On 2015 Oct 21, Chetna Malhotra commented:

Thank you for your interest in the paper. You may also be interested in another paper from our team, using the same DCE, published in Health Policy (2015) comparing patient preferences with those of older adults in the community. A forthcoming manuscript by our team in BMC Palliative Care also uses a similar methodology to assess variation in physician recommendations for end-of-life care.

The attributes for the DCE used in this manuscript were based on a review of literature as well as results from our own focus groups with older Singaporeans (Malhotra et al, 2012), thus enabling us to tailor the DCE to our local context. Due to space considerations, we were unable to present the regression table in the paper, but the rescaled regression coefficients are shown in figure 1 of the manuscript. The figure allows for easy interpretation of the regression results. The DCE experimental design was a main effects design and thus limited any post hoc testing for interactions. This could be an area of future research. Ethical considerations limited our ability to recruit patients who were not aware of their own diagnosis. This would be a common consideration for all palliative care researchers working in similar settings.

On 2015 Oct 13, Cicely Saunders Institute Journal Club commented:

This paper was discussed at a Journal Club at the Cicely Saunders Institute, King's College London, on Wednesday 7th October 2015.

This study is a nice example of how a discrete choice experiment (DCE) can be used in palliative and end-of-life care to assess preferences for aspects of care. It also raises some very interesting questions about the differences in priorities and the extent to which caregivers might be able to act as a proxy for patients, an important consideration for end-of life care. Our Journal Club discussed the work required to ensure sufficient attribute identification for a robust DCE, and wondered if the attributes decided upon in this paper sufficiently captured what is most meaningful for patients and caregivers at the end of life, i.e., there was no mention of a systematic review used to develop the attributes (see Bridges et al., 2011 for an example of DCE reporting guidelines). We also would have liked to see a table of the probit regression output for clarity on how the willingness-to-pay was calculated, and more detail on this in the methods. Furthermore, we found it confusing that the authors state in the discussion that their sample size was too small to explore interaction effects, while it appears they recruited 70% more than their minimum acceptable sample size – some explanation would have been helpful. Lastly, we wondered about the potential risk of bias of only including those patients who knew their diagnosis. This may limit the generalisability of the findings, even to a Singapore context. We enjoyed discussing this paper, and look forward to more papers using DCE methodology in palliative and end-of-life care.

Commentary by Melinda Smith

On 2015 Oct 13, Cicely Saunders Institute Journal Club commented:

This paper was discussed at a Journal Club at the Cicely Saunders Institute, King's College London, on Wednesday 7th October 2015.

This study is a nice example of how a discrete choice experiment (DCE) can be used in palliative and end-of-life care to assess preferences for aspects of care. It also raises some very interesting questions about the differences in priorities and the extent to which caregivers might be able to act as a proxy for patients, an important consideration for end-of life care. Our Journal Club discussed the work required to ensure sufficient attribute identification for a robust DCE, and wondered if the attributes decided upon in this paper sufficiently captured what is most meaningful for patients and caregivers at the end of life, i.e., there was no mention of a systematic review used to develop the attributes (see Bridges et al., 2011 for an example of DCE reporting guidelines). We also would have liked to see a table of the probit regression output for clarity on how the willingness-to-pay was calculated, and more detail on this in the methods. Furthermore, we found it confusing that the authors state in the discussion that their sample size was too small to explore interaction effects, while it appears they recruited 70% more than their minimum acceptable sample size – some explanation would have been helpful. Lastly, we wondered about the potential risk of bias of only including those patients who knew their diagnosis. This may limit the generalisability of the findings, even to a Singapore context. We enjoyed discussing this paper, and look forward to more papers using DCE methodology in palliative and end-of-life care.

Commentary by Melinda Smith

On 2015 Oct 21, Chetna Malhotra commented:

Thank you for your interest in the paper. You may also be interested in another paper from our team, using the same DCE, published in Health Policy (2015) comparing patient preferences with those of older adults in the community. A forthcoming manuscript by our team in BMC Palliative Care also uses a similar methodology to assess variation in physician recommendations for end-of-life care.

The attributes for the DCE used in this manuscript were based on a review of literature as well as results from our own focus groups with older Singaporeans (Malhotra et al, 2012), thus enabling us to tailor the DCE to our local context. Due to space considerations, we were unable to present the regression table in the paper, but the rescaled regression coefficients are shown in figure 1 of the manuscript. The figure allows for easy interpretation of the regression results. The DCE experimental design was a main effects design and thus limited any post hoc testing for interactions. This could be an area of future research. Ethical considerations limited our ability to recruit patients who were not aware of their own diagnosis. This would be a common consideration for all palliative care researchers working in similar settings.