Understanding the Graphs
All graphs included on this website are based on UK data from the National Adult Cardiac Surgery Audit. Initial analysis was sent to hospitals on 7th September 2012 to be validated by 8th October 2012. Final analysis for inclusion on this site was sent to hospitals on 17th December 2012, with a deadline of 4th January 2013 for reporting any issues.
These pages do not display data for congenital or thoracic surgery. For more information about the National Adult Cardiac Surgery Audit please click here.
Case Mix Plots
We have represented the proportion of different operations performed as shown here:
For more information on these procedures please go to the ‘find out about your condition’ section on the right hand side of this page.
These plots enable patients to see how much of the different types of surgery are performed by each hospital or surgeon, and may be useful to help patient choice.
After consultation with our patient representatives, we have chosen to display risk adjusted mortality data in the form of funnel plots. The operations included in these charts are adult cardiac surgery operations on all patients over the age of 18, excluding heart transplants, insertion of artificial mechanical hearts and trauma cases (as these are all subjected to separate analyses). We have also taken the decision to remove emergency operations from the analyses (and by emergency we mean patients who are taken immediately to the operating theatre for surgery, not those who have to wait in hospital until surgery is carried out), as these patients are relatively infrequent and we know that it is very difficult to make appropriate adjustments for predicted risk in this setting.
The funnel plots show how risk adjusted mortality rates of particular hospitals/surgeons compare to the national average, which is the standard that we have set for outcomes. The risk adjusted mortality rates of hospitals/surgeons are plotted on the chart against the number of procedures undertaken. Each hospital is represented by one dot on the funnel. The dot is the risk adjusted mortality, which means that we have adjusted each dot using complex methods so that effectively we show what the mortality would have been had each hospital or surgeon operated on the average case mix.
We use an adaptation of the EuroSCORE risk adjustment model for these analyses, and whilst we know that our model is an accurate predictor overall, and disciminates well in general between patients with higher and lower risk, we also know that it is not designed to adjust for surgeons or hospitals with very unsual casemix profiles, and great caution should be taken in making judgements in this context.
We would expect hospitals to cluster around the average mortality on the plots. As the number of procedures increases the variation of the points should decrease. This is because higher numbers of procedures decreases the likelihood of the mortality rate being high due to chance alone. Similarly, as the number of procedures decrease there will be an increased variation (wider spread) due to sampling variability. The increased clustering around the mean line as procedure numbers grow is what gives the chart its funnel shape.
Using only an ‘average’ line for the standard makes it difficult to tell whether units that are plotted away from it are within accepted limits (as there will always be some variation between hospitals and surgeons due to biological variability). For this reason, the graphs also show control limits, which are represented by red and green dotted lines. For more information on calculating the control limits please click here.
Control limits represent the expected range of values based on the mean. If a hospital or surgeon's risk adjusted mortality rate lies underneath the red control limit, it should be understood as being an ‘expected’ mortality rate. If the risk adjusted mortality rate falls above the red line it means that it is higher than expected, and this may mean a number of things including issues with data quality, abnormal case mix which is not accounted for by the risk adjustment, a chance finding, or poor quality of care.
Below is an example of a graph showing the risk adjusted mortality rate of a surgeon. Parts of the graph are numbered, with explanations of that part of it given underneath, next to the matching number.
1. The green dot highlights the hospital or surgeon whose page you are currently looking at. 'n =' gives the number of procedures that hospital/surgeon has done 1st April 2008 - 31st March 2011. 'Adj mort =' shows the risk adjusted mortality rate for that hospital or surgeon during that same period. The grey dots show all of the other hospital/surgeons included in analysis.
2. The horizontal line along the bottom (called the x-axis) is the total number of cases done 1st April 2008 - 31st March 2011.
3. The vertical line running up the left hand side (called the y-axis) is the mortality rate adjusted for the predicted risk of the patients undergoing surgery by each hospital or surgeon
4. This line represents the ‘standard’, which is the average overall mortality rate in the UK for cardiac surgery over the period of time in question.
5. The highest expected risk adjusted mortality rate is represented by the red dotted line. Graphs for hospitals also show a green dotted line, which represents the lowest mortality rates that we would expect to see.
It is important to remember that the consultant surgeons whose mortality rates are included on this site work as part of a larger clinical team; consisting of anaesthetists, junior medical staff, nurses, perfusionists, pharmacists, and physiotherapists. All of these team members may affect patient outcomes, along with a hospital's facilities.
As a professional society we exist due to the enthusiasm and financial contribution of our members, who are mainly cardiac and thoracic surgeons. The executive committee of the society and the database project are both accountable to our membership. We have analysed all data in the UK National Adult Cardiac Surgery Audit database for all surgery that has taken place in both NHS and private hospitals. As part of our governance process we have looked for mortality rates that are higher than expected and fed that data back to all hospitals and surgeons in line with the methodology described in our publication 'maintaining patients' trust'. We have also informed the Care Quality Commission (the organisational regulator of healthcare) when mortality rates are higher than expected. On this site we have published data for all hospitals that carry out adult cardiac surgery, but we have previously agreed with our membership that hospitals may opt out of publishing at individual surgeon level; they may opt out of publication but cannot opt of out our governance processes. A small number of hospitals have taken up this option. When surgeons have opted out, their 'dots' do not appear on the funnel plots, and no casemix plots or mortality are displayed on the site. We expect that the option to opt out will no longer be available for the next time we publish this data, which is planned for summer 2013.