Display of Summary Quality Data

1
Display of Summary Quality Data

• John Freedman
• Communications and Transparency Committee
• August 6, 2008

2
Acknowledgements

• Linda Green, Marj Radin and their colleagues in
  the DHCFP
• Young Joo, HCQCC policy intern

3
Review

• Many options exist for summary data display
• Summary, or simplified, display intended to
  ensure information is understandable to average
  consumer
• Interest by Council in reducing risk of
  misclassification
• By reducing the number of rating categories
• By applying a statistical test

4
Review

• Today, we will review 4 specific options
• Option 1 4 categories based on percentile
  ranking (15th/50th/85th percentiles as dividing
  points)
• Option 2 3 categories based on percentile
  ranking (15th/85th percentiles as dividing
  points)
• Option 3 3 categories based on 95 confidence
  intervals
• Option 4 3 categories based on 68 confidence
  intervals (equal to 1 standard deviation)

5
Review

• We will show and discuss actual quality data for
  representative quality measures as they might be
  depicted under each of the 4 options
• Stroke mortality (AHRQ/DHCFP)
• Recommended Care for Heart Attack (Hospital
  Compare)
• Recommended Care for Pneumonia (Hospital Compare)
• Todays discussion does not pertain to Leapfrog
  measures except for Patient Safety Score

6
15
50
85
(6.86)
(10.86)
(15.01)
Data Source DHCFP/AHRQ
7
85
50
15
8
85
50
15
9
85
50
15
10
(No Transcript)
11
Stroke MortalityHow Would the Hospitals Rate
Under Each Option?
12
Stroke Mortality
Ranges from 4.2 to 28.9
13
Stroke Mortality
14
50
15
85
(94.4)
(86.2)
(90.8)
Data Source DHCFP/AHRQ
15
(No Transcript)
16
Pneumonia Care How Would the Hospitals Rate
Under Each Option?
17
Pneumonia Care
18
Pneumonia Care
19
15
50
85
(93.2)
(96.6)
(98.2)
Data Source DHCFP/US HHS Hospital Compare
20
(No Transcript)
21
Heart Attack Care How Would the Hospitals Rate
Under Each Option?
22
Heart Attack Care
23
Heart Attack Care
24
Recommendations

• Recommendation Hybrid Method
• Display both percentiles and statistical
  significance based on 95 confidence intervals
• Easy to understand intuitive for
  non-statisticians
• Maintains rank order
• Mitigates effects of tiny differences between
  hospitals
• Allows for indication of statistical significance
• Less discrimination between large and small
  hospitals

25
Recommendations

• Use hybrid method for
• Stroke Hip Fracture Mortality (AHRQ/DHCFP)
• Heart Attack, Heart Failure Pneumonia (Hospital
  Compare)
• Patient Safety Practices (Leapfrog)
• Do not display summary ratings for
• Bypass Surgery Angioplasty (Mass-DAC)
• Total Hip Replacement Mortality (AHRQ/DHCFP)
• Display summary stars only (confidence intervals
  are not available) for
• Heart Valve Surgery Weight Loss Surgery
  (Leapfrog) Patient Experience (Hospital Compare)