Department of Health

Operating Theatre Improvement - Data

  • 05 October 2015
  • Duration: 40:37
  • Well, good morning. And first of all, thank you, Dennis, for inviting me to take part in your Operating Theatre Master Class today. My name is Tom Wainwright and I'm recording this at Bournemouth University in the UK. It's about a week before your seminar. We’ve had a great weekend here, we’ve won the Lions test, we've won Wimbledon, and we’re looking forward to hopefully beating Australia in The Ashes as well.

    Now I’d love to have been with you for the sessions today, not just so I can talk about the sporting results that we've had in the last week, but also so that I could interact and hopefully make sure you got the most out of these presentations. That said, I worked with Dennis on the topics for the three presentations that I’ll be doing today, and I very much hope that they add to the experience and add to your learning at the workshop.

    My background, first of all, is actually in physiotherapy, so what you might not think is a natural sort of background to be talking about operating theatre productivity and quality improvement within our operating theatres. In actual fact, I've been working around operating theatres for about the last nine years. In my role as a physiotherapist I took on a secondary role as an orthopaedic clinical researcher. That was working with a surgeon, a guy called Robert Middleton, he’s an orthopaedic hip surgeon and my job was to run international randomised controlled trials within a hospital. Now, we're able to do this, and we’re able to look at different types of hip prostheses, different types of operations, the use of computer navigation, mostly because I figured out a way of working with theatres which didn't disrupt the normal flow of the operating theatre schedule.

    When you're doing trials like that, it’s necessary for you to understand all of the processes that go on within theatres, so I've got a good knowledge of theatre stores, when we’re getting new equipment in, sterilisation, all the way across to bookings, knowing how the lists were compiled and how the waiting list was managed. Also in the operating theatre itself, working with the surgeon closely, observing how you work with the anaesthetist, observing how they did the surgery.

    Since that time around nine years ago, I’ve also managed a large orthopaedic department down here in Bournemouth where remodelling our operating theatres was a central part of what I did in my time in that role. Over the last four years I've been studying for my PhD here at Bournemouth University and also been completing some consultancy work, both the national body such as the NHS Institute on their productive operating theatre, the Department of Health and also individual hospitals, some large teaching institutions and some smaller district general hospitals. In all of these hospitals I’ve been using quality improvement methodologies to help them improve the running of their operating theatres.

    It’s that experience and also some of the theory and some of the findings from my PhD here at Bournemouth that I hope to share with you throughout the sessions today.

    What do the three sessions look like? Well, Dennis has asked me to talk about three things. The first thing is why do we need data? Well, it’s clear that we need data, and it’s clear that there is a lot of data in operating theatres, but whilst we may have lots of data, we often don't have good information. And it’s information which enables us to run our theatres, to make decisions about changes and to make decisions about how we improve things. In actual fact operating theatres are probably the part of any hospital which are the most data-rich, because of the way timings are captured, because of the way patients are logged, is often where we’ve got most data, but it’s also where we’ve got the poorest information, and that’s something that I want to touch upon as we go through the sessions today.

    I'm not going to be talking about any fancy software and saying that’s the answer and that will make your operating theatres more productive and more efficient. That’s not required. Just an understanding of what you’re trying to achieve and the data that’s going to tell you whether you’re achieving that is what’s needed. In the second session then, I’ll then talk about some of the data that I use. Now this will be examples from some of the hospitals and some of the projects that I've completed over the past four years. The data will be anonymised but in each setting I’ll describe why we collected the data, how we analysed it and what we were hoping to do with it. The idea of that session is really to give you lots of ideas and show you lots of examples about how data can be used.

    You must bear in mind that one solution is not going to be the solution for every hospital or in fact your hospital, but the principles are principles that you can apply to your own setting. And then the third session that Dennis has asked me to talk about is about engagement. That’s not just engagement of your clinical staff, your surgeons, your anaesthetists, but also your staff on the shop floor, your theatre staff and scrub staff, your assistants, and also engagement of the executive board, taking them through a learning process of what you're trying to achieve and also some of the issues that prevent you perhaps from being 100% efficient as they would like to see it on their bar graphs in the boardroom.

    So that's what we're going to plan to run through today. Let’s kick off the first session. So why are we talking about data, and this quote was given to me a few years ago, but I think it sums things up nicely. The data you have is not the date you want. The data you want is not the data you need, and the data you need doesn't exist. And I think we’ve all felt like that many times when we’ve been trying to solve a problem and change and make an improvement.

    And this often leads to lots of common questions that I get asked about data. What data do you collect? Say what data, what data capture points do you collect in operating theatres, entry to anaesthetic room, anaesthetic start, anaesthetic finish, entry to surgical room, knife to skin, closure, exit of theatre, entry to recovery, recovery ready for depart and then recovery exit? They’re just some examples of the time points that I’ve used, but it may be different ones in your hospital depending on the set up. But we’ll talk about that as we go through.

    How do you make sure that it’s good quality data? How do you make sure that the operating theatre nurses are entering it in real time? Do you use bar code scanners? Do you, is it someone's job? Do you have computers in theatres, is it done on paper? All of these questions we can answer. Who analyses your data? This is historically a problem because we’ve got the analysts who sit in a building in another part of the hospital and then get asked to churn out information and reports on theatre efficiency. In actual fact you need those analysts to come into the operating theatre. You need them to understand the practical issues you’re facing. You need to bring that to life for them, so then they can apply what they know about the data and how it can be analysed to solve your problems.

    How much time does it take? Well, really it shouldn’t take any more, it shouldn’t take too much time, but we've got to work through that. We've got to get people to understand the value of data information. Then actually the additional couple of seconds it takes to enter some data, they don't mind. How do you use the data you collect? What’s it used for? Is it used for praising staff? Or is it used for just telling them, sort of telling them off when they’ve had an overrun or a cancellation? It’s important that we balance the routine reporting of data with not just looking at it when things go wrong. How do you collect the data? And what data do you collect coming around the full circle. So through the course of today’s session, I hope we’ll be able to answer these sorts of questions.

    So why do we need data, and why do we need to understand how to use data? Well, if we go back to the context of this, and I’m not going to spend much time on this but because you’ll all be aware of it, but theatre productivity must improve. It must improve in all of our hospitals. In the UK we’re especially feeling this due to the economic issues, the fact that we've got less money and we’ve actually got to do more with the less money and the less resources that we’re now provided. We’ve also got the ever-changing demographics and population changes that you’ll also be experiencing in Australia, an aging population needing more surgical procedures. And mixed in with this we've also got the surgical and technological innovations which mean that there’s new types of surgery, so we're increasing the range of types of surgery that we can offer to patients. So, in short, there's no option, but we have to improve theatre productivity. And the other reason, I suppose we all know, is because we all know in our theatres that there are late starts, there are early finishes, and we could be doing more through the resources that we have.

    It all sounds very easy and sounds like this is the sort of thing that we should be doing, but operating theatres are complex. They’re difficult to manage because they involve lots of different departments converging, they involve lots of different specialties, professions. They’re also busy and they're also at a point where actual care of patients is pretty critical. So operating theatres are complex, but that's actually why a scientific approach is required to help us look through all of this sort of haze, if you like.

    The Institute of Healthcare Optimisation is a body in the USA, and they say that healthcare delivery systems cannot be managed based just on feelings, experience, benchmarking and brainstorming. They take this one step further by posing this question. Which problem is easier to solve? Is it this mathematical equation here? Or is it the question how do we design an effective and efficient operating theatre system? Well, I’m not going to be doing any maths teaching today, and that maths equation is definitely more difficult for me to solve. The thing about the maths equitation is we know we can solve it, we just put in the answer there, and that’s a type of queuing theory formula. But this bottom question, how do we do that? We don't know the answer, and it’s a much harder thing to figure out, and I'm sure that’s something in your teams you are trying to wrestle with at the moment.

    An example of how using data and also information and also the right ways of analysing information are by using queuing theory where it’s appropriate. It can also help us lead to decisions and solutions which are counter intuitive to those that we perhaps would ordinarily jump to and assume. This is another example from the Institute of Healthcare Optimisation, and this is about ICU bed optimisation. But it’s a really nice example about sometimes the answers, when analysed properly with data to provide good information are actually slightly different. It compares two ICUs with the same patient acuity. The first one has five beds and an average length of stay of 2.5 days, and the second one had ten beds and the same average length of stay. The patient demand rate for the one half the size is half the demand rate, one patient per day, and for ten beds it’s two patients per day. The question is, do they have the same waiting times to be admitted to these units?

    Now, you probably say or probably know, because I’m posing this as a question, but when I first looked at this, I thought, well, yeah, you’ve got double the beds, you’ve got double the number of people arriving, so the waiting time should be the same. And these kind of quick decisions based on quick sort of calculations in our head we’re often making, not just in operating theatres, but around the management of a vast range of our health systems.

    The answer to the solution is actually counter intuitive. You see the first ICU, the waiting time is 0.13 days and in the second it’s 0.018 days. That’s actually quite different in terms of a clinical perspective. And it just goes to show that often the answers and the solutions that we should be, and the decisions we should be making about how we change our services, can be actually quite counter intuitive.

    This idea of using evidence-based management, not just for clinical parts of our work, but also for the management decisions we make, is really important. We wouldn’t dream of introducing a new drug without looking at a Cochrane review or a meta-analysis where we’ve looked at stats, where we’ve looked at numbers to check it was the right thing to do for our patients. But we’ll quite often make big decisions about how we run our processes and our systems based on very little sound operations management or management science. This model here from Glaziou in BMJ Quality and Safety shows how the two things are just as important in getting good patient care. Now, good patient care, whether it’s on an individual level for a patient or whether it’s on a population level is really important. It’s about clinical decisions being right, but also the processing system changes being right. So it’s doing the right things, but doing them in the right way. And really, we need to have both of these things, and when we talk about operating theatres, we talk a lot about process and system changes, but how often do we apply the appropriate amount of management science to enable that to happen.

    Quality improvement of, applying management science operation, research and management science to enable us to make improvements to our healthcare is really important and quality improvement within the healthcare setting has been looked at and I'm sure many of you would have been involved in quality improvement projects. The problem we have with quality improvement is that there’s a real publication bias. People never write up things when they go wrong. They just tend to write up when they go well. And this is actually a pretty good systematic review of looking at the best quality, quality improvement articles that have been written, and they looked at which were the key factors for quality improvement success. There’s four things, and they don’t come as any surprise. Leadership both at an executive level, at a clinical level and on the shop floor is extremely important. Organisational culture is also important, the contents, the political will and staff engagement to change, and we’ll talk about that in the third session. The third and fourth are really important – having a data infrastructure and an information system that will enable you to make the changes, and also an experience of quality improvement, and that any of the quality improvement methods or methodologies that you look at will all have data and management of data at their heart.

    And they have it at their heart because whilst every improvement involves change, not all changes are improvements. So you may for instance be doing a project to improve the start times in theatres of the morning. All agree that actually this is a vital part of the work that you should be doing. But improving start times will only be a benefit if you’ve got appropriate measures in place which have shown that that's helping you to improve not just the throughput, i.e. the number of cases you put on this and are able to do, but also avoiding cancellations. So you may well get better numbers and better dates to say, yeah, we're starting on time a lot more often, but unless that has resulted in less cancellations for one example, or increased throughput through that same operating session day, does it really make any difference? If you don’t do any extra cases in that list and you don't get, and you’re not having any less cancellations, then has it made a difference? And that’s because implementing a new process or pathway is not a goal as such, but only a way to achieve a goal. And it’s quite important to think about that.

    In theatres we also need useful data, especially when we’re talking about quality improvement. This is an example of a typical theatre report and theatre reports like this are useful. I'm not knocking them, but for the kind of work that we want to do and you want to do about really improving things, this data is not good enough. It may be adequate for a sort of look at the board level, but that’s it actually telling us? Well, it’s looking at the number of delayed starts, not the number, but how much that is, those delayed starts have cost operating time-wise in hours, and none of those early finishes. This is a hospital I was working in back in 2010/11, a large tertiary referral centre in the UK. You can see each month significant amounts of time were being wasted with late starts and early finishes. This kind of aggregated data is very difficult for us to interpret, and very difficult for us to make decisions about. It may be that in October these 100 hours here was actually half an hour split over 200 different operating sessions.

    Now would that half an hour of each of those sessions, if we'd used it, would we have, would that have been enough time to get another case in? Would it have been enough time to actually have done something useful with? It also doesn't tell us which lists they are, whether it’s afternoon lists, whether it's morning lists. Here with the early finishes it doesn't tell us was there an early finish with enough time left to do something? Was that early finish on the same day as a late start? Was there any early starts? What happened with the, how many late finishes were there? It doesn't give us the level of detail that really enables us to monitor a process or change something.

    And that table that I’ve just shown you can often be better turned into a table such as this one. Again, this was this particular hospital, and these were the reports that were already in place before I started working with them. And this looked at the financial impact of the late starts and early finishes. Now, I'm not saying this isn’t useful, but the costs were based on a 1 200 per hour as per doctor theatre programme guidance, so they're not true costs within the hospital. The second bit is that they’re aggregated costs, so you aggregate the number of hours off and then you multiply that by that number. So we don't know, this isn’t the true financial impact of these starts. So what I’m saying is, this is useful at a top-level for monitoring trends perhaps, but it’s not the kind of data that we’re going to want at that shop floor level when we’re trying to improve processes. We've got to get past this kind of data and the data I showed you before to a more sophisticated level of data.

    Doing that, it’s easiest if we take quality improvement down to three basic questions. What do you want to accomplish? So you’re all stuck here and you want to be more efficient in your operating theatres. But what is that? What does that actually mean? Do you want your theatre utilisation score to go from 80% to 90%? Is that your goal? And if it is, is that an important goal? Or it is something more tangible in that you would like to increase your throughput of surgical cases from 300 to 350 every week? Or is it that you would like to reduce your number of cancellations of cases of the last one on the list from, I don’t know, four or five a week to zero a week, because that’s poor patient experience, as well as a cost to the hospital? What is the actual thing that you want to accomplish? And I would steer away from increasing theatre utilisation. Most of the time it’s about improving and increasing throughput. That’s what we’re after.

    By what method will you accomplished your objective? So if you want to increase throughput of theatres, let’s give that as the example, how are we going to do that? It may be that we have a number of different things, starting a list on time, planning the list better, so providing our admissions clerks with accurate estimated times or even better accurate records of times of how long cases take, so that then they can plan the lists. It may be also that we want to change our staffing, so that we have overlapping staffing shifts, so that if the list does run over because of an unavoidable consequence, we've got the staff in there to cover it, rather than that case being cancelled because all the staff had to go home. It may be changing how you stream cases, so how you dedicated trauma lists as opposed to just mixing them in with elective cases. Lots of different methods, but it's all going back to this, what do you want to accomplish? How will you know when you've accomplished your objective? You’ll only know that if you've got measures in place. Of all these different parts and components that you’re seeking to look at, and then also of your overall goal, looking at that group.

    So once you’ve sort of answered those questions, one of the ways that you can start to understand where you need to focus your work, is by choosing a quality improvement methodology. Now there are lots of different ones that you could use. You may be familiar with lean or Six Sigma, total quality management or perhaps the IHIPDSA cycles. All of them are valid. I use a system that I’m working on in my PhD called the management of variability, and all of the quality improvement models look at reducing variability and standardisation as one of their key tenets. This guy evidently was involved in industry in Japan, a manufacturing industry after the Second World War, and he said own studied variation is the most important aspect when trying to understand the system.

    The only thing is, how do we go about this in the operating theatre department? We’re not in the manufacturing industry. We're not making boxes or widgets. The variation is all around us in healthcare, so what I do is that I classify into two types of variation – what we call natural variation and what we call artificial variation. Now, natural variation is the difference in you and I, different heights, weights, acuity of symptoms, different skills, different motivations, all of those natural things that we can’t control and we never will be able to control. The second thing is artificial variation. These are the rules, these are the processes, these are the way that we do things, the rules that we’ve introduced. We can affect these. There may be things such as Mr So-and-so only has this type of patient on this day, when the sun is shining and the wind is in a certain direction. These are the things we can change, but you need to distinguish between different types of variation if you're going to be successful in improving your processes.

    So, just going a bit further, natural variation is an inevitable feature of healthcare systems. Sources of natural variation includes, or may include differences in symptoms in diseases that patients present with, the times of day that emergency patients arrive, for one example, the socioeconomic or demographic differences between patients, the staff skills, motivation, etc, whereas artificial variation is created by the way a system is set up and managed. This may include the way we schedule services, the working hours and shift patterns, the way staff leave is planned, the order in which we see and treat patients and the processes we design, the steps we decide to employ.

    What is a good thing to do is to look at either a cause and effect diagram or a driver diagram, and identify sources of variation. This is a slide for the NHS Institute, talking about artificial and natural variation, and they say 20% of the variation is natural, patients and staff, and can be managed, whereas 80% of the variation comes from the system. This is the artificial variation and should be eliminated. So in here the orange and black didn’t come out too well, but our processes are unclear. The guidelines differ. There's different ways of setting up things for anaesthetics and complications. These are process issues that we can control, the kit and instrumentation, that again are things that we control. We can eliminate the artificial variation in these aspects.

    So let’s apply this to an operating theatre example, and this is an example that I’ve used and this is why we need data. We basically that, we wanted to reduce cased-on variability, and this is in a joint replacement centre, so this is orthopaedics looking at hip and knee replacement. And we wanted to reduce the variability that cases took. That was so that we could more accurately schedule the right number of cases per list, and also so that we could get more throughput. We didn't want to ask surgeons to operate quicker. We didn’t think that was, well, clinically the right thing to do or too, most importantly, clinically necessary, because actually working with the surgeons and observing surgeons, we actually saw that actually what the surgeons did in theatre was pretty repetitive, was pretty standardised. It varied slightly for different sizes of patients and slightly more, and for the more complex cases there was more advanced techniques required, but actually the processes of surgery are extremely methodical. The things that were causing that variability to increase were the different staff members in theatre, the time it took to go and get a prosthesis when we know what size was required, whether equipment was there, ready and checked. All of these different aspects.

    So what we did was that we went through the process of doing a cause-and-effect diagram, like the one I’ve just shown you, and drive diagram, and we classified the variability, so we decided the project day, i.e. we’re willing to reduce the case time variability for hip and knee replacements, we identified the causes of variability contributing to the current outcomes. Some of those were natural things such as different surgeons, different levels of experience of surgeons, different patients, different complexity of the surgery. Some of those were artificial i.e. the way the theatre was set up, the types of instrumentation or prostheses we used, the type of anaesthetic that was used, etc, etc. We classified the things that we could control and the things we couldn’t as artificial or natural and then we measured those individual causes of variability where possible. We then set about eliminating all of the artificial variation, and we implemented those process changes.

    The result was quite significant and I excuse, this is a very busy slide but we've virtually got over a thousand cases before and after the change and I don’t know why the arrow bars haven’t come out correctly for this second portion, but what you can see is that we got a narrowing of that variability and time, so the total case time here is around about the average and round about sort of two hours, that’s including the anaesthetic time and the surgical time and the sort of non-time we’re in theatre. But we can definitely see that variability has reduced quite significantly.

    Now, part of that will be about data accuracy and ensuring that we've got the right cases coded, but most of it was about reducing those artificial process stats. So when I say that, what did I mean? Well we had sixteen different types of hip and knee prostheses on the shelf. We rationalised down that to three or four. That enabled us for our staff to become much more confident with those sets and instrumentation, much more slick at using them. It also meant that we could have more sets of instrumentation, because we didn’t have to have the storage for all the different type of prostheses.

    It also meant that when we were listing patients, we knew what kind of prosthesis we were going to be using two weeks before the surgery, it didn't change on the day. And because of that, we could get all our kits ready at least 48 hours in advance, not the day before surgery, but two days before surgery. That’s because if we were checking it on the afternoon before surgery, there's not much time to make changes.

    But again, this was all because we removed that variability of having fifteen different kits and also of not knowing exactly which prosthesis the surgeon was going to use on that day. We’ve also introduced checklists, and you’ll know from things like the WHO safety checklist that these are really important, and also very good measures of ensuring standardisation. You can see here this is the kit check. We’ve got clearly defined shelves for where things should be.

    Other examples were that in the eight operating theatres in the hospital, each anaesthetic room was laid out differently. There was different things in the cupboards, different levels of stock, the trays were laid out differently. We talked to the anaesthetic nurses, the anaesthetist, asked them what they want, how they would want it presented, and all the rooms are now identical. The good thing about this is it makes it a lot easier to spot when things are not there, so you’ve not got the patient anaesthetised and then they have to, someone has to run and get something. Everything is always there, it’s where it should be, and it speeds the process up.

    The operating theatres are laid out in exactly the same way. We asked the scrub nurses what’s the best way of lining up all of the trays and all of the tables for these prostheses? What is the best way to train people to do this? They came up with an example and they tinkered with it, it’s changed slightly over time, but what is consistent is that every time it is done the same way in every theatre. This is absolutely essential because these again you may say this is just saving a couple of minutes for a case, but it’s by standardising and reducing the possibility for error that you prevent those things being missing on that one case which can lead to a half an hour or one hour delay in theatres.

    The other thing about rationalising the amount of prostheses you use, you need a lot less options and stock options in your store cupboards, so therefore we're able to organise all of our stock much more easily. This also meant that when someone was having an operation, quite often someone would go away for five or ten minutes in the surgery saying well, where are they? Where are the carts, where are the stems for this patient? And that's because they couldn't find them. Again this is just about eliminating that artificial complexity and variation we introduced to our theatres.

    These all sound very basic things, but you can’t move onto the more complex profiling capacity and demand modelling unless you have got the basics done well. The future of operating planning is using past performance to help you plan your future performance, because you know your capabilities, so using past performance data on timings to help you plan an op and schedule cases effectively. But if you're trying to do that with a large amount of variation in those case times, you’re going to find it very difficult, so reducing the variation of case times will help you not only in the short-term but also in the long-term as you want to do more sophisticated things with the data.

    What we also see here is the effect of case time variability, and this is on a middle grade surgeon, and we’ve got, this is some of the data I’ve just shown you, but broken down per surgeon here. You can see again this narrowing of the variability. The actual average time it’s taken hasn’t changed much at all, but that variability has. That makes scheduling much more easy. It makes the likelihood of having an overrun much less because you haven't got these longer cases.

    We found the same thing in a really experienced knee consultant with over 25 years of operating theatre experience. This is a guy who’s done thousands of knee replacements. Again, you see this narrowing of the variability, so he's not got quicker, the average actually goes slightly by maybe a minute upwards after the change. But what we see is the narrowing of this variability. Again, what this meant for this surgeon was that we could routinely know that we could book an extra case on his list because his operating theatre and the times for his cases were so reproducible.

    The graphs we were just looking at were looking at the total case time. I’m just showing you here 800 cases after we made that change. This is a very busy graph, but I think it just shows how stable processes can actually be. This line here represents the total case time, and these three lines here are constituents, so if you stack them on the top of each other they would equal the purple. This is the total anaesthetic time, this is the total op time. So op time is around about an hour, anaesthetic time is about half an hour. This green is the non-time, the prepping, the draping, there isn’t a knife to skin, and there’s isn’t the anaesthetic start and the anaesthetic finish.

    What you can see is actually the non-time is the smallest component of time and should therefore probably have the less variation time, but there’s very similar variation between those three constituents. And that’s important because quite often in operating theatres an anaesthetist says it’s the surgeon’s fault, the surgeon says it’s the operating theatre staff’s fault, and everyone sort of says, well, we finished late or we didn’t have to cancel a case because of someone else. So being able to break it down and produce charts like I’ve just shown you on those individual bases is also an important thing to do.

    So, this brings to the end the first sort of session that I’m going be doing for you today, and whilst we haven’t gone into lots of examples of data, I hope it’s set the scene and really solved the relationship between quality improvement work and the need for good data. Improving theatre productivity should be regarded as an ongoing quality improvement activity. Successful quality improvement methods are underpinned by using not just the quality improved methodology, whatever it is, lean, Six Sigma or PDSA and supporting this with data turned into information. Managing an operating theatre without the information, for me, is not an option. Moving beyond the data and those graphs that I showed you, looking at aggregated number of hours lost or minutes, we need to move to these more SBC charts, run charts and process charts, and we’ll look at some of these examples in the next session.

Tom Wainwright, UK improvement specialist, on why operating theatre data is important.


NSW Agency for Clinical Innovation.

Reviewed 05 October 2015


Was this page helpful?