The most expensive conclusion available from this paper is that two drugs beat one. It is also the conclusion resting on its weakest evidence, and the authors say so themselves in the last line of their abstract.
How the comparison was built
Sixteen randomized trials with more than a year of follow-up, each reporting outcomes broken down by whether participants happened to be taking an SGLT2 inhibitor or a GLP-1 in the background. [1] Those subgroups were then assembled into a network, allowing comparisons between strategies that no single trial tested directly.
Randomization inside each trial governs who got the study drug. It does not govern who was already on the other one — that was decided by each participant’s own clinician, before the trial, for reasons the analysis cannot see. So the combination is an observational category sitting inside randomized studies.
What the numbers show
Against placebo, every active strategy reduced major cardiovascular events and heart failure hospitalization, all at p<0.05. On composite renal events, the combination reached RR 0.63, 95% CI 0.41 to 0.97, and SGLT2 inhibitors alone RR 0.70, 95% CI 0.61 to 0.79.
Look at those two intervals rather than the two point estimates. The combination’s runs from 0.41 to 0.97 — nearly touching no effect. The SGLT2 inhibitor’s runs from 0.61 to 0.79 and is far tighter. The lower number is the less certain one, which is what you expect when it comes from subgroups rather than from randomized arms.
In pooled head-to-head comparisons the combination beat SGLT2 inhibitors alone on MACE, RR 0.83, 95% CI 0.72 to 0.96, and on heart failure hospitalization, RR 0.73, 95% CI 0.56 to 0.94. Against GLP-1 alone it showed a better kidney filtration slope, a mean difference of 2.29 mL/min/1.73 m² per year, 95% CI 0.14 to 4.44.
What a SUCRA of 1.000 does and does not mean
The combination ranked first across outcomes with SUCRA values from 0.928 to 1.000. SUCRA summarizes how often a treatment comes out on top across simulated orderings. A value near 1 means it almost always ranked best.
It says nothing about by how much. A strategy can rank first consistently while the margin over second place is small, uncertain, or clinically irrelevant — and ranking statistics are the part of a network meta-analysis most often quoted without their intervals.
Where it got weaker
A subgroup analysis restricted to trials in established cardiovascular disease or high cardiovascular risk attenuated the combination’s MACE ranking, though the heart failure ranking held. That is worth registering: in the population where cardiovascular benefit matters most, the case for adding the second drug weakened.
The money version
Two prescriptions cost more than one. Nobody on this roster sells SGLT2 inhibitors, so this is not a purchase available here, and the relevant lesson is transferable rather than immediate: when the evidence for a more expensive option comes from subgroups nobody randomized, the extra cost is certain and the extra benefit is not.
If a prescriber proposes both, that is a legitimate clinical judgment and this analysis is one input to it. What it cannot support is a marketing claim, or a decision made without one — and the honest unit for weighing any of it is how many people must be treated for one event to be prevented, which this paper does not report. The kidney evidence for GLP-1 drugs alone is randomized and worth reading first, and the price check covers only the half of this combination anybody here sells.