Great Myths of Blood Pressure Effect Size in Renal Denervation

Fig. 21.1

Top: The normal distribution of blood pressures from repeated blood pressure measurements in a hypothetical patient presenting to the clinic with a ‘true’ mean blood pressure of 150 mmHg, and a standard deviation of 12 mmHg. Bottom: A graph showing a normal distribution of the changes in clinic-to-clinic blood pressures in the same patient. These changes do not reflect improvement in their hypertension; rather the nature of repeated measurements in a noisy variable. The measurements to the right of the red line represent the 28 % of repeat readings which may be wrongly interpreted as a rise in blood pressure of over 10 mmHg

When given the above question, 86 % of audience respondents replied with D. This re-measurement could well have resulted in a lower blood pressure which would no doubt be documented. This appears perfectly reasonable; indeed, isn’t the second value just as valid as the first?

However, what would these 86 % of attendees have done had the patient not been started on a drug a week earlier? Almost all audience members agreed that the 162/88 could simply be accepted and documented.

What is concerning is the consequences of such behaviour in the context of unblinded and/or uncontrolled trials, where a physician’s expectations of a treatment can influence their measurement strategy.

“Everyone Knows that Office BP Falls More than Ambulatory BP in Drug Trials”

True, but a hilariously self–destructing statement

Office blood does indeed fall more than ambulatory blood pressures in drug trials [7], but only when the office staff are unblinded to the study arm. When controlled and blinded, office and ambulatory pressure fall in direct parallel [1]. The amusing aspect of this myth is that it contains the seeds of its own destruction, by highlighting how powerful bias can be in unblinded blood pressure trials.

It is for this reason that since the turn of the century there has been pressure to abandon bias-susceptible office blood pressure measurement and rely instead on ambulatory monitoring [8], or alternatively to carry out strict blinding.

A good antidote to this myth is to restate it, but rename office blood pressure to “blood pressure documented by a person who knows which arm the patient is in”, and ambulatory blood pressure to “blood pressure automatically documented by a machine which does not know which arm the patient is in”. With such wording the myth is difficult to sustain.

“Office Effects Are Larger than Ambulatory Effects Because the Sympathetics Contribute More to Blood Pressure in the Daytime than in the Night”

Non–sequitur.

While intellectually persuasive in isolation, the reasoning was shown to be fallacious by the elegant work of Felix Mahfoud’s group. Perhaps sympathetics do contribute more to blood pressure in the daytime than in the night, but careful automatic monitoring shows that unblinded renal denervation gives similar reductions in daytime pressure (10.2 mmHg) and night-time pressure (11.1 mmHg) [9].

Proponents of the sympathetic explanation for the human-versus-automatic discrepancy, having lost the battle for the defence of the city of daytime, may well retreat to make a last stand in the inner citadel of the office. However, the history of sieges tells us what happens next.

“RD Is One of the Most Exciting Developments in Medicine in the Last 10 Years”

True, but not only for its contribution to BP control.

Future medical historians may look back on this time as unique. Never before have large proportions of specialists have been so comfortable writing, reading and discussing manually-documented values that overestimated a biological variable to such a large extent in the face of contradiction by automatically-documented data.

A threefold overstatement, in other spheres of medicine, would be serum sodium levels consistently being reported as ~420 mmol/l, body temperatures as ~300°F, or heights as ~5 m.

Threefold overestimation of effect sizes, e.g. 30 rather than ~10 mmHg appears to be unique.

“ABPM Is Unnatural and Weird Because There is a Mechanical Pump Inside, Which Makes a Whizzing Noise”

True but irrelevant.

Any medical assessment, and certainly any measurement of blood pressure, is unnatural. Awareness that it is happening can affect physiology.

Machines lack the natural tendency of physicians to portray an optimistic view of the effect of an intervention they have carried out.

Moreover, machines generally do not have preference for particular digits such as 5 and 0. Humans do, and may be unable to resist rounding up pressures pre-intervention and rounding down pressures after intervention.

It is very rare for automatic equipment to exhibit digit preference. In research where this occurs [10], the effect size has been found to be greatly enhanced to the level seen in manual office measurements. One explanation for this might be mislabelling of office measurements as ambulatory by accident borne of excitement [1].

“I Don’t Care What the Number Is, I Want to Treat Patients [11]”

Up to you.

There may be a group of physicians who do not care whether an effect is 30, 10, 1 or 0.1 mmHg. Should we not all strive to avoid being members of this group?

If the effect size was 1 mmHg (or 0.1 mmHg) nobody would read this book, or even write it.

In medicine almost everything affects everything else. We personally think the only question worth asking is: how much?

“10 mmHg Is Not Enough, It Needs to Be More”

False.

Biological values have no obligation to be any different from what they actually are, in the same way that π does not need to be larger than its current value.

Since politicians in the past have attempted to pass laws to change the ratio of the circumference of a circle to its diameter [12], we should not be surprised that they have medical counterparts attempting to change biological realities. We should ignore both categories of modern-day Canute.

A genuine 10 mmHg drop in BP equates to a 41 % reduction in stroke and a 22 % reduction in coronary heart disease events [13] which would be very useful to have.

“The Only Way to Quantify the Benefit of Renal Denervation on Stroke and Heart Attack Rates Is by Doing Trials with Those as Endpoints”

False.

It is well established in antihypertensive drug trials therapy that successfully and safely lowering blood pressure delivers predictable in cardiovascular events. So strong is the data for this that antihypertensive drugs are accepted on the basis of demonstration of safe reduction in blood pressure in randomised blinded trials. There is no requirement to conduct an endpoint trial.

Only gold members can continue reading. Log In or Register to continue