Audit of clinical activity

This section of the web site posts an audit of my in patient clinical activity since August 1998 when I started to collect audit data on my patients. Before then I relied on the hospital information systems to provide me with feedback on my activity. This was unsatisfactory for a number of reasons. The data was not always accurate, it was incomplete and it was not strictly relevant. For example no information was available on surgical complications including death rates.

So why post this information now, when patients can access information on different surgeons? - see Dr Foster for example (http://www.drfoster.co.uk/). Nowadays a patient is protected by the fact that complications rates are registered with the GMC. Surely 'bad eggs' will be weeded out by the checks and balances now in place in the system. I take the view that a patient has the right to information on a surgeon’s activity. ‘Have you done this operation before?’ ‘How often?’ ‘What are your results?’ ‘What is your complication rate?’ All these are very valid questions and ones which I would ask (and have asked) if I or my own were having surgery. This information should now be available.

However if patients are going to have this information I would like it to be accurate. Hospital information systems remain inaccurate. The best we have achieved to date is a 30% correlation between the hospital information systems and my own data although the absolute numbers of complications correlate pretty well. Since it is the hospital data that is posted with organisations like Dr Foster and the GMC it is all the more important that somewhere there is access to what I regard is a more accurate reflection of my surgical activity.

So why is there this difference between the hospital generated data and my own personal data? The difference lies in the methods of data collection. Hospital data at Frenchay hospital is collected by audit clerks. In practice this means one or two key individuals going through notes after a patient’s discharge to code a patient’s admission. The data is then added to the hospital’s information systems. Our coders have become very good at the job and get some of the data exactly right e.g. overall numbers of admissions to hospital and the reason for hospital admission. However the system does not audit my activity at other hospitals. Coders are not surgeons (or even medical) so they cannot be expected to recognise surgical complications or if they do, apply the appropriate weighting to a particular complication. There is the added difficulty that their assessment is retrospective.

I collect my data prospectively i.e. as I go along. I use a Toshiba Laptop which I carry with me at all times.  Data on all my patients is added to Excel spreadsheets. This method of data collection is directly relevant to my practice. I detect all the complications that are relevant to me as a surgeon i.e. all direct complications of surgery and those non surgical complications that have influenced a patient’s stay in hospital. I do not record minor non surgical complications that do not influence a patient’s hospital stay. The hospital data records some admissions as complications of surgery when this is not the case e.g. patients with a diagnosis of hydrocephalus are often recorded erroneously as complications of shunt surgery. (All attempts to change this anomaly in the data collection system have been resisted for reasons that are difficult to define) My view is that this data compliments the hospital generated data and that both are needed. The hospital data is independent but is retrospective but not strictly relevant to surgery. My data is prospective but is biased towards my view of my practice. This does not constitute ‘cooking the books’, the convenient political view of a surgeon’s own audit but is simply a reflection of a data collection system that is relevant to my day to day practice.

I need to have accurate data to hand, especially if this is less than optimal. I am asked to do twice the amount of work that is recommended as safe by the profession and I am asked to do so with resources that are often inadequate to the task. I am constantly having to tailor the work I do to the resources available to me. I therefore need to monitor my activity very closely. If my complication rates were to change then I would need to know about it as early as possible.

The results reflect the environment I work in. The point of audit is that it is a cycle. Improving my practice will require investment, allowing me to do less cases in a ring fenced environment with proper opportunities for training, continuing professional development and service development. This is the political battle we all have to face. In the meantime I feel it is my duty to be honest about what I can achieve at present.

‘How do you compare with other surgeons?’ Finally I feel I should address this question. The best comparative data between surgeons probably comes from hospital generated databases, bearing in mind that the inaccuracies that I have described above apply to everyone in the department equally. At present there is no uniformity in the way different hospitals collect data so there is no way of comparing surgeons from different hospitals.

At the end of the day I believe one can try to make too much of a science out of this. ‘Do you like your surgeon?’ ‘Do you trust them?’ ‘Have they answered you questions and addressed you concerns thoroughly?’ ‘Do they object or facilitate your request for a second opinion if you so wish?’ ‘Are they confident they have the resources to take your case on?’ The final decision about whom you should let operate on you is often a subjective one. However if you have read this far it will because you want to find hard data on my activity so read on. The raw data is presented in the excel file (click here).

Audit Summary - (August 1998 to April 2008)

I admit in the region of 443 patients per year.  100% of those admissions are now NHS as I have recently stopped operating inthe private sector. 14% of my activity is classified as routine. 47% of my admissions are emergencies and 34% are urgent admissions. 30% of admissions do not undergo surgery. 42% of all admissions undergo cranial interventions, while 29% have spinal surgery. I operate on an average of three glioma patients a month and two meningioma patients a month. I do one cranial epilepsy operation every month and two VNS operations a month. On the spinal side I do an average of three cervical operations and four lumbar operations per month.

Benchmark Audit statistics

Benchmarks statistics for surgical expertise include mortality data, data on patients made worse by surgery, second operation rates, readmission rates and complications. All these benchmarks have to be interpreted appropriately. Mortality varies dramatically with case mix.

 Death rates are very carefully monitored in our department with each case being audited individually and presented at our monthly clinical audit meeting. The overall death rate under my care is 2.5%. The death rate in patients undergoing surgery was 2.8%. Of these 66% were deemed at audit review to be inevitable in that they occurred in patients presenting in extremis. That leaves 23 patients (0.9%) where their deaths could have been avoided. Two had pulmonary emboli. three developed a postoperative haematoma with subsequent complications, three developed a major stroke after coiling of an aneurysm, five suffered a subarachnoid haemorrhage during coiling of their aneurysm, two developed postoperative status epilepticus from which they never recovered, three died from the complications of infection, three suffered multiorgan failure, one patient had an allergic reaction to an anaesthetic drug while one patient had a haemorrhage from his tracheostomy that led to his death. 22 of these cases were urgent or emergency admissions. Only one was a routine admission.

Patients made worse by surgery. 34 patients were made worse by their surgery (2%) 17 patients undergoing glioma surgery had a worse neurological deficit after their surgery (6%). Five patients had neurological complications after extrinsic tumour surgery (2%) Six patients were worse after cervical surgery (2%) Three patients were worse after spinal cord surgery (1%) One patient suffered a stroke after aneurysm surgery (6%) and two were worse after coiling. One patient developed worse pain after injection treatment for trigeminal neuralgia. One patietn had a stroke after her shunt was removed

Other measures of surgical expertise include second operation rate and readmission rates within one month. The crude data on this is not relevant as approximately half the cases were planned. However the complication related second operation rate is 4% and the hospital readmission rate is 4%. I have no idea what should be considered benchmark figures for these statistics.

Surgical infection rates are an important benchmark for a number of reasons. Morbidity from infection is major, prolonging hospital stay and in some cases reversing the benefits of surgery completely. MRSA (hospital 'superbug' infection is life threatening. The established wisdom is that elective surgery done in dedicated facilities is subject to far less infection than elective surgery done using the same resources (beds and theatres) as emergency surgery - the situation at Frenchay. In my patients, the overall infection rate was 2.6%. This is higher than I would like. I would aim for an overall infection rate of <2%. For elective surgery this should be <1%. Interestingly the infection rate for non emergency spinal surgery in this series was 0.8%. (1.7% for emergency spinal surgery). This data is reassuring in the sense that it is proves that, in terms of surgical copmplications elective spine patients are not being put at greater risk by being operated on in an acute / emergency environment as has been suggested in some quarters. In fact the only documented case that I have of MRSA infection in an elective spinal case was in a patient operated on in the private sector (This case resulted in a major review of infection control policy at the Bath Clinic). An extensive review of this case was carried out to establish the cause and we discovered that the patient was a carrier of MRSA organisms acquired in the community.

The situation for implant surgery is similiar with a slightly higher infection rate. A total of 759 procedures were carried out where a surgical prosthesis was implanted. This included the following procedures:

• 219 spinal cord stimulators,
• 20 motor cortex stimulators,
• 18 intrathecal pumps,
• 126 ventricular shunt procedures,
• 47 permanent ventricular access devices,
• 107 spinal implants,
• 197 vagus nerve stimulators
• 15 cranioplasty procedures
• 10 patients who had intracavity chemotherapy (Gliadel wafers).

The overall implant infection rate was 2.8%.

The overall surgical complication rate was 10%. While this in line with other published independent audit studies, it is too high. Improving this will be difficult with a radical rethink of the way surgical care is delivered, in particular ring fencing separate resources for elective and emergency care.