This made the news when the results came out a few months ago. BBC, CNN - it was everywhere. One simple trick to decrease the risk of dying. Just take your blood pressure pills at night.
Could it be true?
A longer exposition of the HYGIA trial (Hermida et al, Eur Heart Jour 2020) and the chronotherapy literature follows, but for the TL:DR; version:
The HYGIA trial was run out of one center, and recruited over 19,000 patients, executed a behavioral change intervention with perfect adherence, and no loss to follow up
No change in nocturnal BP by changing BP medication timing has been reported in other trials, and this intervention does not make pharmacological sense
There was a reduction in cardiovascular death, all cause death *and* non-cardiovascular deaths (typically cancer) from taking BP medications at night instead of morning
The magnitude of effect of taking BP meds at night versus in daytime was greater than the magnitude of taking BP meds versus placebo
Despite lowering BP, the intervention arm did not report any side effects of nocturnal hypotension
Now who does not like an intervention that produces benefits with no downside? Sign me up! But could this all be fool’s gold? Let’s delve deeper.
Circadian rhythms and Blood Pressure
Circadian rhythms are very important - as any frequent flier or new parent will attest. But there are important biological effects beyond sleep - and many cellular activities follow a circadian rhythm. The 2017 Nobel Prize in Physiology and Medicine was awarded to the three scientists who increased our understanding of this topic. In hypertension, blood pressure normally ‘dips’ at night - it is lower when one sleeps. The absence of nocturnal dipping occurs more commonly in individuals with diabetes, obesity, chronic kidney disease (CKD) and other comorbid conditions. Lack of this dipping phenomenon may be due to sleep apnea, increased sympathetic activity, sodium retention, among other factors. More concerning is that it represents a greater hypertension load - and is associated with all sorts of badness: left ventricular hypertrophy, faster progression of kidney disease, higher rates of stroke, and higher rates of cardiovascular death. Some of these might be due to underlying factors (eg CKD, sleep apnea, sodium retention) and some might be from the lack of nocturnal dipping and higher blood pressure load. Hence hypertension nerds love performing ambulatory blood pressure monitoring (ABPMs) where an individual wears a BP cuff attached to a monitor for 24 hours usually and we get detailed and granular recording of their BP over the 24 hour period. Check out the NephMadness coverage or a recent JAMA review (Melville & Byrd, JAMA 2018) for more on this topic.
So lack of nocturnal dipping is bad, and we don’t have a good way to fix it. Yes, lowering daytime BP also lowers nighttime BP but it doesn’t normalize the rhythm. Fixing sleep apnea, improving volume status and reducing sodium retention sometimes helps. But it's a complicated pathophysiology and a complicated mechanism. Could a simple pharmacological trick of timing medications fix this problem? John Hogenesch and his group have been studying the timing of many medications, and they do find that timing is important (Ruben et al, Science 2019). But among their findings is also this: they report a greater effect of timing when using medications with a shorter half life. Not so much with drugs that have a long half life. In hypertension, that means this matters when you use hydralazine or captopril. But no one should use hydralazine or captopril anymore. And we mostly don’t.
Figure from Rubens et al, Science 2019
Pharmacology of BP Lowering Medications
Yes, the drugs that were initially discovered mostly had a short half life, and had to be dosed a few times a day. But that’s from a few decades ago. What do we commonly use now? Amlodipine (half life, t1/2 30 to 50 hours) is the most common calcium channel blocker. For renin-angiotensin system inhibitors we have lisinopril, ramipril, perindopril, trandolapril, telmisartan, candesartan to name a few - all with half lives ranging from a low of 12 hours to 60 hours or more. The preferred diuretics are chlorthalidone and indapamide which again have a long duration of action unlike hydralazine. As an example see this old paper on amlodipine effects (Abernethy, Cardiol 1992). After a few days spent to achieve steady state, the amlodipine one takes in the morning is still very much around at night - the peak:trough ratio is quite high. So how can timing matter?
The HYGIA trial: Conduct
What was HYGIA? The best one can figure out after reviewing the trial, the registration data, the protocol, and the debate with the principal investigator, is that it is a research program, not just one trial (this is similar to what the authors have done before, see MAPEC below). The randomized control trial is one part of a grand research program, being run out of Vigo, involving primary care practices in the region. This is allegedly how they enrolled over 19,000 patients in the trial with a very low level of funding. However, also note:
Few details of randomization procedure are presented
The figure 1 leaves out details of screening, and consent as CONSORT dictates - typically all those who are screened are not eligible and of all those who are eligible few do not consent
The figure 1 also makes the perfect follow up starkly clear - no one was lost to follow up, the 92 patients excluded were from less than 1 year of follow up, not from loss to follow up
The table 1 from any trial is always instructive, and HYGIA is no exception. Of even more interest is what happened to BP with bedtime dosing, and what happened to lab parameters. Have a look.
The participants underwent not just 24 hour ABPM, but 48 hour ABPM (these would be more accurate for sure) - performed from primary care. Typically, not everything goes well when one is doing an ABPM - these are oscillometric monitors - and humans don’t always follow orders. But in HYGIA participants underwent numerous ABPMs throughout follow-up since medications were being adjusted - and still got perfect readings, as far as we can see what is reported. Similarly, adherence to ‘take this pill’ is far from perfect. Now imagine: ‘take this pill only at bedtime’ versus ‘take this pill on awakening’ - how well will this work in practice? The investigators say their advice was to keep pill bottles on the night table next to the bed. And this allegedly worked like a charm with perfect adherence. No one forgot, no one mixed up the instructions. No one had shift work and no one slept in a bed without a table besides them.
The plan from the paper itself suggests that they expected 20% reduction in events (‘morbidity/mortality’) in 5 years, and hence needed 18,300 participants. Most trials have a hard time enrolling to the planned sample size - these investigators overshot the mark, and 19,168 patients and had a median follow up of 6.3 years. Larger and longer than planned, truly an embarrassment of riches. The trial had interim analyses planned - and as you can see below, the difference is so stark, that it should have been halted for clear superiority many years before. If the analyses was actually done as planned.
The HYGIA Trial: Outcomes
The big splash of course was that bedtime dosing reduced cardiovascular death by 64% (HR 0.44) but also all-cause mortality by almost half (HR 0.55). These are absolutely fantastic, earth-shattering, practice changing results. Trials with these findings are typically published in the NEJM, LANCET or JAMA. Perhaps even Nature. Not HYGIA - they remained loyal to their European roots perhaps?
But a deeper dive reveals something more puzzling. The crude numbers are not provided - so with some approximation, we can surmise that the total deaths would be about 957, and cardiovascular (CV) deaths about 310. Hence to explain the HRs reported, there would be 647 non-CV deaths, with an RR of 0.60 for bedtime dosing. Ergo, bedtime dosing reduces non-CV (eg cancer, infectious, trauma etc) deaths. How can that happen? The sheer implausibility of the presented data is mind-boggling. It is believable for the gullible.
The HYGIA Trial: Adverse effects
Lowering BP causes side effects from lowering BP, and side effects from the medications used to lower BP. As the wag says, a drug without side effects has no effect. Like homeopathy. Dare one mention chronotherapy in the same breath? For sure, the HYGIA trial did not need extra BP lowering medications in the intervention, but BP lowering - especially at night - was the purported mediator of benefit. Lower BP at night may not be good if it becomes too low - which will invariably happen sometime in someone with over 9,000 participants. It can cause subendocardial ischemia, watershed infarcts, and normal pressure glaucoma and ischemic optic neuropathy. For the latter, the ophthalmology literature has plenty of case series. Nothing like that happened in HYGIA. No increase in adverse events at all, no hypotension, no falls, no syncope, and a lower rate of ischemic optic neuropathy (HR 0.56). Everything gets better with bedtime dosing.
Other Chronotherapy trials from the same stable
Now this is not their first rodeo. A few years ago, there was similar data published from the MAPEC trial from the same group. It seemed to be a research program, with observational and clinical trials published from the same single trial protocol, and several subgroups presented separately. It’s somewhat byzantine to figure out the different papers.
Hermida et al, Am J HT 2009; Ayala et al Chronobiology Int 2009; Hermida et al, Am J HT 2010; Hermida et al Chronobiology Int 2010; Hermida et al Diabetes Care 2011; Hermida et al, Chronobiology Int 2011; Hermida et al, JACC 2011; Hermida et al, JASN 2011; Hermida et al Am J HT 2012; Ayala et al Chronobiology et al, 2013
But this is not it. The same group has the Midas touch for all their chronotherapy research. They have reported aspirin given at night improves fetal outcomes in pregnancy. They have done a bunch of single drug trial - all of which show remarkably positive results. Nighttime dosing, regardless of drug, population, or outcome, just works in their hands. See an incomplete list here
Contrasting Results
As expected - no one else has been able to replicate these findings. Some notable examples are the HARMONY (Poulter et al, Hypertension 2018) trial.
Another one is a substudy from the AASK trial (Rahman et al, Hypertension 2013).
One important thing to also consider is that we have trials like HOPE: comparing ramipril (which indeed was given at night) versus placebo, in a high CV risk population. Compare the benefit of ramipril vs placebo to that of BP drugs given when awake versus at bedtime (see figure below). How is this plausible? Unless you believe that BP lowering medications given in the morning are somehow harmful!
HYGIA versus HOPE 🤔
Some HYGIA Correspondence
Many people were concerned about the HYGIA hyperbole - and there is correspondence in the pages of the journal, as well as other journals. See Brunstrom et al, Kreutz et al The authors’ replies (eg Hermida et al, Chronobiol 2020, and here) are a snarky screed at best and a word salad at worst.
After a debate with Prof Hermida, we were invited to write a Pro/Con piece - you can read it here (Turgeon et al Hypertension 2021). The critical appraisal of HYGIA occurred on Twitter, so the paper was written by a bunch of us who met on Twitter. Check out a couple of threads here.
A much longer and funny thread in his inimitable style from Prof Francis is here
The way forward
What we really need is good, but properly designed and transparently reported trials. Two such trials are ongoing:
TIME and BEDMED. The results should be available soon.
Time to put this controversy to bed.
Swapnil Hiremath, Nephrologist, Ottawa
Postscript
The results of TIME were presented in Barcelona on Aug 26th. And, as expected, bedtime dosing does not improve CV outcomes. Have a look at some of the results from the #ESCCongress:
The actual paper will be published in a high impact journal soon. Look for the discussion on #NephJC.