rnjbanner
 
Home > RNJ > 2006 > May/June > Developing a Staffing Matrix Using CMI as Acuity Indicator

Developing a Staffing Matrix Using CMI as Acuity Indicator
Diane Romito, BSN RN CRRN

Adequate staffing levels on an acute rehabilitation unit may not be maintained because staffing needs fluctuate according to the needs of patient groupings. Acuity regulations from some state agencies and the Joint Commission on Accreditation of Healthcare Organizations require that staffing address patient acuity needs. Proposed ratio laws either require concrete patient-to-nurse ratios or neglect to consider acute rehabilitation. Neither have practical working tools to support their proposals. The questions of what to measure and how to translate this into effective nursing staffing remain unanswered. At the same time, nursing dissatisfaction grows with increased workloads, overtime, and perceptions of ineffectiveness. This article describes one effort to define and use a working tool for staffing acute rehabilitation units. The study used case mix index as an indicator of nursing time, integrated into a shift staffing matrix. Early results have shown it to be effective, quick, flexible, and efficient. Using this tool, quality patient outcomes within national length of stay benchmarks were maintained and staff satisfaction on this unit improved.

The Scripps Rehabilitation Center, just north of San Diego, is a 30-bed acute rehabilitation unit, serving more than over 500 patients annually; stroke (25%), brain injury (15%), and spinal cord injury (15%) account for the most frequent diagnoses managed. Nursing care is provided around the clock in 8-hour shifts.

Catalysts for Change

In late 2002 and during 2003, a series of events created a need for a revision of the staffing matrix. First, the prospective payment system (PPS) changed from being cost-based reimbursement to case mix group (CMG)-based payment. CMGs are the rehab equivalent of acute care’s diagnosis-related groups. However, CMGs rank both patient diagnosis and the level of functional disability. Units were no longer to be reimbursed based on whatever costs they chose to incur. Each patient’s CMG prescribed the available financial payment. This change created a need for a careful analysis of how much staffing is needed at each level of care. More nurses would be needed for higher acuity patients and fewer nurses for low acuity patients. Efficiency would affect financial viability. Flexibility was key.

Second, the CMG basis for payment also defined which cases used more or less resources based on available data. Some diagnoses require more resources and as a result allow for a higher level of reimbursement. Sicker patients and those with more severe disability were assigned higher CMG ratings and were awarded higher reimbursement payments. This correlated with the need of higher acuity patients for increased hours of nursing care. Because the CMGs included functional levels as well as diagnoses, they more highly correlated to nursing staffing than the diagnostic related groups (DRGs) that had been used in acute hospitals. For the first time, with CMGs, the Scripps Rehabilitation Center could objectively define the weight of its acute rehabilitation patient population.

Third, staffing by acuity became a Joint Commission on Accreditation of Healthcare Organizations requirement, however, a common tool was not provided. Two self-designed tools were tested but found to be unacceptable. One was a medical-surgical tool that captured many nursing activities, but failed to define the functional levels of the patients, which in this context was a significant omission. The other was a more rehabilitation-oriented tool; despite defining nursing time required for various patient levels of function, it was time consuming and only yielded results relatively parallel to those defined by the medical-surgical tool. Neither could be applied to adjusting daily and shift staffing. Most of the time staffing was done by census and when the acuity level was high, the charge nurses requested additional staff.

Fourth, nursing satisfaction scores, reflected in the annual Great Place to Work surveys between fiscal years 2002 and 2004, began to dip. Although this author’s team had always scored high in the unit, they noticed changes in areas of “respect” and “fairness,” and name-protected comments on the survey cited staffing levels and workload as causes. They also noted RN overtime, mostly during times of high patient care need. While retention was not yet a problem, they wanted to proactively respond to the nurses’ concerns.

Finally, the California Nurse Ratio Law was enacted by the state department of health services, with the backing of the nurses’ unions. This law restricts (in most areas of a hospital) the number of patients assigned to RNs or LVNs (but LVNs may not exceed 50% of the mix), although the ratios for rehabilitation units were left undefined and ambiguous.

For example, the law requires set ratios that vary from 1:2 (licensed nurse:patients) in ICU to 1:6 in medical-surgical units. Because rehab units vary widely in the complexity of patients served, this area has been more difficult to define and has been neglected in the law. This gap has led to opinions that vary from 1:3 to 1:12 for rehabilitaion units. There is a need for a well-defined methodology for determining appropriate human resources for specific patient needs.

Searching for Precedents

The puzzle of staffing adequacy has been a topic of ongoing interest to nurses and the regulatory agencies that protect the public welfare. In the 1980s, when DRGs first appeared, they sparked interest in measuring patient needs as related to outcomes, length of stay, and use of nursing resources. Halloran (1985) was one of the first to explore the DRG-nursing equation. Later, Halloran and Kiley (1987) attempted to correlate nursing dependency, DRGs, and length of hospital stay. The later works of Welton, Halloran, and John (1998), more notably with the Minimum Data Set (MDS) of subacute rehabilitation, have suggested the need for a nursing component in determining staffing needs, and suggest that DRG “can be enhanced by including data about the problems of care addressed by nurses.” Mark (2002) explored nurses’ perceptions of staffing adequacy, noting that “perceptions of staffing adequacy were influenced significantly by the hospital’s case mix index…and by patient acuity.” More recently, Neatherlin and Prater (2003) linked nursing time and specific work activities in acute rehabilitation settings.

When considering peer practices, a national consultant from the Hunter Group quoted a 6.5–7.0 nursing hours per patient day (NHPPD) national benchmark. This total included all RNs, LVNs, CNAs and unit secretaries (although unit secretaries are not usually included as direct caregivers in NHPPD and were not included in the proposal). The California Rehabilitation Association Western Alliance did an informal regional survey in response to a query and reported responses of 6.5–7.5 NHPPD with no differentiation for patient acuity. At the American Medical Rehabilitation Providers Association (AMRPA) annual educational seminar, consultants Gill and Basano (2003) recommended 6.0–7.0 direct NHPPD, but stated a dependence on patient mix and acuity. The complexity of diagnoses served was not standard among facilities in the United States.

The most intriguing information came from the Commission for Medicare and Medicaid Services (CMS) in the Final Rule for PPS. Its development of relative weights for each patient CMG included variable functional (FIM) scores, age, diagnosis, cognitive status, complications and comorbidities that complicate care and require more nursing support. The case mix index (CMI) of these groups reflected historic data variances in resource use among the payment groups. Because labor is a large percentage of total costs, and nursing is a large percentage of labor, it was theorized that the CMI might be an effective indicator for patient acuity.

The Tool

Based on the numerical information in the Final Rule for PPS, it seemed feasible to integrate the CMI as an indicator for patient acuity and needed nursing resources, as well as to develop this information into an effective staffing matrix that would vary from shift to shift (based on 8-hour work patterns).

The following were the basic assumptions. A CMI of 1.0 represented the “average patient.” Higher or lower weights would indicate variances noted in acuity and resources needed. This is the basis for the CMG data used by the federal government.

Second, based on national benchmarks available to us, 6.5 NHPPD could be used to reflect the staffing needs of the average patient. Numbers suggested varied between 6.0 and 7.0; the midpoint was chosen. An example of what 6.6 NHPPD represents would be a unit with 10 average patients (with CMIs of 1.0), each warranting 6.5 hours of nursing care in a 24-hour period. This would allow 65 hours of nursing care (10 x 6.5) that would be divided into 8-hour shifts (65/8 = 8+ shifts).This division allows many staff to work in the 24-hour period, such as 3 staff on days, 3 staff on evenings, and 2 staff at night.

A relative ratio of CMI weight to NHPPD was developed. The following are ratio examples: 1.0 CMI:6.5 NHPPD; 1.1:7.2; 1.2:7.8; 1.3:8.5; and 1.4:9.1.For every variation above the average patient, a relative increase in nursing hours would be provided. Lower CMI scores would require relatively fewer nursing hours (Table 1). The information from fiscal year 2003, which compared resources assigned purely by census, was also included.

The next step was creating a patient classification tool that would be quick, accessible, and accurate in determining the average CMI for the day or shift. One was based on an MS Excel(r) format that would be easily accessed by both nursing and staffing (Figure 1). The form lists room, team, patient name, CMG, and CMI relative weights for day staff, evening staff, and night staff. At the bottom of the form the relative weights are calculated automatically, and when the census is entered, the average weight calculates automatically. There was no rounding up or down from the averages. Therefore an average of 1.4111 and an average of 1.4999 both qualify for the “1.4” matrix that allows 9.1 NHPPD.

Data were entered by the charge nurse every shift, based on the CMG/CMI listed by the admission nurse responsible for the IRF/PAI (Inpatient Rehabilitation Facility/Patient Assessment Instrument) data. Although a patient’s function varies from admission to discharge, at any one time on the unit there is a balance between patients who require more and less care. This balance affords the choice of using the same CMI throughout the patient’s stay, and eliminates the need of continuously adjusting each patient’s score. Discharged patients are removed from the list, and admission patients are listed as 1.1 weights until the final CMG is determined by Day 3.

This 1.1 designation was a conservative move meant to encourage the timely evaluation and listing of the correct CMG. However, at an annual review of the system, it was determined to be too low, and now 1.2 is used for admission scores. This matches the overall average CMG admission scores for the previous year. Further development of this area, as well as a time-efficient way to track patient progress and dynamic acuity, is needed.

The final step was to develop an individual matrix for each CMI average (Figure 3). Development was structured according to four tenents.

  1. Days and evenings required more resources than nights.
  2. The mix of RNs, LVNs, and CNAs was adjusted to meet the 1:6 medical/surgical benchmarks in the California Nurse Ratio Law. (This ratio can use up to 50% LVNs in the number of total licensed nurses.) This had the result of increasing licensed nurses and decreasing CNAs, and staff are still evaluating the effectiveness of this change.
  3. RN team leaders can vary assignments between partner couplets and primary care to meet the patient’s needs. Doing so engages a certain level of critical thinking and control which is generally seen as a positive factor by the professional staff.
  4. On a shift-by-shift basis, staffing looks at the census and average CMI and assigns staff accordingly.

Results

A full year of use has yielded positive results. In practice, the tool more effectively defines the variables that affect daily nursing needs. It eliminates rigid, empirical staffing expectations, and is less time-consuming than other systems. Measured outcomes include the following:

  • Consistently more staff when there is an increased number of higher acuity patients—The daily CMI averages varied between 1.06 and 1.6, with a yearly average of 1.26.
  • Maintenance of excellent patient outcomes with a 25-point gain, within federal benchmark LOS listed with CMGs—More serious cases have a decreased LOS. Eighty- seven percent of stroke patients leave within national benchmarks, compared with 56% before adjusted staffing. Although it seems to take longer to accomplish goals when staffing is less adequate, staffing is not the only factor affecting this statistic.
  • Demonstration of cost efficacy—The average NHPPD increased by 0.97, but overtime decreased. Based on this trade off, the change appears to be cost effective and well justified. Between fiscal years 2003 and 2004, cost in actual dollars (calculated by productive hours/units of service) increased by only 0.25%.
  • Higher nursing satisfaction scores from the Great Place to Work survey—Positive comments about the improvement of the staffing situation have been received, and the unit hopes to see this more positive attitude reflected in the survey. All rehabilitation nursing positions were filled at the time this article was written. The authors believe the staff enjoyed participating in the development process, which seemed to validate their rehabilitation contributions.
  • Finally, the unit is better prepared to advocate for rehabilitation nursing in the legislative arena.

In the future, the goal is to further refine the mix of staff needed within the NHPPD. It would also be helpful to validate the actual nursing hours or care provided to support the hypothetical benchmark chosen.

About the Author

Diane Romito, BSN RN CRRN, is director of rehabilitation services at Scripps Rehabilitation Center, Encinitas, CA. Direct correspondence to her at Scripps Rehabilitation Center, 354 Santa Fe Drive, Encinitas, CA 92024 or Romito.Diane@scrippshealth.org.

References

Rules and regulations 152, 66 § 41342–41396 (August 7, 2001). Federal Register (2001), 66(152) Tuesday, August 7. Rules & Regulations, 41342-41396.

Gill/Basano Consulting. (2003, November). Achieving Financial Success through Operational Change: Matching Resource Utilization with Revenue Post-PPS. Paper presented at the meeting of the American Medical Rehabilitation Providers Association 1st Annual Medical Education Conference, Las Vegas, NV.

Halloran, E., & Halloran, D. C. (1985). Exploring the DRG/Nursing Equation. American Journal of Nursing, 85,1093–1095.

Halloran, E., & Kiley, M. (1987). Nursing dependency, diagnosis-related groups, and length of hospital stay. Health Care Finance Review, 8(3), 27–36.

Mark, B. A. (2002). What explains nurses’ perceptions of staffing adequacy? Journal of Nursing Administration, 32, 234–235.

Neatherlin, J., Prater, L. (2003). Nursing time and work in an acute rehabilitation center. Rehabilitation Nursing, 28, 186–190.

Welton, E., John, M., & Halloran, E. J. (1998). A Comparison of Nursing and Medical Diagnoses in Predicting Hospital Outcomes. Unpublished manuscript, University of North Carolina at Chapel Hill, School of Nursing.