Information | Analytics | Expertise

MODELING THE US HEALTH WORKFORCE: SUMMARY OF THE RN SUPPLY AND DEMAND FORECASTING

MEETING (MONTANA, 2016) AND IMPLICATIONS FOR MODELING

OCTOBER 24, 2016

Tim Dall Managing Director, Life Sciences

IHS Markit

+1 202 870 9211

Tim.Dall@ihs.com

IHWC 2016 TECHNICAL SKILLS DAY

RN Supply and Demand Forecasting Meeting

Big Sky, Montana, July 2016

• Meeting hosted by Montana State University, Center for Interdisciplinary

Health Workforce Studies

• Funding from the U.S. Bureau of Health Workforce, Health Resources

and Services Administration

• Meeting brought together ~20 health workforce researchers and nurse

workforce experts

• Goal: improving nurse workforce forecasts to provide information

needed by policy makers, employers, educators, researchers and others

working to assure a strong, appropriately sized, and capable nursing

workforce

• One day focusing on modeling nurse supply

• One day focusing on modeling nurse demand

• Opportunities to share methods and data, and provide constructive criticism

2

Two Demand Modeling Approaches Presented

• Adjusted Risk Choice & Outcomes Legislative Assessment (ARCOLA)

model

• Microsimulation model used to simulate insurance enrollment patterns under

the Affordable Care Act

• Estimated demand for services based on insurance changes → estimate

demand for nurses based on service demand

• Main challenge with this approach is the ARCOLA model is designed to model

changes in insurance coverage; this study was a workforce application

• Health Workforce Simulation Model (HWSM)

• Microsimulation model that simulates health care use for a representative

sample of the population, then simulates demand for health workforce based

on projected demand for services

• Main challenge with this approach is projecting future changes in care use and

delivery patterns under emerging care delivery models

3

Two Supply Modeling Approaches Presented

• Cohort-based model

• Approach models how many nurses from a cohort (specified by birth year) will

remain in the workforce over time

• Approach provides insights to workforce participation rates over time within a

cohort of nurses

• Main challenge of this approach is it does not capture large variation across

cohorts in number of individuals entering nursing as a profession

• Microsimulation-based approach

• Starts with a database of nurses and simulates individual career choices

• Approach to modeling workforce decisions (active in the labor force, hours

worked, retirement) appears to produce aggregate patterns similar to the

cohort-based approach

• Faces many of the same challenges of the cohort-based model: external

“shocks” can cause nurse workforce behavior to deviate from historical

patterns 4

Summary of Goals and Criteria for Building Workforce

Models

• Provide the most accurate projections possible

• Provide flexibility to model wide range of scenarios reflecting new

policies, emerging trends in care delivery, and other (e.g., economic)

factors

• Build on solid theoretical underpinnings

• Build dynamic model: integrate professions and specialties

• Adaptable to different geographic units (national, state, local level)

• Provide platform for continued model improvement; incorporate new

research as it becomes available

• Make model transparent (through reports and presentations)

5

Flow Diagram for the Supply Component of HWSM

6

Starting Year Supply

Demographic and Geographic Characteristics

Data Sources: American Community Survey, association registries, state licensure files

New Entrants

Demographic and Geographic Characteristics

Data Sources: Integrated Postsecondary Education Data System, professional associations

Attrition

Retirement

Career Change

Age/Sex Specific Mortality Data Source: Centers for Disease Control and Prevention

Workforce Participation and Hours Worked

Data Sources: American Community Survey, survey data from state licensure boards, occupation-specific surveys

End Year Supply

By Demographic

and Geographic

Characteristics

Mortality

Data Sources: Starting Supply

• American Community Survey (ACS)

• Active supply defined as nurses working or seeking employment

• Multiple years data used

• Current work using 2014 ACS, with 5-year file (2010-2014) used for some analyses

• Distribution by state, age, sex and education level

• For current work, using licensure data from states that have voluntarily provided

data (GA, OR, SC, TX); ACS data for all other states

• Data strength and weakness

• ACS: Cannot distinguish between nurses working in nursing positions and in positions

that do not require a nursing degree

• ACS: Information on patient care hours not available

• ACS: Small sample size for smaller states

• Licensure files: most states have ‘cleaned’ their data so the data are in good shape;

desire for HRSA supply estimates to use best available source of data and consistent

with numbers published by individual states; shorter time lag between when data are

generated and used

7

7

Trends in Number of US Educated First Time NCLEX-RN

Takers, 2001-2015

8

68,759 70,692

76,688

87,171

99,186

110,703

119,565

129,111

134,727 140,882

144,554 150,266

155,098 157,882

157,957

-

20,000

40,000

60,000

80,000

100,000

120,000

140,000

160,000

180,000

2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Firs

t Ti

me

, U.S

. Ed

uca

ted

Can

dia

tes

Taki

ng

NC

LEX

-RN

Year

Data Source: National Council of State Boards of Nursing, Exam Statistics and Publications,

2001 to 2015 data from various reports.

HR

SA

20

04

re

po

rt

HR

SA

20

14

re

po

rt

HR

SA

20

16

re

po

rt

RN Retirement Patterns

Estimated patterns using 2010-2015 licensure data from Oregon, South Carolina, and Texas; and

2008 Sample Survey of RNs (for nurses under age 50).

9

0

10

20

30

40

50

60

70

80

90

100

25 30 35 40 45 50 55 60 65 70 75

Cu

mu

lati

ve P

rob

abili

ty R

etir

ed

fro

m N

urs

ing

Nurse Age

Texas South Carolina Oregon (Intention, adj) HRSA Model

Summary Regression Results for RNs

Parameter

Predicting

Hourly Wage a

Predicting

Hours/Week a

Predicting Labor Force

Participation, age <50 (CI) b

Intercept -2.67 ** 35.15 **

Unemployment rate (state, year) -0.15 ** 0.05 * 1.03 1.01 1.05

State occupation mean hourly wage 0.85 **

Predicted hourly wage 0.01 0.97 0.96 0.99

Age 35 to 44 3.87 ** 0.26 **

Age 45 to 54 5.21 ** 1.20 **

Age 55 to 59 5.79 ** 0.88 **

Age 60 to 64 5.74 ** -0.31 **

Age 65 to 69 4.70 ** -4.54 **

Age 70+ 2.07 ** -8.57 **

Age 30-34 0.69 0.63 0.77

Age 35-39 0.89 0.79 1.00

Age 40 to 44 0.97 0.86 1.08

Age 45 to 49 1.12 0.99 1.27

Male 1.18 ** 2.78 ** 0.71 0.58 0.87

Age 30-34 * male 2.20 1.59 3.06

Age 35-39 * male 2.81 1.96 4.02

Age 40 to 44 * male 2.63 1.87 3.70

Age 45 to 49 * male 1.94 1.38 2.74

Year 2011 -0.38 ** 0.14 0.93 0.84 1.03

Year 2012 0.39 ** 0.21 * 0.92 0.83 1.02

Year 2013 0.14 0.30 ** 0.93 0.84 1.05

Year 2014 -0.29 ** 0.38 ** 0.97 0.85 1.10

Non-Hispanic black -0.15 2.28 ** 1.32 1.17 1.49

Non-Hispanic other -0.66 ** 1.43 ** 1.23 1.10 1.37

Hispanic 1.12 ** 1.43 ** 1.38 1.19 1.60

Have nursing baccalaureate degree 2.55 ** -0.24 ** 0.98 0.91 1.05

Having nursing graduate degree 4.10 ** 1.56 ** 0.91 0.80 1.03

Population % suburban 12.99 ** 0.73 2.27 1.33 3.89

Population % rural 0.56 1.41 ** 0.77 0.52 1.15

Sample size 150,504 150,504 89,370

R-squared 0.12 0.04

10

Notes: Analysis of

the American

Community Survey; a Ordinary least

squares regression

coefficients.

Statistically

significant at the

0.01 (**) or 0.05 (*)

level. b Odds ratios

and 95%

confidence interval

(CI) from logistic

regression.

Comparison groups

are female,

year=2010, non-

Hispanic white, age

<35 (for wages and

hours) or age <30

(for labor force

participation).

Labor force

participation

regression uses

only clinicians

under age 50.

Comparison of Actual to Predicted Hours Worked by RNs:

Example: Data for the State of South Carolina

11

0

5

10

15

20

25

30

35

40

Less than 35years

35 to 44 years 45 to 54 years 55 to 59 years 60 to 64 years 65 years andmore

Ho

urs

Wo

rke

d p

er

We

ek

RN Age

Actual

Predicted

Projected Percentage Growth in RN Supply & Demand:

Example: Data for the State of Georgia

12

32%

28%

0%

27%

25%

23%

18%

-5%

0%

5%

10%

15%

20%

25%

30%

35%

2015 2020 2025 2030Pe

rce

nta

ge G

row

th in

RN

Su

pp

ly &

De

man

d (

rela

tive

to

20

15

)

Year

Supply: 10% incnew grads

Supply: Ret 2yrslater

Baseline Demand

Supply: Status Quo

Supply: Ret 2yrsearlier

Supply: 10% decnew grads

13

© 2016 IHS

Published RN Supply and

Demand Projections

Forthcoming

14