CASE STUDY – This article reports on the efforts that Erasmus Medical Center in the Netherlands has made to apply Theory of Constraints to patients’ length of stay.
Words: Rienk Gerritsen, senior consultant, Lean Management Instituut; Liesbeth Eijkelboom, senior nurse, and Sander van Gisbergen, unit head, Erasmus Medical Center Thoraxcentrum
According to the Dutch Heart Foundation, the number of people suffering from heart disease is increasing: between 1980 and 2012, the amount of hospitalizations for heart disease went up 20% for men and 43% for women.
Although between 1980 and 2012 the average length of stay in a hospital has decreased by 67%, from 16 to 5 days, for men with a heart disease and by 73%, from 19 to 5 days, for women with a heart disease, shortening the amount of time people with such conditions spend hospitalized is still a critical challenge.
Besides, this has advantages for both the patient and the hospital. For a patient, a hospital is not a nice place to be – it is uncomfortable and from a medical perspective it isn’t the safest environment. Infections and complications are common, especially among people who undergo (heart) surgery. In general terms, it is best for patients to be discharged as soon as they are medically fit. In turn, hospitals are chronically full and free beds come in short supply: the most immediate result of a shorter length of stay is an increase in capacity.
Additionally, costs for insurance companies as well as for society are higher if patients spend more time in a hospital.
Hospitals are under continuous pressure to discharge their patients sooner. The Erasmus MC Thoraxcentre in Rotterdam has been focusing on shortening the length of stay, using the principles of the Theory of Constraints (TOC), in a bid to improve accessibility for patients in need of cardiac care in a specialised hospital.
The Erasmus Medical Center is the largest of eight University Hospitals in the Netherlands. The organization has over 50 departments, each of which contributes to three core activities: research, patient care, and education.
The Thoraxcentre is an integrated organization constituted by the departments of Cardiology, Cardiothoracic Surgery and Lung diseases. Within it, patients are treated for all existing heart and lung diseases.
Cardiac surgery patients recover in the Intensive Care Unit. Within a few days they are transferred to the MC/HC Cardiology ward. The MC/HC Cardio ward is divided in three sub-units: the Medium Care Cardiology, the Heart Transplant Unit (HTU), and the Short Stay Cardiology. Patients that visit the Short Stay Unit usually stay for several hours, and up to 3 days.
The improvement activities you’ll read about were directed at the length of stay at the Medium Care Cardiology Unit.
Patients spend an average of 11 days at the HTU and MCU. This extended period of stay allows for nursing staff to better understand the recovery process of each individual patient. Observations indicate that not every patient is discharged immediately when medically fit. For various reasons patients stay in the hospital for days longer than they need. Nurses become stressed knowing that there is a waiting list while patients that are better off outside the hospital and could be discharged are still occupying the beds.
MEDIUM CARE UNIT PATIENTS
Patients that recover in the MCU suffer from a wide variety of heart diseases. Below is a Pareto of the amount of patients classified per diagnosis.
Statistical analysis (Unifactoral Anova) revealed that patients treated for heart valve failure are discharged significantly more quickly than those given other diagnoses. There exist a causal link between diagnosis and length of stay. The average length of stay for patients with heart valve failure is seven days. For the other patients, it is 11 days.
This classification was further developed based on the experience of some senior nurses, resulting in the following lengths of stay, organized by diagnosis:
- Analysis, Out-of-Hospital Cardiac Arrest – 9 days
- Pre-hydration – 1 day
- Complications after procedures – 3 days
- Surgical procedures – 7 days
- Planned Percutane Valve Replacement – 7 days
- Acute Coronary Syndrome – 5 days
- Activation pacemaker – 5 days
- Analysis arrhythmia – 7 days
- Deccord – 9 days
- Myocardial infarction – 5 days
- Analysis cardiomyopathy – 8 days
- Endocarditis – 49 days
- Anti-clot – 7 days
THEORY OF CONSTRAINTS
The theory of constraints (TOC) is a management philosophy introduced by Eliyahu M. Goldratt in his 1984 book The Goal. It is based on the premise that throughput is limited by at least one constraint. Only by increasing flow through the constraint can overall throughput be increased.
Here are the main steps to follow according to TOC:
- Identify the system’s constraint(s);
- Decide how to get the most out of the constraint;
- Subordinate everything else to the above decision;
- Make other changes needed to increase the constraint’s capacity;
- Warning! If in the previous steps a constraint has been broken, go back to step 1, but do not allow for inertia to cause a system’s constraint.
APPLICATION OF THEORY OF CONSTRAINTS IN HEALTHCARE PROCESSES IN THE NETHERLANDS
There are several hospitals in the Netherlands that are applying TOC to shorten the length of stay. For example, in St. Antonius Hospital, a medical specialist determines an expected date of discharge immediately after admission. This sets the target for all disciplines to discharge the patient no later than that date. If this proves impossible the root cause is investigated. Combining all root causes and analysing the most common one gives an indication of the bottleneck in the process. A team then tackles it until it is no longer impeding the overall patient flow. After this improvement a new cycle of analysis reveals the next bottleneck. This way a continuous improvement cycle is initiated.
This and other examples inspired the Erasmus Medical Center Thoraxcentre to use a similar method in their own process.
“When I started as the head of the department we had 32 nationalities, and using lean as a standard philosophy provided us with a common language. When we began to implement TOC the time was just right. Pressure on staff was mounting and at the same time we had the feeling that length of stay was prolonged by factors that were out of our influence,” Sander van Gisbergen, unit head of the wards, says.
Liesbeth Eijkelboom, a senior nurse, adds: “I coordinate new patient admissions. My phone was ringing constantly as people from the clinic, from other departments, from other hospitals and emergency departments contacted me to ask if they could transfer a patient to my ward. It was frustrating because I had to disappoint them all the time, even though I knew that some patients were ready to go home. I couldn’t even give them an indication of when a bed might become available. We visited another hospital to see TOC in action in an emergency department department. It took some time before we started, but when we did, we learned a lot in a very short period of time. We chose TOC because it helped us to predict delays due to bottlenecks in tests at other departments.
ESTABLISHING AN EXPECTED DISCHARGE DATE (EDD)
In order to be able to analyse the reasons why a certain target was not met, a target had to be in place of course. In applying the TOC, the MCU favoured the idea of having a preliminary Expected Discharge Date (EDD) as a target to work towards.
This prediction of the EDD for every patient at admission was not easy to implement, however. Several discussions around setting guidelines for the EDD took place:
- The difficulty with patients, especially those that are treated on this specific ward, is that the diagnosis in many cases is not specific enough to predict with a high certainty when the patient will be – medically speaking – fit enough to be discharged. And if there is much uncertainty around setting an EDD, then why bother?
- Although the EDD is not meant to put pressure on the recovery time of the patient, it might feel that way if the needed recovery time is not defined yet. Setting a target to get a patient dismissed, no matter what, might send the wrong message;
- Measuring against an EDD might feel as if the accuracy of the prediction and therefore the qualification of the doctor are being assessed.
It helped that the ward had an extensive medical database where diagnosis, date of admission and date of discharge were documented. This enabled staff to decide on guidelines for the EDD based on the statistical variance within and between groups of patients.
In April 2012 the medical head of the department, Dr. Akkerhuis, and unit head Mr van Gisbergen explicitly acknowledged that the EDD was used only for the analysis of bottlenecks in the process and not for any medical decision. They also determined that the EDD should be set within 48 hours after admission. And so the pilot started.
Eijkelboom says: “The Expected Discharge Date is discussed on every ward round. Patient, nurse and doctor constantly remind each other of the EDD approaching. This puts some pressure on the system, which in turn leads to early discharge. There is also much more focus to complete the necessary activities on time. A big change for me was that I can now respond to inquires about submission in a much easier and more detailed way.”
THE DISCHARGE PROCESS
The MCU was already focusing on preparing the patient for discharge in advance. The nursing staff would indicate to the medical specialist that the patient might be fit enough to be discharged within the next three days. During the daily visit, the medical specialist would confirm this indication and, if also agreed on by the supervisor, the discharge process would initiate.
This process is more a checklist of actions that should be performed than a process that is standardized in sequence, process steps and process times. We mapped the process, revealing:
- A minimum of 22 process steps of which many can be performed independently of each other and in parallel;
- Seven disciplines/departments involved;
- At least three official forms and two IT systems required;
- No central oversight or planning system coordinating discharges.
Without any further analysis this process already showed a relative high level of complexity, to which coordination over three days in a three-shift rotating schedule adds to. This observation supports the feeling of the nursing staff that sometimes patients await discharge due to reasons unrelated to being medically fit.
MAKING SURE THAT THE ACTIVITIES START ON TIME
The complete checklist of activities is planned for each patient based on the Expected Discharge Date. The expected time on the ward is divided into three sections and assigned a color. The first 50% of the time is green, the next 25% is yellow and the last 25% is red. If the EDD is not met, the time exceeding the EDD is black.
The activities are called Discharge Determining Tasks: a complete list of these was generated by the department. It was also defined when each task should be performed. This leads to red, yellow and green tasks.
Eijkelboom says: “Depending on the diagnosis, you might have to roughly perform the same tasks. It’s just that these tasks are spread over a longer period of time and for other patients you need to perform them quickly, one after the other.”
The MCU ward used this approach to schedule all activities per patient. The complete checklist was integrated in the software tool the ward used for displaying bed occupancy. This allowed nurses to tick the activities performed.
The system monitors the length of stay of each patient by displaying a traffic light mark on the left side of the EDD. In the bed occupancy overview the EDD is displayed next to the patient name and bed number. If the patient reaches the 50% mark relative to the EDD, the traffic light on the screen lights up in yellow. If the patient reaches the 75% mark relative to the EDD the traffic light will change to red. If the patient exceeds the EDD text will turn red. If actions are not completed in time, an exclamation mark will be displayed per phase.
The primary objective of planning for each patient is to start activities on time. The TOC is used to identify the bottleneck in the process by analysing activities that are not completed on time. If activities are not started promptly they might not be finished on time, although the capacity of that activity is not a bottleneck.
Eijkelboom comments: “In general the process was not clear, so the clarity this tool provides was very welcome. It helps me to keep an overview and check that all tasks are being performed. There was no resistance on this at all. It fulfilled another need: to have a centralized checklist that helps with the shift transfer. Everybody welcomed having this transparency.
“Establishing a strong relationship between diagnosis and Expected Discharge Date does result into ownership naturally shifting towards the nursing staff. The doctor establishes a diagnosis and the nurse enters the EDD in the system. In fact, the nurse has more responsibility for meeting the EDD than the doctor does.”
DOCUMENTING AND ANALYSING REASONS FOR NOT MEETING EDD
The hypothesis is that the activity that is most frequently responsible for not meeting the EDD represents the bottleneck in the process.
In order to determine the current state, a standard for documenting the reason for early or late discharge was required. The MCU produced a list with the most likely reasons for delays in discharge based on the experience of nursing staff and medical experts. Whenever a patient does not meet the EDD, the relevant reason is selected from this list.
In the period between February 14th, 2013 and October 1st, 2013 in the Medium Care Ward 72% of patients were discharged before or on the EDD (the exact number is 418).
For the 115 patients (28%) not meeting the EDD the top three reasons for late discharge were identified as:
- Not yet fit for discharge (35%);
- Conflicting lab result (23%);
- Not defined, to be investigated (16%)
Remarkably, the failure to meet the EDD is almost never associated with occurrences like waiting for results, waiting for treatment, or waiting for an X-ray.
This analysis revealed no specific bottleneck. Therefore no specific improvements were performed.
Sander van Gisbergen explains: “Although strictly speaking the TOC analysis did not reveal any specific bottlenecks in MRI or other diagnostic tests, during the implementation of TOC we did improve these factors. Everybody could see that having MRIs only accessible once a week tent to delay discharges and because we had much more focus on the subject we did discuss the topic and decided to make the accessibility to MRIs daily. In the same way, we immediately reserved a number of slots in the planning of the operating room for pacemaker procedures. So we have essentially improved several crucial bottlenecks even before they became apparent from the analysis.”
Eijkelboom adds: “The reasons indicated above correspond to the nature of the patients that are treated here. We are a university hospital, which means that procedures tend to be more complicated and that patients often have more complex or multiple conditions. This in turn contributes to a more frequent deviation from the EDD.”
IMPROVEMENT IN LENGTH OF STAY
Although no specific improvement was performed on activities that were expected to appear as bottlenecks, the length of stay was shortened quite significantly, going down 20% from 4.9 days to 3.9 days.
Eijkelboom comments: “The most evident decrease in length of stay was achieved in TAVI (Transcatheter Aortic Valve Implantation) patients, one of our largest patient groups. EDD was originally set to eight days after admission. This date was frequently not met. Recently we set it to six days after admission and we are frequently reaching five or even four days.”
(Note: The average length of stay as displayed above includes the short stay department. At the same time the TOC was implemented in the short stay cardiology unit, as part of another project – PCI in day-care – that had the effect of decreasing the average length of stay.)
There is now better control over the activities that need to be performed and there is more room for transparent and fact-based discussions between nurses and medical specialists. This facilitates the communication between the disciplines.
EXTRA PATIENTS, DECREASE OF WAITING LIST
Another welcome result is in the ward’s waiting list. The time people spend on the waiting list has decreased by 50% and gone from three weeks to 10 days. The number of patients on the waiting list is also lower. The accessibility of the department has increased.
It was expected that if the length of stay is decreasing and the waiting list is shortening, the amount of patients treated would increase. We were not able to establish this relationship, because during the implementation the number of beds available in the ward decreased by 12%.
INFORMATION TO THE PATIENT
The EDD is communicated to patients, who seem to appreciate this transparency and involvement.
The amount of complaints has decreased from 12 per year to nearly zero over the course of the last three to four years. The downward trend had already started before implementing TOC but the consensus is that this methodology contributes or at least does not negatively influence the number of complaints.
In the spirit of improving communication, every patient discharge includes a discharge conversation. Patients often comment that the information supplied is very good. This positive trend is also visible in the responses to the continuous patient satisfaction survey (Project Coolsingel). In general patient satisfaction showed a positive trend.
Eijkelboom says: “I think that younger patients especially appreciate greatly the information we give them throughout the process. Some of them are in bed with their laptop surfing the internet to learn about their conditions, and they like knowing about their treatment. They often challenge the medical staff on decisions regarding the EDD.”
INVOLVEMENT OF NURSING STAFF
There is nothing suggesting that nurses and doctors are experiencing increased pressure as a result of the TOC implementation. In fact, the opposite is true.
“All tasks are included in standard checklists now and follow a specific timeline. This provides all employees with the tools to work to a plan’s specifications. Because of all these standards we now have an opportunity to focus on deviations,” says Sander van Gisbergen.
Eijkelboom also commented: “The process is much clearer now. Although, there are more patients flowing through the system, it still feels like there is less pressure. Also, knowing that you can influence the process positively is a good feeling that contributes to making work more enjoyable. I will definitely take part in the next project.”
The Theory of Constraints improves system performance by identifying and improving a bottleneck. The Medium Care ward of the Erasmus Medical Center Thoraxcentre implemented its principles to improve the length of stay. They implemented an Expected Discharge Date, started to plan activities using a structured timeline and to classify the reasons for not meeting the EDD.
Implementing TOC helped the ward to achieve the following:
- The average length of stay was shortened significantly;
- The waiting list was also shortened significantly;
- Patient satisfaction increased;
- Employee satisfaction increased, especially among project participants.
Rienk Gerritsen has been a consultant with Lean Management Instituut for the past five years. Prior to this, he worked as a production manager at a bio-pharmaceutical production plant of Johnson&Johnson who consistently used lean thinking to improve operations.