Comparing the Effectiveness of Carotid Stent Systems versus Endarterectomy

1
Comparing the Effectiveness of Carotid Stent
Systems versus Endarterectomy
Peter W. Groeneveld, MD, MS Assistant Professor
of Medicine Philadelphia Veterans Affairs Medical
Center University of Pennsylvania School of
Medicine
2
Acknowledgements

• Feifei Yang, MS
• Alexis Greenhut, MPH
• Christine Camacho, BA
• Janell Olah, MFA

3
Carotid Stent Systems

• Catheter-based treatment of carotid stenosis
• Distal embolic protection to prevent stroke
  during procedure
• Alternative to vascular surgical procedure
  (carotid endarterectomy)

4
Carotid Stent System
5
Carotid Stent Systems Recent Events

• FDA approved first CSS in 2004
• CMS covered CSS in March, 2005
• Clinical trials have reported conflicting results
  about the efficacy of CSS compared to CE
• CREST trial (2011?) pending

6
Research objective

• Leveraging the limited clinical availability of
  CSS (n951 hospitals) compared to endarterectomy
  (n2229 hospitals)
• What were the observed differences in outcomes
  between CSS recipients vs. endarterectomy
  patients in the year after the 2005 coverage
  decision?

7
Methods/1

• Fee-for-service Medicare inpatient claims from
  2005-2006 (MEDPAR)
• Patients over age 65
• Identified all recipients of CSS or
  endarterectomy
• Fit propensity score model predicting receipt of
  CSS, using patient-level demographic, clinical
  variables, and hospital-level variables, as
  predictors

8
Methods/2

• CSS patients matched to CE patients during same
  time period (4/05-3/06 contemporary controls)
• Separately fit PS model and matched historical
  CE patients (Apr 2004-Mar 2005) to CSS recipients
• Fit proportional hazards and logistic regression
  models for mortality and combined death/stroke/AMI

9
Methods/3

• Outcomes determined via death indicators in the
  Medicare enrollment database (Denominator) and/or
  by view of subsequent hospitalization records in
  the 270 days following the carotid procedure

10
(No Transcript)
11
Alternative Approach

• HRR strongly correlated with CSS receipt, but
  weakly correlated with other factors influencing
  mortality
• Used Hospital Referral Region as instrumental
  variable for CSS receipt
• Fit probit models where mortality or
  mortality/stroke/AMI at 90/270 days were the
  outcomes
• Implemented with STATAs ivprobit

12
Study Cohorts

• 8,306 CSS recipients matched to 8,306
  contemporary and 8,306 historical
  endarterectomy controls
• 66,457 endarterectomy patients and 8,306 CSS
  patients included in instrumental variables
  analysis

13
Match quality/1
Characteristic CSS n8,306 CE Before Match n66,457 Std Diff CE After Match n8,306 Std Diff
Logit ps (SD) -1.8 (0.75) -2.5 (0.63) 70.5 -1.8 (0.75) 3.5
Age, mean (SD) 77 (7) 76 (6) 6.7 76 (6) 0.3
Female 3416 (41) 28,658 (43) 4.0 3,406 (41) 0.2
Black 387 (5) 2,272 (3) 6.3 396 (5) 0.6
14
Match quality/2
Characteristic CSS n8,306 CE Before Match n66,457 Std Diff CE After Match n8,306 Std Diff
CHF 994 (12) 5,354 (8) 13.1 963 (12) 1.3
Periph. vasc. dis. 2,553 (31) 10443 (16) 36.2 2,483 (30) 2.1
Hypertension 4,779 (58) 45,051 (68) 21.3 4,847 (58) 1.7
CAD 4,431 (53) 29,614 (45) 17.6 4,441 (54) 0.2
Stroke 159 (2) 992 (2) 3.3 149 (2) 0.9
Trans isch. attack 194 (2) 1,697 (3) 1.4 212 (3) 1.4
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
(No Transcript)
20
Conclusions

• Propensity-score matched comparisons of CSS and
  endarterectomy patients suggest CSS associated
  with worse outcomes
• However these findings were not duplicated in the
  IV analysis, suggesting unobserved heterogeneity
  among patient populations

21
Conclusions/2

• Although inferiority of CSS cannot be disproved
  by these results, we did not find a clear case
  supporting the superiority of endarterectomy
• Confidence boundaries suggest differences in
  270-day mortality rates are less than 3.3
  percentage points

22
Impact on Future Research

• If the difference in outcomes were only 2, 3,202
  patients would be needed in a clinical trial to
  detect it with 80 power
• Unlikely that the ongoing CREST trial (goal
  enrollment1200-1600 patients, anticipated
  completion date2011) will be able to
  definitively determine superiority of one therapy
  versus the other

23
Impact on Policy

• Our study provides evidence that in the carefully
  selected clinical settings at which Medicare
  beneficiaries can receive CSS, outcomes of care
  for CSS patients are comparable to those achieved
  by patients receiving endarterectomy