because, as Morgan Meyer and others have argued, knowledge is rarely used in its original sense.

In a previous blog I discussed Gabbay and Le May's work in general medical practices that led them to understand that doctors create their own "mindlines". I have also read more recently about the SECI model proposed by two Japanese researchers in the 1990s, Nonaka and Takeuchi. They were studying the creation of knowledge within commercial organisations and sought to include the tacit knowledge we all gain through doing and observing but which may be hard to explain, with the explicit knowledge shared through discussion and reading.

This latter model reinforces what Gabbay and Le May found, that doctors' knowledge developed not only through the use of explicit information in the form of guidelines but also the tacit knowledge gained through observing how each other worked and their own interactions with patients and people or organisations outside their practice.

The SECI model stands for: socialisation, externalisation, combination and internalisation. [The diagram here is reproduced from the Gabbay and Le May book].

The idea is that we learn tacit knowledge through a process of socialisation. This involves us observing others working. In an organisation where knowledge creation takes place, they argue, this tacit knowledge needs to be made explicit. This is done by individuals using metaphor and analogy to explain their experiential learning. In turn the externalised knowledge is combined with other information sources.

From my point of view, looking at how to help clinicians use research in decision-making, this includes using research but also using other information sources. The combined knowledge might then be expressed as a lecture, a practise protocol or a guideline generated from this process. Finally, we internalise the combined knowledge as we make sense of it and, if we use it, introduce it in some form into our practise.

Much of the knowledge translation world has ignored, it seems to me, not only the other sources of information that go into making a decision but also the tacit knowledge we all have that may never be expressible. If we take account of pre-existing knowledge - and value this as much as the scientific knowledge we gain - we take a constructivist philosophical view. If we take the view that only scientific knowledge gives us true knowledge then we take a positivist view.

I have moved progressively from the positivist to the constructivist over the last few months as I have learnt more about the way we actually function. It seems to me that even if we wanted to perform every activity according to a positivist stance that everything can be explained by some physical or social law, we would never succeed because we inherently value our own knowledge and experience.

Moving forward, my feeling is that if we - those interested in helping useful and appropriate research to be incorporated into clinical decision-making - are to serve any function then we need to  respect more the complex decision-making and knowledge creation that all clinicians (and human beings generally) are engaged with daily.