Rodney P. Bensley, Rob Hurks, Ruby C. Lo, Jeremy D. Darling, Elliot L. Chaikof, Allen D. Hamdan, Mark C. Wyers, Marc L. Schermerhorn
Surgery, Beth Israel Deaconess Medical Center, Boston, MA.
OBJECTIVES: Administrative data (AD) have been used to compare CEA and CAS. However, there are limitations in defining prior stroke/TIA and complications that may vary by procedure. Therefore, we did a direct comparison between administrative data and MD chart review for CEA and CAS.
METHODS: We performed an outcomes analysis on all CEA and CAS procedures from 2005-2010. We obtained ICD-9 codes from hospital records regarding prior stroke/TIA, symptom and high-risk status, and complications. An MD performed a chart review of the same patient records and the results were compared.
RESULTS: We identified 983 patients: 876 CEA and 107 CAS. Prior stroke/TIA (>6 months old) was overestimated with AD (8.5% vs. 5.9%) while the proportion of symptomatic patients was grossly underestimated (8.4% vs. 31.6%). Physiologic high-risk status was also underestimated (5.0% vs. 9.1%), but individual physiologic high-risk variables were overestimated. Anatomic high-risk status was unable to be determined with AD. The identification of perioperative strokes was identical at 2.1%, but with AD there were 6 false positive strokes and 6 false negative strokes identified giving an accuracy of 56%.
CONCLUSIONS: Administrative data are useful, but subject to coding accuracy. Administrative data are unreliable for determining preoperative symptom status and postoperative stroke complications when analyzing outcomes of CEA and CAS.
AUTHOR DISCLOSURES: R. P. Bensley, Nothing to disclose; E. L. Chaikof, Nothing to disclose; J. D. Darling, Nothing to disclose; A. D. Hamdan, Nothing to disclose; R. Hurks, Nothing to disclose; R. C. Lo, Nothing to disclose; M. L. Schermerhorn, Medtronic, Consulting fees or other remuneration (payment), Boston Scientific, Consulting fees or other remuneration (payment), Endologix, Consulting fees or other remuneration (payment); M. C. Wyers, Nothing to disclose.
Comparison of outcomes analysis using administrative data vs. MD chart review in patients who have undergone CEA or CAS
|
Administrative Data |
MD Chart Review |
|
Administrative Data |
MD Chart Review |
Prior stroke/TIA |
8.5% |
5.9% |
Physiologic high-risk |
5.0% |
9.1% |
Symptomatic |
8.4% |
31.6% |
Age >80 |
19.3% |
19.3% |
Anatomic high-risk |
0% |
4.8% |
MI within 30 days |
12.6% |
0.2% |
Stroke complications* |
2.1% |
2.1% |
Unstable angina |
0.1% |
1.0% |
True positive strokes |
15 (71.4%) |
21 |
CHF class III/IV |
9.3% |
0.3% |
False positive strokes |
6 (28.6%) |
0 |
Hemodialysis |
8.9% |
1.1% |
False negative strokes |
6 (28.6%) |
0 |
Pulmonary dysfunction |
16.7% |
1.0% |
*The accuracy of identification of postoperative strokes with administrative data is 56%
Posted April 2012