Dear Rodolfo                                                                                           7th February 2007

 

Reflections on the Scientific Method in Medicine

 

I have just finished reading the Essay you sent me in January.  Before I started, I thought it might be quite a difficult piece to read, and in the sense that every word and sentence contained a separate thought, it did need a lot of concentration.  But it was a splendidly constructed and argued story, and I have thoroughly enjoyed reading it.  I was very impressed by the width of your reading and knowledge, and you have written in quite splendid English style as well – congratulations!

 

I thought I would reply this way rather than by email, so that I can edit freely as I go along.  As I read your paper, I made marginal notes, and I am using these to try to construct a reply which picks up points and issues that you raise, but also tries to synthesise my own position at the same time.

 

It would be easy to say that I agree with all you have said, and in a way that is true as I completely agree with your definition of the problems we face as clinicians with a commitment to research, and also agree with your Summary and your Conclusions and Recommendations.  But I am going to take a different view about how to cope with these problems.  It may seem on the surface to be a rather pessimistic and nihilistic view, but it reflects my own experiences of some 40 years in the world of research and ideas, and in many ways I feel quite positive and comfortable with my rather different approach to coping with the same dilemmas.  I will try numbering my paragraphs to keep ideas and comments apart, and will try not to repeat myself as I go along.  Some paragraphs will follow the order in which points are made in your paper, whereas others are ‘asides’ as I go along.

 

1. As I said above, your knowledge of such a wide philosophical literature puts me to shame.  Having said that, I am a huge fan of Alvan Feinstein, and have quoted from his Clinical Judgement and from other writing many times.  I was also greatly influenced by Thomas Kuhn’s ‘Structure of Scientific Revolutions’ (I think that is more or less the correct title), which I am sure you know of although you don’t quote from it.  The other writer I want to throw in is Sir Peter Medawar, who I will refer to later.

 

2. Your Summary on page 1 is an excellent review of the problem we are discussing.  I particularly agree about the ‘too little theoretical research realm’, which has meant that too much/most research (certainly in our field of general practice/population medicine) does not really seem to come from anywhere logical or end going to somewhere really useful.  I have always tried to make sure that any work I was doing personally had two levels to it: one, relating to informing a clinical or clinical-related topic; and two – usually as important or more important – taking forward our understanding of some theory or generalisable phenomenon. (I will refer to that later too).

 

3. My first problem with your essay is its conclusion that there is a single integrative solution to the problem we agree exists; my second problem is knowing how to present my case for that position as eloquently as you have presented yours!

 

4. I suppose I start from two premises.  The first (here I agree strongly with you) is the usefulness of trying to ‘model’ concepts (a position you clearly share).  I use two models.  The first model (if that isn’t too grand a term) relates to what I understand is the representation of ‘systems theory’ in medicine (you talk about this in several places).  This is in the form of a triangle, split into layers, with the cell at the top, and the community at the bottom.   It is clear that the approaches to researching in these different layers (of which I recall there are six or seven) are significantly different;  but so also are the beliefs and values of those who work in them.  It is certainly important that there should be dialogue, understanding, empathy and support amongst those in different areas.  There will also be times when collaboration across boundaries is desirable and indeed is and achieved.  But to attempt to integrate everything is not necessarily appropriate – and almost certainly not likely to be achieved.

 

5. Going back to Kuhn, ‘paradigm change’ occurs when the accepted theory driving research in a domain of activity proves not to adequately inform further researches in that domain.  Thus it seems desirable to let each section of the ‘systems theory’ triangle go about developing its own paradigms, keeping all the time an eye open for when paradigms can be shared across different strata in the model.  In one of my more reflective articles (and Ian McWhinney has argued similarly before me) I suggested that general practice research with its focus on the person would become the new ‘medicine’ paradigm as biomedicine proved less and less able to improve human health.  Then along came the human genome project, and my argument began to look like a nonsense.  However, it is clear that the human genome project may not deliver all that much for most of society for most of their lives, and perhaps my cause will in due course prevail.  But for the present, it seems better that each paradigm develop on its own, rather than try to artificially weaken both by trying to integrate them.

 

6. The second model I use when approaching problems (particularly of relationships in families and in communities) is to draw a series of circles, each circle representing an interest or a person.  Then move the circles around so that they overlap where appropriate.  Very soon a complex web emerges.  It is quickly clear that often values and priorities in the different circles are so different that consensus is unachievable.  However, it is still possible for those in fundamentally different circles to be persuaded to understand, tolerate and even sometimes value and respect different positions.

 

7. The circles of medical research are relatively easily modelled in sets of overlapping circles, and indeed the thrust of your paper is suggesting this as a basis for integrating methods.  But medicine has overlaps with all areas of human endeavour (that is with ‘politics’ in the societal sense), and there comes a point where the forces driving social phenomena (for example poverty) and those influencing, say, medical schools (for example, the need to generate research revenue) are so far apart that integration is neither practical nor probably meaningful.

 

8. My second premise (where I take a different position from yours) is built on Sir Peter Medawar’s advice to researchers to ‘tackle the most important problem that you think you can solve’.  I am not sure that you have spelled out enough what the ‘most important question’ you think your new model would answer.  Yes, you describe modelling drug trials, but that is not really a general-practice/public health type of question.  You talk of the need to assimilate modern statistical expertise, but my experience of statistical opinion as applied to much primary care research has not always been helpful.  In our field it is often judgement on the meaning of data that is called for, as well as – of course – rigor in constructing data collection and genuine attempts to identify important variables and confounders.  In general practice research, the individuality of patients may be a given on paper, but we really have to take on board just how difficult it makes our research to do, and how easy it is for statistics to mislead instead of help when we are analysing results.

 

9. As an aside, I often wonder at meta-analyses bringing together dozens of papers and ending up by concluding that we can be increasingly confident that there is a minimal benefit to a population of doing x or y.  Does this really advance population well-being?  My concern is that the integrative model you are proposing would, were it to prove feasible, only increase the focus of funders on promoting work of the kind that we together would feel there is more than enough of already, at the expense of people and population centred enquiry.

 

Perhaps that is enough to stimulate a reply.  In short, I admire your essay;  I totally agree with the nature of the problem you have described; but I am not sure that your solution is the one I would go for!

 

Do write again!

 

Best wishes

 

John