Association Between Primary Care Practitioner Empathy and Risk of Cardiovascular Events and All-Cause Mortality Among Patients With Type 2 Diabetes: A Population-Based Prospective Cohort Study

Thank you for taking the time to provide a thoughtful and extensive commentary on our paper.

We agree with the title of your response 'results do not show an effect of empathy on mortality'. Our paper is an observational design and as such, it cannot determine effect or causality. Accordingly, we have not at any point reported an effect in our paper. We have been clear in reporting an 'association' rather than any effects. In fact, we have avoided any absolute terminology as all observational studies require cautious interpretation of associations. We feel we have been rather cautious with the terminology used and alongside our results, we have provided an extensive limitation section which the reader has kindly relisted for us. Our limitations already acknowledge the challenges of interpreting observational designs.

The reader does, however, pose an interesting discussion about the association being variable between tertiles which raises the clinical question about whether there is an optimal level of empathy which is beneficial beyond which clinical impact might be limited. Could too much empathy be problematic through collusion or less biological focus on disease management perhaps?. An alternative and more likely explanation are that our findings were underpowered or the clustering or continuous variables by tertiles may have produced spurious results. However, we can not get away from the fact that we did observe an association between empathy and our outcomes that was consistently in the same direction regardless of the model used. This can not be ignored by a purely statistical approach.

The reader would prefer a rigid statistical interpretation of our results and has listed preferences for the phrasing and terminology to be used in our paper. Although we feel that this is one approach that could be taken, we also think that rigid statistical language alone without clinical language is not our preferred approach. We note that the readership of this journal are predominantly primary care doctors and so too are we as the authors of this paper.

However, we are also interested in engaging in discussions with our colleagues from around the world who might have experience in this type of novel research that meticulously captures the human elements of care and relates this actual clinical outcome over an extended follow-up period. Accordingly, we would be keen to open discussion with researchers who might offer alternative approaches to epidemiological methods and statistical analyses. To that end, we would encourage the reader to contact us directly to share his previous research and knowledge in this field.

Competing interests: None declared