Intended for healthcare professionals

Views & Reviews Personal View

Publishing individual surgeons’ death rates prompts risk averse behaviour

BMJ 2014; 349 doi: https://doi.org/10.1136/bmj.g5026 (Published 12 August 2014) Cite this as: BMJ 2014;349:g5026
  1. Stephen Westaby, consultant cardiac surgeon, Oxford, UK
  1. swestaby{at}ahf.org.uk

Most deaths are related to team dynamics and hospital infrastructure rather than surgical error, says Stephen Westaby, and publishing surgeon specific mortality data is unhelpful for the profession—and patients

A surgeon was recently castigated in the press for allegedly manipulating operative data.1 Failing to ask why, journalists missed the bigger picture—that of a surgical profession under stress. Repeating the US experience, publication of surgeon specific mortality data (SSMD) has shifted emphasis from patient care to self preservation.2 3 Meanwhile, surgeons struggle to maintain low death rates in outdated facilities with inconsistent teams and without circulatory support devices available in other European countries.4

In the 1980s the US Health Care Financing Administration collected but did not disclose individual death rates for New York state cardiac surgeons.3 A newspaper sued and then published the information, but was criticised, and soon afterwards, risk averse behaviour and gaming with risk stratification were widely documented.5 The answer was to avoid high risk patients. In Massachusetts, the risk profile in centres with higher mortality fell.3 Cardiologists struggled to obtain coronary artery bypass grafts for comorbid patients, but not in cases exempt from reporting. Published outcomes for primary percutaneous coronary angioplasty showed that patients with cardiogenic shock were less likely to receive treatment.3 Risk adjustment provided little consolation. Patients do not understand or engage with risk algorithms.6

Risk averse behaviour exists. It is a dynamic process that affects individuals, teams, and whole departments. The drivers are self protection or the prestige and private practice that come with topping the league. Fictitious surgeons A, B, and C each perform 100 operations (see figure). A is inexperienced, with a low risk patient cohort, but is adept at maximising risk algorithms. He has two deaths from intraoperative events but low mortality overall. B is skilled and experienced with a high risk patient referral base. Two patients die from severe comorbidity and three from postoperative management errors. His mortality rate is questioned. He must now decline high risk patients until his outcome data revert to acceptable limits. C is concerned about his borderline status and knows that his senior colleague is unhappy. He will not accept more high risk patients but makes plausible arguments to justify his decisions. To protect his professional standing he adopts A’s approach to manipulating risk factors. His outcome data remain within acceptable limits. With fewer high risk patients requiring prolonged intensive care the hospital saves money. After 100 more low risk cases, B’s mortality falls and everyone is happy—except the sick patients denied access to surgery.

The actual risk of surgical mortality depends on patient factors and hospital structure and processes.7 8 Donald Berwick, president of the US Institute for Healthcare Improvement, has argued that negative outcomes result from failures of process and systems and not individuals.9

In 2007 the US Society of Thoracic Surgery switched emphasis from SSMD to composite measures of access, process, safety, outcomes, and patient experience, leading to so called star ratings with voluntary but universally adopted publication at hospital level (see table 1).5 The most informative new outcome measure was “failure to rescue” (FTR).7 All centres have the same incidence of complications after operation, principally related to patient comorbidity. Recognising and tackling critical events determines survival, distinguishing high from low mortality centres.7 Team consistency, intensive care staffing levels, patient to nurse ratios, nurse education, job satisfaction, and burn out all affect FTR.10 11 Poor human resource management and high FTR rate are closely aligned.12 In 2008 the American College of Cardiology advocated publication of scientifically valid performance measures at hospital level, with transparency about the dangers of public disclosure.3

Table 1: Contrasting approaches to outcomes disclosure and quality improvement

View this table:

The Society of Cardiothoracic Surgery of Great Britain and Northern Ireland established the National Adult Cardiac Surgery Registry in 1994, but it was a whistleblower that instigated the Bristol Inquiry.13 14 Bristol changed everything by turning the public, politicians, and the press against the medical profession. The Guardian used the Freedom of Information Act to access the coronary artery bypass graft mortality rates of UK surgeons. This was done with punitive intention and without acknowledging the adverse consequences in the US. Unconvincing efforts were made to dispel the spectre of risk averse behaviour.15

To the public, SSMD reflect surgeons’ technical competence. Yet in reality team consistency and operating theatre practice supersede individual performance.16 The FTR concept is paramount in the UK where cardiac centres often rely on temporary staff.17 A recent analysis of cardiac surgical mortality in an NHS hospital identified FTR as the predominant mode of death.18 Similarly, the National Confidential Enquiry into Patient Outcome and Death in elderly patients highlighted not surgical error but problems with infrastructure and process.19 Only 30% of the deceased patients had received satisfactory postoperative care.

Because few deaths are related to surgical error, publication of SSMD diverts attention from deficiencies in NHS infrastructure. An understanding of why patients die allows something to be done about it; attributing a pile of bodies to an individual surgeon does not. Risk averse behaviour benefits neither patients nor the profession. Surgeons will endorse truly transparent outcome data that contribute to patient safety, such as the star rating system.5 In Berwick’s commissioned review of the NHS after the Francis inquiry, he reiterated that hospitals not individuals must be held accountable for poor outcomes.20 Surely the surgical royal colleges knew this. Soon afterwards, SSMD were released for other surgical specialties and interventional cardiology.21 Those who failed to submit information were named and shamed, and the press denigrated individuals on the basis of hastily assembled data. Many specialties lacked infrastructure to review outlying performance.

Much has been learnt from Bristol and Stafford but the ghosts of these scandals persist in the generation of political rather than evidence based policies.13 22 A letter from the Society of Cardiothoracic Surgery executive to the membership on 16 June 2014 finally acknowledged “widespread discomfort within the profession and the inevitable impact on patient care.” Publication of SSMD has not improved patient choice, waiting lists, equipment, staffing, or team consistency. Defensive practice and the European Working Time Directive combine to decimate surgical training.23 24

The sickest patients consume disproportionately high financial resources. It makes economic sense to discourage intervention in high risk patients and SSMD achieves this. Public disclosure of SSMD is not a benign matter: it exposes the profession to serious conflicts of interest. More comprehensive systems of outcome reporting have greater likelihood of improving quality without negative effects.25

Patients must continue to come first.

Notes

Cite this as: BMJ 2014;349:g5026

Footnotes

  • Competing interests: I have read and understood BMJ policy on declaration of interests and declare the following: I was a member of the Society of Thoracic Surgeons US and American College of Cardiology during the outcome disclosure debate.

  • Provenance and peer review: Not commissioned; not externally peer reviewed.

References

View Abstract