The Nursing Model Components

Published 2016

In this module:

Modeling Supply

Estimating Base Year Nurse Supply

For most states, estimates of the current supply of RNs and LPNs came from the pooled 2010-2014 ACS files. Data were combined to increase the sample size to provide stable state-level estimates by education level, age, sex and race/ethnicity. (Race/ethnicity is a new component added to the supply model.) The ACS sample weights from the 5-year file were recalibrated to sum to the state totals of RNs and LPNs in the 2014 ACS. HWSM is designed to use data from state licensure files as data become available for use instead of ACS data.

Georgia, Oregon, South Carolina, and Texas provided licensure data. For these four states, starting supply is based on licensure data instead of the ACS.1 Nurses are included in the licensure files if the nurse is licensed and active in nursing in the state being modeled. Licensure files verify the nurse is licensed in the state. With the ACS data, licensure is implied by the ACS respondent self-reporting activity status and occupation as a nurse.

The ACS estimates extrapolated to 2015 averaged 5-8% higher for RNs-LPNs compared to estimates from the 2015 licensure files. Differences varied by state-occupation combinations.2  A comparison of ACS and licensure files for these four states suggests that:

  • RN estimates from the ACS appear to be more consistent with licensure files than are the LPN estimates from ACS. This likely reflects that the LPN sample size is smaller in ACS compared to sample size for RNs.
  • If, at the national level, ACS overestimates FTE supply of nurses, then estimates of national demand based on ACS also might be overestimated by a similar percentage.
  • Information from additional states would help determine to what extent the ACS accurately reflects estimates of supply from state licensure files.

Modeling New Entrants to the Nursing Workforce

New entrants reflect nurses entering the workforce for the first time upon completion of a nursing program. They also include individuals who migrate mid-career from one geographic area to another (discussed later). HWSM used first time, U.S.-educated candidates taking the National Council Licensure Examination (NCLEX) as the starting point for estimating the number of new entrants to the nursing workforce. In 2014, there were 157,882 first-time, U.S.-educated takers of the NCLEX-RN.3 Of these, 70,857 nurses had completed a baccalaureate degree. Another 87,025 had completed a diploma or an associate degree.4 There were 55,489 first time takers of NCLEX-LPN in 2014. We assume that nurses who initially fail the NCLEX will retake the test at least twice. We also assume an eventual pass rate of:

  • 96% of RNs trained at the associate level
  • 98% of RNs trained at the baccalaureate level
  • 95% of LPNs

We assume that these nurses will enter the workforce.

For modeling future supply under a Status Quo scenario, HWSM assumed that annually the number of nurses passing the NCLEX includes 69,440 RNs at the baccalaureate level, 83,540 RNs at the associate or diploma level, and 52,720 LPNs. The new entrant statistics for RNs include the estimated 16,000 LPNs who further their education and become RNs each year. Alternative supply scenarios modeled include training 10% more or 10% fewer nurses relative to current numbers. This illustrates the sensitivity of supply projections to the number of nurses being educated each year.5

The National League of Nursing survey contains data on students enrolled in entry level nursing programs. In 2014, the survey showed that 91% of LPNs are female, 86% of RNs in associate or diploma programs are female, and 85% of RNs in baccalaureate programs are female.6  Estimates of the age distribution for new nurses come from analysis of the 2008 National Sample Survey of Registered Nurses (Exhibit 45). Limited data are available on the age distribution of new LPNs. National League for Nursing data from 2008-2009 suggest that the age distribution for LPNs is similar to the age distributions for Diploma and Associated Degree RNs.7  Hence, for modeling, we use the age distribution of Diploma and Associated Degree RNs as a proxy for the age distribution of LPNs. The race and ethnic distribution of new nurses varies widely by state. We use the race/ethnicity distribution of nurses age 30 or younger in the 2010-2014 ACS as a proxy for the age distribution of new nurses (Exhibit 46).

  • 1These state licensure data were 2015 data, while the ACS data was 2014 data. Consequently, to obtain 2014 estimates for these states we projected backwards based on projected graduates and attrition from 2014 to 2015.
  • 2For example, ACS estimates for RNs in Georgia and South Carolina appear to be similar to estimates from state licensure files. For Oregon the ACS estimate is smaller, and for Texas the ACS estimate is larger. For LPNs, the Georgia and Texas estimates are relatively consistent with estimates from state licensure files. For Oregon and South Carolina, the ACS estimates are much larger in percentage terms than are estimates from state licensure files.
    Source GA OR SC TX 4-State Total
    RNs: 2015 projected from 2014 ACS 85,600 33,700 47,900 222,300 389,400
    RNs: 2015 Licensure files 84,600 37,600 49,400 200,700 372,200
    Difference (ACS-Licensure) 1,000 (3,900) (1,500) 21,600 17,200
    % Difference 1% -10% -3% 11% 5%
    LPNs: 2015 projected from 2014 ACS 29,800 4,400 12,900 78,200 125,400
    LPNs: 2015 Licensure files 27,900 3,400 8,600 76,500 116,400
    Difference (ACS-Licensure) 1,900 1,000 4,300 1,700 9,000
    % Difference 7% 29% 50% 2% 8%
  • 3Foreign-educated NCLEX takers are excluded from this analysis because there is no evidence that employers currently are relying on foreign-educated nurses to fill nursing vacancies. Many foreign educated nurses take NCLEX but do not come to the U.S.
  • 4National Council of State Boards of Nursing. 2014 Nurse Licensee Volume and NCLEX® Examination Statistics (PDF - 1 MB).; 2015.
  • 5Additional scenarios modeled include ±5% change in nurse productivity levels and ±2 years earlier or delayed retirement.
  • 6National League for Nursing. Nursing Student Demographics: 2013-2014.; 2014.
  • 7National League for Nursing. Biennial Survey of Schools of Nursing, 2008-2009.; 2010.

Modeling Nurse Workforce Participation

Nurses might temporarily leave the labor force due to family, education, economic or other considerations. Permanent departure from the labor force might be due to retirement, career change to another occupation, or death. Permanent departure can also result from emigration (moving away from that geographic location to work elsewhere). This section describes permanent attrition from the workforce modeling, labor force participation, and weekly hours worked. Modeled hourly wage also is described. This is one input used to model labor force participation and hours worked patterns.

Attrition Patterns

In this section, we describe analyses and assumptions regarding nurses who permanently leave the nursing workforce. This differs from temporary departures such as for child rearing, illness, or other reasons where the nurse intends to eventually return to employment.

We modeled a small amount of attrition each year for nurses under age 50. The preliminary RN supply projections assumed that about 97% of RNs taking the NCLEX exam for the first time would eventually pass and enter the workforce. We then modeled labor force participation rates using the ACS. We estimate that about 92-95% of RNs would be active in the workforce through age 50 depending on age. After age 50, we model attrition from the workforce as nurses age.

A challenge with ACS data is that ACS does not collect occupation data if an RN has been out of the workforce for five or more years. However, if the RN remains in the workforce but in a non-nursing position, their occupation will not indicate RN. Instead, it will indicate the current occupation. While our starting supply of RNs will be accurate, our labor force participation rates will not reflect some younger RNs permanently leaving nursing.

HRSA’s 2008 NSSRN indicates that a small percentage of RNs under age 50 intends to leave the workforce. In addition, a small percentage of recent graduates are not employed in nursing.8  However, this snapshot for 2008 was in the middle of a national recession. Also, of nurses not working in nursing, many plan to return to nursing. One challenge is that the survey question asks if a nurse intends to leave nursing in the next 3 years. It is unclear from the response whether the intention is to permanently or temporarily leave nursing. The survey indicates that for nurses under age 50 who are not working in nursing approximately 57.5% have been out of nursing for 0-4 years. Presumably, most of these nurses are represented in the ACS unless they are working in a different occupation so their occupation code changed. An estimated 42.5% of nurses who have left nursing have been out of nursing for five or more years. These nurses would not be represented in ACS as a nurse. Therefore, the ACS likely understates the number of trained nurses who are not active in nursing by a few percentage points. Some nurses who indicate they are not in nursing are in other health occupation or government jobs. It is possible that these nurses still are working in a role that requires a nursing background or degree even though the nurse is not practicing in a traditional nursing role.

According to the 2008 NSSRN, by age 30-34, approximately 8.7% of nurses are not employed in nursing. This increases to about 10% from age 40-49.8  Analysis of the ACS indicates about 92-95% not employed in nursing (with the percentage not employed varying by age). Based on the data in Figure 3-4 and Figure 3-24 of the 2008 NSSRN report, we make the following assumptions:

  • From when the nurse initially enters the workforce through age 39, there is a 0.33% probability of leaving the workforce each year.
    • Say a nurse enters the workforce at age 25. By age 39, they have a cumulative 3.3% probability of having permanently left nursing (on top of an approximately 5% probability of being out of the workforce).
  • Between age 40 and 49, we calculate an estimated 0.42% probability of leaving the workforce each year.
    • By age 49, a nurse who entered the workforce at age 25 has an 8.8% cumulative probability of having permanently left the workforce (on top of an approximately 5% probability of being inactive).

To summarize the modeling assumptions:

  • Approximately 3% of nurses who graduate from a nursing program do not pass the NCLEX and enter the workforce.
  • There is an 8.8% probability of leaving nursing by age 49.
  • There is also a 92-95% employment rate for those in nursing through age 49.
  • From age 50 and older, the nurses have a probability of permanently leaving the workforce that increases with age (as described later).
  • For each 100 nurses graduated from a nursing program at age 25, we calculate by age 49:
    • Approximately 84 of these nurses would be working in nursing
    • 3 never entering the workforce
    • 8 having left nursing altogether
    • 5 currently out of the workforce

Multiple approaches have been explored and used to estimate nurse attrition patterns. Prior to 2016, ACS-derived labor force participation rates by age and sex for RNs age 50 and younger were used. For RNs over age 50, labor force participation rates for college educated men and women over age 50 were used as a proxy for labor force participation rates for male and female RNs over age 50 with a similar education level. As noted above, ACS does not capture occupation for individuals out of the workforce for five years or more.

Refined estimates of nurse attrition patterns are used in the updated supply projections based on licensure data from Oregon, South Carolina and Texas (Exhibit 47). Multiple years of licensure data (2010-2015) were analyzed. These licensure data do not contain individual identifiers to link nurses across years. Therefore, we compared the age distribution of active RNs in one year to the expected age distribution in a subsequent year if all RNs active in prior year had remained active. The gap reflects net attrition from the workforce (including mortality, retirement and net migration out of the state).

The Oregon data reflected a survey question about intention to retire within the next three years. Staff from the Oregon Center for Nursing provided informal guidance on this question. According to staff, approximately a quarter of all nurses who in 2010 had expressed an intention to retire within the next three years were still in the workforce in 2014. Therefore, we adjusted the estimated attrition patterns to reflect Oregon’s previous analysis that intention to retire might overstate actual retirement. Also, we added mortality patterns to the intention to retire patterns to estimate overall attrition rates.

The supply projections are based on the average attrition patterns estimated across the three states. Attrition patterns differ by age and nurse type (RN or LPN) but do not differ by other nurse characteristics. This is an area of ongoing research. Also, attrition patterns used in the model reflect input from participants in a recent nurse workforce retreat sponsored by HRSA.9

For nurses age 70 and older, the sample sizes in the state licensure files are small. Estimates of attrition patterns fluctuate accordingly. Therefore, we assume that starting at age 70 the annual attrition rate is 50%. In addition, we model that annually approximately 16,000 LPNs become RNs. Also, approximately 16,200 RNs leave the RN workforce each year to become nurse practitioners. This reflects that close to 15% of NPs remain practicing in a traditional RN role.

The approach used for modeling attrition patterns reflects limitations with data sources such as ACS. If a person has been out of the workforce for 5 years or more, then ACS does not collect information on prior occupation. Likewise, if a person left nursing for a career outside of nursing, then the ACS captures data on the current occupation but there is no indication of previously having been working in nursing. Hence, estimates of attrition patterns based on ACS can understate true attrition patterns.

  • 8 a b Health Resources and Services Administration. The Registered Nurse Population: Findings from the 2008 National Sample Survey of Registered Nurses (PDF - 3 MB). U.S. Department of Health and Human Services; 2010.
  • 9In July 2016, HRSA and the Montana State University Center for Interdisciplinary Health Workforce Studies sponsored a 3-day meeting of nurse workforce researchers. The meeting goal was to critique alternative approaches to modeling nurse workforce supply and demand and to provide input on HRSA’s workforce modeling assumptions, inputs and methods. One outcome of this meeting was to incorporate workforce attrition probabilities among younger nurses. Another was to adjust estimates of the number of RNs being trained as advanced practices nurses to reflect that some RNs become trained as APNs but still continue to practice in a traditional RN role.

Hourly Wages

Earnings potential (modeled in terms of hourly wages) is modeled as a function of nurse characteristics and external factors as summarized in Exhibit 48. The equations to predict hourly wages were estimated separately by nursing occupation using data from the 5-year (2010-2014) ACS for individuals who are currently employed. Hourly wages were calculated by dividing estimated weekly earnings by estimated weekly hours. We omitted records where hourly wages were below the 5th percentile or above the 95th percentile. Estimated hourly wages for these omitted records were outside the plausible range. State mean hourly wage for that occupation from the OES data was included as an explanatory variable. Mean wage varied across states and years. Both occupation mean hourly wage and each person’s hourly wage (i.e., the dependent variable in the regression) were adjusted to 2015 dollars using the Consumer Price Index and adjusted to a national average using a state cost-of-living index.10

For the nursing occupations modeled, individual wage is highly correlated with occupation mean wage in that state. Wages tend to increase for those early in their career and rise more slowly above age 35. Male nurses tend to earn higher hourly wages. Wages vary by race/ethnicity. Hourly wages rise with the percentage of the population living in suburban areas. As with many cross-sectional analyses using person-level data, the R-squared values for these equations are low. This reflects that these regressions explain only a small portion of cross-sectional variation in hourly wages worked.

Hours Worked

Forecasting equations related average hours worked to nurse age, sex, education level, state overall unemployment rate, and average wage in the occupation. Data for all variables came from the ACS with the exception of average wage. Average wage was obtained from the BLS. To convert average hours worked into Full Time Equivalents (FTEs), an assumption needed to be made about the average number of hours worked per week by a full-time nurse. Analysis of the ACS suggests that the average hours worked per week is 37.3 among nurses working at least 20 hours per week for both RNs and LPNs. For modeling purposes, HRSA is now defining an FTE as 40 hours per week. This measure can remain constant over time and across health occupations. While workforce projections published before 2017 used different hour estimates to define an FTE, 40 hours was used from 2017 onward.

Ordinary Least Squares regression coefficients highlighted a number of interesting data points:

  • Average weekly hours worked declined substantially among older nurses (Exhibit 49).
  • For both RNs and LPNs, weekly hours worked decline rapidly from age 60 onward.
  • On average, male RNs work 2.78 more hours and male LPNs work 1.77 more hours than their female counterparts.
  • Hispanic RNs work 2.28 hours more than non-Hispanic white RNs
  • RNs with a baccalaureate or graduate degree work 1.43 hours more than RNs with an associate or diploma degree.
  • RNs and LPNs in states with a larger proportion of the population residing in rural areas11  tend to work more hours.
  • Hours worked per week by RNs and LPNs rise slightly with the unemployment rate.

Activity Status

Activity status for nurses is modeled using prediction equations derived from ACS (2010-2014) data. This analysis focused on nurse clinicians under age 50 (as the activity status for clinicians over age 50 is modeled as attrition). The dependent variable is whether the nurse is employed or unemployed. The unemployed population is everyone currently not employed but whose most recent employment in the past five years was in nursing. Explanatory variables are the same used to model hours worked.

The overall activity rate for RNs and LPNs under age 50 was, respectively 95% and 91%. The odds of being employed vary by nurse demographics—in particular age (Exhibit 50). A higher overall unemployment rate slightly raises the odds of RNs being employed. Higher earnings potential is associated with a slight decrease in the odds that RNs are employed. Interaction terms for sex and age group are included to reflect that labor force participation differences between men and women might differ by age group. To compare male RNs age 35-39 versus female RNs of the same age, one multiplies the odds ratios for male and the male-age interaction. For example, male RNs age 35-39 have twice the odds (0.71*2.81=2.00) of being active in the nursing workforce as do female RNs of the same age. Male RNs age 45-49 have odds of being active in the labor force that are 1.38 times the odds for female RNs of similar age.

Compared to non-Hispanic white nurses, the odds that an RN is active in nursing is 38% higher for Hispanics, 32% higher for non-Hispanic blacks, and 23% higher for non-Hispanic “other race” RNs. Non-Hispanic black LPNs have 42% higher odds of being active in nursing compared to non-Hispanic white LPNs.

Cross-state Migration Patterns

Previous nursing projections for HRSA modeled two migration scenarios. First, newly trained nurses remain in the state in which they are trained. Second, nurses completing training migrate across states based on the relative distribution of growth in employment opportunities. Under this second scenario, states with faster employment growth might experience a net inflow of nurses trained in other states with fewer employment opportunities.

For this update, we assume that nurses will initially enter the workforce in the state where they took the NCLEX exam. We then model cross-state migration based on prediction equations estimated using logistic regression on with the 5-year (2010-2014) ACS file. Cross-state migration models whether a person moves out of a state. It then models whether a person moves into a state. Of 134,593 RNs in the 5-year file (with different nurses surveyed each year), 2,526 (1.9%) indicated working in a different state compared to a year ago. Of the 34,555 LPNs in this file, there were 495 (1.4%) who indicated working in a different state compared to a year ago.

Analysis of nurse cross-state migration patterns suggests that:

  • The probability of migration declines with age. Nurses age 30 and below have the highest probability of migrating to another state.
  • Male RNs are more likely to move than female RNs.
  • RNs whose predicted hourly wages (a continuous variable) exceeds the national average wage are less likely to migrate to another state.
  • RNs with higher levels of educational attainment (bachelors and graduate-level degrees) are more likely to move across state.
  • Non-Hispanic White RNs are more likely to relocate compared to other race/ethnicity groups (Exhibit 51).

Using the ACS sample weights, this analysis suggests that annually approximately 59,802 RNs and 12,220 LPNs change states. When modeling cross-state migration patterns, HWSM uses the above equations to generate a probability that each nurse will migrate out of the state. This probability is then compared to a random number between 0 and 1 using a uniform distribution. If the random number is below the estimated probability of moving, then the nurse is moved out of that state.

We ensure that the national number and characteristics of nursing moving out of states matches the number and characteristics of nurses moving into states. When a nurse is simulated to move out of state, that nurse is reassigned to another state using the distributions in Exhibit 52. Between 2010 and 2014, of the estimated 59,802 RNs who moved to another state each year approximately 1% moved to Alabama and 8.1% moved to California.

Over time, projections of number of nurses exiting a state change based on the characteristics of nurses in that state and overall number of nurses. The variation across states and across years reflects the modeling of migration determinants. It also reflects the use of a random number generator to move nurses across the various states based on the geographic distributions described previously. As illustrated in Exhibit 53 and 51, Alaska is projected to have a net import of 179 RNs per year and 51 LPNs per year. That is, more nurses will move into the state each year than move out.

Modeling Demand

The projected demand for nurses is derived from the common model outlined in Modeling Demand for Health Care Services and Providers. Predicted probabilities were applied to the simulated micro-data set for future years to obtain projected service use specific to the settings that employ nurses. For example, projected growth in hospital inpatient days and emergency visits is used to project growth in demand for RNs and LPNs employed in hospitals. For work settings outside the traditional health care system, HWSM used the size of the population most likely to use those services to project demand (Exhibit 55).

HWSM used provider staffing patterns to project demand for health care workers by delivery setting based on the demand for health care services. As illustrated in Exhibit 55, nurses were found in almost all care delivery settings. Nurse staffing patterns were calculated using the portion of national FTE nurses providing care in each setting. These were then divided by current estimates of the workload driver in that work setting. The baseline demand projections assumed these ratios remained constant over time. The demand for nurses in academia is based on the estimated number of nursing graduates, assuming current ratios of nurse educators-to-students remained constant. Estimates of the distribution of nurses across employment settings came from analysis of the 2015 OES. We used data from the 2008 National Sample Survey of Registered Nurses to break our hospital totals from the OES data into inpatient and emergency departments. We also used NSSRN data to break out nurses in education to those providing school health and those in nursing education.12

National staffing ratios by care delivery setting at baseline were applied to the projected service use to obtain the staffing requirement by setting. These were aggregated to obtain the total demand for nurses. Projections were made at the state level and summed to produce national estimates.

Baseline and Alternative Nursing Workforce Projections

Supply Projections

HWSM can project future nurse supply under multiple scenarios. These illustrate the sensitivity of the models to the continuation of trends in key supply determinants. The Status Quo scenario models the continuation of current numbers of nurses completing their nursing education and current patterns of labor force participation. Labor force participation is affected by attrition, being temporarily out of the workforce, and hours worked patterns. Labor force participation varies by nurse demographics, education level, and other characteristics of the nurse or community. The Status Quo scenario models the continuation of these patterns taking into account the changing demographic and changing education levels of the nursing workforce.

Alternative supply scenarios modeled include the impacts of:

  • retiring two years earlier or delaying retirement by two years, on average
  • graduating 10% more or 10% fewer nurses annually than the Status Quo
  • a gradual 5 % increase or 5% decrease in average nurse productivity levels

The early or delayed retirement scenarios simply shift workforce attrition patterns for nurses age 50 and older by ±2 years. For example, a nurse who would have retired at age 65 under the Status Quo scenario would now retire at age 63 under the Early Retirement scenario. That nurse would retire at age 67 under the Delayed Retirement scenario.

The ±5% change in productivity scenarios assume that each year between 2014 and 2030 there is small (about 0.31%) change in nurse productivity. The cumulative impact of those annual changes reaches ±5% impact by year 2030 versus year 2014. Productivity is defined for supply modeling as the number of patients that can be treated by 1 FTE nurse over the course of a year (as defined by the staffing levels in Exhibit 55). Productivity changes could occur because of changes in technology or practice patterns, or through changes in average hours worked. A ±5% productivity change is equivalent to ±5% change in FTE supply.

Demand Projections

The Status Quo scenario for modeling demand assumes that recent (2009-2014) patterns of care use and delivery will remain unchanged. It does, however, consider population growth and aging as well as expanded insurance coverage that has occurred and is projected to occur under the Affordable Care Act. Care use and delivery patterns likely will change over time. There is limited published information or data to use for modeling how care use patterns might change over time. There is also limited information on the nursing workforce implications of changes in care use or delivery. Using information from several published demonstrations of emerging care delivery models, we simulated the potential impact of such changes on the nursing workforce.

The following examples combine information from the published literature with HWSM. They illustrate the changing roles of RNs and LPNs within a care coordination model. These models are currently part of ongoing studies on nurse utilization in coordinated care settings. Each pilot study utilizes RNs in roles such as nurse care managers working with other staff to coordinate care, improve patient self-education, and adhere to treatment plans.

The pilot studies also illustrate how RN care managers coordinate with pharmacists, behavioral health providers, and licensed clinical social workers. These and other emerging care models focused on improving population health highlight the shifting roles of RNs. Service demand is reduced and redirected from higher cost hospital inpatient and emergency department settings to more clinically appropriate outpatient and community-based care settings. As a result, some future reductions in clinical RN staffing in hospital settings are possible.

The Camden Coalition (Camden, New Jersey) provides health services to a patient population that experiences multiple social barriers to accessing health services.13  RNs are utilized in care manager roles to support and oversee patients’ transition into primary care. Camden Coalition’s RN model focuses on patient engagement. Patient care is tailored to the specific needs of each patient to ensure a more effective transition into primary care. To date, hospital admissions by “super users,” or patients who frequently utilize hospital services, declined by 57%. Emergency department visits declined by 33%, and the cost of care decreased by 56%.14  Implementing such a model at the national level could reduce demand by about 158,000 RNs and 14,000 LPNs in hospital settings in 2030. This assumes super users account for 4% of all visits to the emergency room15  and 14% of inpatient hospital days.16

CareOregon (Portland, Oregon) is a non-profit Medicaid managed care plan. It serves 128,000 low-income residents representative of one-third of the state’s Medicaid enrollees.17 Two-thirds of patients have one of 12 common chronic conditions including diabetes, depression, and chronic heart failure. Two-thirds of the health plan members are children. More than 5,700 adults are dual-eligible for Medicaid and Medicare services. CareOregon provides two health care tracks:

  • Primary Care Renewal (a patient-centered medical home initiative) works through safety net clinics
  • Care Support, a multidisciplinary management program for members with high risk of poor health outcomes

Both health care tracks utilize nurse care managers on care coordination teams. They work with social workers and care coordination assistants to monitor patients and identify risks before health crises occur. Nurse care managers coordinate services, patient education, and treatment adherence. Care Support reported decreases in non-obstetric hospital admissions and emergency department visits of about 34%. Offering such a model to all Medicaid beneficiaries nationally could result in lower hospital-based RN and LPN FTE demand in 2030 by about 151,000 and 11,000, respectively. This results from lowered levels of service use in the inpatient and emergency settings.

Community Care of North Carolina (CCNC) (Raleigh, North Carolina) utilizes nurses as managers in the provision of services for chronically ill patients.18  The patient population includes the Aged, Blind, and Disabled sub-population. This sub-population accounts for nearly 70% of the service dollars but fewer than 30% of program recipients. Nurse care managers work with physicians and pharmacists to provide coordinated patient care. Duties include but are not limited to:

  • medication reconciliation
  • coordination with medical homes and primary care providers providing patient care
  • coordination and with community agencies and other local resources providing support services for the Medicaid population

CCNC reported that admission rates decreased by 21% and emergency department visits decreased by 32.8% between 2006 and 2011. Implementing this program for a similar national Medicaid population could reduce the projected 2030 FTE demand for hospital-based RNs and LPNs by about 103,000 and 7,000, respectively.

These pilot studies illustrate that while demand for nurses might rise for some roles (e.g., care management), the overall demand for nursing services could fall in hospital settings. In general, the literature suggests that the decline in nurses resulting from lower health care utilization will exceed the increase in demand for nurses for care management. Hence, the demand projections presented in this report might be high. The demand projections could understate projected surpluses if current supply trends continue.

Population Health

The above pilot studies focus on the short-term implications of care utilization and staffing among select high-utilization subsets of the population. There are broader trends in population health that have long-term implications for the nurse workforce. New policy guidelines, provisions in the ACA, and new reimbursement models are designed to promote preventive care. These have the potential to improve the health of the entire population (beyond just high risk, high utilization subpopulations). One example is guidelines and reimbursement for counseling and treatment to promote a healthful diet and physical activity to individuals at high risk for developing cardiovascular disease or diabetes. Two other initiatives target smoking cessation and controlling blood pressure, cholesterol levels, and hemoglobin A1c levels.19 20 21

We modeled the potential long term health impacts and nurse demand of achieving certain health goals. These models built on a recently published study22 and employed a Markov-based microsimulation approach described in detail elsewhere.23 24 25  The following population health goals were considered:

  • Sustained 5% body weight loss for overweight and obese adults: Counseling and pharmacotherapy have been shown to reduce excess body weight by 5% or more. This lowers the risk for diabetes, cardiovascular disease, and other morbidity.25
  • Improved blood pressure, cholesterol, and blood glucose levels: Published trials examined patients with elevated levels. These trials show counseling and pharmacotherapy can improve cholesterol, blood pressure, and hemoglobin A1c levels.26 27 28
  • Smoking cessation: Smoking cessation can reduce risk for cancer, heart disease, and other morbidity.29

The model’s prediction equations came from published clinical trials and observational studies. The simulation was conducted using a nationally representative sample of adults. This sample was constructed using the 2013-2014 National Health and Nutrition Examination Survey combined with national population projections. Outcomes from this model were then used in HWSM to model the demand for health care services and nurses.

Between 2015 and 2030, achieving these population health goals could reduce:

  • cases of heart disease by 10.2 million
  • stroke incidence by 3.2 million
  • myocardial infarction incidence by 3 million
  • incidence of cancer and other diseases.22

This reduction in incidence/prevalence would reduce demand for nurses. However, the improved health of the population would also reduce mortality. If the modeled goals were achieved, the projected size of the population in 2030 would be 6.3 million higher than current Census Bureau projections. These additional 6.3 million people would be primarily elderly. This includes about 2.9 million age 75 or older, 2.3 million age 65 to 74, 1 million age 45 to 64, and approximately 30,000 adults under age 45.

Compared to the baseline demand scenario, by 2030, national demand for RNs and LPNs under this population health scenario would be higher by approximately 105,800 FTEs and 69,500 FTEs, respectively. This increased demands would be to support the larger population even though per capita use of nursing services would be lower. This scenario suggests that efforts to improve population health might reduce demand for nurses in the short term. Overall demand for nurses could rise in the long term to the extent that preventive care increases longevity.

Modeling Supply and Demand by Metropolitan versus Nonmetropolitan Location

State-level indicators of metropolitan/nonmetropolitan for modeling nurse supply in 2014 came from analysis of the ACS. Using USDA 2013 Rural-Urban Continuum Codes (RUCC), we classified each county or county subpart in a PUMA as metropolitan or nonmetropolitan.30 Metro and non-metro county classifications are based on Office of Management and Budget (OMB) delineation as of February 2013. OMB defines metro counties as counties with RUCC values of 1, 2, or 3. All other counties are defined as non-metro. The RUCC file was merged with the PUMA-county crosswalk file available through the Missouri Data Center. This allows us to map PUMA to a county. We were able to assign a PUMA as either metro or non-metro based on the RUCC definitions. Finally, we merged the PUMA-county crosswalk file including the metro/non-metro indicator with the ACS file to generate statistics by metro and non-metro.

Indicators of metropolitan/nonmetropolitan to model demand for nurses is based each person’s metropolitan status as indicated in the BRFSS. Metropolitan status is based on the “MSCODE” variable in the 2013-2014 BRFSS survey data. Based on the BRFSS variable, metropolitan area is defined where one of the following criteria is fulfilled:

  • In the center city of an MSA
  • Outside the center city of an MSA but inside the county containing the center city
  • Inside a suburban county of the MSA
  • In an MSA that has no center city

The population living in metropolitan areas would utilize approximately 83% of the nation’s RN services. An estimated 85% of FTE RN supply is in metropolitan areas. Though the 83% and 85% are similar, many patients in nonmetropolitan areas might travel to metropolitan areas to receive specialized care. Nurse staffing patterns could differ between metropolitan and nonmetropolitan areas. This could reflect differences in patient acuity levels and differences in productivity due to patient volume.

  • 13Camden Coalition of Healthcare Providers. Outreach to High Utilizing Patients – Basics of Care Management and Care Transitions in Camden, NJ.
  • 14Hong CS, Siegel AL, Ferris TG. Caring for high-need, high-cost patients: what makes for a successful care management program? Issue Brief (Commonw Fund). 2014;19:1-19.
  • 15Castillo EM, Brennan JJ, Chan TC, Killeen JP, Vilke GM. 85 Factors Associated With Frequent Users of Emergency Department Resources. Annals of Emergency Medicine. 2012;60(4):S32. doi:10.1016/j.annemergmed.2012.06.364.
  • 16Pennsylvania Health Care Cost Containment Council. Pennsylvania’s “Super-Utilizers” of Inpatient Hospital Care (PDF).; 2014.
  • 17Dorr D. Teamwork and Medical Home in Rural Settings; A Case Study with Care Management Plus. Presented at the: Presentation for the Oregon Health and Science University; September 2008.
  • 18Dobson LA, Hewson DL. Community care of North Carolina--an enhanced medical home model. N C Med J. 2009;70(3):219-224.
  • 19U.S. Preventive Services Task Force. Healthy Diet and Physical Activity for Cardiovascular Disease Prevention in Adults With Cardiovascular Risk Factors: Behavioral Counseling.; 2016.
  • 20Centers for Medicare and Medicaid Services. Medicare Preventive Services. Accessed July 18, 2019.
  • 21U.S. Preventive Services Task Force. Final Recommendation Statement: Obesity in Adults: Screening and Management.; 2019.
  • 22 a b IHS Markit. The Complexities of Physician Supply and Demand: Projections from 2015 to 2030: 2017 Update (PDF - 1 MB). Report prepared for the Association of American Medical Colleges. Published 2017.
  • 23Dall TM, Storm MV, Semilla AP, Wintfeld N, O’Grady M, Narayan KM. Value of lifestyle intervention to prevent diabetes and sequelae. AmJPrevMed. 2015;48(1873-2607 (Electronic)):271-280.
  • 24Su W, Huang J, Chen F, et al. Modeling the clinical and economic implications of obesity using microsimulation. JMedEcon. 2015;18(1941-837X (Electronic)):886-897.
  • 25 a b Su W, Chen F, Dall TM, Iacobucci W, Perreault L. Return on Investment for Digital Behavioral Counseling in Patients With Prediabetes and Cardiovascular Disease. Preventing Chronic Disease. 2016;13. doi:10.5888/pcd13.150357.
  • 26Taylor F, Huffman MD, Macedo AF, et al. Statins for the primary prevention of cardiovascular disease. CochraneDatabaseSystRev. 2013;(1469-493X (Electronic)):CD004816. 
  • 27Baguet JP, Legallicier B, Auquier P, Robitail S. Updated meta-analytical approach to the efficacy of antihypertensive drugs in reducing blood pressure. ClinDrug Investig. 2007;27(1173-2563 (Print)):735-753. 
  • 28Sherifali D, Nerenberg K, Pullenayegum E, Cheng JE, Gerstein HC. The effect of oral antidiabetic agents on A1C levels: a systematic review and meta-analysis. Diabetes Care. 2010;33(1935-5548 (Electronic)):1859-1864. 
  • 29Yang W, Dall TM, Zhang Y, et al. Simulation of quitting smoking in the military shows higher lifetime medical spending more than offset by productivity gains. Health Aff(Millwood). 2012;31(1544-5208 (Electronic)):2717-2726.
  • 30Centers for Disease Control and Prevention. 2013 NCHS Urban-Rural Classification Scheme for Counties.
Date Last Reviewed: