Journal of NeuroEngineering and Rehabilitation

3. Are the methods clearly described? 4. Is the data adequate to support the authors interpretations/inclusions? 5. Are the results generalizable? These questions are each given a score according to the answer, where with 0 = no, 1 = in part, 2 = yes. These point values are then added for a score out of 10 where 7 – 10 is high , 4 – 6 is medium , and 0 – 3 is low , quality. Each article was assessed independently by two graders experienced in scoring articles with this tool and scores were agreed upon by a consensus of the multi-disciplinary team. Empirical evidence statements were then synthesized by the group and given research grades according to the criteria provided in the UK-NSF document [ 14 ] . Statements with Research Grades of A or B were incorporated into the resultant CPG. Results Out of the 323 potentially relevant studies, 308 were ex- cluded mostly due to lack of an exercise testing modality other than simple ambulation or those not related to an outcome variable associated with prosthetic rehabilita- tion. The remaining 15 full-text manuscripts were read and a further five eliminated due to being solely a validation study with no useable data, incorrect applica- tion of an outcome measure, or inclusion of only able-bodied individuals as the sole subjects. The PRISMA diagram is shown in Fig. 2 . Of the remaining 10 studies, eight studies were determined to be high-quality, one medium-quality, and remaining study was rated as low-quality. In terms of quality, five were prospective trials, four were retrospective cohort or mixed cohort designs, and one systematic literature review without data aggrega- tion was included in this review. Seven of the 10 articles utilized captive samples from inpatient prosthetic re- habilitation programs in university or institutional med- ical systems with one study utilizing a convenience community sample. The remaining two studies did not report sample selection procedure. Half of the articles were written by a group of authors from a university hospital in Japan while two were written by one author in Slovenia and the remaining three articles originated from two groups in the Netherlands. The sample includes 448 subjects included in the ag- gregation of data. Seventy-eight of these subjects were extracted from the included systematic review to include only those who had completed an exercise testing mo- dality [ 15 ] . Attrition included 40 subjects for an overall Fig. 2 PRISMA flow diagram Klenow et al. Journal of NeuroEngineering and Rehabilitation 2018, 15 (Suppl 1):64 Page 14 of 72