Monitoring the Quality of Invasive Cardiac Services: The Unintended Consequences of Public Reporting Frederic S. Resnic, MD MSc, FACC Brigham and Women’s Hospital and Harvard Medical School March, 2010
Apr 01, 2015
Monitoring the Quality of Invasive Cardiac Services:
The Unintended Consequences of Public Reporting
Frederic S. Resnic, MD MSc, FACCBrigham and Women’s Hospital andHarvard Medical School
March, 2010
BrighamandWomen’sHospital Case Summary
• Mr. H. is an active 67 year old, with history of hypertension, coronary artery disease and dilated cardiomyopathy, who presented in acute pulmonary edema to an outside hospital. He had previously refused coronary angiography.
• On presentation, patient was profoundly hypotensive and dyspneic and required mechanical ventilation and support with multiple vas0-pressor agents
• Urgent catheterization revealed left main with severe three vessel CAD and a thrombotic (acute) lesion in right coronary artery. PCWP=38, pH=7.09
• Underwent successful emergent PCI of RCA with IABP support. Echo demonstrated EF=15% with global hypokinesis and inferior AK.
BrighamandWomen’sHospital Case Summary
• Transferred to BWH CCU for urgent consideration of CABG.
• Unable to wean IABP; continued pressor dependence, worsening O2 requirements, worsening renal function.
• Deemed not surgical candidate by two staff cardiac surgeons due to excessively high perioperative risk.
• Family sought “everything that can be done”
• Referred for high risk PCI of unprotected left main coronary artery, LAD and LCx to potentially allow wean from IABP and pressor support.
BrighamandWomen’sHospital
BrighamandWomen’sHospital
BrighamandWomen’sHospital Hospital Course
• Remained on pVAD support for 5 days with reduced vasopressor requirements.
• Myocardial function stabilized enough for pVAD to be removed; however continued pressor dependent and CVVH required for volume balance.
• Progressive multi-system organ failure with ARF, ARDS and progressive liver failure. No clear neurologic recovery despite weaning all sedation.
• Ultimately, patient made comfort measures only and expired peacefully on hospital day 20.
Issues RaisedBrighamandWomen’sHospital
OverviewBrighamandWomen’sHospital
• Defining “Quality” in Cardiac Surgery and Angioplasty
• Benefits and risks of public release of individual quality monitoring results
• Evidence for unintended consequences
• Strategies for a more comprehensive approach to quality monitoring
Interpreting Mass-DAC ReportsBrighamandWomen’sHospital
Source: 2006 PCI in MA – www.massdac.org
Mass-DAC uses “Standardized Mortality Incidence Rates” (SMIR) to compare hospital risk adjusted in-hospital all-cause mortality as a measure of overall quality.
2008 No Shock and No STEMI Risk Model
BrighamandWomen’sHospital
Source: 2008 PCI in MA – www.massdac.org
2008: No Shock and No STEMIBrighamandWomen’sHospital
Source: 2008 PCI in MA – www.massdac.org
2008 results indicate all centers performed within expectations.
2008 Shock or STEMI Risk ModelBrighamandWomen’sHospital
Source: 2008 PCI in MA – www.massdac.org
2008 Results: Shock or STEMIBrighamandWomen’sHospital
Source: 2008 PCI in MA – www.massdac.org
Again, no institutions identified as statistical outliers….
Cardiac Quality: The Big PictureBrighamandWomen’sHospital
ClinicalOutcomes
ProcessMeasures
Appropriateness
Access toHealthcare
Cardiac Quality: The Big PictureBrighamandWomen’sHospital
ProcessMeasures
Appropriateness
Access toHealthcare
ClinicalOutcomes
Benefits Risks
• Promotes Informed Consumer Choice
• Hawthorne Effect
• “Teeth” for Quality Monitoring
• Accelerates Adoption of Best Practices
• Transparency
BrighamandWomen’sHospital Trade-Off’s in Public Reporting
Benefits Risks
• Promotes Informed Consumer Choice
• Hawthorne Effect
• “Teeth” for Quality Monitoring
• Accelerates Adoption of Best Practices
• Transparency
BrighamandWomen’sHospital Trade-Off’s in Public Reporting
Outcomes Trends in MABrighamandWomen’sHospital
Adapted from www.MassDac.org cardiac surgery and PCI reports 2002-2005
Unadjusted mortality has declined for both CABG and PCI treated patients in Massachusetts.
Volume and Mortality Trends 2003-2007BrighamandWomen’sHospital
Source: 2007 PCI in MA – www.massdac.org
Statewide results indicate a 7.5% per year reduction in elective (non Shock or STEMI) volume since 2003.
Continued reduction in mortality of high risk group may indicate growing risk aversion by PCI operators.
Benefits Risks
• Promotes Informed Consumer Choice
• Hawthorne Effect
• “Teeth” for Quality Monitoring
• Accelerates Adoption of Best Practices
• Transparency
BrighamandWomen’sHospital Trade-Off’s in Public Reporting
• Over-emphasis on MD
• Emphasis on Low Risk Cases
• Risk Avoidance of High Risk Cases
• Up-coding and Gaming
• Unmeasured Quality Parameters Ignored
NY State PCI Mortality TrendsBrighamandWomen’sHospital
Adapted from: Annual Angioplasty Quality Reports 1997-2004 available from: www.health.state.ny.us/statistics/diseases/cardiovascular/
In-hospital mortality declined by 29% between 1998-2004, but was accompanied by a 43% reduction in the PCI treatment of cardiogenic shock.
NY PCI Mortality: 1998-2004PCI for Cardiogenic Shock
1998-2004
Survival with Cardiogenic ShockBrighamandWomen’sHospital
Hochman J et al. The SHOCK Trial 1999
Immediate revascularization confers sustained survival benefit is similar whether PCI or CABG is used.
Risk Avoidance: Lessons from NYBrighamandWomen’sHospital
Michigan, with no public reporting, was compared to NY State for PCI risk factors and outcomes.
Adapted from: Moscucci et al. JACC 45(11). June 2005.
MI Shock: 2.56%
MA Shock: 2.28%
NY Shock: 0.38%
MI Shock: 2.56%
NY State in the SHOCK TrialBrighamandWomen’sHospital
Apolito RA et al. Am Heart J February 2008
• Investigators explored practice patterns of participating centers from NY State and all other U.S. enrolling centers in the SHOCK trial.
• NY State was only state mandating public release of risk adjusted outcomes.
• Provided a contemporaneous comparison with rigorous data collection and follow-up of high risk patient population in NY as compared with other regions.
NY State in the SHOCK TrialBrighamandWomen’sHospital
Apolito RA et al. Am Heart J February 2008
After institution of public reporting, centers in NY demonstrated lower rates of emergent revascularization as compared to non-NY centers.
Time to CABG:
NY = 101.2 hrNon-NY = 10.1hr
NY State in the SHOCK TrialBrighamandWomen’sHospital
Apolito RA et al. Am Heart J February 2008
Selective utilization leads to decreased mortality for PCI and CABG in Shock patients….However, overall mortality is increased in NY as compared to other states.
NY State in the SHOCK TrialBrighamandWomen’sHospital
Apolito RA et al. Am Heart J February 2008
Comparing NY and MABrighamandWomen’sHospital
Analysis based on data excerpted from public cardiac reports and U.S. census data
Comparison of 2003 revascularization rates for cardiogenic shock demonstrate a 2-fold difference between the States.
Cas
es o
f c
ard
iog
enic
sh
oc
k tr
eate
d p
er 1
00,0
00 p
op
ula
tio
n
35.7 cases
71.3 cases
Risk Adjustment SpecificityBrighamandWomen’sHospital
Resnic FS and Welt FG Public Health Hazards of Risk Avoidance - JACC 2009
We reviewed over 5,000 consecutive PCI procedures at BWH to assess the adequacy of data collection systems and risk adjustment algorithms for predicting mortality post-PCI.
DefinitePCI Related
PossiblePCI Related
NOTProcedureRelated
MA Public Reporting: So What?
Outcomes Trends in MABrighamandWomen’sHospital
Adapted from www.MassDac.org cardiac surgery and PCI reports 2002-2005
Unadjusted mortality has declined for both CABG and PCI treated patients in Massachusetts.
Decline of rate of revascularization in Cardiogenic Shock in Massachusetts
BrighamandWomen’sHospital
Source: Mass-DAC Data Review. November 2007
37%
43%
Between 2003 and 2005, the rates of revascularization in Massachusetts declined 37-43%
Cardiac Quality: The Big PictureBrighamandWomen’sHospital
ClinicalOutcomes
ProcessMeasures
Appropriateness
Access toHealthcare
Impact on Access to CareBrighamandWomen’sHospital
Source: Werner RM, Asch DA and Polsky D. Circulation March 2005
Disparities in access to CABG increased in NY, relative to other states, after the release of report cards
ReducedAccess
ImprovedAccess
Operator Volume and PCI OutcomesBrighamandWomen’sHospital
Source: Moscucci et al. JACC August 2005
Exploration of Michigan data revealed a consistent trend toward improved risk adjusted outcomes with increasing operator volumes.
Operator Volume and PCI OutcomesBrighamandWomen’sHospital
Source: Moscucci et al. JACC August 2005
…. This trend was preserved within each expected risk quartile. Even in the lowest risk patients, low volume operators conferred twice the risk for death than high volume counterparts.
Cardiac Quality: The Big PictureBrighamandWomen’sHospital
ClinicalOutcomes
ProcessMeasures
Appropriateness
Access toHealthcare
Incremental Patient Health Benefit
↑ Patient Benefit
↑ Survival
↓ Patient Benefit
↓ Survival
↑ Patient Benefit
↓ Survival
↓ Patient Benefit
↑ SurvivalPhysician Preference
Patient Benefit
Appropriateness and Case Selection Creep
BrighamandWomen’sHospital
Acu
te R
isk
of P
roce
du
re
BrighamandWomen’sHospital
Incremental Patient Health Benefit
Acu
te R
isk
of P
roce
du
re
Appropriateness and Case Selection Creep
50yo STEMIin Shock
Focal CAD w/ Angina
Focal CAD w/o Angina
65yo withAnt. STEMI
75yo STEMIin Shock
75yo ST Δ’sw/ Sepsis
BrighamandWomen’sHospital
Minimal CAD w/o Ischemia
Incremental Patient Health Benefit
Acu
te R
isk
of P
roce
du
re
Appropriateness and Case Selection Creep
50yo STEMIin Shock
Focal CAD w/ Angina
Focal CAD w/o Angina
65yo withAnt. STEMI
75yo STEMIin Shock
75yo ST Δ’sw/ Sepsis
BrighamandWomen’sHospital
Minimal CAD w/o Ischemia
Incremental Patient Health Benefit
Acu
te R
isk
of P
roce
du
re
Appropriateness and Case Selection Creep
Public Reporting can promote a
Perverse Incentive
Acu
te R
isk
of P
roce
du
re
BrighamandWomen’sHospital
Incremental Patient Health Benefit
Appropriateness and Case Selection Creep
Improving Risk AdjustmentBrighamandWomen’sHospital
Decline of prevalence of Cardiogenic Shock in PCI and CABG in MA
BrighamandWomen’sHospital
Source: Mass-DAC Data Review. November 2008
Between 2003 and 2005, the rates of revascularization in Massachusetts declined 37-43%.....
Intro Comp Use Criteria
Outcomes of CU AdmissionsBrighamandWomen’sHospital
SOS Admissions, first PCI Compassionate Use Only SOS No CU SOS Total p-value
N Percent N Percent N Percent
All Cases 96 100.0% 5492 100.0% 5588 100.0%
Successful Procedure 76 79.2% 5176 94.2% 5252 94.0% <0.001
Post-Procedure Cardiogenic Shock 6 6.3% 148 2.7% 154 2.8% 0.035
New Renal Failure 7 7.3% 68 1.2% 75 1.3% <0.001
Any Bleeding Complication 14 14.6% 417 7.6% 431 7.7% 0.011
Bleeding - other/unknown source 8 8.3% 159 2.9% 167 3.0% 0.002
Any Vascular Complication 2 2.1% 48 0.9% 50 0.9% 0.212
Blood Products 25 26.0% 643 11.7% 668 12.0% <0.001
In-Hospital Death 67 69.8% 245 4.5% 312 5.6% <0.001
Primary Cause of Death Cardiac 46 47.9% 185 75.5% 231 4.1% 0.000
Neurologic 15 15.6% 12 4.9% 27 0.5% <0.001
Death in Lab 13 13.5% 27 0.5% 40 0.7% 0.356
Source: Mass-DAC October 2009
Improvement in Mortality Prediction Model (Shock/STEMI)
BrighamandWomen’sHospital
Source: Mass-DAC October 2009
BrighamandWomen’sHospital
Improvement in Mortality Prediction Model (Shock/STEMI)
ROC Area: No CU: 0.87w. CU: 0.90
P<0.01
Why? Reclassification of CasesBrighamandWomen’sHospital
Source: Mass-DAC October 2009
BrighamandWomen’sHospital Reclassification of Cases with CU
Source: Mass-DAC October 2009
Decline of rate of revascularization in Cardiogenic Shock in Massachusetts
BrighamandWomen’sHospital
Source: Mass-DAC Data Review. November 2008
Between 2003 and 2005, the rates of revascularization in Massachusetts declined 37-43%.....
Intro Comp Use Criteria
Additional Physician InputBrighamandWomen’sHospital
Outcomes Trends in MABrighamandWomen’sHospital
Adapted from www.MassDac.org cardiac surgery and PCI reports 2002-2007
Unadjusted mortality has declined for both CABG and PCI treated patients in Massachusetts.
ConclusionsBrighamandWomen’sHospital
• Monitoring the quality of cardiac procedures is essential, given the cost and consequences of these services.
– Historical failure of physicians to adequately police the process
• MA has the most statistically rigorous methods to evaluate risk-adjusted mortality, and is viewed as a model by other states
• Rigorous review of high quality risk-adjusted mortality data is necessary, but not sufficient, to assess the quality of cardiac care delivered in Massachusetts.
• Beyond risk-adjusted mortality, quality must also account for appropriateness of care, access to care, additional health related outcomes of care, and evaluate key processes of care delivered
Comprehensive Cardiac QualityBrighamandWomen’sHospital
ClinicalOutcomes
ProcessMeasures
Appropriateness
Access toHealthcare