They did not know they were taking part in a trial and that the focus of the study was on their alcohol consumption. The trial has low risk of bias from randomisation sequence generation, allocation to experimental groups and blinding of intervention facilitator and outcome assessor, as outlined in the Methods. The protocol was published to guard against reporting bias and the high rates of follow-up minimise the likelihood of response bias. This trial was designed to minimise the impact of assessment on alcohol consumption, with the 3-item AUDIT-C questionnaire the only alcohol-related measure used at baseline. Collection of AUDIT-C data was necessary to establish eligibility for the trial, therefore reactivity of assessment was minimised rather than eliminated and may have been responsible for the slight reduction in score within both groups at follow-up. Although this trial did not quite meet its pre-defined sample size, the 95% confidence intervals around the XAV939 primary outcome were narrow and excluded the 20% reduction in alcohol intake used to power the sample size calculation. Therefore, the study was sufficiently powered and we can conclude with some certainty that there was no evidence of a difference between groups. The study team worked closely with the organisation’s communications team to advertise the health check on the company’s web-portal, in-line with their standard procedure; the trial could not have been conducted without the support and guidance of the occupational health lead. Only 3% of the total workforce took part in the health check. Comparing the health behaviours of the participants in this study with the general adult population in England suggests that fewer of the participants were smokers, but median number of portions of fruit and vegetables was lower than average, as was median number of minutes of physical activity. In contrast, the proportion of employees exceeding the threshold for alcohol misuse was higher than the national average. Whilst there was a large proportion of employees who exceeded the drinking threshold, the average score on the AUDIT-C was low. It is therefore unsurprising that there was no difference in alcoholrelated harm when this population was unlikely to be experiencing problems at baseline. We do not know whether the health behaviours of the participants in this study are representative of this individual organisation. Long-term evaluation of between group differences was not possible in this trial due to the wait-list design and the imminent launch of the company’s alcohol campaign, which would include a designated website with online tools for reducing drinking, opportunities for screening and feedback and possibly a road show. This campaign would have contaminated the study, therefore follow-up data was collected before the end of 2012 so that the company could launch their alcohol campaign in early January 2013. Epidemiological studies continue to report a high prevalence of obesity amongst the leisure population of horses and ponies in the UK. The well-documented negative impacts of obesity on health and performance have led to obesity being considered one of the major welfare issues in horses and ponies facing industrialised nations today.