Surgeon specific outcomes – an unexpected journey

Posted 28 June 2013 in

No Bilbo, No Gandalf, but still unexpected and some journey.

Let’s start at the very beginning. The fundamental purpose of the Association is to advance the management of patients with colorectal disease. You cannot advance if you don’t know where you are and hence the very early development of colorectal cancer audit traced back to the 1990s and the Wessex Audit, the Trent and Wales Audit and a series of Scottish Audits that all described patient results after surgery for colorectal cancer. From that experience we got the ACPGBI’s minimum dataset in 2000 and so began systematic but voluntary collection of colorectal surgical activity and associated patient outcomes.

Using voluntary clinician participation the Audit got a view of 8000 cases from 261 surgeons in 2002 and 10600 cases in 2004 but could not achieve a truly national audit by relying simply on enthusiasm. Furthermore, there were other difficulties -money to pay analysts as well as complex data protection issues; both solved by taking the Queen’s shilling and establishing a true national audit with the Health commission. A decision that quickly and comprehensively magnified the power of the audit to describe cancer outcomes based on 18,600 patients in 2006, 24,000 in 2010, 28,000 in 2011, 29000 in 2012 and this year – 29,445 people diagnosed with colorectal cancer between April 2011 and March 2012.

So on the upside the Audit delivers a truly comprehensive and contemporary picture of colorectal cancer care – the biggest in the world. The significant downside is that this very comprehensiveness has been achieved by bypassing the clinician and feeding on administrative data routinely collected by hospital cancer management teams, whose focus is slavish devotion to cancer waiting times. As is the case with all NHS data, what you gain in overall coverage of the clinical picture, you then lose in the fine detail or “granularity” of the clinical pathway. A compromise – but one that makes it possible to describe with reasonable accuracy, outcomes such as postoperative death at national, network and unit level. Indeed so useful is this approach, that the cancer audit specification review meeting chaired by Mike Richards in October 2012 emphasised that this model of audit based on routine NHS data collection (like cancer waiting times) was the future of NHS cancer audit and that bespoke data collection just for audit purposes should be avoided.

The Unexpected Bit – four months ago, in February 2013, at the Royal College of Surgeons of England, the NHS Medical Director informed us that he is now seeking surgeon specific outcomes such as death rates for colorectal cancer surgery. Now whether a doctor or a patient, if you have been involved with colorectal cancer you cannot fail to note that the care given in 2013 is a truly multidisciplinary activity that involves surgeons, nurses, oncologists, radiologists, anaesthetists and critical care. Assessment of co-morbidity, selection of management pathway and ensuring good postoperative care being at least as important as who mobilised the splenic flexure or fired the staple gun. But put that to one side – the directive from on high was clear: surgeon specific outcomes (death rates) and four months to deliver it.

Problem one – the bowel cancer audit didn’t have a GMC number for 30% of cases in the audit – never mind if we even had the right GMC number attribution. And then there was clinician consent. The legal advice we heard at later meetings was that clinician consent to data publication was required. The rest of the story you know – many, many hundreds of man and woman hours feeding data backwards and forwards between the Audit and Trusts and trying to get a responsible consultant GMC code with consent beside each elective colorectal cancer resection. The fundamental problem being that while deaths (the numerator) is easy to define as this comes from the Office for National Statistics, the Cancer Management Team-Somerset-Open Exeter data upload can play havoc with the Denominator. And it is the Denominator that defines the “death rate” – cases lost for any one of several data process reasons artificially drives down the GMC number attributable denominator and drives up the GMC number attributable death rate. For this reason many Trusts and surgeons have decided to redo the whole data submission exercise for a second time. It is only just starting to dawn on our political masters the enormity of the task in hand and the potential for confusing patients and erroneously harming reputations.

Transparency of outcome that helps a patient make a treatment decision is the very ethos of ACP. With all the caveats expressed above, if it is surgeon specific outcomes patients want (and it is unclear whether this question has even been asked) then ACP has to be central to an accurate and readily understood answer. On this basis we have to get better at making sure that each colorectal cancer case in the Audit has the correct consultant responsible GMC number attribution and the correct details of procedure and risk adjustment. In this quest we cannot abandon the comprehensive cancer coverage given by using Trust cancer management team data through systems such as Somerset and go back to the world of 8000 enthusiast cases.

For this reason the NHS Information Centre will seek to develop mechanisms by which the individual surgeon can get an early look at the data uploaded by their cancer team through Open Exeter to the Audit. If in the future cancer management teams can be persuaded to do data uploads on a monthly basis – then this should enable you to see any discrepancies and prompt your cancer team to improve the accuracy of your data before any analysis. These developments are necessarily complex and this has to be seen as work in progress; we cannot transform a system of data collection that delivers 30,000 colorectal cancer cases per year overnight. But looking to the future there appear to be opportunities within NHS Information Centre that may yet allow greater real time scrutiny of clinician specific data by clinicians – although the responsibility for data upload would have to remain with Trust cancer management teams.

Although the last four months have been difficult and traumatic – there is a silver lining. The impact of clinician-specific outcomes is that there has to be detailed clinician engagement in colorectal cancer data. On this basis we can swing the current surgeon outcome focus back onto the transparency issues that really matter such as – who and why are certain patients not having a resection, the standard of postoperative care, the access to stenting and emergency theatres. Unexpected as the journey has been, there is the prospect of a real prize – securing better patient outcomes from better data quality.

[Thanks to James Aitken, Colin McCardle, Chris Marks, Jeff Stamatakis, Mike Thompson, Paul Finan, Jason Smith for showing the way] Nigel Scott & Graham Williams