Dr Christmas is quite correct in reiterating the uncertainty we expressed in our discussion about the placebo condition in our study. This does indeed have very significant implications for the interpretation of our results. It is for this reason that we suggested a number of different interpretations for our findings in the Discussion, including the possibility that light therapy ‘may, therefore, be an effective treatment for symptoms of low mood in epilepsy at lower intensities than those typically used to treat seasonal affective disorder’. We also discussed the possibility that this could indeed be a negative finding or that the results we found could be due to other factors unrelated to light therapy, such as the establishment of fixed morning routines.

Dr Christmas is correct in that in the original protocol for the study the control arm should have been receiving 100 lux. The modifications to the original protocol were submitted with the paper as an online appendix, to conform to the CONSORT guidelines for reporting trials. In this case, when the light boxes were modified to 100 lux, the disparity in intensity was very obvious and we did not feel that the study would conform to the important double-blind aspect of the design. It would have been very clear to any patient who received the 100 lux box that they had been assigned to the low-intensity arm of the trial. We therefore modified the boxes to administer 2000 lux at 20 min in the low-intensity arm. The boxes appeared bright, but literature on seasonal affective disorder indicates that this would not be a therapeutic dose within this time frame, whereas 10 000 lux at 20 min would be a therapeutic intensity/dose.

As we stated in the introduction to our study, the primary outcome measure for this trial was seizure control. We have reported these results separately^{Reference Baxendale, O'Sullivan and Heaney1} and that paper is fully referenced in our study. Although it is possible that bright light therapy may result in an increase in seizures for some patients, this was not a statistically significant finding in our previous study and, as yet, the risk remains theoretical. Clinicians will be aware that seizure control should be carefully monitored following the introduction of any new treatment offered to people with epilepsy.

In presenting the results of our study for publication we have sought to provide as clear an account of the data as possible. The results are by no means clear-cut or definitive. However, there are some interesting aspects to the data that suggest that this may not be a dead end in terms of a treatment option for some people with epilepsy. This study stands as a guide for future research. We hope that its limitations, which we fully acknowledge and have set out at length in the Discussion, will serve as a useful guide for future research in this area.