Department of Health

Operating Theatre Improvement - What data do you need?

  • 05 October 2015
  • Duration: 44:04
  • Well, welcome to the second session. In this session I’m going to be talking about the data that I use, and this is the data that I’ve used over my time as an external consultant, helping different hospitals improve outcomes and productivity within their operating theatres. Also the data that I have used routinely when managing my own orthopaedic unit and also some of the data that I’m looking at and exploring around variability within my academic studies in PhD.

    So in outline, this particular session. The first rule I’m going to talk about accurate data, understanding how it is collected, collated and analysed. This is so, so important. Anything that we do with our data and our information is completely devalued and completely undermined unless we are absolutely sure that the data is of good quality. The quality of data in will affect the quality of data out, and subsequently the information that you can glean. Lots of reports are churned out, and I'm going to show you a few examples about where data errors or missing data can have such a big effect on how we interpret those graphs that we see at board meetings and at departmental meetings about theatre utilisation.

    The second point I’m going to talk about the importance of understanding capacity and demand. We haven’t got time to do a full workshop on capacity demand, that’s a day in itself, but really optimising throughput in our theatres is about understanding the demand that we’re seeking to address with our current capacity and utilising the capacity and demand so that we don't get mismatches, so that we use our capacity to the best effect.

    I’m going to give you some examples of how you can present referral data and demand data and also give you some examples of how you can present data to explore and understand your current capacity. Then I'm going to look at how you create your own data for smaller improvement projects. These may be things looking at specific elements of the processes within theatre. I'm going to look at how I analyse start and finish times, and how I believe that we should be aiming for a set time, that we should be aiming for a window of time, a range within which we can expect a list to start or finish. And the last point I'm just going to look and touch on how understanding variability of case times can perhaps lead to ways of scheduling theatre lists that may be slightly counter intuitive in terms of what would maximise throughput.

    So, the following slides are going to be talking from a, talking about a project I did, and this is a large, tertiary centre in the UK. This is back in 2011 and I was asked to help as part of a team to improve the productivity and throughput within the theatres. So the first thing we were looking to do was some diagnostics, looking at the data. In the sort of three month period that we looked at there were fourteen thousand procedures. The theatre times were recorded in full for just seven thousand and forty patients, round about 50%. Frankly for a leading centre within the UK, this was not sufficient, and it made our jobs of actually providing an accurate diagnostic extremely difficult.

    Now what’s that percentage in your hospital? The hospital where I was a manager, it was 100%. And I know it was, because I checked it every week. The good thing about that was that all the staff know that I’ve checked it, so they would complete that data on an ongoing basis. They also knew that I used that data and presented it back to them and made it helpful to them, so there was a vested interest in why they entered that data. But this figure, 50%, what’s your figure in your hospitals? Do you know? Have you understood how date is collected? Have you really got down to the nitty-gritty?

    The lack of data severely limited what I wanted to do. I wanted to get a quick snapshot of some of the issues that I was hearing from the interviews with members of staff in an operating theatre. I wanted some data to send to check whether some of the problems they were reporting were real problems or whether they were perhaps more hearsay. So I went to look at start and finish time variability. This may have included looking at a specific surgeon and his analysis, so say on a normal day, Tuesday, list for Mr X in theatre 1, what were the range of start times and finishing times each week over those months?

    Let’s look at a day and week analysis. I went to look at late start times due to spinal MBT on Tuesdays or Wednesdays. This was something that was being flagged up as a problem, but because of the lack of accurate data I couldn’t draw any conclusions to corroborate whether the stories and whether what the staff were telling me were actually true. I also wanted to look at turnaround time. A lot of the staff were saying when they did Saturday lists, additional lists, turnaround times were a lot quicker, a lot slicker, and they got a lot more done, and they wanted to work like they did on Saturdays in the middle of the week. This, I’m sure, will be something that chimes familiar with all of you, but because of a lack of data, I couldn’t compare that, and I couldn’t compare across teams, surgeons, or procedures with different specialties either. So unless you’ve got data, I can talk about all the data and how I use it, but step one, the most fundamental step that you could do, was ensure that you've got 100% accurate, timely data capture and completeness.

    We'll just talk through some of this data in this particular example. Here I’ve got some commentary, and I’ve provided the commentary in quite a lot of detail for these slides, so that you can look back and read it, and I’ll provide the slide decks to you. There’s a lot of information to take in. So in the first three months of this year only 54% of patients had theatre time. It’s a slightly different time period from the one I showed in the slide before. We need to have data into every patient if we are to look at start times, finish times and turnaround times accurately. So you can see here, for each theatre, the number of recorded procedures, the known data to calculate the op time. And these are large amounts of patients where we couldn’t tell you.

    And this has implications, because accurate operations not only enable us to run on this better and to look back at it, it was just to plan them, because we were wanting to use this data to advise them how to plan operating lists in the future. We could take a certain procedure code in a certain type of operation and then look back at the last 50 procedures of that time, and then provide an average or perhaps a mean or a standard deviation around that average, for them to then plot what is going to make an appropriate list going forward. Can we get the right number of cases on it?

    The next slide that’s, it’s some examples of why we need complete data sets. Missing data will skew calculations, queries and look-ups. You may think I’m labouring this point, but all of your who, I'm sure, have theatre software, that software will have drill downs and provide loads of graphs and bar charts and tables that either your managers or yourselves may use, and it will often present data in very certain ways, and we’ll look at an example of that in a moment. But when you drill behind it, it presents data in these certain ways about overruns, late starts and early finishes, but it doesn't tell you about how much data was missing, or how it calculated this data. So this list is a good example, so this is one list on one day in one theatre. Tuesday all day list with Mr X.

    So when did the last case finish on this list? Was it 14:25? Meaning a very early finish? But it’s impossible to determine that from the data, because the data analysis that many of our theatre systems will use will just look at that day and it will assume the latest time on each list is the last case. That will then skew the information. And this is a common problem I find in hospitals, especially the theatre operating systems and analysis. You’ve got to ensure you know the data behind all these drill down tables and graphs that you're presenting and you’re make decisions from.

    Again, this is another example, missing data will skew calculations, queries and look-ups. This time it’s start times. We’ve just looked at finish times which are a big thing in theatres, this time it’s start times. So this is Prof, this is his theatre session, theatre 6, all day session, and it’s 10:40. So when did the first case start on this list? Was it 10:40 because he was late doing other things, having a teaching session, MBT meeting or something else which gets sort of told to me in interviews? Or was it one of these missing data sets? He may have done two cases already before 10:40? It is possible to determine from the data? And the data analysis on these theatre operating systems will have assumed that the earliest time on each list will be the first case. So if that list was due to start at eight o’clock, when that data is fed forward in your theatre management suite and operating system, it will show you that we had a late start by two hours forty minutes. That will then get aggregated up to the monthly totals and we’ll have two hours forty minutes potentially of unused time which we’re then looking at, but we don't know for certain, because we haven’t got the data.

    Again, this then looked at March 2011, and these are examples of the level of detail that I would want to be looking at when I’m looking at theatre data. The key thing here is, I also need to understand locally what your staff understands by anaesthetic starts or anaesthetic finish or time into theatre. So having these time entry points and having them collected 100% of the time is good, but you need to ensure that everyone knows what they mean. So time of arrival, let’s be specific, time of arrival, for me that’s time of arrival into the anaesthetic room. Anaesthetic start is when the anaesthetic starts. These definitions can change but they need to be agreed locally. Time into theatre is when they’re wheeled through from the anaesthetic room into the theatre. The operation start, that is the knife to skin, that’s the first cut of the operation. The operation end is closure. And I’m labouring this, it doesn't really matter about the definitions that I use. The important thing is the definitions you have are locally defined and that all of your operating staff and all the people entering the data understand them and share the same understanding. Really important.

    The second thing is, is it entered in real time? There’s often a computer in the theatre these days and it can be entered in a matter of seconds by one of the ODAs who’s circulating and doing the running in theatres. So not only do we need that for starting, looking at late starts and early finishes, but if you want to look at actually detail, granular detail about anaesthetic time and about operation time and surgical time, we need to ensure that we’ve got 100% of all these data fields captured. You can see here it really is quite poor in this particular hospital I was working at. Again, I’d ask you what’s that in your hospitals?

    So going forward I was talking about these graphs that the theatre data management suites sort of bring out. These ones here, were they late starts or were they due to missing data? You'd look at that in a boardroom and say well, they’re late starts, or they didn’t, but actually, given the data that we know about, that could well be due to missing data. The only way to check this is line by line, and so for one week in March I did that, and only 21% of this had fill-up breaking times for every page, so it’s actually imperative. I’ve spent about five minutes talking about this now, but anything we do from here on in isn't worth a grain of salt unless we know the data is important. Sorry, unless we know the data is accurate, and we also understand the importance of that. You must question the information produced by these theatre report packages. You know, if you’ve got a theatre report package and it’s producing you data like this, where is the data being pulled from? It must be the same sources as the data I’ve just been describing when we’ve got all those data errors. If so we need to be very careful with the conclusions that we’re drawing.

    This was a particular situation at this hospital and all of that data I’ve just shown you with all of that, sort of, all of those errors and all of those missing fields, was actually presented in this graph, and this was a theatre management suite called Insight, and I’m not picking on Insight by any means, but this is just an example. And here it says theatre time completed, anaesthetic time completed, recovery time completed. So here it’s saying that there’s 95% completeness for theatre data. Now is that per list? Is that per patient? I know that there’s 50% of patients where we didn’t have any data, so we’ve got to be really careful about understanding these management suites that we all rely on for our data. We really have to understand the nuts-and-bolts of where they get the data from, and how that data is processed and analysed to produce these kinds of graphs.

    For me, the best way is to get the raw data in an Excel file downloaded from your hospital pad system and do your own analysis with things such as SBC charts and your own formulas. You then understand the process with which that data is gone through. Much, much more safe and also can be much more specific to the problems that you're trying to solve. These packages tend to point you towards presenting data in a way it’s set up to present. That made not enable you to answer the question that you’re trying to solve within your theatre.

    I'm now going to talk a bit about capacity and demand, and as I mentioned just before, this is a big subject. But what I've done is I’ve provided Dennis with an Excel model, and that’s, he's got that and I’ve given it to him and he could make that available to you if you want, and it’s something that I’ve used with a lot of trust and I’ve shared with a lot of NHS colleagues, and it was produced by the NHS Institute to sort of help other people. And really, what it does is, it looks at the capacity gap, it looks at the gap between your current capacity and the demand that you have. And this was a big issue in the UK because we’ve had, not only did the increasing demands due to changes in population, but we had an increasing demand because of a government-led initiative to reduce waiting times. So, going back a few years, we had a big effort to understand this gap between demand and the capacity that we currently have. The model is used extensively as a diagnostic and planning tool and it helps to bring clarity around the complex curing behaviour and relatively inflexible capacity built into operating theatre management systems. So it really is sort of a self-build.

    What it does is six steps. You want to define the modelling period, you then want to estimate the capacity required during the modelling period to achieve a balance with your demand, and then you need to estimate the available operating capacity during that modelling period. It then, when you enter all the core data so that it can do this, estimates the backlog reduction required within the modelling period to achieve the target maximum wait time. Step 5 then estimates the gap between your chosen scenarios by relating to variations in demand, perhaps doing less management. And then six is really the last bit closing the gap. What you could do there is you can effectively choose different options of either increasing your capacity or increasing your productivity in order to reduce demand. It can give you a few scenarios with which you can put your data in and play out what you’d have to do. So you use your current local data in the model to then understand what you're going to have to do to balance your forward demand for your services. It’s a really useful tool. I’ve provided it as an Excel file to Dennis. I’m sure he can share it with you. It’s got the instructions within that. This is a page from the Excel model. It’s giving you clear instructions on the left hand side, and what it does, it seeks to recommend a sustainable waiting list size for each queue within the system.

    I’m not going to talk too much about the detail of that. That’s the kind of thing that you do on an individual unit bases and if Dennis is working with your particular hospital, that's the kind of thing he can do in the field with you. But the two key elements of this model are capacity and demand, and I’m going to talk and give, and show you a few examples of how you present and understand your capacity and understand your demand locally.

    So, understanding demand. I'm going to show you some examples, or exemplars of analytical methods to look at referrals, the balance, i.e. the additions to the waiting list each week, understanding your urgent versus emergency demand in certain specialities, the backlog, how to look at reducing your total waiting list size and then looking at your planned or readmissions commitment and your waiting time targets.

    So this is a very simple wave that I look on a week to week basis at referrals. Again this is really useful at diagnostic stages to understand how things go, but also when I was managing a unit, this is the kind of thing that I’d look at on a weekly basis to look at the ongoing trend. All demand for the operating theatre capacity originates in the referral of numerous individual patients to name consultants or specialties and then some of those will then go on to require surgery and some of those a series of operations. So our hospitals receive referrals, and I’m talking mainly about elected care, but this kind of modelling could also be done for non-elective care.

    So here, just really simply, you can see there are some changes and no prizes for guessing, this is the sort of week of Christmas here, but you can see at the end here going into the New Year, you can see that there are eight consecutive above-average weeks. Now, this is what we call a run chart and if you have sort of eight in a row over the average then you know there’s probably a change to that process and that’s not just by luck that you’re getting that change. So that’s an observable trend and change in referral patterns. So why is that interesting? Well that tells me that we're going to have to look if that continues at what we’re supplying capacity wise in the future, otherwise our waiting list is going to go up and we're not going to meet our current waiting times.

    This is an example of how we look at the balance, so the simple run chart of additions on the last page can be converted into bars showing the weekly cohorts, with a snapshot of their current status. Hence those added to the list long ago, i.e. to the left side of the graph have mostly been treated or removed for other reasons. We’ve got the colour block here saying the reds are those that are waiting, the blues, TCR, those have got a “to come in date” or a TCR date. The sort of pale blues are removed, the yellows are offer pending, and these are discharged or omitted. So most of the ones on the left hand side have been discharged.

    Those patients added more recently to the right may have been admitted and discharged quickly, i.e. there may have been an urgent case, or they may have been on a shorter queue, or they actually may have been booked in for their surgery out of turn and put them sort of put in quickly to plug a gap or they’re available as a late cancellation. These profiles you can also do at a consultant level as well as a service level, but it does give you a snapshot and bear in mind in England we work to what’s called the 18 week referral rule so these ones here that are still waiting along here with the reds, I would be questioning why they’re still on my waiting list. What’s happened to those patients? How have we lost those in the system? Again, it’s a really visual way of looking at where everyone is on your waiting list. Simple to do, the kind of thing that I look on a week to week basis when I’m managing a unit. The other thing which would be really useful for you to do as a diagnostic within your own hospitals.

    This here looked at urgent versus emergency demand and this was in a specialist orthopaedic hospital, so this is interesting because what we’re seeing is, as well as we’ve got one queue coming into our hospital, which is all of the referrals and then that then divides into different specialties, other queues, and then within each specialty we then have our different consultants, which is another queue, and the consultants then divide their cases into different queues, the ones that they’re only doing themselves, the ones that their fellow will only do, or the ones that a registrar will only, and then further to that they’ll then say well these are either urgent, routine or whatever in terms of their demand. This is interesting because actually the more queues you introduce to a system the harder it is to manage. When we’re managing these referrals and we’re trying to poke into this, the more rooms that your surgeons and your critical staff put on, the harder it is for your admissions staff to put the pieces of the puzzle together so that you have a good and well-maintained operating list. So a large part of what I do and what I would suggest you to do, is look at all the different queues, so look in to see how many different rules are there about what can go on what list and in what order. The more things you can remove, the easiest it is on a scheduling perspective in order to use your theatres optimally.

    So this was just looking at urgent, soon and routine. Again, I would get away from this. Obviously there will be some cases such as the sarcomas on here, while there will be a lot of urgents, but that’s whereby you may want to have dedicated lists once you understand that demand. This is a similar kind of thing with understanding trauma lists and trauma cases where there'll be some cases in your trauma that you have to see soon, and that’s where you need to perhaps remove the queues in some cases but also put some people in certain queues, such as specialised ruptured neck and femur lists, or other examples.

    This list, this looks at the total elective waiting list by weeks waited from original decision to admit date. And this is very interesting. Just by looking at the backlog, the waiting list in the snapshot can also highlight relative shapes of past demand and booking behaviour and consultant level charts can also be produced to look at this. This is really interesting, so by the number of weeks waited, how many patients are waiting, have a TCR date suspended or offer pending? So you can here we have an 18, we call in England, there’s still patients after 18 weeks that haven’t got a date or haven’t been booked. Again, interestingly on zero one you’ve a lot of patients here that have got their date booked, and there’s still a lot here that have not got their date booked. Now, are these patients being booked out of turn for a good reason, or are they being booked because of some rule that we’ve put in, or just because the systems within your admissions and booking class aren’t robust enough to ensure that everyone gets booked in turn?

    This next slide looks at the total planned list, and this is very sort of top-level. High levels of priority are given to urgent and 18-week referral patients, so schedules quite often struggle to book planned patients who are already in the system. So this is where you could look at this example in terms of your trauma and elective list, so when you have to balance that, if you’ve got mixed lists, how would you put the urgent trauma cases in with elective cases that have already been booked? And this is where putting dedicated trauma list against dedicated elective list is really important. Here again with your weighted TCR, booked, suspended or offer pending, but this is those patients that have planned patients or planned dates.

    So these were some examples of just how you have snapshots in from a diagnostic point of view or a weekly point of view, of how you can look at demand. Capacity is also something you can look at and I would recommend it highly, and I’m just going to give you some examples of analytical methods for capacity. This one here is really good. This is week by week and this is again from a project that I’ve completed. And this looks at scheduled hours of operating times per week. And this is for patients with booked TCR dates, so you can see as time goes along here on the side you’ve got elective waiting list patients, planned patients and unbooked time. So you’ve got significant unbooked time as opposed to what the operating schedule is on paper.

    Now, what you don't know is how much of that scheduled operating time is actually available. It may be it doesn’t include weekend sessions or agreed overruns, it doesn't take into account bank holidays or late starts due to another meeting, so getting a diarised assessment of future capacity and whether that can be staffed and whether it’s got the right clinical people available to run a list is absolutely essential. And again, you can look here on a very quick basis about how much time you’ve got booked out. But again this relies on you having accurate data about how long your operations take, so the ideal would be to have all of your patients on your waiting list and all of them to have their procedure code and their name procedure alongside that, and for that particular surgeon you then having a look-up column in your Excel sheet that tells you what the average time is for that surgeon to do that operation. That means each surgeon can have a waiting list, a waiting list not just of numbers of patients, but also a waiting list generated in terms of what is the likely time required in operating theatres to complete that waiting list. That is extremely valuable, because then you can then understand about how this capacity and this demand is either in sync or out of sync.

    This is another way that I look at it and this is a really usable way of analysing a specific list, and I often do this with individual surgeons or surgical teams. Really easy to do, and you just plot it in a really visual way. It’s a good way of engaging with clinical staff and also a good way when you're going through the change process to get them to help comment on why things didn’t run smoothly, to get to sort of some of the bones of what's going on. So really simple. The red bits is essentially gaps either when the theatre isn’t being used, the yellow is anaesthetic time, the green is op time. So you can see here, this list had four patients on again, it’s an orthopaedic list, but it could be applied to anything. And you see in terms of understanding the intensity of utilisation of available theatre time you have to go down to this micro-level in order to almost derive sensible consistent statistics for reflecting the realistic maximum amount of time which can be utilised.

    Now, what’s interesting here is sometimes things such as the bone tumour incision it could be hard to understand how long that operation is going to take. Longer or more complex operations may have a larger degree of variability in how long they take. We never know for sure, so you can always use average times, looking at past data, but another way is to get surgeons to estimate the required operating theatre time for each case. And I’m actually pretty much in favour of this and I think this holds up very well against using average times. Surgeons have a pretty good idea of how long they think something is going to take. Sometimes they may be out because of unforeseen circumstances or complexities, but experienced surgeons having done a lot of cases can normally quite accurately understand this. So getting them to mark down on the TCI card is actually potentially more powerful than just looking at an average from a coding point of view.

    What’s even better is if you present the data back to surgeons so they’ve got their average times, so that they can compare that with what their sort of brain would think and their internal computer would think. Just by feeding it back to them on a weekly basis, monthly basis, they don’t have to say anything more, intuitively they will get better at predicting the times for their surgeries. If they’re then marking that down we can have greater accuracy. Now, if it was that actually this shoulder assessment here, it was a very quick operating time, but actually the surgeon now anticipated that it was going to take longer than that, we need to try and capture that because that then means that that list was appropriately planned because the surgeon was planning for how long it may well take. If it then finishes quicker, well that’s good, because the patient’s got a good result, but we shouldn’t then beat the surgeon up, because to the best of their ability they plan their list appropriately. Clearly here, their op time overrun of 80 minutes, that’s sort of interesting because we’ve got so much lost time in there. So this way of presenting day-to-day visual, I find it works really well using it with surgical teams to get them to come up with ideas about how they could do things better.

    This graph here is also useful because this looks at capacity, but it looks at effective late starts on overruns and cancelled patients due to overruns. And so for each different consultant it looks at the number of late starts, lists with overrun cancels and the number of patients cancelled. So you can see the more late starts you have, those with higher numbers of late starts, also have larger numbers of cancellations. And on the maroon it’s the total number of cancellations. You’ve got to be careful here, and you’ve got to present this data carefully, but it does show that if you’re not starting on time, then you’re going to be more likely to cancel patients at the end of the list. That’s pretty obvious to all of us, but again comparing surgeons within your department, if you’re brave enough, that can be a really good way of eliciting change, but you’ve got to make sure your data’s accurate, because if it’s not accurate they’ll be quick to tell you.

    This table is an interesting way of looking at cancelled theatre cases, and I’ll just sort of come back to the talk I did a bit about the artificial and natural variation. What we're looking for here is not the natural sort of reasons, the natural variations if you like. So if a patient is medically unwell on the day of their planned surgery, there’ very little that we can actually do about it. Well, if there’s an admin error, or if there’s no equipment available that's something we could have done something about. And that’s something that we should really work hard to prevent happening. It’s unacceptable because we can’t organise equipment in time, that we fast the patient unnecessarily and then cancel their operation, waste those resources and also give a poor patient experience. So it’s those things that we need to focus on.

    I’m just going to talk briefly now about generating your own data. Sometimes you have a problem and the data isn’t collected anywhere, so you need to get your hands dirty and generate it. In one particular site I was looking at, the sort of theatre stores and the equipment guys were really sort of hacked off, because they estimated that 50% of the loan kit which they ordered, processed and which was ordered, and then they had to process and pass through their sterile services, which was actually never used. Now, they had no system for recording the data I collected, I did it manually by looking through the paper diary that they kept and also looking through all the loan kit request forms. The key thing here was a lot of the request forms were poorly filled in, not dated, there were high numbers of requests made for loan kit inside 48 hours before the op day. So this is poor planning. When the patient is listed for surgery, it’s likely that we’re going to know if they’re going to need any loan kit or any other additional equipment or services.

    So, what did I find? Well, I had to do this manually but we found a number of days notice given between requests for loan kit. I looked at the last 98 cases, and the vast majority were inside 96 hours. And an awful lot of them had no date at all, probably because they were inside 48 hours from the surgical procedure. So this was creating a lot of work at very short notice, making it highly stressful for sterile services, that are having to prioritise this ahead of other kits and instrumentation which was then slowing down their sterilisation. These things are important so I looked at the last month, two months and said how many loan kits used per day? And there's a lot of loan kits in these particular orthopaedic theatres. So if you’re using this many loan kits, I have to ask, okay, it may be that it’s a very specialised centre, but probably isn’t it worth getting some of these kits in on consignment having them available because you’ve got a lot of then additional rework and late work that the guys were having to do. So this is an example where we may not have data, but with a quick hour and a half of looking through some books you can generate data which can be quite powerful and showing this to the clinicians did change their views on how they order it, and they were much more considerate to the guys working in the instrument and sterile services section.

    So late starts. This is an interesting one. Who causes late starts? Well, if you ask the surgeons it’s not their fault. If you ask the ward staff, it’s not their fault because theatres didn’t send the flooring time, and then theatre staff said well we sent for the patient two hours ago but the ward haven’t brought them up, and the anaesthetist, well they’re going to blame the surgeon because the anaesthetists are always ready to start the list, but then the surgeon is never there. So what we’re talking about here is everyone blaming everyone else. They key thing is to make it a combined team responsibility and I hope in your hospitals that is the case. There'll always be natural variability in start times, so this is how I look at them and I’m looking and I want to see a particular theatre and I want to do it by theatre or by known list, and I want to have a look at that form and save the time. I don’t mind that it’s maybe eight o’clock here, it’s going to be a little bit earlier or a bit late, we’re not working bang on the clock, we know we’re not sort of machines, we’re not mechanics. And actually, anywhere in there, I’m comfortable with it, within that window. I think that’s a good enough variability.

    But understanding process control charts is these ones, it’s when we have the outside of these points that I start to get worried, or it’s when we have a run of seven or eight below or above the line. We’ve got a row here below the line, and that’s actually good, I was quite happy with that. It’s only these ones that I’ll investigate. It’s only these ones that I’ll look at and I’ll look at these in real time at the end of each day then I’ll know what I’m going to do the next day and I’ll look up and try and investigate it. If you leave it too long, if you leave it to the quarterly meeting or the monthly meeting everyone will have forgotten. It’s also important if you’re managing a unit and you are running this improvement process, you need to be the one getting to know and understanding it. You need to want to be talking to people. And always say don't always take the first answer they give you as the reason. Use the five whys. Why was that? Are you sure? Why? Was it really that reason? Because quite often people put something down, something that actually is that you have to think a bit deeper to find the true reason.

    So, there is variability, but it’s quite tight. There’s also variability in finish times, but here you’ll notice the variability is greater. Well actually, that’s not because we’re any worse, but obviously as you’ve gone through the day you’ve had a number of cases that you’ve done in sequential order and each of those cases will have a variability about how long they can take to perform. So your natural variability here, it will be slightly larger, but again I'm happy with that. I want this list to finish between a window, it’s due to finish at half four, but around about half four is the average, and it’s a window, 45 minutes or so either side.

    What's important here is your staff understanding this. If they think it’s bang on half four and it goes over then they’re going to sort of down tools. You need to understand this window and again I’m only going to investigate when it’s wildly outside of the control limits, or where I have a run of a number of cases that are outside it. So this is a really simple way. You’re looking at your finish time, so it’s the exit of the last patient from theatre, remember we've got 100% data accuracy from what we spoke about earlier, and then were just plotting it in a really simple format on an SBC chart that by applying statistical rules of SBC charts and process control charts, we can see whether that late finish was worth investigating or not. The problem is I find that there’s a lot of unnecessary work in a lot of hospital trying to find reasons for all of these late finishes or these early finishes. It’s wasted time. Whatever we do, we’ll always have late or early finishes and that’s because of the natural variability that we get between different case times.

    Now talking about that natural variability of case time, you can see here, this is just a normal distribution, this is case time. I think this is just for hip replacement and this is around 60 minutes and in most surgical procedures that line will perhaps become a bit longer on this side. And that’s a normal distribution. It’s unrealistic to expect it to take exactly the same time every time for every surgeon. We know it's going to be around that. It’s harder to manage the wider that is, but if it’s quite narrow like that, you can plan cases quite appropriately.

    So in summary, what data do I use? Well, it depends whether it’s measurement for judgment, diagnosis, improvement or sustainability. The judgment I'm not so interested in in this workshop. I'm more interested in diagnosis and improvement, so getting you to understand what is wrong, what is going on with your theatre processes, or what is going right and I'll show you some examples of diagnoses for capacity in demand and also measurement for improvement, we’ve looked at some of those SBC charts for monitoring late starts and early finishes. We’re going to talk about sustainability in the next presentation, and we’ll talk about how you can use data to ensure that your checking things are being maintained and also to motivate and sort of reward your staff.

Tom Wainwright, UK improvement specialist, on what operating theatre data is needed.


NSW Agency for Clinical Innovation.

Reviewed 05 October 2015


Was this page helpful?