Methodological issuesThe completion rates of the outcome measures were very high. Ninety-four percent of trial participants provided data at 12 weeks and 92% at 24 weeks, although as researchers often were needed to aid completion, the time allocated to this task requires consideration. In general, these participants took longer to perform the TUG time/functional ability assessment than that reported for healthy community dwelling subjects 65–84 years [52]. Additional outcome measures, such as frailty, may be included in future, in order to indicate whether exercise is maintaining the level of resilience, even if it does not lead to an improvement. This may be more ppropriateand realistic for an older population with many comorbidities. Further, a longer follow-up period of 12 to 18 months would also be recommended in future studies to explore effects of discontinuation and longer term effects (30). Within the VIOLET feasibility study there was no assessment of visual impairment. This was largely a pragmatic decision; however, this is recommended for future studies as it would allow an assessment of whether a VIOP might be able to join a mainstream class, or whether they might require one to one intervention, or more intensive supervision. The inclusion criteria did not allow screening out for ‘deafness’. People who are profoundly deaf and have a VI are difficult to accommodate in a group setting.The assessment of whether potential participants were ‘physically able to take part in a group exercise class’ category was, on the whole, carried out when face to face, and thus open to individual interpretation. This was also the case when assessing the participant’s ability to walk indoors and outdoors with or without aid. Mixing together able and less able participants, impacts on the challenge and potential effectiveness of the programme. Stratification by functional ability, as well as falls risk, is recommended for a definitive trial.There were a number of potential participants who ‘self-reported’ an uncontrolled medical condition as reason for exclusion, however the degree of concordance with a GP was not assessed. If recruitment were to be carried out in primary care in a future study, the assessment would be carried out by a GP, rather than by a participant or a researcher, who may not be medically trained, although our method replicates current practice in falls services. Although the progression criteria to judge the feasibility of progressing to a full trial were all met, suggesting that this intervention could be taken to a full study, recommendations from the research team and participants suggested that most VIOP in the pilot trial could have been integrated into a mainstream class.Training (CPD) of instructors to accommodate a range of VIOP in their mainstream sessions is minimal and easy to facilitate through online training so would be affordable and have sustainable reach. Only those with multiple comorbidities (such as extreme deafness or extremefrailty) required significantly more supervision. Mainstream classes can be much larger, so if VIOPs were to join, extra supervisors may be necessary.