thiazides: best monotherapy for htn??

A recent huge real-world study (LEGEND-HTN) compared the effectiveness and safety of the major first-line antihypertensive drug classes initiated as monotherapy, finding that thiazide or thiazide-like diuretics were better than ACE inhibitors, and that non-dihydropyridine calcium channel blockers fared the least well (see htn monotherapy hctz lancet2019 in dropbox, or doi.org/10.1016/
S0140-6736(19)32317-7).

Details:
-- the authors developed a comprehensive framework for real-world evidence from several large databases, including six administrative claims and three electronic health record databases: IBM Market Scan databases for US employer-based private payers, Medicare supplemental beneficiaries, and multi-state Medicaid; Japan and South Korea national health databases; and the Optimum Pan Therapeutic US health systems, IMS/IQVIA disease analyzer from Germany, and the Columbia University Medical Center electronic record databases
-- 4.9 million patients were included; the authors generated 22,000 calibrated, propensity-score-adjusted hazard ratios, comparing all classes and outcomes across databases
-- they limited their search to those initiating monotherapy for hypertension
-- hypertension therapy initiated:
    -- ACE inhibitor: 48% of patients, of which 80% lisinopril
    -- thiazide or thiazide-like diuretic: 17%, of which 94% hydrochlorothiazide (HCTZ)
    -- dihydropyridine calcium channel blocker: 16%, of which 85% amlodipine
    -- angiotensin receptor blocker: 15%, of which 45% losartan
    -- non-dihydropyridine calcium channel blocker: 3%, of which 62% diltiazem
-- initial blood pressure was highest in those put on thiazide or thiazide like diuretics (hereinafter called "thiazides") at 142/88 mmHg, dihydropyridine calcium blockers (DHP-CCBs) 141/88, ACE inhibitors 140/84, ARBs 138/32, and non-dihydropyridine calcium blockers (nDHP-CCBs) 133/80
-- median on-treatment time at risk for patients varied by drug class and database between 1-7 months, though in most databases 25% were exposed to the first drug class for more than one year, and median overall follow-up for patients was more than two years (25% had more than five years of follow-up)
-- primary effectiveness outcomes: acute myocardial infarction, hospitalization for heart failure, and stroke
-- six a secondary effectiveness and 46 safety outcomes were assessed

Results:
-- comparing thiazides to ACE inhibitors:
    -- 16% fewer myocardial infarctions, HR 0.84 (0.75-0.95)
        -- adjusting for baseline blood pressure: HR 0.85 (0.70-1.03), no longer achieving statistical significance though still likely clinically significant
    -- 17% decrease in hospitalization for heart failure, HR 0.83 (0.74-0.95)
    -- 17% decrease in stroke, HR 0.83 (0.74-0.95)
    -- safety profiles favored thiazides over ACE inhibitors: the major differences between the groups were that hypokalemia was much more common in the thiazide group, angioedema and cough in the ACE inhibitor group, end-stage renal disease in both calcium blocker groups
    -- seven of the other 9 effectiveness outcomes presented found a significantly decreased HR in favor thiazides versus ACE inhibitors
-- non-dihydropyridine calcium blockers were significantly inferior to the other classes
-- all of the other comparisons between the medications were not statistically significant

Commentary:
-- this study found that adverse cardiovasc effects were not much difference between the drug classes, except that thiazides versus ACE inhibitors did have a significant lower risk of seven cardiovasc effectiveness outcomes including the three major ones, and the non-dihydropyridine calcium blockers did the worst.
    -- they calculate that if 2.4 million ACE inhibitor initiators had instead started a thiazide, more than 3100 major cardiovascular events might have been avoided: 1.3 cardiovascular events avoided for every 1000 patients who initiated the thiazide
--there are no direct RCTs comparing many of the desired head-to-head medication comparisons: e.g. between thiazides and ARBs, or even ACE inhibitors vs ARBs. And no new RCTs have been found since 2009, per the 2017 Cochrane Review. This current type of large-scale real world study allows for indirect comparisons with mathematical modeling to attempt to equalize the differences in patients involved. 

--my major critique of the study is that there is no information about the dose of the HCTZ, the predominant thiazide in the study.  And there are really important concerns about this, enumerated in a blog on 5/31/19, which also commented on the increased risk of melanoma with HCTZ (see http://gmodestmedblogs.blogspot.com/2018/05/hctz-and-melanoma-risk-and-not-such.html) and quoted here:

--there are essentially no clinical outcome data on HCTZ 12.5 mg, a very commonly used dose now. It is notable that the old JNC8 "recommends thiazide or thiazide-like diuretics", and for HCTZ, "using 12.5 to 25 mg/d as initial dose", but the new ACC/AHA guidelines state that chlorthalidone is the preferred diuretic, and HCTZ should be prescribed at least in the 25-50 mg dosages (see http://gmodestmedblogs.blogspot.com/2017/11/new-aha-hypertension-guidelines.html ). the old studies finding actual clinical efficacy for thiazides were done using higher doses
--there are even some data that combinations of other antihypertensives with low dose HCTZ are inferior to combinations with calcium-channel blockers (CCBs): eg the ACCOMPLISH trial found that there was a morbidity/mortality advantage of the combination benazepril/amlodipine over benazepril/HCTZ
--there are med adherence data suggesting that HCTZ is among the worst (eg, see htn med compliance 2010 in dropbox, or Friedman O. Am J Med 2010: 123; 173)
--an article by Messerli looked at data on 24-hour ambulatory BP monitoring by several different antihypertensives, also finding that HCTZ in doses of 12.5 to 25mg had 1/2 the blood pressure reduction of ACE-I, ARBs, b-blockers, and CCBs (6.5/4.5 mmHg, vs 12/8 mmHg reductions in the other meds) though in-clinic BP recordings all have about the same BP reduction: those on HCTZ had similar office blood pressures as with the other drugs (ie, it looked good to us), BUT the 24-hour average blood pressure was much higher with HCTZ. And one concern is that the diurnal variation of stroke incidence, for example, is in the early AM hours, when the effect of HCTZ has largely worn off (see htn HCTZ Meta-anaylsis messerli am j cardiol 2011​ in dropbox, or Messerli F. J Am Coll Cardiol 2011; 57: 590). [a couple of additional comments: several studies on ACE inhibitors have found higher stroke rates, which goes along with their weaning effectiveness in the early AM associated with their shorter half-life; and there are ample studies finding that 24-hour ambulatory blood pressure is the best predictor of future cardiovasc events: eg see http://gmodestmedblogs.blogspot.com/2015/01/uspstf-recs-on-ambulatory-blood.html
--there are some additional studies finding that blood pressure variability may be a predictor of cardiovascular events (see http://gmodestmedblogs.blogspot.com/2016/09/blood-pressure-variability-increases.html  ). a recent study suggested that amlodipine was somewhat better than other meds, though chlorthalidone and losartan were just about as good (see htn bp variability amlodipine htn2017 in dropbox, or DOI: 10.1161/HYPERTENSIONAHA.117.10087, as well as bp variability amlod better JASH2014 in dropbox, or doi.org/10.1016/j.jash.2014.02.004)

    -- So, it would be an unfortunate interpretation of the current study to necessarily assume that low-dose HCTZ is also clinically effective in preventing major cardiovascular events

-- one concern with this type of study, combining large databases from different countries and different groups of people is selection bias for medications, which might be very different in different areas. And there may well be unanticipated residual confounding which limit the accuracy of their results. For example, ACE inhibitors tend to be used preferentially in those with diabetes or kidney failure, and there may be other associated conditions not accounted for in the mathematical model that lead to an inferior outcome compared to thiazides. And the doses and choices of medications may be very different in different cohorts
    -- and they had to do significant mathematical modeling: for example the thiazide group was 61% female and 39% male, whereas the ACE inhibitor group was 38% female and 62% male; 5% of diabetics were on thiazides whereas 18% were on ACE inhibitors, 26% of those with hyperlipidemia were on thiazides versus 36% on ACE inhibitors. overall those put on ACE inhibitors were more likely to be male, older, have diabetes, hyperlipidemia, atherosclerosis, or heart disease relative to those put on a thiazides (and might therefore have had more cardiovasc events). And those put on non-dihydropyridine calcium blockers had higher prevalence of atrial fibrillation or other heart diseases than on the other classes of meds, and women on non-DHPCCBs were more likely to be pregnant, than those on ACE inhibitors or ARBs
     --These were corrected for through their mathematical modeling, but, again, potential unanticipated differences could remain and affect the results

-- another issue is that blood pressure lowering tends to be greater in those with higher blood pressure, and in this study those on thiazides did have the highest initial blood pressure, though we do not have data on the achieved blood pressure reduction. And, in the above study, controlling for initial blood pressure did dilute the effects (making them non-statistically significant) specifically for acute myocardial infarction. the initial blood pressure was much less in those on nonDHP CCBs (though statistical correction for the initial BP did not matter much).

-- The advantage of this type of study is that it is huge, involves people of different ages and in different countries, and it involves an unfiltered population likely to better represent the real-world than a restricted study population
-- some limitations include the absence of important measurement such as blood pressure in some of the databases, the lack of data tracking the results of blood pressure changes associated with the dosage of medications given, the combination of very different studies in very different populations, reliance on significant mathematical modeling to attempt to equalize the populations, the short-term followup for clinical outcomes

So, an interesting study raising the very real possibility that thiazides are better than either ACE inhibitors or non-dihydropyridine calcium blockers for initial therapy of hypertension. But, there really are some important caveats:
-- many people were likely assigned to their chosen antihypertensive for other reasons, such as rate control by non-dihydropyridine calcium blockers in those with atrial fibrillation, or the potentially protective effect of ACE ihhibitors for those with diabetes or chronic kidney disease or heart disease. And, in the global assessment of the patient, these choices may well have been totally reasonable and appropriate, independent of the results of this study
-- I still have significant concerns about thiazides, given the above comments on there being less effective at low-dose. It still seems to me that chlorthalidone is a better choice for a diuretic given its superior 24-hour effectiveness
--interestingly, ARBs did much better than ACE inhibitors in comparaison to thiazides, with no significant differences in outcomes.  should that be our preferred renin-angiotensin-aldosterone system blocker??

 geoff​

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@uphams.org

to get access to all of the blogs:

1. go to http://gmodestmedblogs.blogspot.com/ to see them in reverse chronological order

2. click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

3. or you can just click on the magnifying glass on top right, then  type in a name in the search box and get all the blogs with that name in them

please feel free to circulate this to others. also, if you send me their emails, i can add them to the list