Hospital Readmissions Research: In Search of Potentially Avoidable Costs

Slide Presentation from the AHRQ 2008 Annual Conference

On September 9, 2008, Bernard Friedman, Ph.D., made this presentation at the 2008 Annual Conference. Select to access the PowerPoint® presentation (63 KB).

Slide 1

Hospital Readmissions Research: In Search of Potentially Avoidable Costs

Bernard Friedman, PhD
Center for Delivery, Organization, and Markets
Agency for Healthcare Research and Quality (AHRQ) Conference, 2008

Slide 2

Agenda

• Brief overview of some AHRQ internal research on readmissions over the past several years.
• Then a more detailed presentation on recent work in progress.
  • If voice doesn't hold out, there is a poster on this project in the Cafe.
• Finally, a few words on how the Healthcare Cost and Utilization Project (HCUP) team is trying to make new tools available to outside analysts to study readmissions.

Slide 3

Published Studies

1. Joanna Jiang was the lead author at AHRQ on several published studies of diabetes discharges.
   • One finding was that half of the discharges or hospital costs in a year are for people with multiple discharges for diabetes and its complications.
2. I examined (with Joy Basu) all readmissions within 6 months for people with 16 Potentially Preventable initial admissions.
   • Large variety of principal diagnoses for the RE-admissions.
   • Just the Potentially Preventable RE-admissions within 6 months had a projected national cost of about $1.4 Billion in today's $. This covered 4 states with 15% of the U.S. population.

Slide 4

Published Studies (Continued)

3. A recently accepted paper (with Joanna Jiang and Anne Elixhauser): "Costly Hospital Readmissions and Complex Chronic Illness".
   • Shows importance of the number of different chronic conditions in predicting readmission rates and annual cost.
4. Bill Encinosa and Fred Hellinger recently published "The Impact of Medical Errors on 90 Day Costs and Outcomes: An Examination of Surgical Patients".
5. All projects except the last one used our HCUP databases at AHRQ:
   • We receive statewide discharges from 40 Partners, all-payers covered.
   • A dozen Partners have provided encrypted patient identifiers that we refine by checking the age and gender of each supposed re-hospitalization.

Slide 5

Do patient safety events contribute to readmissions?

• Ongoing study for presentation in more detail.
• Under review at a journal. Already had a revision, but we'll be happy to have more suggestions.
• B. Friedman, Joanna Jiang, William Encinosa, Ryan Mutter.

Slide 6

Objectives

• To report 1-month and 3-month hospital readmissions, as well as deaths, after major surgical procedures in adults using a large multi-state and multi-payer database in 2004.
• To test whether 9 selected patient safety events contribute to these outcomes after controlling for measures of severity of illness and the presence of unrelated chronic conditions.

Slide 7

Background/Motivation

• A meta-analysis of small scale studies using clinical chart review found that better quality of care was associated with reduced readmission rates (Ashton, 1997).
• Health plans and many patients would benefit from a reduction in safety events and readmissions.
• BUT, hospitals and physicians do not always have an incentive to reduce readmissions (especially in Medicare and Medicaid). And there is a question if hospitals yet have adequate incentive to reduce safety events. (Mello et al., 2007).

Slide 8

Timeliness

• Starting with Fiscal Year 2009, the Centers for Medicare and Medicaid Services (CMS) will be collecting data on some safety events and other "never events."
• "Voluntary" to be used for public reporting.
• Several AHRQ Patient Safety Indicators. Some events measured differently.
• When affect Medicare payment? Only postoperative infections so far.

Slide 9

Study Design

• Healthcare Cost and Utilization inpatient discharge databases for 7 dispersed states: CA, FL, MO, NY, TN, UT, VA in 2004.
• Adults in surgical diagnosis-related groups (DRGs), not related to pregnancy or delivery.
• Remove any rehospitalization that was birth-related or due to trauma.
• Multinomial logistic regression model for 3 mutually exclusive outcomes: death, readmission, or discharge without readmission. The model yields simultaneously a relative risk of death and a relative risk of a readmission.
• Control for:
  • Severity level (using all patients refined (APR)-DRG software).
  • Unrelated chronic comorbidities (downloadable software from AHRQ).
  • Payer group.
  • 15 common DRGs at the initial admission.

Slide 10

Selected Safety Events in Surgical Patients

• Excluded safety events with more than a third of instances that were "present on admission" in two states with such data. [Houchens, et al., 2008].

Example: Iatrogenic Pneumothorax

• Numerator:
  • Discharges with ICD-9-CM code of 512.1 in any secondary diagnosis field among cases meeting the inclusion and exclusion rules for the denominator.
• Denominator:
  • All surgical discharges age 18 years and older defined by surgical DRGs, subject to exclusions below.
• Exclude cases:
  • MDC 14 (pregnancy, childbirth, and puerperium).
  • With diagnosis code of chest trauma or pleural effusion.
  • With an ICD-9-CM procedure code of diaphragmatic surgery repair.
  • With any code indicating thoracic surgery, lung or pleural biopsy, or assigned to cardiac surgery DRGs.
• Full specifications of all Patient Safety Indicators used in study:
  http://www.qualityindicators.ahrq.gov/psi_overview.htm

Slide 11

Selected Patient Safety Risks

The table presents the results for "Patients at Risk" and "Safety Event" for the category, "Patient Safety Event."

Slide 12

Key Findings

• The 3-month readmission rate was less than 17% for those with no safety event but 24.8% when a safety event occurred.
  • 2/3 of readmissions within 3 months occurred within the first month.
• The relative risk ratio (RRR) for readmission due to any safety event, adjusted for all other factors was 1.20 (1.165-1.235), P<.001.
• The in-hospital death rate was 1.3% with no safety event but 9.2% with a safety event. RRR=1.654 (1.562-1.752), P<.001.
• Medicare and Medicaid patients were more likely to have readmissions than privately insured patients: RRR about 1.5 in each case.

Slide 13

Multivariate results: Relative Risk Ratios

The table shows the results for "Patients at Risk," "Relative Risk of Death," and "Relative Risk of Readmission within 3 Months."

Slide 14

Discussion

• Hospital readmissions are one way that safety events can have costly consequences, in addition to deaths or more expense at the initial stay.
• A simultaneous multiple-outcome model makes sense (deaths tend to reduce readmissions) and is feasible.
• The study suggests that extensive risk adjustment does not eliminate the contribution of safety events to readmissions (surgical patients, at least).

Slide 15

Final notes

• Although safety events were found to contribute to readmissions:
  • The problems of effective management of chronic illness are probably a more important determinant of readmissions overall.
• This type of research is the tip of the iceberg made possible by a decade of development of safety indicators and risk adjustment by AHRQ staff, contractors and consultants.
• Ongoing infrastructure development for outside analysts to use with HCUP databases (Claudia Steiner and ThomsonReuters).
  • We hope this will make it easier to analyze readmissions for large databases.
  • Will require permission from more Partners to release encrypted patient identifiers.