RESEARCH FOR PRIMARY GENERALISTS: IMPOVERISHED DATA FROM SIGNIFICANT NUMBERS, OR RICH DATA FROM “INSIGNIFICANT” PEOPLE?

Julian Tudor Hart

Discussion paper presented to Conference of the European General Practice Research Network, University of Bologna at Bertinoro, Italy 9^th May 2009.

Before I left school in 1945, I was already committed to a Marxist world view, but in a romantic, adolescent, comic book version.  I entered medical school intending to serve poorer people, who seemed to understand the world better than I did.  I liked simple messages.  Conflicting evidence or doubts seemed to distract me from my purpose.  Young activists like simple messages.  So even do some who have retired.

In 1956, when all political certainties fell apart for anyone able to think for themselves, I was forced to grow up.  Simultaneously with the collapse of the British empire in Suez and the Soviet empire in Budapest, a first large crack appeared in imperial concepts of disease.  As a young family doctor in London, I followed closely the work of Sir George Pickering on the distribution of arterial pressure in whole populations,[1] and his ideological struggle with Sir Robert Platt.[2]  After a lifetime researching the nature of essential hypertension, Pickering concluded that it was not a disease, but a continuously distributed risk for several serious vascular outcomes.  Sir Robert Platt contended that there were two populations – diseased people with hypertension, and normal people without hypertension.  The first group was a concern for doctors, the second was not.

Fortune favours the prepared mind.  Because I was a Marxist, it seemed obvious that Pickering’s view opened up a huge new territory for research and clinical action – not in hospitals, but in primary care, where we could study whole registered populations leading their widely diverse real lives, at home and at work.  Every young doctor entering general practice after hospital training faced the same huge problem: we all found that very few of the patients we saw in primary care had any of the well-defined, isolated and readily classified problems we had studied in teaching hospitals.  What we actually found in primary care was a chaos, in which the official classification of disease was useful for only a small minority of patients.  In other words, for most of the people we saw, existing knowledge was not useful.  It followed from this that there was a great deal we didn’t know.  If you don’t know, you must organise to find out.  So research in primary care seemed both necessary, and potentially fruitful. 

Study of hypertension seemed a good place to start, because its continuous distribution, and continuous relationship to risks of lethal or mutilating outcomes, seemed to resemble other major health problems which together accounted for most of the serious illness seen in primary care: chronic respiratory impairments, type 2 diabetes and obesity, recurrent depression or demoralisation, alcohol dependence, and many other processes in which at an early stage in their development, there was no qualitative division between normal and abnormal people.  If we could get it right for control of hypertension in whole populations as they actually lived and worked, perhaps we might develop good models for health conservation, maintenance and recovery in other dimensions.

I left my practice in a slum district of London, first to get more experience as a junior hospital doctor, then as an apprentice in epidemiology, first to Sir Richard Doll, then to Archie Cochrane.  The connection was political.  Most post-war pioneer epidemiologists were socialists of some sort, either Labour reformers or Communist revolutionaries, so I had the advantage of being in an Old Comrades’ Network, equivalent to the Old Boys’ Network.  Little remains of the OCN today, rather more of the OBN. 

Most research is still centred on teaching hospitals, most of them serving plebean populations.  There has always been an implicit contract: plebs get free state-of-the-art care, and in return they allow themselves to be observed, taught upon and experimented upon.  Cochrane was doing research on coal mining communities outside hospital, where medical care was provided by general practitioners working in conditions and with resources, which had hardly changed since the first world war, with consultation times averaging around 2 minutes.[3]  I returned to family practice determined to provide state-of-the-art primary care as a precondition for research and teaching in the real world outside hospitals, just as this was a precondition inside hospitals.  We needed to start building peripheral foci of excellence, as well as centres of excellence.

I married Mary Thomas, one of Cochrane’s staff, trained in his very high standards of research methodology.  She understood the nature of research teamwork in data collection, and the full implications of randomisation for experimental studies.  Together, we eventually developed just such a peripheral unit of excellence, resourced by the Medical Research Council and supported by a central team of epidemiologists, statisticians, laboratory workers, and data processors (our work preceded the era of desktop computers). 

For the MRC, we did pilot projects to pave the way for large multicentre studies thoughout the United Kingdom.  For ourselves, we pursued the course indicated by Pickering’s work – to explore the full implications of looking at health impairments as they are distributed throughout the whole lives of whole populations, sick or well.   Most epidemiologists were collecting small data from large populations – so large, that the individual people contributing those data lost their names and became numbers: anonymous, depersonalised, and therefore impoverished data.  We had only about 2000 people in our practice population.  Our largest studies, including the whole upper Afan valley and part of the Llynfi valley, reached at most about 5000 people.  However, for these people we had an immense wealth of data accumulated more or less continuously over many years, including records of those who had moved or died.[4]  These were real people, not numbers.  They were typical of people everywhere with greatest and most complex needs, least resources for care, and highest event-rates.  For experts commanding hundreds or thousands of attenders at hospital clinics linked in multicentre trials, and for all who prosper from the world as it is, evidence from our small population seemed insignificant – too small, and too complicated, to be useful within their reductionist assumptions.  But for jobbing clinicians just trying to bring the world we actually have a bit closer to the world which rational application of knowledge would make possible, its significance was obvious and exciting.  When experts suggest that experience in primary care can never be large enough to get beyond anecdote, we should reply by organising our collected anecdotes into communal stories.  These epics could be far more compelling than anything we can now read in our (frankly, rather boring) medical journals.

As time went on, we became increasingly aware of another dimension, equally important.  Single health problems are rare, and most of those presenting in primary care involve more than one medical or social category.  Health care systems based on guidelines for management of single problems cannot be made to work efficiently by simply adding together these guidelines for single problems, when most patients in fact have multiple problems.  Guidelines become vertical tubes designed by and for specialists, without generalists to recognise and maintain the horizontal, interdependent connections which still exist in the real world.[5]  As Barbara Starfield has pointed out, this is a huge field for primary care research, which we have hardly even begun to explore.[6]  The economic consequences of such research will mostly be bad news for the shareholders of pharmaceutical companies, which thrive from multiplying mandatory indications for continuous medication.  The fragmentation of care and demoralisation of medical and nursing staff compelled to perform standardised tasks in isolation, rather than think creatively, imposed by a commercial and industrial model on all health care processes since the 1980s, has suppressed curiosity, blunted social responsibility, promoted greed, and corrupted data.  More data is now recorded than ever before, but it is hugely exposed to bias and even to outright falsification.[7]  Even when we return to the NHS gift economy whose development was interrupted by the commercialising offensive in the 1980s, we shall still need fiercely to defend the independence of clinical judgement and data collection.  Without respect for empirical data and professionalism, we open the door to the vulgarised pseudoscience of Lysenko and the anti-intellectual persecution of Mao and Pol Pot. Renewed exploration of the real world will require struggle and political understanding, of a kind fundamentally different from those of the 20^th century.

When Mary and I retired in 1992, our MRC Glyncorrwg Research Unit died.  As in nearly all deaths, there was no single cause, but the dominant factor was a lack of political or philosophical leadership in the new era after 1985, when 20^th century concepts of socialism died also.  Motivation and leadership are central: not only their motive power, but also their social direction.  However, you are a new generation.  You will recover from the disillusion and demoralisation of your parents.  Where there’s death, there’s hope.

ACKNOWLEDGEMENTS

I am most grateful to Prof. Pasquale Falasca and Dr Franco Delzotti for encouragement and support, and to Lyvia Lo Presti for her excellent translation. 

REFERENCES