Impact of Socioeconomic Status on Unplanned Hospital Admissions and - - PowerPoint PPT Presentation

www.rti.org

RTI International is a registered trademark and a trade name of Research Triangle Institute.

Impact of Socioeconomic Status on Unplanned Hospital Admissions and Readmissions in Medicare

Yan Tang, PhD Christopher Beadles, MD, PhD Amy Mills, BS Musetta Leung, PhD

Acknowledgements and Disclaimer

• Acknowledgements

RTI International

– Olivia Berzin – Anna Prigozhin – Pavel Semukhin

• Disclaimer

– This project was funded by the Centers for Medicare & Medicaid Services

under contract no. HHSM-500-2011-00152G.

– The statements contained in this presentation are solely those of the

authors and do not necessarily reflect the views or policies of the Centers for Medicare & Medicaid Services. RTI assumes responsibility for the accuracy and completeness of the information contained in this report.

Background

• Unplanned hospital admissions and readmissions are prevalent and

expensive

• Empirical evidence suggests health care utilization may differ by

socioeconomic status (SES) (DHHS, 2016; Lewis et al., 2017;Singh and

Siahpush, 2006)

• Despite efforts to improve health outcomes for low SES individuals,

research suggests gaps likely remain (Kochanek et al., 2015; Singh and

Siahpush, 2014)

• Understanding the relationship between individual SES and health

care utilization is important to designing innovative interventions to deliver high-quality health care

Research Objectives

• To describe socioeconomic status using a composite SES measure

based on neighborhood SES status from the US Census data

• To examine the association between the neighborhood

socioeconomic status on unplanned hospital admissions and readmissions in a Medicare population

Methods

• Data

– 2015 Medicare fee-for-service claims data for beneficiaries assigned to the

Shared Savings Program Accountable Care Organizations (ACOs)

– 2010-2014 American Community Survey 5-year estimates data

• Study sample

– Four cohorts (≥65 years old):

• Diabetes
• Heart failure
• Multiple chronic conditions (MCC)
• ≥2 of eight chronic disease groups: acute myocardial infarction, Alzheimer’s

disease, atrial fibrillation, chronic kidney disease, COPD and asthma, depression, heart failure, stroke and transient ischemic attack

• Primarily defined by CMS’s Chronic Condition Data Warehouse (CCW)
• Hospitalization

Methods (cont.)

• Outcomes (patient-level)

– Three unplanned admission measures:

• Dummy variable indicating any unplanned hospital admission for individuals with

diabetes, heart failure, or MCC, respectively

– One unplanned readmission measure:

• Dummy variable indicating any unplanned readmission within 30 days of

discharge from the index hospitalization

• Identification of unplanned admissions

– Admissions for some types of care are always considered planned

• Transplant surgery
• Maintenance chemotherapy, radiotherapy, immunotherapy
• Rehabilitation

– Otherwise, a non-acute admission for a scheduled procedure is considered

planned

– Admissions for acute illness or complications of care are unplanned

Methods (cont.)

• Key independent variable

– ZIP code-level SES index score according to the Agency for Healthcare

Research and Quality (AHRQ) SES algorithm (AHRQ, 2008)

• Normalized score based on high occupancy housing, property values, poverty,

median household income, education, and unemployment rate

• Ranges from 0-100, with higher score indicating better SES status
• Control Variables

– Individual demographic information (e.g., age, sex) – Comorbidity status (e.g., CMS Condition Category groups)

• Statistical analysis

– Generalized linear mixed models with a binomial distribution and a logit

link, with ACO-level random effects

Characteristic Diabetes Heart Failure MCC Hospitalization Sample size 750,620 584,127 1,195,909 1,041,289 Female (%) 53.8 54.8 58.4 57.4 Hospital admissions/readmissions per 100 individuals 29.1 38.9 34.1 15.7 SES index score distribution

Results

Characteristics of the Study Sample

Results (cont.)

Measure Only SES Index Odds Ratios (95% CI) SES Index + Covariates Odds Ratios (95% CI) Admissions for individuals with diabetes 0.985 (0.984 - 0.986) 0.987 (0.986 - 0.988) Admissions for individuals with heart failure 0.988 (0.987 - 0.990) 0.988 (0.987 - 0.990) Admissions for individuals with MCC 0.986 (0.985 - 0.987) 0.989 (0.988 - 0.990) Hospital readmissions 0.983 (0.982 - 0.984) 0.994 (0.993 - 0.995)

Regression Results: Estimated Effects of SES on the Outcomes

All odds ratios were significant at p<0.001 level

Results (cont.)

28.7 27.2 26.3 25.2 5 10 15 20 25 30 Quartile 1 Quartile 2 Quartile 3 Quartile 4

Unplanned Admissions per 100 Individuals Quartile of SES Index

Adjusted Admissions for Individuals with Diabetes, by SES

Measure was adjusted for individual demographic information and comorbidity status

Results (cont.)

39.7 38.2 36.9 36.2 5 10 15 20 25 30 35 40 Quartile 1 Quartile 2 Quartile 3 Quartile 4

Unplanned Admissions per 100 Individuals Quartile of SES Index

Adjusted Admissions for Individuals with Heart Failure, by SES

Measure was adjusted for individual demographic information and comorbidity status

Results (cont.)

34.5 33.0 32.0 31.5 5 10 15 20 25 30 35 Quartile 1 Quartile 2 Quartile 3 Quartile 4

Unplanned Admissions per 100 Individuals Quartile of SES Index

Adjusted Admissions for Individuals with MCC, by SES

Measure was adjusted for individual demographic information and comorbidity status

Results (cont.)

13.4 13.0 12.6 12.6 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Quartile 1 Quartile 2 Quartile 3 Quartile 4

Unplanned Readmissions per 100 Individuals Quartile of SES Index

Adjusted Hospital Readmissions, by SES

Measure was adjusted for individual demographic information and comorbidity status

Summary of Key Findings

• Higher SES index score was associated with lower unplanned

hospital admissions/readmissions

• The magnitude of the associations between SES index score and the

four measures was small

• Risk adjustment for individual comorbidities generally mitigated this

association

• Limitations:

– ZIP code-level SES index score serves as a proxy for individual’s

socioeconomic status

– Unmeasured potential confounding (e.g., provider characteristics)

Implications

• The observed associations between SES status and hospital

admission and readmission measures were explained in part by differences in patient medical complexity

• Quality improvement programs intended to enhanced care delivery

for individuals with low SES status may promote the delivery of high- quality health care

• These findings are among Medicare beneficiaries assigned to SSP

ACOs; it is unknown whether other accountable care organization models would have similar findings

Thank you! Questions?