Your editorial last week commenting on the Cochrane review of randomised trials of treatment for 'mild' hypertension[1] prompts two important questions: Why has it taken more than 30 years to reach this conclusion, when it was already evident from any careful and critical reading of the trials claimed originally to justify interventions in the diastolic range 90-100 mm Hg? And what should we do now to get practice onto a more rational footing?

Whole community control of high blood pressure is both possible and effective. In 1974 I described how we organised this in Glyncorrwg, measuring pressure proactively three times in 100% of men and 98% of women, reviewing defaulters at the end of every blood pressure clinic, and allocating staff responsibility for proactive follow-up.[2] We found surprisingly high prevalence of dangerously high blood pressure, much of it in men under 40.[3] This must have contributed to the 28% difference in mortality under 65 over five years between the Glyncorrwg population and a neighbouring control population.[4] Throughout this work we used the threshold for treatment validated by the three United States Veterans Administration trials. These were the best evidence available in 1968, when we began proactive case-finding.

In 1983 the World Health Organisation held the third of a series of international symposia on mild hypertension in Switzerland. Prior to the meeting, a letter was circulated from its chairman, Prof. Austin Doyle, main author of the Australian National Blood Pressure Study (ANBPS), which together with the United States' Hypertension Detection and Follow-up Program (HDFP), provided such evidence as there then was, from which to define an evidence-based intervention point. This letter invited those invited to attend the symposium to endorse its conclusion, namely that medication should start at 90 mm Hg. In my reply I refused to do this, as we had been told that the symposium’s purpose was to discuss the then available evidence, and thereafter try to agree a conclusion.

Thoma Strasser, the WHO organiser of the symposium, appealed to me to do what all other participants had done, and sign the statement, reminding me that three transnational pharmaceutical companies were sponsoring all three symposia and had a right to expect results. I still refused. A committee meeting was called to discuss the statement, chaired by Doyle, to which I was invited, surrounded by most of the international great names in hypertensology. He asked me why I wouldn't sign. I replied that I had not yet seen convincing evidence that medication was justified from a diastolic level of 90 mm Hg, without other major indications for intervention; and that the workload consequent on taking blood pressure control seriously for the whole population, not just for customers in medical trade, would present huge logistic problems which should be considered. We knew that within the next few months we would have better evidence from the UK Medical Research Council trial of mild hypertension. Probably we should await that before reaching a conclusion. To which Doyle replied: "Fuck the MRC trial. Do we always have to wait for the fucking British?" Bill Miall, who led that trial, was sitting next to me. Like everyone else at that meeting, he had nothing to add, and advised me to sign the document like everyone else. Which I then did. I thought I had reached the limit of what a mere GP could do without becoming hopelessly isolated. In a letter to the Lancet, a few weeks later Bill withdrew his signature, on the grounds that it might compromise the then still unpublished conclusions of the MRC trial. Even when those came, I was never convinced that the very small reductions in cardiovascular and cerebrovascular event rates justified the conclusion, except in diabetics.

At that same symposium, W McFate Smith, who had led the inconclusive Multiple Risk Factor Intervention Trial (MRFIT) in USA, revealed the pressure to find positive results experienced by organisers of all such large scale trials. MRFIT cost the US government $115m.[5] When President Nixon had agreed that then very large sum, he had emphasised forcefully that in return he expected big positive results. There were none, not because it was badly organised, but because practice in USA already generally preceded evidence, so that no valid and stable untreated control group was to be found. Smith concluded that, realistically, medical scientists might have to accept that some degree of "quasi-science" was inevitable and therefore justifiable, for obtaining necessary state funding. Conclusions must sometimes be bent to our market. This won him sardonic remarks from Ian Robertson, when he closed the symposium.[6]

If we really want a free, comprehensive NHS as a public service, the scope of its work cannot include a market for medication accepting all the plausible mythologies lobbyists can maintain among the lay public, media editors and politicians. I hope schoolchildren learn that sucess in experiments is measured not just by showing that new methods are useful, but equally by showing that they are no better than old methods, or perhaps don't work at all. All three outcomes represent worthwhile gains in knowledge. Why can’t adults learn the same? Yet it is still much harder to get studies with so-called negative results published, than those with positive results, however small their advances may seem, even at their birth, before experience has time to reveal their limitations. In all three editions of my book on community control of high blood pressure, I advised primary care units not to start antihypertensive medication without repeated careful measurements over three months or more, confirming diastolic pressures at or over 100 mm Hg, unless there was already evidence of organ damage, diabetes, or strong family histories of cerebrovascular disease.[7] All that did was to isolate my book from its market. We all like to hear good news.

Julian Tudor Hart

[1] Cochrane Database Systematic Review 2012;8:CD006742, CD007653, CD008893.
[2] Hart JT. The marriage of primary care and epidemiology: continuous anticipatory care of whole populations in a state medical service. (Milroy lecture) Journal of the Royal College of Physicians of London 1974;8:299-314.
[3]Hart JT, Edwards C, Haines AP, Hart M, Jones J, Jones M, Watt GCM. Screen detected high blood pressure under 40: a general practice population followed up for 21 years. BMJ 1993;306:437-40.
[4] Hart JT, Thomas C, Gibbons B, Edwards C, Hart M, Jones J, Jones M, Walton P. Twenty five years of audited screening in a socially deprived community. British Medical Journal 1991;302:1509-13.
[5] Multiple Research Group. Multiple Risk Factor Intervention Trial: risk factor changes and mortality results. JAMA 1982;248:1465-77.
[6] Robertson JIS. Concluding remarks: cum grano salis. In Gross F, Strasser T (eds) Mild hypertension: recent advances. New York: Raven Press, 1983, 413-8.).
[7] Thresholds for follow-up and medication. In, Hart JT. Hypertension: Community Control of High Blood Pressure. 3e. Oxford: Radcliffe Medical Press, 1993:76-8.

Competing interests: None declared

I thank Ms Lenzer for expanding on the issue of the authors’ conclusion and feel that it raises interesting questions about the interpretation of results when one author is recommending dramatic changes to current practice and another is being rather more prosaic.

There is one point that I feel I didn’t make strongly enough in my initial response and so would like to expand further. In my view, the Cochrane review adds nothing to the current evidence base on how to treat Stage 1 hypertension. The question it asked can be summarised as ‘if patients with Stage 1 hypertension are inappropriately prescribed outdated medication regimes do they demonstrate a reduced number of cardiovascular events?’ In fact, due to the inclusion of some patients with high cardiovascular risk and the low number of events recorded, even this question is not satisfactorily answered.

Competing interests: None declared

I would like to thank Drs Floyd, Clark and Ferro for their comments, however, I must clarify something that may not have been evident in the article; it was not the "authors’ conclusion" that "the analysis should lead to dramatic changes in the way doctors treat mild hypertension," as Dr Floyd states, rather it was ONE author's conclusion - that of Dr Cundiff. Indeed, as a group, they simply called for further study.

While I applaud Dr Wright, as a senior author on the paper, for not insisting on a single unified interpretation of what to do about the data, I did question him about Dr Cundiff's interpretation and all Dr Wright would say is that the data suggest the need for more study.

I think this difference of opinion about what to do in the face of uncertainty raises a fundamental question about what one promotes in the absence of positive evidence of net benefit: Does one say, "Yes, let's do it anyway even though we lack firm evidence of net benefit," or do we say, "Wait up. Let's see what's proven that we can do (e.g., Mediterranean diet, exercise, weight loss, quit smoking) until something else is proven,"?

If you are to say "Take the drug because we believe it works - even in the face of suggestive evidence that it might not work" then we might as well promote homeopathy and remote prayer in my mind. We are beginning to understand that disease creep, the expansion of treatment groups to include low- and even very low-risk individuals - is too often far better for pharmaceutical sales than for the public health.

I thank Drs Wright, Cundiff, Floyd, Clark and Ferro for bringing this question to the fore; a question I’d really like to see the medical community address: In the face of uncertainty, on whom does the burden of proof rest?

Competing interests: author of BMJ news article