Across 59 real-world studies and 24,859 patients, the pill form of semaglutide lowered blood sugar by about as much as the clinical trials had promised.
That is the finding of a meta-analysis published in May in touchREV Endocrinology ↗ by Sweekruti Jena, Deep Dutta and colleagues across India and Bangladesh. It pooled the studies that have tracked patients actually taking oral semaglutide ↗ in clinics, rather than in the controlled conditions of a trial. Oral semaglutide, sold as Rybelsus, is the only pill in the GLP-1 receptor agonist ↗ class approved for type 2 diabetes. Everything else in the class, including the injected version of the same drug, is a shot.
The numbers
Patients' HbA1c, a blood test that reflects average blood sugar over roughly three months, fell by 1.11 percentage points at six months (95 percent confidence interval 0.86 to 1.37) and 1.19 points at a year. More than half of them, 54.7 percent, got below the 7 percent line that doctors treat as controlled diabetes. Weight dropped an average of 4.38 kilograms (about 9.7 pounds) at six months and 5.96 kilograms (about 13 pounds) at a year. Blood pressure and cholesterol improved too. The authors report the same pattern held across Asian and non-Asian groups, which matters because much of the real-world data on this drug comes from Asia.
Those figures land close to what the registration trials showed. The headline of the paper is not that the pill is strong. It is that the pill is strong outside the trial, where patients are messier.
Why "real-world" is the whole point
Trial patients are selected. They tend to be more adherent, have fewer competing illnesses, and get more support than the average person picking up a prescription. Oral semaglutide is also fussier than most pills: it has to be taken on an empty stomach with a small sip of water, and the patient has to wait at least 30 minutes before eating, drinking, or taking anything else, or the drug barely absorbs. A reasonable worry was that those rules would quietly erode its effect once it left the trial. They mostly did not, and that is the useful piece of information for a doctor deciding whether to start someone on it.
There is a catch worth stating plainly. The studies disagreed with each other a lot. The statistical measure of that disagreement, called I-squared, ran near 99 percent, which means almost all the spread between studies came from real differences in how they were run and who was in them, not from chance. A pooled average sitting on top of that much variation is a blurry composite, not a precise number. The honest read: the direction is solid, the drug works in ordinary practice, but the exact size of the benefit for any one clinic or patient is soft.
The platform angle
Semaglutide is the most-covered molecule in this section, almost always through its injected outcomes, from heart and kidney trials to the cirrhosis database studies. The oral version is the quieter member of the family, and the question hanging over it has always been absorption: a peptide swallowed as a pill is fighting an uphill battle against the gut, which is built to digest peptides, not deliver them. This meta-analysis is the cleanest answer yet that the workaround holds up at scale. The molecule sits on the platform as a live card ↗ against the GLP-1 receptor ↗, the same target every drug in the obesity and diabetes race is built around.