Okay, that was a great depiction of the reasons, five reasons where evidence may go from high quality through the four grade stages, because with grade things can be high quality or certainty, moderate, low or very low. But you were also starting with the best starting point, I think, which is randomized trials. And we don't always have randomized trials, which start as high certainty and then as you very nicely described, move down moderate low or even very low if they have all these problems that you listed. But we don't always have randomized trials, do

we? We don't. And I'm always pleased to say that, well, you could actually have an observational study. You could have a registry of patients with diabetes, and that's your best evidence. You still use grade, and you could actually end up with high certainty evidence. Because there's a way also to systematically appraise an observational study or a systematic review of observational studies with grade, it's just that you start low in terms of certainty. But there are reasons why we may end up with moderate or high certainty, but it's always also a great opportunity to talk about confounding and the issues of risk bias and reasons why most observation studies are not well suited to answer questions about treatment effects. They were all about randomized controlled trials. That you could also look at observation studies and even in some cases, and we know about some cases according to that we've used, where you have high certainty, for example, about rare harms. Real world evidence is to me a very problematic term, but you could argue that for some outcomes, observational studies may be the best source and they may actually really help us inform treatment effects as well. Okay,

great.