For more information on this article, contact Stephan J. Silich at firstname.lastname@example.org.
Introduction: In response to concerns regarding delays in transferring critically ill patients to intensive care units (ICU), a quality improvement project, using the Six Sigma process, was undertaken to correct issues leading to transfer delay. Objective: To test the efficacy of a Six Sigma intervention to reduce transfer time and establish a patient transfer process that would effectively enhance communication between hospital caregivers and improve the continuum of care for patients. Methods: The project was conducted at a 714-bed tertiary care hospital in Staten Island, New York. A Six Sigma multidisciplinary team was assembled to assess areas that needed improvement, manage the intervention, and analyze the results. Results: The Six Sigma process identified eight key steps in the transfer of patients from general medical floors to critical care areas. Preintervention data and a root-cause analysis helped to establish the goal transfer-time limits of 3 h for any individual transfer and 90 min for the average of all transfers. Conclusions: The Six Sigma approach is a problem-solving methodology that resulted in almost a 60% reduction in patienttransfer time from a general medical floor to a critical care area. The Six Sigma process is a feasible method for implementing healthcare related quality of care projects, especially those that are complex.
A delay in transferring newly critically ill patients to an intensive care unit (ICU) may lead to an unfavorable impact due to the suboptimal environment for delivering the appropriate treatment (Beckmann, Gillies, Berenholtz, Wu, & Pronovost, 2004). For example, patients with septic shock had a significant delay in receiving intravenous fluid boluses and inotropic agents on a medical floor as compared to those in an ICU (Duke, Green, & Briedis, 2004). A 6-hr transfer delay of critically ill patients from the emergency department to the ICU was shown to increase hospital and ICU lengths of stay (LOS) and in-hospital mortality (Chalfin, Treciak, Likourezos, Baumann, & Dellinger, 2007). A greater mortality was also observed in patients requiring mechanical ventilation or renal replacement therapy when ICU admission was delayed (De Feo & Barnard, 2005).
At our hospital, ICU transfer delays resulted in (1) poor utilization of physician resources (e.g., residents being utilized to observe the cardiac monitor of upgraded patients) and (2) increased nursing demands to provide intensive care to these patients without a change in nurse:patient ratio. This further affected the physician and nursing staff's abilities to care for other patients. Transfer delays beyond 5 hr were not uncommon and resulted in decreased patient, family, and staff satisfaction. In response to these issues, the institution assembled a Six Sigma Team.
Six Sigma can be described as a management philosophy that focuses on developing and delivering near-perfect products and services (Pyzdek, 2003). It was originally developed by Motorola in 1986 and further enhanced by General Electric (Jiju, 2004). “Sigma” is a statistical term that measures how far a given process deviates from the mean (Lundberg et al., 1998). There are “six” standard deviations in variable performance of a given process. The central idea behind Six Sigma is that if the “defects” of a process can be measured, then solutions can be designed to eliminate them. A defect is anything that could lead to customer dissatisfaction (Fairbanks, 2007). Six Sigma defines quality as having less than 3.4 defects per million opportunities (DPMO). The Six Sigma score correlates with the number of defects; less defects yield a higher score.
Six Sigma Teams consist of people from different departments within an institution who are involved in the process that needs improvement. The leadership and technical roles of Six Sigma are organized in hierarchical fashion. “Master Black Belts” are experts in Six Sigma that assist in data calculations and function as resources to the team. The team is led by “Black Belts” who have prior experience with Six Sigma and can function as a leader. “Green Belts” are team members that have some experience in Six Sigma and have been selected by the institution to become more familiar with the Six Sigma process. “Yellow and White Belts” are relatively new to Six Sigma (Sonnenfeld, 1985).
A major tenet of Six Sigma is that the process must be organized and data driven. Six Sigma members use the five-step DMAIC (define, measure, analyze, improve, and control) approach, which is an acronym for defining the major issues; measuring the system process or practice prior to any interventions; analyzing the initial data to develop a root-cause analysis; improving the system process through intervention; and finally the control phase where data are collected to assess the impact of the intervention. We hypothesized that using this approach would establish a more efficient practice that would significantly decrease the average transfer time.
This study was performed at a 714-bed, tertiary care, teaching hospital located in New York City.
Scope of the Project
The team consisted of a Physician Sponsor, Chief Medical Resident, Director of Bed Management, Patient Care Unit Manager, Charge Nurse, Unit Clerk, Director of Environmental Services, Assistant Director of Health Information Management, and Manager of the Transport Department. A hospital administrator was the assigned “Black Belt” responsible for team building and project management. Three team members (Director of Bed Management, Assistant Director of Health Information Management, and a hospital administrator from the North-Shore Long Island Jewish (NSLIJ) Healthcare System) underwent “Green Belt” training. Several members had prior experience through other Six Sigma projects (Yellow Belts) and some members were hearing about Six Sigma for the first time (White Belts).
The initial step was the development of the “Project Charter” to clearly define the problem statement, business case, and goal and scope of the project, which was to include all patient transfers to the ICU and the cardiac care unit (CCU).
A high-level process map (Figure 1) was created to graphically display and better understand the major events that were occurring. At our hospital, physicians (mostly residents) were being utilized to personally observe the heart monitor in nontelemetry settings. It was also learned that some floor nurses had difficulty administering care (e.g., intravenous pressure agents) that was unfamiliar to them. Family complaints also arose when there was a perceived, overly long wait to transfer patients to the ICU/CCU. The objective of this Six Sigma deployment was to improve the patient transfer time, which would in turn have a beneficial impact on quality by providing critical care in the appropriately monitored setting, improving the utilization of the residents, improve patient safety by not having untrained staff administer intensive drug regimens, and increase patient, family, and staff satisfaction.
The financial impact of this project was deemed too difficult to accurately measure. However, it was recognized that quickly freeing up residents allowed them to return to caring for other patients, thus limiting potential delays in LOS. Limiting potential errors, by having critical-care-trained staff provide the care, could decrease resource wasting and avoid potential malpractice suits. Limiting patient complaints and improving patient/family satisfaction can generate potential income because complaint investigations are costly and higher satisfaction can better ensure that patients will want to return and refer others to the hospital. Finally, improved staff satisfaction can help limit staff turnover, which limits recruitment and training costs.
The most clearly defined, objective measurements identified were the time stamps at different steps generated by the computerized programs used in the transfer process. The clarity of these measurements facilitated the collection of data and easily identified which steps generated the most amount of time. The objectiveness of using time stamps ensured that the measures could not be easily refuted.
A data-collection plan was created to include eight identifiable phases for time measurement: (1) bed management notified via phone or transfer order; (2) bed management assigns bed; (3) bed management faxes transfer request to the sending unit; (4) environmental services flag the bed “clean” (ready/available bed); (5) sending unit informs the receiving unit; (6) sending unit clerk inputs transfer order into the computer; (7) transport department dispatches transporter; and (8) receiving unit clerk inputs electronic transport order as completed. The process and measurement also included a breakdown of the three different work shifts, as well as the number of beds involved in the transfer (one-, two- or three-bed transfers). For example, if a patient was transferred from the floor straight to an awaiting ICU bed, that was considered a one-bed transfer. If there was no available bed in the ICU and an existing patient in the ICU had to be moved out so a patient could be moved in, that would be a two-bed transfer and so on.
Preliminary measurements revealed that the average time for a patient to be transferred from a floor bed to a critical-care bed was 214 min, with a maximum delay time of 420 min. Additional data showed that the amount of variation in the process (assessed by the standard deviation) was 170 min. Initial capability analysis revealed 423,728 DPMO. The sigma score was only 1.6. The performance goals recommended by the Master Black Belt were to reduce the standard deviation by 50% and raise the sigma score to approximately 2.2, thus decreasing the DPMO to 242,000. Subsequently, the goals for this project were set at 90 min for average transfer time and an upper specification limit (USL) set at 180 min for a maximum individual transfer time. The USL of 180 min was largely determined by the fact that the measurement phase showed that the maximum number of beds that needed to be cleaned in any one unit transfer was three. One hour per each patient transferred in a three-bed transfer would allow for proper cleaning of the room, transfer orders to be written, proper communication handoffs between hospital personnel, and safe transfer of the patients and their belongings. The USL is an upper limit above which the process performance is deemed unacceptable (a “defect”). The lower specification limit (LSL) was set at 0 min, which limited analysis of transfers that went exceptionally well. The “customers” (representative residents, nurses, bed management personnel, transporters, etc.) of this transfer process, with the exception of the patients, agreed to these goals/limits.
The data collected in the Measure Phase were analyzed to create a list of process steps and identify sources of variation in the process. Complex processes often have a myriad of definable steps. Identifying the few vital steps, or “vital Xs” as they are often referred to, will help in avoiding the natural tendency of trying to manage every process step. By determining the vital Xs, it becomes possible to focus on only those that are critical to producing the desired outcomes.
The first tool used was a “Fishbone Diagram,” a cause and effect illustration that enhances identification of potential factors causing an overall outcome—in this case, the delay in transfer (Figure 2). The statement of the problem was placed in the box at the “head” of the diagram. The remainder of the fishbone consisted of one line drawn across the page, attached to the problem statement, and several vertical lines or “bones.” These vertical branches, chosen as subcategories of the major categories of influence, were labeled with the specific “cause and effect” titles. The specifics of the fishbone diagram were developed by group discussion. When completed, the diagram provided a visual understanding of the root causes of the problem and allowed the brainstorming for possible solutions to begin.
Next, a “Failure Mode and Effects Analysis” (FMEA) was done (Figure 3). The FMEA identifies potential and actual points of failure, as well as corrective actions. In particular, this tool identifies an effect (outcome) and quantifies it based on the level of “severity” (using a scale of 1–10). It shows how likely an effect is to occur. The likelihood of effect or the frequency of “occurrence” is used to describe how often the outcome is initiated by the root cause. The process of stopping the unwanted outcome is referred to as “detectability.” Thus, the resultant value is the risk priority number (RPN), which is computed by multiplying the “severity” by the “occurrence” by the “detectability.” Of the eight steps identified, the FMEA yielded four critical steps with high RPNs.
Finally, a series of “Hypothesis Testing” (Figure 4), which uses statistics to determine the probability that a given hypothesis is true, was undertaken. In brief, a series of various hypothesis tests were examined by calculating a “p-value,” which is also known as the observed significance level or the probability value. The p-value helped delineate the causes that were “vital,” which focused the determining of the potential specific causes for the differences.
After careful analysis of these three tools, it was determined that the increased turn-around time centered on Bed Management's ability to “assign” a clean, ready bed. Thus, there needed to be an “available bed” in order for Bed Management to facilitate this process. The turn-around time greatly increased depending on the number of bed transfers needed. When the assignment involved a one-, two-, or three-bed transfer, the average turn-around time was 126, 249, or 404 min, respectively.
It was also discovered that there was an increase in turn-around time related to how long it took for the sending unit to communicate the order (via phone/fax) with the receiving unit, which was due to poor communication and too many process steps. However, there was no statistical significance in turn-around time in relation to the shift time, the day of the week, whether it was a phone or fax order, the specific unit the patient was transferred to and from and whether or not the transport department was utilized.
Although it was not originally deemed a vital X, it was agreed upon by the team that the medical resident's completion of the transfer orders was a key step. It was found that there was no standardized process for a resident-driven completion of the transfer orders. Some residents completed their orders immediately, others completed them later. Also, there appeared (by direct observation) to be poor communication between the physicians and the nursing staff in the critical-care areas.
The following critical elements were recognized: (1) poor process flow; (2) inconsistent communication; (3) no standardized order writing process; (4) overutilization of remote cardiac monitoring; and (5) lack of understanding at the staff level of the importance of this issue. Next, a specific solution plan was developed.
One new process was to pilot having a clean, ready bed always available in a large room (ICU Annex) used for equipment and device storage that is located directly across the hall from the ICU entrance. The environmental services director (ESD) and the ICU charge nurse would have accountability for ensuring that a clean, ready bed was always available. The ICU director personally educated all supervisors on the new policy. This solution eliminated the need for the units to call the ESD for bed delivery.
Another improvement was the creation of an “electronic” bed assignment notification via the installation of Tele-Tracking software in the ICU/CCU and Telemetry Unit. The head of Bed Management installed the software, educated all personnel, and ensured that it was utilized on all shifts. Bed Management would notify the receiving and sending units via Tele-Tracking. Additionally, a “notification alert” would now be utilized so that when Bed Management assigned a bed, it would flag as such in both the sending and receiving units, notifying the respective clerks of the bed assignment. This eliminated multiple process steps (i.e., the need to fax, phone, and page notifications) and resulted in less work for the nurses in the ICU/CCU. Also, the “ready to move” function in Tele-Tracking was instituted by the sending unit clerk. This provided “real time” notification that patients were ready to be moved.
The process for writing transfer orders out of the ICU/CCU was also changed. The goal was to ensure that transfer orders were completed immediately after rounds. All residents were instructed to “flag” the patients’ charts for discharge/transfer to alert the unit clerk to place the transfer order, which notified bed management. This solution would expedite the transport of patients out of the ICU/CCU to make beds more quickly available for incoming patients.
A fourth new procedure called for the accepting critical-care physician to determine whether or not a remote cardiac monitor was to be placed on the patient awaiting transfer to the units. It was realized that while some upgraded patients (e.g., those ruling in for a myocardial infarction) required constant cardiac monitoring, others did not. Once a monitor was placed on a patient in a nontelemetry ward, a resident physician had to be assigned to the room to constantly observe the patient for fatal arrhythmias. If the cardiac monitor was safely deemed unnecessary by the attending intensivist, then this freed up the resident to facilitate the transfer, as well as care for other patients.
Finally, the project itself called attention to the importance of quickly moving critically ill patients to the critical-care areas. Because all departments that shared a role in this process were part of the Six Sigma Team, new education and enhanced teamwork skills developed from this project.
In this phase, most of the Six Sigma Team becomes disbanded. Constant data tracking and documentation were done by the process owner (in this case, the Director of Bed Management) and the Black Belt to measure any improvements and ensure that they would be sustained. In addition, the team sponsor and the nursing and physician staffs were updated on a monthly basis with on-going data.
After implementation of the new processes, data were collected and analyzed on patient transfers over a period of 1 year for 462 consecutive patient transfers to the ICU/CCU (Figure 5). The target of decreasing the average transfer time to less than 90 min was immediately approached and then finally attained by the fourth month. In the first 6 months, there were still rare instances of individual transfer times exceeding 180 min, which only allowed the sigma score to reach the mid-three range (but it did break the 2.2 goal). However, by the eighth month, there were no defects and a sigma score of six along with a yield of 100% were reached and maintained for the remainder of the control phase.
For the entire control phase, the mean time for the transfer of patients from a floor to a critical-care bed was 84 min as compared to the initial mean (preimprovement analysis) of 214 min; a marked reduction in the transfer time of 138 min (Figure 6). Additionally, the standard deviation in the transfer time was reduced by 135 min. The standard deviation is one of the most common measures of variability in a data set; as it gets smaller, the process capability gets better. The postimprovement data showed a standard deviation of only 35 min. The overall sigma score was raised from 1.6 to 3.8 and the yield, which represents the percentage of the process that is acceptable to the customer, was raised from 54% to 98.9%.
After seven consecutive months of no “defects,” the project was turned over to the process owner in December 2010 and the team was disbanded. A project summary is depicted in Figure 7.
The improved process aligned with the hospital's strategic business objective, set forth in the Project Charter, which outlined the following goals and standards of the project:
Customer satisfaction Patients, their families, residents, and staff all experienced timely transfers, which led to increased satisfaction.
Operational excellence Improved utilization of residents and nurses enhanced operational excellence.
Quality Better communication procedures led to a decrease in the risk of adverse events for a patient transferred to a monitored bed.
Economic profit Though not directly measured, immediate and delayed financial benefits (see Section “Define Phase”) were likely realized, as well as unnecessary costs were avoided (e.g., complaint investigations).
Six Sigma provided a comprehensive analysis of the patient transfer process prior to implementing new solutions. Six Sigma utilizes data, the voice of the customer(s), and statistical analysis to determine the factors that are most critical to quality improvement. It also requires accountability and constant evaluation after implementation of new solutions (a control phase), which fosters sustainability. Furthermore, the use of the Six Sigma jargon provides for a universal language that can compare and contrast the effectiveness of different projects.
A very key step was to set realistic improvement goals that were measureable. This cannot be overstated. The analyze phase helped to understand what would be a realistic goal for individual and average patient transfer times.
Interviews with staff to find out their concerns and insights were very helpful. Assembling a team of individuals who performed integral roles of the patient transfer process was important. This ensured “buy-in” prior to the implementation phase and served as the basis for creating the “fishbone” (cause and effect) diagram and conducting the FMEA.
One limitation of our study is that improvements could have been secondary to the Hawthorne effect, which postulates that processes being watched improve because they are being watched (Tennant, 2001). Nonetheless, we believe the changes that were made to the overall process lead to the significant results. Another limitation was that we could not pilot the new solutions in one area of the hospital while continuing the old process, to serve as a real-time control in another. However, a true historical control was used, which was measured in the months just before the implementation phase. Lastly, the financial impacts of the new processes were not directly measured.
Although the NSLIJ system has employed and trained “Six Sigma experts,” other organizations can still benefit from using the various tools often implemented in a Six Sigma project, even without the specific Six Sigma resources and experts. For example, an organization can assemble a team of various disciplines to define, measure, analyze, improve, and control a fragmented process. They can create a cause and effect diagram, run an FMEA, identify key steps (the vital Xs), brainstorm, and formulate practical solutions and measure the outcome of the implemented strategy. For this specific problem (ICU transfer delays), implementing a computerized bed-tracking software program, having a clean, ready bed near the ICU, improving the efficiency and communication of transfers into and out of the ICU and determining the need for cardiac monitoring prior to transfer led to almost immediate, major reductions in transfer times, which were sustained over 1 year.
Stephan J. Silich, JD, is the Director of Business Development and Six Sigma Certified Blackbelt for Staten Island University Hospital in Staten Island, New York.
Robert V. Wetz, MD, is the Associate Chairman of Medicine & Residency Program Director for the Department of Medicine at Staten Island University Hospital in Staten Island, New York.
Nancy Riebling, MS, MT, is the Director of Operational Performance Solutions and Six Sigma Certified Master Blackbelt for North Shore-LIJ HealthSystem in Long Island, New York.
Christine Coleman is the Director of Bed Management for Staten Island University Hospital in Staten Island, New York.
Georges Khoueiry, MD, is the Associate Director of Student Education for the Department of Medicine at Staten Island University Hospital in Staten Island, New York.
Nidal Abi Rafeh, MD, is a cardiology fellow at Staten Island University Hospital in Staten Island, New York.
Emma Bagon, RN, is the Patient Care Unit Manager in the Intensive Care Unit at Staten Island University Hospital in Staten Island, New York.
Anita Szerszen, DO, is the Director of Geriatric Research and co-chair of the Research Division for the Department of Medicine at Staten Island University Hospital in Staten Island, New York.