Letters Publishing individual surgeons’ death rates

Author’s reply to Bolsin and Colson, Nelson and colleagues, and Vallance-Owen

BMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g6032 (Published 09 October 2014) Cite this as: BMJ 2014;349:g6032

Gaming of Clinician level Mortality data

I hope that Stephen Westaby is not inadvertently suggesting that a high proportion of overseas doctors in a speciality is indicative of poor quality of care.[1]. But I agree with his view that holding individual surgeons culpable for hospital system failures is not a productive endeavour.

Bolsin and Colson’s suggestion that mere publication of morality data somehow forced the " bad surgeons" to dramatically improve overnight is farfetched. The improvement in mortality is more a reflection of “enforced gaming" behaviour by vast majority of good surgeons.[2].

Will-Rogers phenomenon need to be taken into account when analysing trends in risk adjusted mortality data. When the detection and coding of risk factors gets better, the relative surgical outcomes improve in all groups, without any absolute improvement because of this statistical quirk. [3]

In my speciality of oncology, if clinician level mortality data following chemotherapy are published and individual oncologists are held culpable, it would almost certainly lead to reduction in use of palliative chemotherapy. While some might feel that's not a bad thing, surveys indicate majority of cancer patients are willing to try toxic chemotherapy for palliative benefits and minimal survival benefits.

Deaths following chemotherapy, as we quite often find out in our monthly mortality and morbidity meetings, are often due to disease progression in patients who are keen on having more palliative chemotherapy. Even chemotherapy related deaths such as neutropenic septic deaths are often due to systems failure such as inadequate acute oncology service and failure to rescue septic patients.[4]. An unintended consequence of publishing chemotherapy mortality data would be more patients dying in intensive care unit rather than in hospice as is the case in USA.[5].

References:
1 Westaby S. Author’s reply to Bolsin and Colson, Nelson and colleagues, and Vallance-Owen. BMJ 2014;349:g6032.
2 Bolsin SN, Colson M. Publishing performance data is an ethical obligation in all specialties. BMJ 2014;349:g6030–g6030. doi:10.1136/bmj.g6030
3 Feinstein AR, Sosin DM, Wells CK. The Will Rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med 1985;312:1604–8. doi:10.1056/NEJM198506203122504
4 Systemic Anti-Cancer Therapy: For better, for worse? A report by the National Confidential Enquiry into Patient Outcome and Death. 2008. http://www.ncepod.org.uk/2008report3/Downloads/SACT_report.pdf (accessed 18 Oct2014).
5 Wright AA, Zhang B, Keating NL, et al. Associations between palliative chemotherapy and adult cancer patients’ end of life care and place of death: prospective cohort study. BMJ 2014;348:g1219–g1219. doi:10.1136/bmj.g1219

Competing interests: Member of NHS England Cancer Drug fund (CDF), Member of NHS England Specialised Urology Clinical reference group (CRG), Registered stakeholder for Chemotherapy CRG and East Midlands Urological Cancer Research lead.

18 October 2014
Santhanam Sundar
Consultant Oncologist
Nottingham University Hospitals NHS Trust
Nottingham. NG5 1PB.