WHY HAVE TRIALS OF PREVENTIVE MEASURES AGAINST PREMATURE CARDIOVASCULAR DEATH APPEARED TO BE SUBSTANTIALLY EFFECTIVE ONLY IN SMALL “NAÏVE” POPULATIONS?

For the past three decades at least, we have known apparently causal and modifiable risk factors predictive for about half of all premature deaths from stroke and coronary heart disease.  Though premature mortality from these has fallen substantially over the same period, large and costly whole community studies have failed to demonstrate more than marginal outcome benefits from proactive campaigns promoting preventive interventions, compared with control communities left to ordinary care (including no care at all).  Some barely positive conclusions have seemed to depend more on faith than scientific reasoning.  Funding on the scale required by these studies has to come from very large government, charitable or commercial sponsors.  These seldom seem to understand that negative answers may reduce our ignorance no less effectively than positive ones.   Unless they get at least some indication of what they think they paid for, the next project may never be funded.  This may be hard for researchers to ignore.

On the other hand, some small studies on so-called naïve populations – peasant, industrial and post-industrial working class communities with little previous access to high quality care, and high cardiovascular event rates - have shown apparently substantial positive effects.  These seem usually to be dismissed as too small to prove anything much, with their positive results attributed to a combination of chance and publication bias.  That seems to have been the fate of our project in a Welsh coal and steel community, where we combined clinical care with public health aims and epidemiological research from 1961 to 1987. [1]  

It took us 8 years to bring reactive care to a standard where planned proactive care, and data collection to measure progress, became feasible.  Starting in 1968, to reactive clinical care we added systematic proactive search for reversible risks throughout our entire community, with sustained proactive follow-up for all cases.  Compared with a nearby control community with virtually identical social indices and ordinary reactive care of better than average quality, we showed a 28% lower all-causes mortality under 64 over the last 5 years of our cumulative 20-year programme.  Control of smoking and blood pressure formed its core, using Veteran’s Administration criteria for treatment.[2] The largest differences in premature mortality were for cardiovascular, respiratory and stroke deaths.  Despite strong editorial endorsement from the Lancet, British Medical Journal and British Journal of General Practice when it appeared, and visible evidence that some of its lessons influenced government policy in England and particularly in Scotland, this study has rarely been cited since, either by clinicians, epidemiologists, or health policy specialists.

Did our results depend only on chance?  In our target village there were 21 deaths under 64, compared with 22 expected at UK rates.  In the control village, there were 48, compared with 30 expected.  Certainly there was scope for wide chance variation at this level of numbers.  I have never believed that this difference was likely to be sustained over another 5 years.  But would it have fallen below conventional levels of significance?  I don’t think so.  We have no equivalent figures for the control population, but smoking in hypertensives aged 20-65 at entry fell prodigiously, from 56% to 20% by 1989 (verified by urine cotinine), and their group mean blood pressures fell from 186/110mmHg at entry to 146/84mmHg, using random-zero sphygmomanometers throughout.  For our diabetics, also identified proactively, smokers fell from 44% to 12%, and group mean blood pressures from 171/93mmHg to 155/81mmHg.  For the adult population as a whole, impact was much less but still substantial, at least for male smokers, who fell from 61% to 42% (women showed no significant net change: older ones smoked less, but younger ones smoked more). 

I think we may plausibly suppose that our programme of proactive, community-wide anticipatory care for several common chronic conditions, built around premature arterial senescence, chronic obstructive pulmonary disease and diabetes, actually was effective, with a sustainable difference around 10 to 15% between this and good but solely reactive care.  That would be a much larger health gain than anything claimed for any of the giant mass interventions.   In which case, what were the important differences, other than scale, between our study and the large community studies which virtually monopolise the literature?

I was trained as an epidemiologist, and my wife Mary was trained as a team organiser, by Archie Cochrane and Bill Miall in the MRC epidemiology unit at Cardiff.  They gave us obsessional respect for accurate measurements, high response rates,[3]and sceptical interpretation of results.  Cochrane also taught us how to approach industrial working class communities as sustained participants in producing new knowledge.  Because I already had 5 years experience as a GP in London, already excited by George Pickering’s paradigm shift in understanding of the nature of arterial pressure as a continuously distributed variable, when I returned to primary care I could combine public health aims, epidemiological methodology, and clinical self-respect with acquired appreciation of the delicate and often conflicting relation between population wants and needs, from which to develop a project sustained over a working lifetime, rather than the one, two or at most five years of the big studies.  I suspect that to apply such knowledge as we have (always much less complete than either professionals or populations imagine) effectively to the real world of human behaviour, it usually takes at least five years even to establish the mutual respect and trust on which high and sustained responses depend.  We lived inside the community we studied, our children went to their schools, and limits to what we did were set not by socially remote ethical committees, but our own assessments of how far our entire community had progressed so far, and how fast it was willing to move in the future.  We could measure that from response rates.  We watched the funerals, and saw enough astonishingly high-risk survivors, to remember both the seriousness of our work, and the still huge extent of our ignorance.  Above all, we had an integrated view of our work, and our patients’ continuing life stories were the centre of interest for our entire team.  Rapid industrialisation of primary health care, dividing it into numerous tasks by a divided, often unstable workforce with narrow skills, confining judgement to remotely devised guidelines derived from studies of behaviourally if not biologically different populations, has little in common with the work we did.  It probably does ensure that tasks are done (boxes are ticked), but whether they will be integrated to yield higher productivity of health gain seems extremely doubtful, and they take all the fun out of work for everyone.

Effective health care should be based on evidence from professional literature, but also from the people we serve, about how they actually live their lives, what they think about themselves, and what they think about us.[4] When we all understand that, I doubt if anyone will describe any peasant, industrial or post-industrial population as “naïve”.  That term might apply better to those who use it.

I am grateful to Prof. Lennart Råstam at Lund University for informing me of this wonderful website, and to Henry Blackburn and Geoff Rose for writing the book on which all our early work was based.  A fully referenced version of this text is available at www.juliantudorhart.org.