Computer modelling for mediation and negotiation

In my years as a consultant working in hospitals I used what I had learnt at the gynaecology department. Each time I entered another part of the hospital, I knew little about every health care process at the start. That was a great advantage. I knew from previous experience that data are hard to find and that I would have to ask doctors and nurses to get a clue of how things really worked. For them to do be willing to do tell me, I had to be a recognized and accepted outsider. And being young helps too. To the people in the hospitals it was natural that I did not know a lot about their work. And so they told me everything.

I started to see that using computer models is effective, not so much as a projection of how things should work, as an ideal situation, but as a tool to show what was going on. A common view on the logistical system makes people involved solve problems by themselves. I will give an example.

In a hospital  the eye doctors and board of directors were not on speaking terms on the facilities available to the eye outpatient clinic. The eye doctors and the adjacent dental surgery clinic both argued that they had too little space and demanded more of it. The hospital board argued that the hospital as a whole already had too much space for what they could afford. The hospital board added that maybe the priority for the doctors should be the reduction of the number of waiting patients. The board seemed to view it as a zero sum game: if the eye doctors needed more space, the dental surgeons would have to give it to them. They had to basically figure it out themselves. The result was they were all stuck and no progress was made on the matter. The views of the stakeholders were different, contradictive and yet all true to some extent.

The simulation model built showed the patient processes as they were. This was all based on factual input provided by the doctors, nurses and assistants supported by planning and production data. At this point we are talking logistics and systems engineering. But instead of calculating the ideal situation the model ‘fact checked’ the perceived problems. We, of course, first had to know what the perceived problems were and what people thought were the causes. Were the waiting patients really there? How come? Is space the real bottleneck?

The model that we made highlighted all the concerns and assumptions that people had. The making of the model already made people aware of what was happening. By cross checking stories told and verifying these with data, people started to adjust their perception of reality or they introduced other factors that were relevant. The process of making the model together with those who are concerned was, again, very effective.

It turned out that lack of space was indeed one of causes of the problems that the eye doctors experienced. Not to the degree they had claimed though. The study also showed that 10 % more patients were planned into the scheme than what was even possible. No wonder there were patients waiting. And the fact that the doctors worked in a room of their own, resulted in patients having to walk from one room to the other, with delays and space unavailability issues as a consequence. It also turned out that waiting is part of the eye patient process – when they get eye drips and have to wait for it to work. And elderly patients who tend to show up early for appointments, was another relevant factor.

The model presented a balanced and objective part of reality for all parties. It brought all stakeholders back to the negotiation table and they were able to solve the space issue themselves. Every stakeholder had cards in their pockets, that they did not want to play before, but now they did. Whether they negotiated the perfect solution according to the model, was not important anymore. They had found an effective solution for now. It was a boost for the relationships and mutual trust increased.  This was a more fruitful ground for perhaps more optimization in the future.

Having had these experiences on a relatively small scale of one or a number of outpatient departments, the next step was to apply this method on a hospital wide scale. This is where my PhD questions were born as this turned out to be a lot harder.

 

Absence of data – a blessing in disguise?

My first scientific research was in 1999 in one of the major hospitals in The Hague. I worked there for about 6 months doing my Master thesis research. I studied the process of Gynaecology patients and more specific the patient file processes. Patient files, still on paper at the time, were often missing. Besides solving this problem, my main question was whether computer simulation added value in the problem solving process.

My main memory of this experience was that I spent 3 months collecting data and that it was, in Trump terms, a total disaster. Instead of knowing the patient and the administrative processes in detail after three months I only collected fragments but could not put them together. There were data on patients being in the outpatient department and there were data on patients being in the nursing department, but the time in between visits or steps was unknown. The same applied for the process of patient files. It was impossible to describe and quantify the processes from start to end in an objective way.

My last resort was to go to one of the gynaecologist and ask him. He was most helpful and we worked together for some weeks describing the patient processes and quantify these. It turned out that there were 8 patient types with more or less the same process. A process consisted of a number of activities that took place in a certain order and in an often predictable rhythm. In the process of making the model it became clear that patient files underwent activities in a rhythm that was not synchronized with the patient process. In other words, it was almost a miracle if the file was present at the right location when the patient showed up for an appointment.

Many people in the hospital already knew this. The patient file was often on the doctor’s desk, waiting to be processed. But assistants and other administrative staff did not feel comfortable to address this issue with doctors. The gynaecologist that helped me was quite happy to see the bigger picture and impressed by the fact that with his help we were able to make a ‘fancy’ computer simulation model. Besides it seemed like a useful exercise, it was fun for him. He accepted the analysis without any hesitation and was willing to change his own habits and behaviour to solve the problem. I was happy with the result, knowing I had solved the problem and wrote my thesis. But then the main lesson was still to come.

I presented my research to the entire group of gynaecologists. They listened politely but discarded the story as ‘organisational chit chat’, not really an issue that was of their concern. Then the gynaecologist that I worked with stood up and started a monologue to his colleagues about what needed to be done and that their own way of working needed to change. From then on I knew that making simulation models in a hospitals is only effective when the people working in the processes provide the data input themselves. They need to be involved in the making of the model, otherwise they won’t accept the outcome.

To be able to  work together with doctors and nurses it is important to understand the medical profession and what the ways are of doctors and nurses. To learn more about the doctors two thesis are worth reading: Doctors Orders by Karen Kruijthof and, only available in Dutch I think, Doctors in Charge by Yolande Witman. Both books provide an inside look on how doctors think, behave and what their beliefs are. Facts and evidence based measures are important factors in making a change that includes doctors and nurses. Simulation models could be a useful tool, if not merely made and presented by the outside analyst or data engineer, but developed and used by the people involved.

‘Nice job colleagues, but does it also work in theory?’ – part 2

I had quite a warm-up period before I went into PhD research. Not that I was aware of that at the time. Over the years, working in hospitals, I learned bits and pieces of the puzzle. I analyzed many outpatient processes, some in general, some in detail like those of eye patients, gynaecology and oral surgery. People gave me several tours of the Operating Theatres, the morgue, the laboratory, the Central Sterilization Department. I observed cardio vascular surgery. I visited a supplier of ready made dinners for patients, I spent a couple days with Oncologists talking to cancer patients every 10 minutes. I studied planning schemes of all sorts of departments, I talked and observed to logistics staff working in the central storage and internal transport. I guess I had more than 100 interviews and a similar number of workshops with pharmacists, doctors, nurses and managers. For over 3 years I worked on the logistics of the OR in several hospitals and in my Master thesis I spent 9 months in the Gynaecology related deparments. I understood that all these processes are connected somehow. But how?

I learned that hospital processes are extremely diverse and there are many flows going round. I once did a study of a hospital with two locations and at some point I had modelled 18.000 unique movements per week, including patients, materials and staff. These 18.000 flows consisted of 36 different flow types, mainly goods, but also, for example, bed patients. The flows crossed, they arrived on the same day on the same nursing deparment, sometimes even 30 (!) times a day.  Flows were, whether it be Cardiology patients or linnen, all (if at all) managed individually. Data on these processes was often not available or hard to find, mostly in the head of the experts: the doctor, planners and the people who do the job. The fragmentation seems so obvious and the optimization options endless but at the same time it’s crazy to even get started with that. My laptop even crashed trying to calculate stuff with these data.

Not surprisingly some people simply groan by the sight of such complexity. Me too I must admit. Data collection was like Sisiphus labour. It made me think of other ways to do it though. In my Master thesis research I did three months of data collection – only to conclude that that was a dead end road. Very frustrating. But I then discovered something I had not really thought of initially: the hospital is full of work process experts. Doctors, nurses and other staff know how things work and flow. Let’s use that. That turned out to be fruitful in several ways. More on that in my next blog.