“…Data came from patient records or clinical-administrative databases ( n = 27) [ 5 , 60 , 73 , 74 , 78 – 81 , 83 – 89 , 91 , 94 , 98 , 101 – 105 , 108 , 110 , 112 , 114 ], or a combination of these methods, and from patient self-report ( n = 17) [ 53 , 55 , 58 , 62 , 63 , 65 , 68 , 71 , 72 , 77 , 82 , 90 , 96 , 100 , 107 , 109 , 113 ] in many studies. The remaining studies measured outcomes from self-reported data only ( n = 5) [ 50 , 54 , 67 , 69 , 95 ] or did not clearly describe the data source ( n = 17) [ 51 , 52 , 56 , 57 , 59 , 61 , 64 , 66 , 70 , 75 , 76 , 92 , 93 , 97 , 99 , 106 , 111 ]. Regarding the proportion of trauma and orthopaedic surgical patients still using opioids at follow-up, the proportion of patients using opioids at 3 months as reported in some studies varied from 12 to 30% in trauma patients [ 63 , 86 , 100 ] and from 10 to 50% in elective orthopaedic surgical patients [ 67 …”