“…Seventeen studies (46 percent) used a difference-in-differences specification that controlled for unobservable confounders. 12,16,[27][28][29][30]32,[34][35][36]38,40,[42][43][44][45][46] Twelve studies evaluated the impact of tort reform on defensive medicine by assessing health care utilization 20,22,23,25,26,28,37,38,47,49,51,52 of which two (17 percent) used a difference-in-differences specification. 28,38 Thirteen studies studied quality of care 16,20,21,26,28,29,[33][34][35]37,41,48,54 of which five (38 percent) used a difference-in-differences specification.…”