Estimating Needed Capacity of Nursing Home and Hospital Beds Presented by Megan Stratman and Matt Spellman
Motivation To build a statistical model to estimate needed capacity of nursing home beds and hospital beds within Metropolitan Statistical Areas (MSA) To use this model to determine whether the Eugene- Springfield MSA has too many or too few nursing home and hospital beds
Introduction Current hospital situation in Eugene-Springfield Implications of Certificate of Need (CON)
Outline of Presentation Review of Literature – to help understand what past studies’ have determined Building our regression models Estimating and interpreting coefficients Applying regression results to Eugene-Springfield MSA to determine whether below or above needed capacity of nursing home and hospital beds
Review of Literature Nursing Homes Age Gender Marital Status Functional Impairments Educational Attainment Income Market Competitiveness
Review of Literature Hospitals Birth Rate Death Rate Constraint Function: “Break-even” Demand (Population, Insurance Coverage, etc) Government Regulation
Review of Literature Certificate of Need (CON) Ambiguous effect on supply of beds Proponents: contains costs and maintain quality Opponents: restricts competition
Methodology and Hypotheses Building Two Models Dependent variable: Model 1: Nursing Home Beds Model 2: Hospital Beds Independent variables in each model: Demographic Presence of CON regulations By building these models and running regression analyses, we can determine which variables impact needed bed capacity
Methodology and Hypotheses Nursing Homes—Selected Variables Variables Hypothesized Signs Rationale Total Population 65+ (in 1,000s) + Larger population of elderly, larger demand Males 65+ (in 1,000s) - Relatively shorter lifespan, wife able to provide in-home care Presence of CON (1=yes) - Market distortion restricts supply 65+ Below Poverty Level (in 1,000s) - Unable to afford nursing home care; ambiguous effect: Medicaid, past studies Disability 65+ (in 1,000s) + Increased functional impairments increase need for living assistance Some College 65+ (in 1,000s) + Undetermined interaction; higher education may imply longer lifespan
Methodology and Hypotheses Hospitals—Selected Variables Variables Hypothesized Signs Rationale Total Population (in 1,000s) + Larger total population, larger demand Total Population 65+ (in 1,000s) + Larger population of elderly, larger demand Females (in 1,000s) + Child-bearing age increases demand of beds Presence of CON (1=yes) - Market distortion restricts supply Population Insured + Insured are more likely to use hospital facilities
Methodology and Hypotheses Data Collection National data according to Metropolitan Statistical Areas (MSA) Sources: U.S. Census Bureau (demographics) (nursing home beds) American Hospital Association (hospital beds) Kaiser Family Foundation (insurance coverage)
Regression Results Nursing Homes Explanatory Variables Base ModelAdd CONAdd PovertyAdd DisabilityAdd College Total Pop 65+ (in 1,000s) *** (29.24) *** (29.42) *** (44.29) *** (40.70) *** (41.42) Males 65+ (in 1,000s) *** (69.29) *** (69.73) *** (96.31) *** (91.79) *** (101.19) CON 1=yes * (127.03) ** (133.77) ** (126.27) ** (121.22) Below Poverty Level (in 1,000s) ** (45.63) (85.57) (89.42) Disability 65+ (in 1,000s) (50.69) (53.96) College 65+ (in 1,000s) 9.34 (26.43) R-squared
Methodology and Hypotheses Interpreting the Coefficient on Total Males 65+ Coefficient = -455 Hold Total Population Constant Increase Males by 1,000 Decrease Females by 1,000
Methodology and Hypotheses Interpreting the Coefficient on CON Dummy variable used to control for qualitative data Presence of CON = 1 No CON = 0 If CON is present, insert 1 into equation nhbeds = 228(totalpop65)– 455(male65) – 214(CON) nhbeds = 228(totalpop65)– 455(male65) – 214(1) nhbeds = 228(totalpop65)– 455(male65) – 214 If CON is not present, insert 0 into equation nhbeds = 228(totalpop65)– 455(male65) – 214(CON) nhbeds = 228(totalpop65)– 455(male65) – 214(0) nhbeds = 228(totalpop65)– 455(male65) Coefficient on CON has no impact on regression estimate
Regression Results Hospitals Explanatory Variables Base Model Add Total Pop 65+ Add Females Add Insurance Add CON Total Pop (in 1000s) 2.713*** (0.255) 1.372*** (0.337) (0.295) (3.340) 1.371*** (0.335) Total Pop 65+ (in 1000s) *** (3.152) *** (2.937) *** (2.721) *** (3.139) Females (in 1000s) 5.420*** (1.685) Population Insured (2.967) CON 1=yes (99.629) R-squared
Application of Models Final Regression Equations Nursing Home Beds Hospital Beds
Application of Model MSA Total Pop 65+ Male 65+ Presence of CON 1=yes 65+ Below Poverty Eugene-Springfield42,95418,14013,149 Medford-Ashland28,99112,63511,944 Spokane, WA51,94921,19814,021 Albuquerque, NM80,42133,99507,213 Nursing Homes Estimated # of Beds Actual # of Beds 1,6831, ,4283,686 3,1772,302
Application of Models Nursing Homes (actual – estimated)
Application of Model MSATotal Pop Total Pop 65+ Eugene-Springfield322,95942,954 Medford-Ashland181,26928,991 Spokane, WA417,93951,494 Albuquerque, NM712,73880,421 Hospitals Estimated # of Beds Actual # of Beds 1, ,5671,625 2,3461,970
Application of Models Hospitals (actual – estimated)
Review of Regression Building Estimated two models Nursing home beds Hospital beds Using the models, we estimated needed capacity of beds in the Eugene-Springfield MSA
Conclusions Nursing Home Beds: CON regulations have an impact by restricting supply of beds, relative to supply in a competitive market Hospital Beds: CON mimics how a competitive market would function State fixed effects may help explain the differences in bed supply; note Washington vs. Oregon Influencing factors: preferences of residents; substitutes and alternatives; others unable to be captured empirically
Thank You Any Questions?