Tag Archives: Blood pressure

A Very Clever Study Using Instrumental Variable Analysis

In observational studies, adding a third or fourth anti-hypertensive medication leads to rapidly diminishing marginal improvements in blood pressure and an increase in side effects. However, this finding could originate from a type of selection bias called bias by indication; the harder cases get more treatments. To get around this problem an Instrumental Variable (IV) analysis was carried out on a dataset generated by a randomised trial of intensive versus standard blood pressure target.[1]

In the IV analysis, where randomised group was the instrument, the incremental effect of adding a further anti-hypertensive was maintained across all orders of increment. Not only was there an improvement in blood pressure, but also a reduction in cardiovascular events. There was only a small increase in adverse events. This proves that there is a bias by indication in observational studies, which reduces the apparent marginal benefit and increases marginal harms. This is a neat study and a fine example of IV analysis.

— Richard Lilford, CLAHRC WM Director


  1. Markovitz AA, Mack JA, Nallamothu BK, Ayanian JZ, Ryan AM. Incremental effects of antihypertensive drugs: instrumental variable analysis. BMJ. 2017; 359: j5542.

Raising Blood Pressure in Sepsis Patients

I never cease to be amazed at the number of treatments that were received wisdom, but which have been shown to be harmful – sometimes thoroughly harmful.

I well remember my professor of surgery extolling the virtues of completely restoring blood pressure in patients who were bleeding heavily. It turns out that this sensible sounding treatment is plain wrong. One should raise the blood pressure sufficiently to keep the patient awake and the kidneys perfused, but no more. Likewise, I was always taught that in cases of septic shock, fluid replacement should be sufficient to restore blood volume. The latter idea was critically questioned after a randomised trial of a bolus of fluid for critically ill children [1] (which we featured in the quiz in our last News Blog). Here, the fluid bolus was associated with a striking increase in the risk of death.

Now a somewhat similar trial has been carried out among critically-ill adults.[2] The study was carried out in Zambia among patients with septicaemia. Over 200 patients were randomised to receive fluids (and sometimes drugs) to restore blood volume and raise the blood pressure versus less intensive therapy. The results of this trial among adults with sepsis are striking; there was a considerable increase in death rates among those in the intervention group. The difference was considerable at 15 percentage points. Patients in the intervention group received a mean of 3.5 litres of intravenous fluid compared with only 2 litres among controls. Further, 14% received a medicine to support blood pressure in the intervention group compared to only 2% in the control group.

Not surprisingly most of the patients in the study were HIV positive, but there is little reason to think that these results cannot be generalised more widely. A picture is starting to emerge in the literature in favour of not trying to completely restore blood volume in critically-ill patients, at least in African settings. There is a single RCT in North America that produced contradictory findings.[3] It is hard to explain why treatment should produce such different findings across African and North American settings.

— Richard Lilford, CLAHRC WM Director


  1. Maitlan K, Kiguli S, Opoka RO, et al. Mortality after Fluid Bolus in African Children with Severe Infection. N Engl J Med. 2011; 364: 2483-95.
  2. Andrews B, Semler MW, Muchemwa L, et al. Effect of an Early Resuscitation Protocol on In-hospital Mortality Among Adults With Sepsis and Hypotension. A Randomized Clinical Trial. JAMA. 2017; 318(13):1233-40.
  3. Rivers E, Nguyen  B, Havstad  S,  et al.  Early goal-directed therapy in the treatment of severe sepsis and septic shockN Engl J Med. 2001; 345(19): 1368-77.

More Bad News for Unfermented Milk

A recent News Blog conveyed evidence that milk consumption is associated with osteoporosis in a dose responsive manner.[1] Contrary to that, observational studies have shown that milk consumption is associated with slight reduction in blood pressure. The osteoporosis data is consistent and the effect is large, while the hypertension evidence is inconsistent, and the magnitude of effect is rather small. Anyway, a recent large re-analysis of observational evidence from cohort studies,[2] using an instrumental variable, finds no improvement in blood pressure in association with increasing dairy product consumption. The instrumental variable is genotypes associated with lactase deficiency. Not surprisingly, people who cannot digest lactose consume less dairy products. Randomised trials also show no association between dairy intake and blood pressure, although follow-up is limited in these studies. Unsurprisingly, the quoted study is not able to distinguish between fermented and unfermented milk, and hence does not shed light on the galactose hypothesis. This hypothesis says that galactose is somewhat toxic in adults who would be better off sticking to fermented milk products where galactose has been converted to lactic acid.

— Richard Lilford, CLAHRC WM Director


  1. Lilford RJ. Two Provocative Papers on Diet and Health. NIHR CLAHRC West Midlands News Blog. 12 December 2014.
  2. Ding M, Huang T, Bergholdt HKM, et al. Dairy consumption, systolic blood pressure, and risk of hypertension: Mendelian randomization study. BMJ. 2017; 356: j1000.

Recent History of Anti-Hypertensive Treatment Trials

An interesting essay by Pfeffer and McMurrary traces the recent history of trials of anti-hypertensive therapy.[1] The authors chronicle the scientific evidence that has accumulated to show that treating hypertension lowers the risk of stroke, heart disease and death; that systolic blood pressure is very important; that outcomes are improved, even if the starting blood pressure is only slightly raised (particularly in diabetes); but that it is also possible to try too hard to control blood pressure. No mention here of salt, which is important in the genesis and maintenance of hypertension. However, this is a readable account of how the scientific and ‘social’ discovery of how to treat the ‘silent killer’ that is hypertension progressed hand in hand.

— Richard Lilford, CLAHRC WM Director


  1. Pfeffer MA, & McMurrary JJV. Lessons in Uncertainty and Humility – Clinical Trials Involving Hypertension. New Engl J Med. 2016; 375: 1756-66.

Effects of Salt in Diet

Well there you have it, salt really is bad for you – based on a study of 107 randomised comparisons on the effects of reduced sodium intake on blood pressure, 99 cohorts on the effects of blood pressure on cardiovascular mortality, and a survey of the salt intake of three-quarters of all adults in the world.[1] It is hard to argue with the results of a study like that. Modelling based on the data from the above study, funded by the Bill and Melinda Gates Foundation, concludes that one in every ten deaths from cardiovascular disease can be attributed to a high salt intake. Incidentally, having 107 randomised comparisons makes it possible to examine the correlation between the reduction of salt intake in those trials, and the reduction in blood pressure. It turns out to be linear.

— Richard Lilford, CLAHRC WM Director


  1. Mozaffarian D, Fahimi S, Singh GM, Micha E, Khatibzadeh S, Engell RE, Lim S, Danaei G, Ezzati M, Powles J. Global Sodium Consumption and Death from Cardiovascular Causes. NEJM. 2014; 371: 624-34.

Do the Right Thing?

This is my first venture into the world of blogging, so don’t be surprised if the tone of this entry differs to those of more polished writers. Given a platform to sound off I feel a bit spoiled for choice. But here goes.

In Philip K Dick’s 1968 dystopian novel “Do Androids Dream of Electric Sheep?” the protagonist, wrestling with his conscience, is told “Go and do your task, even though you know it’s wrong.” As an academic in public health I find myself in a dilemma. I have an interest in prevention of cardiovascular disease. My reading of the research evidence impels me to some quite clear conclusions about how to accomplish this. But I also have to teach. And I have to teach an orthodoxy that I don’t believe. I invite my readers’ thoughts on this dilemma.

The orthodoxy is that there is a condition called hypertension. We know there must be such a condition because there are societies and guidelines which embed this condition in very real societal structure.[1] [2] We have hypertension clinics and hypertension specialists. We have drugs licensed for hypertension. This condition must be diagnosed, which means categorising individuals as either hypertensive or normotensive. The most recent guidelines emphasise the need for technological gee-whizzery in the shape of 24 hour ambulatory blood pressure monitoring to assist this process. The condition must be treated: essentially with drugs. The drugs to be used are individually tailored to patients according to their age, sex and sometimes race. The patients are monitored and the drugs further tailored in the light of their response to treatment. The University of Birmingham helpfully runs a week long course for GPs to help them follow this orthodoxy. Case vignettes provided examples for clinical practice. These vignettes helpfully included discussion of the best treatment for a 22 year old woman with systolic blood pressure of 160 mm Hg. Multiple choice questions assessed the participants’ knowledge. I contribute to the teaching. Herein lies the dilemma.

Unfortunately most of the orthodoxy is nonsense. There is not really a condition called hypertension. Every living human being inhabits a continuum of risk of cardiovascular disease and blood pressure is just one of the factors influencing the continuum of risk.[3] Blood pressure alone does not offer any special insight into the chances of developing cardiovascular disease. The categorisation of individuals by risk would be arbitrary, but at least it has a clear relationship to something that matters: their chances of getting cardiovascular disease. The categorisation of individuals by their blood pressure is not helpful at all. We can trace this error back to the misty origins of medicine itself. Doctors categorise: sick and well. The categories are reified into diagnoses and shackled to treatments through the concept of “indications”. Specialists prosper. Indications become regulatory frameworks for medications and treatments. Pharmaceutical empires rise.

This matters. Nobody should care about their blood pressure. They should care about their chances of getting heart disease. Nobody should be given drugs to lower their blood pressure. They should be given drugs to reduce their chances of getting heart disease. It makes a big difference.

So what do we really know about blood pressure? We know that higher blood pressure is associated with a greater chance of getting cardiovascular disease. We know that drug treatments reduce that chance. But we also know blood pressure alone is not a great indicator of risk. Age is the strongest indicator of risk. If her blood pressure is really 160 mm Hg our 22 year old female has a probability of getting cardiovascular disease in the next 10 years of less than 0.2%. If treatment reduces her risk by a quarter we will treat 2000 such women for 10 years for one to benefit.[4]

Blood pressure is intrinsically variable. In essence this means that it is not measurable with any great degree of precision. So any categorisation of blood pressure also must include a measurement of mis-categorisation.[5] When high blood pressure is uncommon, the most likely explanation for a measured high blood pressure is chance variation. So anyone aged under 35 whose measured systolic blood pressure is >160 mm Hg after an average of three readings at separate clinic visits is much more likely to have had three chance high readings than to have truly raised blood pressure. The problem is slightly reduced, but not at all resolved by using ambulatory blood pressure monitoring.[6] First do no harm. There is no need to measure blood pressure in otherwise healthy young adults. It does little except put them at risk of misclassification. So it seems highly unlikely that our 22 year old female even has high blood pressure at all.

Put bluntly, our 22 year old probably does not have high blood pressure; any benefits from treatment are likely to be trivial; she would likely not have chosen to take treatment if anyone had told her this. Needless to say this was not the right answer helpfully provided by the hypertension specialists.

Blood pressure is hard to measure because it is so variable, but it is almost impossible to monitor reliably. By measuring blood pressure at two, three four or five consultations before treatment and again after treatment in the same individual, it is not possible to reliably know the effects of that treatment or even if the patient is taking their treatment.[5] [7] This is important. It means that every time blood pressure is measured and treatment changed as a result, chance has played at least as great a role in the change in treatment as any other factor. Yet the fact that we can’t monitor treatment accurately does not even matter much, because the effects of drug treatments vary little from one individual to another. The differences between different treatments at standard doses are trivial.[8] When it has been investigated it appears that almost all of the effects of treatment are attributable to drug dose and pre-treatment blood pressure and almost none to individual patient variation.[9] So the effects of treatment may be unmeasurable, but paradoxically they are largely predictable. As for tailoring treatments by age and sex and race, this hardly matters. The differences in response to treatment are too small to be detectable through measurement. Since almost everyone can benefit from two or three drugs it is irrelevant which drug class we consider “first line” since everyone will get drugs from two or three classes.

The solution is simple. Remember our purpose: to prevent cardiovascular disease, not to normalise blood pressure. We can identify who is most likely to be at risk using the most reliably recorded information we have: age and sex. These are untroubled by measurement error. We can refine that risk estimate with additional information: smoking and diabetic status. So far all of this is already recorded in the patient’s records. We can refine it further with more unreliable measurements like blood pressure and cholesterol levels, but often this is not necessary. Those unlikely to be at high risk (otherwise healthy young adults) need no blood pressure measurements, as no level of blood pressure will be worth treating. Those likely to be at high risk will benefit from treatment. We should prioritise them for action. We just need to reassure ourselves that their blood pressure is not so low that treatment will render them symptomatic. Precision is not necessary. Only those in the intermediate risk group might benefit from a more accurate estimate of their risk. Finally we can forget about all that monitoring of treatment effects. The drugs will work if patients take them. The measured blood pressure will vary during treatment. This makes for a rather different therapeutic relationship with the doctor.

And back to my dilemma. Do I continue to teach something I don’t believe? In the novel, the protagonist tries to evade the problem only to be told “You will be required to do wrong no matter where you go. It is the basic condition of life.” So what do I do? Should I go on teaching what I don’t believe or should I rock the boat?

–Tom Marshall, Deputy Director CLAHRC WM, Prevention and Detection of Diseases

[1] British Hypertension Society. British Hypertension Society. 2012.
[2] National Institute for Health and Care Excellence. NICE Guidance – Hypertension (CG127). 2014.
[3] Lewington S, Clarke R, Qizilbash N, Peto R, Collins R, Prospective Studies Collaboration. Age-specific relevance of usual blood pressure to vascular mortality: a meta-analysis of individual data for one million adults in 61 prospective studies. Lancet. 2002; 360: 1903-13.
[4] ClinRisk Ltd. QRISK®2-2013 Web Calculator. 2013.
[5] Marshall T. When measurements are misleading: modelling the effects of blood pressure misclassification in the English population. BMJ. 2004; 328: 933.
[6] Marshall T. Measuring blood pressure: the importance of understanding variation. Rev Bras Hipertens [Braz J Hypertens]. 2005; 12(2): 75-82.
[7] Hayen A, Bell K, Glasziou P, Neal B, Irwig L. Monitoring adherence to medication by measuring change in blood pressure. Hypertension. 2010; 56(4): 612-6.
[8] Law MR, Morris JK, Wald NJ. Use of blood pressure lowering drugs in the prevention of cardiovascular disease: meta-analysis of 147 randomised trials in the context of expectations from prospective epidemiological studies. BMJ. 2009; 338: b1665.
[9] Bell KJ, Hayen A, Macaskill P, et al. Monitoring initial response to Angiotensin-converting enzyme inhibitor-based regimens: an individual patient data meta-analysis from randomized, placebo-controlled trials. Hypertension. 2010; 56(3): 533-9.