Staff members of the Center for Health Workforce Studies have been working with the Lewin Group on the update of the HRSA Nurse Supply Model (NSM) and Nurse Demand Model (NDM). Although the exact analyses included in the NDM could not be replicated at the county level due to data constraints, the basic logic employed in the NDM was very useful in thinking about demand for RNs.

The project staff decided to apply a simplified version of the NDM logic to: 1) estimate health care utilization in different settings for counties (e.g., inpatient days); 2) estimate current national RN staffing by setting (e.g., RNs working in inpatient units); 3) calculate national RN staffing intensity for each setting (e.g., RNs per inpatient day); 4) apply national RN staffing intensity ratios to measures of utilization for each county; and 5) sum estimate demand for each setting to produce overall RN demand for individual counties. Each step is summarized briefly below.

A. Estimate Health Care Utilization

The data on county-level health care utilization primarily came from the Area Resource File (ARF). The ARF included data on:

• Short-term inpatient days (non-psychiatric hospitals)
• Long-term inpatient days (non-psychiatric hospitals)
• Psychiatric hospital inpatient days
• Nursing home unit inpatient days (hospitals)
• Outpatient visits (non-emergency)
• Emergency department visits

The number of (non-hospital) nursing home residents in a county was obtained from the 2000 Census. This was based on the Census short-form data, which is theoretically obtained from 100% of the U.S. population.

The number of home health patients per county was estimated using the age and gender distribution of the population, based upon national age-specific and gender-specific utilization rates from the Centers for Disease Control and Prevention (CDC).

Although this estimate was based upon population characteristics rather than actual use of services, home health patients by definition were receiving services where they live, so this was somewhat less problematic than estimating other types of utilization based upon population characteristics.

B. Estimate Current National RN Staffing

Data for current levels of RN staffing by setting were taken from the 2000 NSSRN, which included data on the number of RNs employed in the following types of care:

• Short-term inpatient (non-psychiatric hospitals)
• Long-term inpatient (non-psychiatric hospitals)
• Psychiatric inpatient (non-Federal)
• Nursing home unit (hospital)
• Outpatient (non-emergency)
• Emergency outpatient
• Non-hospital nursing home
• Home health
• Nurse education
• Public/community health
• School health
• Occupational health
• Non-hospital ambulatory care
• Other nursing care

These numbers were combined with the national utilization data described above to compute national levels of RN staffing intensity for the various types of care. The national ratios are shown in the table below.

Table 14. National RN Staffing Ratios by Type of Care

C. Estimating RN Demand by County

These national staffing ratios were then applied to the utilization rates for each county. For example, the national ratio was 4.97 RNs working in hospital inpatient units per inpatient day. If County A has 12,000 inpatient days per year, their demand for RNs in inpatient units is estimated at 59.6 (4.97 x [12,000/1,000]).

Overall RN demand for the county was obtained by summing RN demand in the county across all settings. (This procedure also opens the possibility of comparing setting-specific demand to setting-specific supply if data on RN supply by setting are available at the county level.)

D. Use Supply of RNs to Estimate RN Shortages

RN shortages were thus measured as follows:

RN shortage = Estimated demand for RNs in the county
minus the number of RNs in the county
(adjusted for commuting patterns).

Raw shortage estimates were then standardized as a percent of demand. The results are presented for all counties in the U.S. in the map in Figure 8. The counties with the greatest shortages are shaded black. The full technical report has separate maps for all 50 states, along with a table with the numerical scores.

This method has advantages over any of the other methods examined in this study, especially in relation to the guiding principles initially proposed for the study:

• It uses nationally available data that is periodically updated.
• It uses actual health care utilization patterns by county.
• It accounts for multiple types of care (including non-clinical services).
• It accounts for differences in RN staffing intensity across settings. Some limitations persist, however. It does not account for county or state variations in health systems (e.g., HMO penetration, use of LPNs), and does not account for patient acuity within types of care. Furthermore, it assumes current RN staffing levels were adequate at the national level in 2000, which may not have been the case.

The NDM uses factors such as HMO penetration and LPN staffing in regressions to adjust estimated staffing intensity and make it specific to each county rather than applying national ratios. A similar procedure might eventually be used to do the same thing here.