case study

DECISION SUPPORT SYSTEM CURES FOR HEALTH CARE

The Problem

Avantas is an Omaha, Nebraska-based company recognized for proven best-practice work strategies for the health industry. These work strategies were developed to deliver immediate margin recovery and long-term financial savings by focusing on those areas in which health care organizations can realize continual and quantifiable improvement by creating value with regard to staffing and scheduling activities. Blending state-of-the-art technology and sound business processes, Avantas creates a precise balance of labor supply to patient demand across the client enterprise, regardless of the hospital's size or geographic location.

Avantas works with a variety of hospital systems to better manage their nursing resources. In this specific situation, Avantas worked with a client to help improve staffing decisions. The client needed to decide whether the five-hospital system needed to hire additional temporary staff to cover staffing demands during the upcoming flu season, a time of high patient volume. The five hospitals are located in the same metropolitan area. This particular exercise focused on the inpatient units (general medical/surgical and intensive care units) at each hospital.

This particular application of Planners Lab^TM addressed staffing needs of a hospital system during a typical high-demand period. Typically, the flu season strains hospitals' nursing resources. Not only do the hospitals need more nurses to handle the increase in patients, but the nurses also have an increased rate of unplanned absences due to the flu. This can cause major staffing problems for t hospital, including the possibility of turning away patients. In an effort to better prepare for the increased demand. The hospital system decided to look at current staffing and forecasted hiring levels to prepare for any potential weeks or days when staffing could be critically low. This high-demand time generally begins in January and continues through the end of March. The problem was analyzed at the end of October, allowing time for adjustments before patient volumes increased.

Solution

Decision makers in this case were the hospital system's CFO, the chief nursing executive, unit managers, and the human resources director. They wanted to adopt an easy-to-use tool. Mantas decided to use Planners Lab after previous attempts at analyzing this situation in Excel. Planners Lab offered an easy-to-use alternative to Excel to create "what-if" scenarios.

The hospital system's human resource staff had established short-term hiring goals to increase the number of nurses on staff. They had to keep in mind that these nurses would need at least 12 weeks of orientation before they would be available to take patients without assistance. The model also considered the current turnover rate. Another variable that could be adjusted was the number of temporary nurses contracted to work during the high-volume period. These variables could he adjusted to better accommodate the approaching high-demand time. The hospital system needed a software tool that would make it easy to adjust these variables and view the effects of its decisions.

The model is structured on a biweekly basis in order to incorporate the number of hours worked by core and contingency staff, with a lag of 12 weeks for training new hires. The basic model includes each hospital within the health system, a node for the entire health system, a node for human resource variables, and a node for different adjustments to explain the behavior of new hires. The variables within this model are primarily quantitative and include the following:

• Total hours demanded for 2009 flu season
• Total hours from 2008 flu season
• Total hours from 2008 flu season after new hires
• Gap between hours demanded and actual hours from 2008
• Gap between hours demanded an actual hours from 2008 after new hires
• Total required hires to meet demand

• Total remaining hires to meet demand after new hires

In terms of complexity, the model is straightforward and uses few advanced functions within the Planners Lab program. However, if the model had been completed in Excel, it would have easily made a 30-hour project a 160-hour project in terms of the advanced Excel techniques and expertise needed, such as macro programming. Furthermore the plain-English assumptions in the Planners Lab model are easily explained to others.

The following example shows how this model can be interpreted, changed, and applied by analyzing the impact of recruiters on the number of new hires. The analysis initially assumed that there were six resident nurse (RN) recruiters who were projected to recruit six individuals per week. Only a few of these new recruits actually make it past the interview stage of the hiring process, which lowers the net number of recruits after turnover is considered. Consequently, if the number of new recruits is less than the turnover, the vacancy rate increases and each hospital faces staffing problems. Figure 3.1 shows the desired vacancy rate versus the projected vacancy rate.

Thus, with all other assumptions holding true, the vacancy rate will be higher than desired, meaning that contingency resources will be needed to help staff the hospitals. Contingency resources include the following types of staff; agency nurses, core staff working overtime, traveling nurses, and other float resources in the hospitals. Contingency resources are more expensive than core staff, so the goal is to reach the desired turnover rate to limit use of these resources.

What happens if the number of hospital recruiters is increased from 6 to 12? Figure 3.2 illustrates such a change.

By changing the assumption from 6 recruiters to 12, one can project that toward the end of

January the desired vacancy rate is reached and the need for contingency resources is averted. Other variables, including the Number of Average Recruits per Recruiter per Week, Actual Employment Ratio of New Recruits, Total Resident Nurse (RN) Staff, Current Resident Nurse (RN) Openings, and even the Desired Vacancy Rate, can be adjusted to reach desired scenarios as well.

Without increasing the number of recruiters to 12, the desired vacancy rate would not be achieved in the next 3 months. However, the primary concern was to determine if the hospital system would have enough staff to treat the approaching high-demand flu season. In Figure 3.3, we can see how the additional hires made by the hospital system impacted the total actual hours worked versus the number of hours demanded during the 2009 flu season. The figure shows that the current hiring effort will provide enough staff for most of the flu season to treat the forecasted level of patients with the current levels of contingency resources being utilized. Temporary staff will be needed for a few weeks in the early part of the year, but additional short-term contracts are all that will be necessary.

Results

The hospital system's final decision was that current hiring goals were sufficient to address the upcoming high-demand time. Short-term hiring goals were continually monitored and adjusted if it appeared they were not attainable. The hospital system continued to use the Planners Lab model on a weekly basis to update the hiring goals with the actual hires as they became available. Planners were also able to identify two or three periods of time when staffing was too low, and extra efforts were made to bring on additional staff members in advance of those times.

The Planners Lab model provided the information the decision makers needed to make informed decisions on how to proceed. In the past, these decisions would have been made in the dark, because the decision makers did have time to create a complicated spreadsheet. The Planners Lab model was ideal because it was easy to understand and maintain.

Questions for the Opening Vignette

1. Why would this decision involve senior executives in an organization?
2. What types of decision parameters were used in making this decision?
3. What other modeling tools could be used for developing a model?

4. Besides the parameters considered by the modelers in this case study, what other situations could a staffing projection model include?
5. Why is this model a good example of a DSS?

What We Can Learn from This Vignette

A key component of DSS is the ability to visualize the results. It is important for executives to visualize the results of modifying assumptions, as discussed in this example.

Oftentimes, decision makers are forced to make a decision without understanding their true options. This is the reality of business decisions that have to be made in a timely manner. When deadlines are short and analytical resources are tight, companies are left with limited options to thoroughly understand the decisions they need to make.

In this example, decision makers were able to quickly use the Planners Lab tool to set up a DSS to model the complex scenario they faced. Not only did this save the hospital system money, because it required few resources to generate the analysis, but, more important, it enabled the executives to make informed decisions on how to proceed with confidence. Without this tool, a decision could have been made to bring on a number of expensive temporary nurse contracts, which the model showed was not necessary. This is an example of a DSS application. In this chapter, we will see many other related DSS applications. We will also learn to build a DSS using the Planners Lab software employed in this vignette.