“…Just under half of studies (n=22) 28,30–36,38,40–44,46,49,53,55,63–66 use regression adjustment as their statistical analysis approach (Appendix Table S6). Other popular approaches include difference-in-difference analyses (n=10) 39,47,48,51,52,54,56–59 , instrumental variables (n=8) 27,60,62,67–71 and regression discontinuity (n=3) 26,37,50 . 17 studies 27,28,30,31,33,37–39,45,47,48,50–52,54,57,62 use the natural experiment as “intention to treat”, e.g.…”