To the Editor,
Any credible Lean presentation, including application in healthcare and hospitals’ EDs [emergency departments], should clearly state what key Lean concepts were used and, specifically, how these concepts were implemented.
Some key Lean concepts include standardized work; reducing batch size/load leveling or sequencing; stop at abnormality; pull work only by the end customer; error-proof implementation; and other. Further, any credible Lean should describe which of 7 types of wastes were eliminated and how it was done.
The point: All of the above basic requirements for any credible Lean presentation were completely missing in Dr. Kulkarni’s editorial, leaving readers to guess what the ED “Go Lean” means, what was actually done, and how it was done.[1]
Lean is not the only or even the best tool for every situation that requires process improvement and streamlining of the patient flow. Although some Lean tools, such as Value stream map, may be great for relatively streamlined manufacturing processes, it is not too efficient to apply simple lean concepts for ED patient flow that have patient-to-patient variability, multiple information streams, shared resources, and interdependencies with other hospital departments. Other methodologies, such as process flow modeling and simulation, could sometimes be more appropriate.
The point: There is no one ideal methodology to improve operational efficiency and patient flow. Lean concepts could be an appropriate methodology or not, depending on the complexity of the process. It is not nearly enough to claim being Lean with simple housekeeping and injecting some discipline in the execution of the process.
Reporting numerical change in terms of only centricity metrics, such as median (or average) change without information on data variability, is usually highly misleading. If the data variability is high, such a change could not be statistically significant. It was easy to construct an example to demonstrate that the change of the median time to triage from 60 minutes to 15 minutes could not be statistically significant.
The point: A basic requirement for any credible quantified reporting of the change should include a summary of the test for statistical significance of this change.
Moving a bottleneck from one part of the system (the ED) to another part of the system (floor nursing units) does not help to move patients faster, or gain an operational efficiency. Only alignment of patient flow for different subsystems for the entire hospital could help to attain the overall goal of operational efficiency. Individuals involved in operational improvement activities should be qualified professionals with expertise in various quantitative process improvement methodologies.
The point: Because the ED is not an isolated stand-alone department but is a part of the entire hospital system, quantifiable gains of the ED operational efficiency could be attained only with basic engineering principles of complex system analysis.
Alexander Kolker, PhD
Milwaukee, Wisconsin
[email protected]
Reference
1. Kulkarni RG. Going Lean in the emergency department: a strategy for addressing emergency department overcrowding. MedGenMed. 2007;9:58. Available at: http://medgenmed.medscape.com/viewarticle/566882 Accessed January 9, 2008.
Author’s Reply
Thank you for your important comments to the editorial.
Although the original submission for the video editorial admittedly did not address all of the important comments in your letter, it would be fair to state that much of the information requested was indeed in the first draft submitted to the editors of the journal. Unfortunately, strict word requirements limited the amount of information that could be communicated in the editorial itself. The opportunity to expand and address the important questions you bring up in this response is welcomed.
Muda is the Japanese term for any activity that does not add value to a process, and thus by definition is wasteful. As you point out, there are in fact 7 specific types of waste: defects, overproduction, transportation, waiting, inventory, motion, and overprocessing. A more detailed description of these wastes can be found at: http://en.wikipedia.org/wiki/Muda_%28Japanese_term%29. In addition, some add other types of waste, including — but not limited to the following: skill, safety, information, material, and breakdown.
Our process, although addressing several different types of waste, focusing on the “door-to-triage” metric in the first instance, achieved the greatest gains in eliminating motion and overproduction for our staff and waiting for the patient. Specifically, we found, for instance, through direct observation generated “spaghetti” maps that our triage nurse — a valuable resource — spent a significant amount of time walking around the treatment areas looking for open patient care areas and locating and preparing stretchers instead of doing their primary function, which was to triage patients. The solution was to maximize the time spent by the triage nurse to actually stay in the triage area and utilize a different, less valuable resource to identify open patient treatment areas and to locate empty stretchers and prepare them for use by patients. From the customer’s or patient’s perspective, the primary waste eliminated was waiting. In the course of our event, we noted that patients had to “run a gauntlet” of encounters in a sequential manner when they arrived at the door of the department: greeter, security guard, registration, triage nurse, treating nurse, and resident or midlevel provider before finally interacting with the individual who they had primarily come to see — the supervising staff physician. We changed the model to be “patient-centric” by specifically addressing the goals of this Lean event to have the patient see the triage nurse as the first person when they walked through the door, and furthermore beyond the scope of this event, to be brought back to the treatment area immediately if a treatment space was available with other activities, such as registration, coming to the patient through the use of mobile workstations instead of the patient being required to endure a necessary encounter from the hospital’s perspective. Our initial graphic analysis of simple box plots of the door-to-triage time for pre- and postevent conditions was strikingly different; of note, there was no overlap of the Q3-Q1 interquartile range between the 2 datasets. Our subsequent statistical analysis of the median door-to triage time for pre- and postevent conditions was limited to using the Mood’s median test because the 2 data sets were not normally distributed and the variances were not equal. The results from our analysis allowed us to conclude, with 95% confidence, that we could reject the null hypothesis that the 2 medians were equal (P < .05) and that the changes that we made to the process significantly improved the door-to-triage time.
The other metric that we focused on was the “decision to admit-to-floor” time split. Specifically, we looked at the timepoint at which the emergency medicine physician made the decision to admit a patient to the hospital to the time when the patient actually left the department for their hospital bed. Although it is certainly true that the emergency department is not an island unto itself in the hospital system, this time split similar to the previous metric we chose was under significant control of the emergency department; inpatient bed availability was not under the control of the department, but after notification that the inpatient floor was ready to receive the patient (ie, the inpatient bed was clean and ready) the emergency department was responsible for completing a series of steps to prepare the patient to go to his/her inpatient bed. A series of 37 steps was initially identified as part of the process contained within this metric and included several emergency department-dependent functions, such as completion of an electronic nursing report, completion of the physician paper documentation, booking of transport in computer application, waiting for transport to arrive from a location in the hospital not in the department, paging and waiting for a call back from the inpatient admitting physician, “medical holds” preventing transport, and locating necessary equipment and identifying appropriate personnel for the patient during transport (eg, portable monitor, oxygen tank, nurse for patients with cardiac problems); all were contributing factors to an initial length of about 3 hours for this metric.
As part of this intervention, we not only reduced the number of total steps by 30% (to a final count of 26 steps), but also eliminated several specific types of waste from the process, including motion — implementation of a voicemail dictation system for the emergency department physician to avoid an initial and often repeated pages to the inpatient physician; waiting — creation of an intradepartmental location for transport personnel and improved communication with radiology for expediting completion and interpretation of pending radiology studies and overprocessing; elimination of outdated requirements for transport (eg, portable cardiac monitor not necessary for transport of “low-risk” cardiac patients); and creation of specific guidelines for “medical holds” for patients in the department. Real-time simulation showed a 54% reduction in process time (pre- 248 minutes, post- 114 minutes; n = 80, P < .01). On 6-month follow-up, we observed year-on-year reduced process time of 40.1% (pre- 197 minutes, post- 118 minutes; P < .0001) as well as an overall reduction in length of stay of 19.1% (pre- 355 minutes, post- 288 minutes; P < .0001).
The Lean methodology was used simply as a tool to assist us to achieve the gains highlighted in the video editorial. In fact, other tools, such as computer-based modeling using simulation software, have also been employed as justification for trials for varied staffing models and other interventions and have also proven successful. Any efforts in a highly complex interdependent system, such as a hospital, can only be taken in context of the course of the implications for downstream resource strain and utilization and the bottom line — the quality and safety of the care delivered to a patient. The interventions and changes that we implemented were only possible through a highly collaborative effort. The team that we put together included not only representatives from the emergency department, such as nursing, midlevel staff, staff physicians, resident physicians, greeters, registration, security, and technicians/patient care assistants, but also hospital leadership and representatives from inpatient physician and nursing leadership as well as general hospital departments — such as transport and environmental services. On balance, we believe that the Lean methodology is an important and useful tool for emergency departments to take advantage of in a shared goal to improve the overall value delivered to the patients who they serve.
Rick G. Kulkarni, MD
Medical Director
Adult Emergency Department
Yale-New Haven Hospital
Assistant Professor of Surgery
Section of Emergency Medicine
Department of Surgery
Yale University School of Medicine
[email protected]
http://www.medscape.com/viewarticle/569069?src=mp
