Comment by soligern

You may have a leg to stand on if these were small groups being studied. Those underlying studies have ns of between 5000-100,000 with significant concurrency in results. You can’t just hand wave that away.