rossj

Report
Hospital Volume and 30-day
Mortality following Hospitalization
for Acute Myocardial Infarction
and Heart Failure
Joseph S. Ross, MD, MHS
Mount Sinai School of Medicine
James J. Peters VA Medical Center
Background
• For numerous surgical conditions and
medical procedures, admission to higher
volume hospitals has been associated with
lower mortality rates.
• Strongest associations for cancer and AAA
surgeries, more modest for PCI and CABG
and orthopedic surgeries.
Background
• Fewer studies of medical conditions.
• Conceptually:
– For surgeries and procedures  practice
makes perfect
– For medical care  less routinization;
organizational structures and processes
Background
• Care for medical conditions is common
and costly:
– HF is most common admission, 2nd most
expensive for Medicare
– AMI is 4th most expensive for Medicare
• Drive to improve health care quality – is
volume a marker?
Background
• Two studies focused on AMI treatment.
– Farley & Ozminkowski (Medical Care, 1992)
used HCUP data from 1980-87, didn’t adjust
for invasive capacity: 10% increase in hospital
volume decreased mortality 2.2%.
– Thiemann et al. (NEJM, 1999) used CCP data
from 1994-5, prior to key advances, but
adjusted for invasive capacity: HR=1.17 (1.091.26) [lowest quartile to highest quartile]
• No studies focused on HF treatment.
Research Objective
• To examine whether admission to a higher
volume hospital is associated with lower
mortality rates for AMI and HF.
Data Source
• Medicare Provider Analysis and Review
(MEDPAR) claims data from all FFS
beneficiaries hospitalized from 2001-3 in
U.S. acute-care hospitals.
Study Population
• FFS patients hospitalized for AMI and HF
identified using ICD-9-CM codes.
• Transfers linked into a single episode of
care; outcomes attributed to index
hospital.
• Excluded patients admitted to hospitals
with 10 or fewer admissions, admissions
<24hrs not AMA.
Main Outcome Measure
• 30-day risk-standardized all-cause
mortality rates (RSMR).
Primary Independent Variable
• Hospitals were categorized by conditionspecific volume quartile (prior to
application of exclusion criteria):
– Low (Q1+Q2)
– Moderate (Q3)
– High (Q4)
Statistical Analysis
• Weighted hierarchical model that included
patient variables (1st level) and hospital
variables (2nd level):
– CABG surgery/PCI capacity
– Teaching status
– Ownership status
Results
• From 2001-3:
– 801,307 AMI hospitalizations in 3,978 hospitals
– 1,245,564 HF hospitalizations in 4,328 hospitals
Mean Condition-Specific Volume
Hospital Volume
Low
Moderate
High
AMI
41
149
647
HF
100
312
1031
% of Patient Hospitalizations
AMI
HF
Low
4%
5%
Hospital Volume
Moderate
19%
22%
High
77%
73%
Patient Characteristics by Volume
(For AMI)
Hospital Volume
Low
Moderate
High
Sociodemographics
Age, Mean
Female, %
Past Medical History
81
57
80
54
79
51
Prior MI, %
Valvular heart disease, %
Htn, %
12
12
33
12
13
36
14
16
49
DM, %
PVD, %
25
15
27
16
33
19
Hospital Characteristics by Volume
(For AMI)
CABG surgery capacity, %
Hospital Volume
Low
Moderate
High
2
10
59
PCI capacity, %
COTH member, %
Teaching affiliate, %
3
1
6
17
3
13
57
17
44
Public ownership, %
36
17
9
Volume & Observed AMI Mortality
30%
23.9%
20.9%
20%
17.2%
10%
0%
Low
Moderate
High
Volume & AMI RSMR
• Admission to both high and moderate
volume hospitals was associated with
lower AMI RSMRs when compared with
low volume hospitals:
– High: OR=0.82 (0.79-0.85)
– Moderate: OR=0.89 (0.86-0.93)
Volume & Observed HF Mortality
20%
12.6%
12.1%
Low
Moderate
11.4%
10%
0%
High
Volume & HF RSMR
• Admission to both high and moderate
volume hospitals was associated with
lower HF RSMRs when compared with low
volume hospitals:
– High: OR=0.85 (0.82-0.89)
– Moderate: OR=0.93 (0.89-0.96)
Conclusions
• Hospital volume was associated with lower
risk-standardized odds of death after
admission both AMI and HF among FFS
Medicare beneficiaries.
• For high volume hospitals, 18% lower
odds for AMI, 15% for HF.
Limitations
• Focused only on mortality, not other
important dimensions of quality.
– i.e., processes of care, patient experiences.
• May not be generalized to other conditions
or to care provided in ambulatory settings.
• Observational study – can not rule out
confounding of hospital volume by other
unmeasured variables.
Implications
• A relationship between volume and
outcomes may exist for some medical
conditions, as well as for surgical
conditions and procedures.
• Provides some reassurance as quality
organizations begin to use volume as a
surrogate for quality.
Study Team
Yale University/Yale New-Haven Hospital
• Yun Wang, PhD
• Jersey Chen, MD
• Judith H. Lichtman, PhD, MPH
• Harlan M. Krumholz, MD, SM
• Entire CORE team
Harvard University
• Sharon-Lise T. Normand, PhD
Sunnybrook Health Sciences Centre
• Dennis T. Ko, MD, MSc

similar documents