This is a study about treating blood pressure in which both groups ended up with the same blood pressure. What happened afterward was not the same at all, and that gap is worth more attention than the headline result — it is the clearest case this desk has met of a comparator deciding a finding.
What was compared
Adults with overweight or obesity and resistant hypertension — pressure still uncontrolled on an ACE inhibitor or ARB, a calcium antagonist and a diuretic — who started a fourth drug between 2017 and 2025 across 67 US healthcare organizations. [1] Those starting semaglutide or tirzepatide were matched against those starting spironolactone or eplerenone, the guideline-recommended option. Of 213,309 eligible patients, 22,694 started a GLP-1 and 5,673 an MRA; propensity matching left 4,153 in each group.
Over a median 1.4 years the GLP-1 group had lower risks across the board: MACE HR 0.63, 95% CI 0.52 to 0.78. Cardiovascular events HR 0.74, 95% CI 0.59 to 0.92. Major adverse kidney events HR 0.64, 95% CI 0.46 to 0.88. Acute kidney injury HR 0.62, 95% CI 0.46 to 0.83. All-cause mortality HR 0.34, 95% CI 0.21 to 0.55.
And the blood pressure
Systolic pressure fell 5.7 mmHg on the GLP-1 and 6.3 mmHg on the MRA at twelve weeks, with overlapping intervals. On the thing both drugs were prescribed to do, the two groups were the same — and the MRA arm did marginally better.
What separates them is probably the healthcare system
Consider what it takes to be in each group. Spironolactone is a decades-old generic costing a few dollars a month; a clinician can add it to anybody’s regimen this afternoon. Getting semaglutide or tirzepatide approved as a fourth-line blood pressure drug requires an insurer’s cooperation, a prescriber willing to pursue it, documentation, and often the patient’s own money.
The people who clear that are, on average, better insured, more engaged with their care, seen at better-resourced practices, and well enough to be considered for an elective addition. Those characteristics predict survival on their own. Propensity matching adjusts for what was recorded, and none of that is recorded — it is the shape of the confounding that a national registry showed from the access side, arriving here as an apparent treatment effect.
What the study is still good for
Two things. The kidney findings are consistent with randomized evidence in other populations, so they are plausible rather than merely observed — though the size is suspect for the same reasons. And the blood pressure comparison itself is useful: a GLP-1 achieved about the same reduction as the guideline drug in a population where pressure is hard to move.
The authors say prospective studies are needed to determine whether these drugs should be part of treatment for resistant hypertension, which is the correct conclusion from their own data. The paper declares no funding.
Why this matters outside cardiology
Because the same structure appears everywhere in this market. An expensive drug is compared against a cheap one, and the people who obtained the expensive one were already different — richer, better covered, more closely followed. The resulting difference gets attributed to the molecule, and matching on recorded variables cannot repair it.
Nobody on this roster sells anything for resistant hypertension. What sellers do publish about who they will and will not prescribe to is thin, and the disclosure scorecard measures how thin — access is the variable nobody records and it shapes every real-world result this field produces.