Better demand forecasting saves millions on expensive agency nurses-while contributing to an improved bed fill rate.
At a Glance
Keys to reducing premium labor costs:
- Create a centralized staffing office.
- Develop a predictive model, such as one using census points, to forecast demand.
- Train staff on the new processes.
- Monitor performance monthly.
- Define key staffing processes for core and seasonal staffing needs.
By developing processes for forecasting nurse staffing needs, Lee Memorial Health System (LMHS), was able to save more than $11 million in one year alone. The savings are due to the elimination of agency nurses, or travelers, as of May 2008. The new predictive model has also helped LMHS refine its bed management approach at our five acute hospitals, leading to a 20 percent daily increase in the number of beds filled across the system. Just as important, nurse leaders have gained confidence in the new process thanks to a collaborative approach to training and implementation. With labor costs on the rise, many health organizations should consider optimizing their human resource consumption based on predicted volumes.
The Roots of the Problem
Back in 2000, the number of nurse travelers-agency nurses with 13-week contracts-hit a historical peak at LMHS. Based in Fort Myers, Fla., the public health system has substantial seasonal volume, which has helped traveler numbers continually inch up over several years. LMHS also has had significant nursing vacancies. Following the 2006 acquisition of two area hospitals, 350 travelers were on the combined payroll at the system's five acute care facilities. Each traveler cost the health system twice as much as a staff nurse.
Part of the problem was that the health system didn't have the right processes in place for unit directors to hire temporary nursing staff. LMHS had established tight controls for units to employ new staff nurses, but the recruitment process often took months. On the other hand, unit directors could bring in experienced traveler nurses in two weeks. For many units, using travelers became the path of least resistance.
In addition, some units suffered from scheduling overkill. Directors overstaffed their units, fearing last-minute sick calls and spikes in census. It was the human resource equivalent of supply hoarding. Call it staff hoarding. The nursing unit directors did not have a good predictive model for forecasting demand. They needed a tool that would help them predict their staffing needs by day and shift, based on historical information.
Thanks to a collaborative effort among its staff, LMHS pinpointed five strategies for resolving the situation:
- Retool the centralized staffing office.
- Implement a forecasting model to predict demand, identify staffing needs, and guide bed management.
- Engage and train staff.
- Regularly monitor performance.
- Define processes for use of contract staff.
Three groups at LMHS played key roles in this collaborative effort: nurse leaders, the centralized staffing department, and the system's operations improvement (OI) team. The OI team consists of nurses and other professionals who are trained in Lean Manufacturing processes, certified as Six Sigma Black Belts, and hired to work with LMHS departments to improve productivity and efficiencies.
In 2007, LMHS moved from a largely ad hoc nurse staffing process in each hospital to a more centralized and formalized system approach. The goals were as follows:
- Define, standardize, and operationalize key staffing processes.
- Retain existing staff and increase and coordinate the number and availability of in-house float pool and pro re nata staff (PRNs)-or as needed, per diem staff-to help reduce the use of agency nurses.
- Minimize cancellation of scheduled PRNs who are year-round system employees. Some PRNs are full-time, benefited employees, while the majority do not receive benefits and are not guaranteed a particular amount of hours.
- Fill core nursing positions so that the health system can avoid paying premiums for temporary staff.
Key to this centralization effort was retooling the health system's centralized staffing department. In the past, staffing procedures were not standardized. It became apparent that the processes and procedures needed to be brought under one umbrella. LMHS developed teams to address key goals, such as scheduling and planning and the deployment of PRNs. LMHS has 115 registered nurses (RNs), 10 licensed practical nurses (LPNs), and 27 certified nurse aides in its PRN pool. Having a centralized office laid the groundwork for process changes focused on matching staffing to demand.
Predict Demand, Beds, and Staff
The next step was to develop an accurate, predictive model that would help unit directors determine their staffing needs and use in-house PRN nurses from the centralized staffing department to fill "holes" in a more efficient way. The tool that LMHS chose was a model that uses census points, or forecasted admissions (see the sidebar below).
Census points are derived from using actual census figures and applying current and historical trends to determine the following month's census-by shift, by day of the week, and by unit. LMHS's OI team collects census data from about 50 units at the five hospitals and plugs these data into spreadsheets, which generate frequency distribution models. (Hospitals with automated bed tracking would likely have an advantage in collecting and tracking data needed to use census points.)
Census points are like targets. Eighty percent of the time, a unit should be at-or under-its census point within a calendar month. For example, if a unit's census point is 30, the unit is estimated to be at 30 or fewer patients, 80 percent of the time. Twenty percent of the time, the unit is expected to be above 30. Unit directors need a plan to flex their staffing up during high census times.
On a monthly basis, census points are used to determine staffing needs and the ideal number of beds to keep open on all nursing units. The goals are to maximize productivity and efficiency and improve nursing's ability to meet census fluctuations. This means better use of all nursing staff.
Coordinating beds. Bed placement is now accomplished to maximize each unit's census point. The hospital bed coordinators open beds in a specific order, which is determined each month. For instance, all medical/surgical units receive admissions until all of their census points are reached and current staff productivity is maximized. Then, if there are additional medical/surgical admissions, a unit is chosen to accept new admissions and bring on additional staff.
The goal: All of the open beds on the first unit will be filled before the next unit opens additional beds. All units that go beyond their census point will be expected to return to their census point as soon as possible as discharges or transfers occur.
Matching staffing and census points. Forecasting scheduled needs by shift and day of the week required a philosophical change across the organization, as well as collaboration across the nursing, staffing, and OI teams. It required unit directors to match their staffing based on forecasts, rather than average daily census. The steps involved:
- Estimating census per day, per shift using historical data
- Summarizing data by day of the week and shift
- Converting census data into the number of staff necessary to meet the scheduled need and distributing data to directors
- Loading scheduled need values into scheduling software
About four weeks before the start of a schedule, unit directors receive their forecasted census and staffing needs from the centralized staffing department. Directors review the census points and must suggest any changes to the forecast within two to three days. For example, a director would want to point out if a physician is on vacation and won't be admitting patients to a unit, thus decreasing the estimated census point and reducing the scheduled staffing need.
Keeping it balanced. To balance their schedules, unit directors must plan for the number of RNs and LPNs they need to cover their anticipated census on a given day of the week (see exhibit) . A weekly schedule is considered unbalanced if there is an opportunity on any day to shift staff to fill needs.
View Exhibit 1
Reports indicating the current "balancing" on the unit's schedule are distributed by the centralized staffing department. Directors use these balancing reports as guides to complete their unit's schedule-in other words, to better match staffing to anticipated demand.
Piloting the program. LMHS chose to test the new approach in some of the most difficult units to staff: ICUs. The system's ICUs had large swings in staffing from day to day.
An OI coach and the director of the system's staffing department (who is an RN) met several times with the ICU directors about the initiative. They walked the directors through how to use census points to staff their units, how to balance their schedules, how to read the balancing reports, and how to make schedule corrections. After a nine-month trial, ICU directors had revamped their scheduling practices and gained confidence in the collaborative program.
After the ICU pilot, LMHS began rolling the program out to other service lines in the system. Currently, three of the health system's acute care hospitals actively use census points to adjust staffing to demand. The remaining two facilities will be transitioning to the process later in 2009.
Train and Monitor
Making the transition to the new census point model required collaboration and buy-in from nursing leaders throughout the organization. In other words, it required that LMHS introduce the changes using a "pull," rather than a "push," philosophy.
Gaining nurse executive-level support early on in this collaborative process was critical. The chief nursing officer and nursing vice presidents have helped articulate the advantages of census point predictions and balanced schedules to unit directors. The RN director of LMHS's centralized staffing department has also been instrumental in developing nursing consensus for the new processes, working directly with the OI coach to ensure that the processes developed are practical for unit directors.
Training. Providing education to nurses through meetings and seminars has been a big part of LMHS's "pull" approach. The health system has now reviewed how census points are derived, as well as the principles behind the process, with all of our nursing vice presidents and unit directors. Nursing leaders understand that open beds with no demand are wasted beds, and that the LMHS wants to minimize unfilled beds.
All in all, the OI coach and system staffing director have provided about one week of training to leaders. LMHS has provided specific training modules to nursing leaders on:
- How to balance unit schedules
- How census points are used in that process
- How to use the scheduling software
- How to book RNs and LPNs to cover anticipated census on a given day and shift each week
After a year of trial and error, unit directors trust the process and know that if they need staff in a pinch, they can get additional RNs, LPNs, and certified nurse's aides from the centralized staffing department. In other words, the processes work to ensure that staffing needs are met and patient safety and quality care are consistently maintained.
Monitoring. A scorecard was developed to monitor unit efforts to balance their schedules, along with other key staffing indicators (such as, request fill rates and timely submission of schedules). LMHS leaders use the scorecard to analyze trends and set goals for system improvements.
For example, during the ICU pilot, the OI coach audited each week of a four-week schedule to ensure that ICU directors balanced their staff appropriately. The coach scored the directors' ability to match staff to anticipated need. During the pilot in the fall of 2007, only 16.7 percent of schedules across all six ICUs were balanced. In June 2008, the six ICUs scored 42.7 percent, a 26 percent improvement.
Address Long-Term Staffing Needs
As part of its staffing process changes, LMHS needed to establish better control of long-term staffing practices. Today, a systemwide, long-term scheduling coordinator is the go-to person for all unit directors with staffing needs that exceed two weeks. The health system also established new processes for requesting contract staff.
A process for core contract staffing requirements. Requests for long-term contract staff must be sent via e-mail to our system's long-term scheduling coordinator. The coordinator reviews the request and sends the request to a review committee. If approved by the committee, the coordinator works with the centralized staffing department to determine direct placement opportunities, such as the deployment of a PRN to a specific unit for a four-week assignment. If there are no direct placement opportunities, the coordinator explores the in-house traveler pool, composed of system employees. If none are available, a traveler nurse may be hired.
If nursing directors provide their required needs in advance, on the dates communicated to them, the approval process does not increase the length of time to receive the traveler. However, if it is an immediate need, then the approval process will add a few days to the time it takes to secure travelers and have them arrive.
A process for seasonal contract needs. Working with finance, nursing enters seasonal contract dollars and FTEs (determined by annual traveler need analysis) into the budget for seasonal census spikes in January, February, and March. Unit directors send the long-term scheduling coordinator all seasonal staff requests. The coordinator gathers all relevant unit financial information, census points, and current staff schedules. Then, a small committee reviews all documentation and decides whether a staffing need is warranted, with the system staff director making the final decision.
In the beginning of this initiative, LMHS also worked closely with nursing leaders to purposefully reduce its traveler numbers. Unit directors received biweekly traveler analysis summaries. In addition, the system staffing director held biweekly phone calls with each vice president and their directors on how to keep the traveler numbers below budget.
Early Signs of Success
LMHS has seen immediate results from its staffing improvements, including an $11 million savings in nurse labor costs. Twelve months into this initiative, LMHS was using 50 percent fewer travelers. After 16 months, the number of travelers was zero. In May 2008, the health system had enough confidence to cancel travel contracts, replace travelers with internal staff, and ultimately bring itself in under budget by $6 million in contracted labor.
In addition, LMHS has seen the following positive changes:
- A 20 percent improvement in daily bed fill rate
- A 200 percent increase in direct placement PRNs
- A 23 percent drop in PRN cancellation rates after one year
- Better staff coverage on units
- Reduced nursing overtime by 3 percent at three campuses
- An initial decrease in PRN cancellations
Staff nurses also seem satisfied with the changes. LMHS has decreased turnover slightly from 14.3 percent to 13.1 percent in two years. In FY08, nursing vacancy rates were down to 5.4 percent, as compared with 7.4 percent in FY00.
Although LMHS still has work to do, it has come a long way, thanks to the combined efforts of staff from various departments. Anticipating demand helps units provide better staff coverage each day. For example, knowing ahead of time that a 36-bed unit will have a census point of 30 patients allows unit directors to grant time off and reduce the use of costly staffing options like overtime or agency nurses. In other words, it helps the units plan ahead.
It also helps LMHS use its human resources and beds more efficiently. Knowing the anticipated needs of the units across the health system helps staff open beds and place supplemental staff as needed.
John Wiest is chief financial and institutional services officer, Lee Memorial Health System, Fort Myers, Fla. and a member of HFMA's Florida chapter (email@example.com)
Kristie Huff, RN, MSN, is system director for resource management and staffing, Lee Memorial Health System, Fort Myers, Fla. (firstname.lastname@example.org).
Kristi McMillan is operations improvement coach, Lee Memorial Health System, Fort Myers, Fla. (email@example.com).
Using Census Points to Match Demand to Capacity
- Census points are derived from using actual census figures and applying current and historical trends to determine the following month's census-by shift, by day of the week, and by unit.
- Eighty percent of the time, a unit should be at-or under-its selected census point within a calendar month. For example, if a unit's census point is 30, the unit shouldn't exceed 30 patients, 80 percent of the time. Twenty percent of the time, the unit is expected to exceed 30 patients, in which case, the unit director needs to flex staffing up, if needed.
- Daily census point predictions are compared with a unit's staffing matrix to determine the closest census that maximizes staff productivity.
- The monthly figures across units are compared to ensure that there are enough beds predicted across the system to meet demand.
- Census points are distributed to all necessary personnel, and bed placement is managed to maximize each unit's census point.
- Census point levels can be used to determine monthly staffing needs. For example, if a unit's capacity is 40 but the census point is 36, the unit director knows that she can grant some level of paid time off and still have enough staff available on a given day to meet demand.
Publication Date: Wednesday, April 01, 2009