Swick, Maureen PhD, MSN, RN, NEA-BC; Doulaveris, Phyllis MSN, RN, NEA-BC; Bagnall, Timothy BS Systems Engineering; Womack, Dana MS, RN
The healthcare system in the United States is in transition. Macro-level forces including an expected shortfall of nurses, economic uncertainty, and evolving federal policies create a need for optimization of professional nursing resources. Anticipating this, the American Organization of Nurse Executives (AONE) has developed guiding principles for the role of the nurse in future patient care delivery.1 These principles outline the knowledge, skills, and abilities that nurses require to continue to provide first-rate healthcare in the years to come and to significantly enhance nurses’ role, responsibility, and stature. The nurse of tomorrow will be an architect of patient care, creating a design for healthcare through the integration of available medical knowledge and advanced technology, coordination of care plans through multidisciplinary rounding, and the provision of expert advice and advocacy. The authors, in conjunction with the nurses of the pilot unit, designed a care delivery model incorporating the AONE guiding principles.1 Lean healthcare concepts and healthcare computer simulation were used to support the design of the new model.
The term Lean was coined in the late 1980s to describe Toyota’s revolutionary manufacturing system.2 As Lean became prevalent in manufacturing businesses worldwide, a few visionary US hospitals began to apply the concept to healthcare delivery as a means to achieve meaningful process improvements.
Lean is a production methodology that supports adding value and reducing waste through a culture empowering frontline staff to make operational decisions regarding how to improve processes. In the hospital setting, registered nurses (RNs), licensed practical nurses (LPNs), and unlicensed assistive personnel (UAP) are intrinsically sensitive to the needs of the patient.3 Hospitals that embrace Lean principles recognize that frontline employees have valuable insight about how to add value or reduce waste in a system and use interdisciplinary teams to achieve improvements. The literature contains many examples of Lean applications in healthcare, including the use of time studies, value stream maps, and rapid improvement events to significantly improve adherence to clinical guidelines3 and reduction of steps required to prepare operating room case carts.4
Embracing the principles of Lean, the authors, hospital executives, and RNs and UAP of the pilot medical unit assembled to plan a new care delivery model in several Kaizen (Japanese for “change for the better”) events. The team focused on key issues such as role clarity, nursing delegation, communication, and trust.5-9 To prepare for Kaizen events, the authors used computer simulation technology to identify candidate care delivery models that enhance the role of the professional nurse while maintaining or reducing operating expenses. Through simulation, the authors were able to put conventional nursing delivery model hypotheses to test and determine care delivery model alternatives that offer desirable improvements.
Applications of computer simulation in healthcare date back to the 1960s with a body of supporting literature containing numerous reports of successful application to improve healthcare decisions.10-12 From enhancing emergency department flow to optimizing operating room department schedules, computer simulation provides a validated method for improving patient care and hospital operations.13
Similar to the way an architect builds small-scale models to portray and evaluate new building designs, computer simulation provides virtual representation of systems for analysis. Simulation provides decision makers with robust operations research functionalities, including insight into the current operational state. This research can demonstrate how value is presently delivered to customers and where waste exists along with forecasting into proposed operational states. The ultimate goal is to predict how value will be delivered. A compelling reason to use computer simulation is the capability to experiment and test proposed design alternatives at minimal cost and risk. Experimentation can take place virtually, without impacting the real system or incurring expensive capital expenditures.
In the current simulation study, the authors tested 129 hypothesized care delivery models with varying staffing levels, skill mix, and professional practice techniques in a 2-week simulated scenario. The goal of the study was to test the impact of differing staffing levels, skill mixes, and professional practice technique (whether RNs focused only on RN-specific services) on patient wait times, RN service proportion (proportion of services provided by RNs that require an RN-specific skill set), percentage RN time spent in value-added and wasteful care activities, and cost. Performing experimentation of this scale in the real world would not be viable logistically or financially. Figure 1 illustrates the place of simulation in relation to healthcare’s process improvement creed of plan, do, check, and act.14 In this context, simulation is organized into 4 phases: design, verify and validate (V&V), experiment, and analyze.
Phase 1 involves the design and development of a mathematical representation of care delivery in the simulation software. Design and development is where the simulation analysts and key stakeholders diagram relevant events to represent the system virtually and provide sufficient detail for answering underlying questions. The scope of the study dictates the complexity of the simulation.
Verify and Validate
Phase 2, V&V, involves verification that simulated processes behave as intended (verification) and that the simulation produces realistic results (validation). For example, for verification, the simulation analyst would ensure that each time an RN needed to gather supplies, time spent walking to the supply holding location occurred. To validate, the simulation analyst would ensure that the simulation realistically represents the process. If it takes 2 minutes to gather supplies and the simulation predicted 3 days, the simulation would not provide results of any value to stakeholders. For improved V&V, this study’s simulation was reviewed by an independent, third-party healthcare simulation consulting firm.
In phase 3, experimentation, stakeholders identify alternative system designs for investigation. Because the team was interested in designing a care delivery model that enhanced the role of the RN, it evaluated models with variances in skill mix (RNs, LPNs, and UAP), staffing levels (patient-to-staff member ratios), and professional practice techniques. By testing these 3 categories, simulation provided comparative information to evaluate proposed care delivery models with regard to enhancing the role of the nurse and maintaining or reducing operating costs.
Analysis is the fourth and final phase of simulation. The authors compared the predicted results of patient care and staffing performance metrics, including the mean time patients waited to receive care (indicator of patient satisfaction), the RN professional service proportion (indicator of role enhancement), staffing cost, and nursing time spent performing value-added and wasteful care activities.
The patient care unit that served as the basis for the simulation design was a 52-bed, adult medical unit. Common diagnoses for the unit include esophagitis, cellulitis, septicemia, kidney and urinary tract infection, pneumonia, and metabolic disorders.
Figure 2 depicts the unit design. Two long corridors, approximately 200 ft in length, provide direct access to patient rooms with health information system (HIS) terminals located along the periphery. Support service space exists between the 2 corridors and is composed of a medication room, kitchen, staff break room, soiled holding room, clean holding room, clean supply room, secretary station, physician workspace, and staff locker room. The preproject care delivery model consisted of a staffing mix of RNs and UAP. The RNs were responsible for approximately 5 patients, assisted by UAP who floated throughout the unit. Time and motion studies confirmed that RNs on the study unit spent a significant portion of their time (>20%) performing activities that did not require professional RN knowledge, skill, or ability.
Inova’s Lean Healthcare Simulation Design
The team took a decidedly Lean approach (Figure 3) in the simulation design. In Lean, “value” is commonly defined as what a customer is willing to purchase. The antithesis of value is “waste.” Services that do not provide value but cannot be eliminated from a system are considered to be “necessary waste,” whereas services that do not provide value and are unnecessary constitute waste that should be eliminated. For the purposes of this study, the team chose to use 3 Lean categories, 1 for waste and 2 for value, with value categorized as either “bedside value” or “nonbedside value.” For RNs, value activities were further qualified as those activities that require an RN-specific skill set (eg, medication administration, assessment, education). Services performed away from the bedside (eg, preparation and conclusion services associated with value activities, care planning, and coordination) could arguably be considered necessary waste in Lean terminology because they do not provide value as a patient-facing service and potentially could be streamlined. However, the team acknowledged that RN-specific services conducted away from the bedside provide value to the patient and chose to define these activities as nonbedside value. The team defined waste as services that do not provide value or activities performed by an RN that could be performed by others.
The patient care scenario rectangle in Figure 3 helps illustrate the concept of Lean value categories. In the figure, assessment is the first service the patient receives. To provide this service, the RN must first prepare (nonbedside value), then perform (value), and finally conclude (nonbedside value). “Preparation” may involve a review of notes before entering the patient’s room. “Performing” includes the direct patient care received during assessment. “Concluding” entails charting the assessment. This same pattern of preparing, performing, and concluding applies to all services of the simulation.
In Figure 3, the “variables rectangle” simplifies the parameters that the team configures to test alternative care delivery models. The patient presentation rate and mean time on unit allow the analyst to manipulate patient census. Increasing the presentation rate or time on unit will increase the patient census. Required care allows the analyst to vary the amount and type of care received by patients such as blood sugar checks, assessments, baths, meals, blood administration, IV starts, linen changes, medication administration, specimen collections, turns, vital signs, and wound administration. Shown in the RN and UAP section of the variables rectangle, the team may also configure the skill mix and staffing level, the professional practice technique, and hourly wage.
Once the team has set up the appropriate experiment variables and executed the simulation, results are available for analysis. Results include the mean patient wait time for each service and an aggregate for all services, RN professional service proportion, operating costs, and time spent in the Lean value categories.
To help visualize the simulation, the team included an animation in its design (see video, Supplemental Digital Content 1, Lean Healthcare Animation, http://links.lww.com/JONA/A65). The animation shows RNs and UAP providing care to patients and the simulated real-time performance of key metrics. Animations help verify and validate a simulation and serve as a powerful presentation medium.
To set simulation variables to realistic values, the team conducted a time and motion study. Professional RNs on the team observed RNs and UAP for 25 individual 4-hour sessions (18 RNs, 7 UAP). Using data from the time and motion study, the team devised a representative patient and staff day outlining patient and unit services (eg, case conferences). For each service, the team fit the recorded duration samples to statistical distributions so the simulation could make realistic service duration estimates.
The 18 RNs observed in the time and motion study were selected because they represent the age and nursing experience range that exists on the study unit. The nurses in the observation phase came from a total possible sample of 52 (34.6%) nurses. They ranged in age from 27 to 73 years (mean, 39 years) with between 2 and 38 years of nursing experience (mean, 11 years). Ten (55.6%) were educated at the bachelor of science in nursing level, 6 (33.3%) were associate degree prepared, and 2 (11.1%) were diploma prepared. The 7 UAP who were observed were from a possible sample of 16 (43.7%) with 1 to 8 years of experience (mean, 4.8 years). Four RN observers, employed in the organization in advanced practice roles, were selected for their familiarity with nursing processes and nursing research principles. Registered nurse observers participated in a 1-hour training session, conducted by the study’s primary Lean consultant, in data collection techniques.
For the time and motion study, the team used a standard template to collect information, including time, service, location, and notes. Table 1 shows an example of a medication administration observation using the template. The populated template allowed the team to decipher service duration, walking paths, and whether the service is value added or waste.
Analysis included the development of a coding schema to capture service types that represent the greatest percentage of nurses’ time. Service observations were coded by the observers and were validated by an RN reviewer. Durations for value-added services, including preparation and conclusion times, for recorded episodes of admission, discharge, assessment, and medication administration are displayed in Table 2. The nurse manager of the study unit heuristically validated the mean duration times for each coded nursing service.
The team mapped each service to a Lean value category (Table 3) to identify the percentage of time spent performing direct care services (value), indirect and necessary services (nonbedside value), or services that do not provide value or could be performed by another care team member at a lower cost (waste). The team determined that some direct care services provided by the RN were waste (eg, blood draw) because they could potentially be provided by an LPN or UAP, freeing the RN to focus on RN-specific services.
Nurses spent the largest percentage of their time providing services that could have been provided by UAP (18.43%), administering medications (17.98%), coordinating patient care (9.52%), and conducting patient assessments (8.03%) (Figure 4). Examples of services that could be provided by UAP include vital signs, blood sugar checks, and simple dressing changes.
Summarizing by Lean value categories, nurses spent 22% of their time in direct care services (value), 49% of their time performing indirect services (nonbedside value), and 29% of their time performing unnecessary activities or services that could be performed by others (waste) (Figure 5). The 29% waste signifies opportunity for care delivery model redesign to allow nurses to practice at their full potential and reduce the cost of services provided by overqualified personnel. The 49% nonbedside value also signifies redesign opportunities to streamline services. Less overall waste means more time to provide value-added services.
The spaghetti diagram maps staff walking paths performed during observation and provides insight regarding wasted motion (see video, Supplemental Digital Content 2, which depicts the Spaghetti Diagram, http://links.lww.com/JONA/A66). Walking is a non-value-added activity that should be reduced to the extent possible. Point of use supplies—those located at the patient’s bedside—offer an alternative to central supply storage to reduce nursing staff walking waste.
The team ran 129 two-week simulated experiments testing various skill mixes, staffing levels, and professional practice techniques against typical patient volumes on the pilot unit. The goal was to identify if alternative care delivery models provided more value to the patient, enhanced the role of the professional RN, and maintained or reduced operating expenses.
Although the simulation recorded more than 150 resulting variables for performance comparison, the team focused on 6 key metrics: mean patient wait time, RN professional service proportion, cost, RN bedside value-added time, RN nonbedside value-added time, and RN waste time. Patient wait times, including service preparation times, are reported as the mean patient wait for all services. In the simulation, each RN service was assigned a priority so that time-sensitive patient care needs (eg, blood sugar checks, medications) were addressed before less time-sensitive services (eg, baths). Whereas detailed simulation results itemize wait time for each service, the mean wait for all services provides an overall metric by which to assess the timeliness of value-added services. Registered nurse professional service proportion (ranges from 0.0 to 1.0) provides insight into the percentage of RN-provided services that require RN-specific skills—an index to show if skills are being used to their fullest compliment. Higher relative numbers indicate that RNs spend more time performing highly skilled services and, in the authors’ opinion, a role enhancement in line with AONE’s guiding principles. “Cost” estimates the staffing cost for the simulated period. The percentage of RN time spent in the Lean categories of bedside value and nonbedside value, similar to the RN professional service proportion, is an indicator of skill use and job satisfaction. More time spent performing value-added services and less time spent performing wasteful services foster a rich professional environment.
Figure 6 shows abridged results for 9 experiments run by the simulation team to test a patient census of 45. Experiment 126.96.36.199. Baseline is the benchmark—a predicted representation of the existing performance of the unit—against which to compare the other 8 care model design alternatives. Grey-shaded cells indicate predicted improvement and black-shaded cells indicate predicted deterioration. In the benchmark, 10 RNs and 3 UAP provided care to 45 patients for the 2-week simulated period. The benchmark’s estimated mean patient wait time was 0.51 hours (31 minutes) and the RN professional service proportion was 81%.
In the simulated 2-week experiments of the care delivery model alternatives, 4 of the 8 care delivery model alternatives predicted an improved mean patient wait time, RN professional service proportion, RN time spent on bedside and nonbedside value-added services, and lowered cost. The remaining 3 care delivery model alternatives predicted an improvement in mean patient wait and cost but deterioration in RN professional service proportion and an increase in RN time spent on activities defined as waste. In all cases of deteriorated service proportion and increased waste, the professional practice technique had been set to simulate a scenario in which RNs provide all services rather than focusing on services that require RN-specific skills. Consequently, the RN professional service proportion and waste worsened over baseline as RNs provided more general services. Improvements in value-added time, both bedside and nonbedside, for each alternative came about primarily for 2 reasons. As the staffing mix changes to include more UAP and fewer nurses, nurses spend greater percentage of their time in value-added activities and less time on services defined as waste. As services that do not require RN-specific skill sets are taken on by UAP, nurses also have more available time to spend in value-added services such as patient education, midshift coordination, and care planning.
When LPNs are added to the staffing mix, LPNs provide bedside value-added services (eg, medication administration), allowing RNs to spend more time on nonbedside value-added work. However, the increase in value-added work fluctuates with the professional practice technique and the number of LPNs and UAP in the staffing mix. For instance, experiment 188.8.131.52. All predicts only an increase of 0.3% and 6.9% in bedside and nonbedside value-added time, respectively. This is a modest improvement relative to other alternatives.
Although the experimentation results shown in Figure 6 are based on a mean patient census of 45, the simulation predicted similar outcomes for other typical patient volumes. In these other tested volumes, the key metrics similarly resulted in an improvement over the respective baseline. The simulation results underscored an opportunity for an alternative care model strategy that could meet the AONE guiding principles to enhance the RN role and maintain or reduce costs.
Simulation using the principles of Lean offers healthcare organizations a proven method for strategic planning and process improvement. The authors used this simulation to identify alternatives that support AONE guiding principles regarding the future role of the professional RN to help plan a new care delivery for a medical-surgical unit. Recognizing the opportunity for an improved care delivery model, a team consisting of hospital executives, administrators, RNs, and UAP gathered to plan details. The opinions of the RNs and UAP were weighted heavily during planning. The project (Kaizen) team discussed detailed aspects of the design including role responsibilities, delegation, communication, and trust. Because of this study, the organization redefined its care delivery model staffing level and skill mix to create an environment where RNs can focus on RN-specific services. The organization also created a 2-day delegation and communication training course for RNs, updated its UAP job profile to reflect the newly identified role responsibilities, and established an academy for the development of UAP. Simulation proved to be a useful decision aid that allowed decision makers to predict the impact of alternative care delivery model changes, select a model that allows RNs to spend their time in a manner that best uses their professional training, adds value to the patient, and reduces waste in the system.
Although the study unit has many commonalities with other medical patient care units across US hospitals, practice, differences in physical layout, staffing practice, care policies, or differences in patient or staff populations may limit the applicability of simulation outcomes to other medical units. The time and motion observations reported in this study were performed during day shift hours. Future simulation studies are needed to test care model design alternatives specific to off-shift hours.
The future of the US healthcare system and beyond will require different care delivery models to meet the expected shortages of nurses, uncertain economic climate, and projected increases in the volume of persons requiring care. Care delivery models that enhance the role of professional nurses by maximizing their knowledge, expertise, and time are likely to be the most successful and yet remain, for the most part, unexplored. Nursing executives may find Lean and Lean healthcare computer simulation to be helpful tools in the design of care delivery models that meet future demands, enhance the role of the professional RN, and enhance patient care delivery.
We thank Linda Sallee for assistance in data collection, analysis, and review and Heidi Sitcov and Jacqueline Peterson for data collection.
© 2012 Lippincott Williams & Wilkins, Inc.