Bratzler, 1996⁴⁷

Retrospective, observational study using Medicare claims database and medical records

8 hospitals in Oklahoma (OK)

OK Medicare beneficiaries

1993-1994

All CEA cases from the OK Medicare claims data; hospital selection not specified; all surgeons performing CEA in the 8 study hospitals

813 CEAs
(774 patients)

Median-73 yrs
% White NR
% Female NR
26% DM
20% COPD
10% CHF
67% CHD
71% HTN
26% Smoke
98% had stenosis > 60%

Cebul, 1998⁴⁸

Retrospective, cohort study using Medicare provider analysis and review files

115 hospitals/478 surgeons in Ohio (OH)

7/93-6/94

Random sample of 700 of 4,120 non-HMO Medicare beneficiaries in OH (18 patients had no medical record; 4 had stroke; 3 had bilateral carotid procedures during same hospitalization); hospitals performing CEA in OH

678 patients

Mean-73.1 yrs
94% White
46% Female
26% DMB
16% COPD
9% CHF
% CHD NR
71% HTN
31%^b Smoke
% stenosis NR

Halm, 2003⁷
Rockman, 2005⁵⁵
Halm, 2005⁵³
Press, 2006⁵⁴

Cross-sectional study based on medical record review of in- and outpatient records

4 university hospitals, 2 community hospitals served by 67 surgeons

1/97-12/98

Reviewed 2,365 of 2,390 CEAs based on hospitals' databases.  Cases excluded if re-operation, surgery combined with other major procedure, no CEA performed or missing data. 

Each hospital contributed 130-583 cases.

2124

Mean-72 yrs
87% White
43% Female
29% DM
9% COPD
8% CHF
% CHD NR
73% HTN
% Smoke NR
96% had >50% stenosis

Halm 2007⁴⁹

Retrospective, observational study using New York State Medicare claims database and medical records

1/98-6/99

Reviewed 10, 817 eligible cases (94.8%). Excluded reoperations, CEA combined with CABG, or no CEA performed.  551 cases excluded because of missing data. 

9588

Mean-74.6 yrs
93% White
56% Male
30% DM
19% COPD
10% CHF
 62% CHD

VA-NSQIP
Samsa, 2002⁴⁶

Secondary analysis of  VA NSQIP data

132 VA medical centers

Patients undergoing surgery at a VA medical center

1994-1995

1996-1997

94% of those available for assessment included in database, most excluded because of multiple index operations; 5 of the 123 VAs assessed <80% of eligible cases.

All VA hospitals performing major surgery; all surgeons performing surgery at VA hospitals.

7,842

Mean-68 yrs
91% White
2% Female
17% DM
17% COPD
2% CHF
% CHD NR
% HTN NR
% Smoke NR
% Stenosis NR

Horner, 2002⁴⁵

Secondary analysis of data in VA NSQIP examining differences in CEA outcomes by ethnic group

132 VA Medical Centers

Patients having CEA

10/94-9/97

Limited to men having CEA

6551

20% >75  yrs
91% White
0% Female
29% DM
12% COPD
2% CHF
% CHD NR
% HTN NR
% Smoke NR
% Stenosis NR

Karp, 1998⁵⁰

Retrospective, cross-sectional study

Medicare beneficiaries who underwent CEA in Georgia 

1993

Excluded 35 cases due to missing data.

1945

Mean-72.3 yrs
91% White
47% Female
22% DM
24% COPD
8% CHF
% CHD NR
% HTN NR
% Smoke NR
69% >75%

Kresowik, 2000⁵²

Retrospective, observational study using Medicare database and medical records

30 hospitals in Iowa

Iowa Medicare beneficiaries

1994 & 6/95 to 5/96

All CEA cases from the Iowa Medicare claims database (Part A & B); all hospitals in Iowa performing CEA on Medicare patients; all surgeons in Iowa performing CEA on Medicare patients

2063

Median-74 yrs
% White NR
40% Female
% DM NR
% COPD NR
% CHF NR
% CHD NR
% HTN NR
% Smoke NR
% Stenosis NR

Kresowik, 2001⁵¹

Retrospective, observational study using Medicare database and medical records

10 states

Medicare beneficiaries

6/95-5/96

Random sample of 10,561 from 28,083 procedures identified from the MEDPAR Part A claims.

10,030 patients

Mean-73.6 yrs
% White NR
43% Female
% DM NR
% COPD NR
% CHF NR
% CHD NR
% HTN NR
% Smoke NR
% Stenosis NR

Kresowik, 2004⁴⁴

Retrospective, observational study using Medicare database and medical records

10 states

Medicare beneficiaries

6/98-5/99

Random sample of procedures identified from the MEDPAR Part A claims.

9,945 patients

Mean-NR
% White NR
43% Female
% DM NR
% COPD NR
% CHF NR
% CHD NR
% HTN NR
% Smoke NR
% Stenosis NR

Bratzler, 1996⁴⁷

347
(43%)

Not reported

Overall=3.7%

High volume hospital (>100 cases/year) = 3.5%
Low volume hospital=5.2%
3% HTN
2% wound hematoma
2% pneumonia

Data collected from medical record and claims database

Reviewer blinding not discussed

No comprehensive evaluation, outcomes determined by coding or documentation in chart

Generalizability low, select population

Good

Cebul, 1998⁴⁸

167
(25%)

Not reported

2.4%

Hospital-specific stroke/death rates inversely related to the number of procedures, ranging from 7.7% lowest quartile to 2.5% highest quartile

Asymptomatic patients at higher-volume hospitals (greater than median) had no strokes or death at 30 days compared to 4.9% and 4.6% in lower volume hospitals.  Outcomes did not differ significantly by surgeon volume.

Undergoing surgery in a  higher volume hospital was associated with a 71% reduction in risk of stroke or death at 30 days, after adjusting for patient characteristics (OR=0.29 (o.12-0.69)).

No assessment of patients, outcomes determined from readmission data; study did not include outpatient visits

Predominantly white population

Fair

Halm, 2003⁷
Rockman, 2005⁵⁵
Halm, 2005⁵³
Press, 2006⁵⁴

1413
(65%)

Not reported

Asymptomatic with no co-morbidities=1.28%
Low comorbidity (1 cardiac risk factor) =  2.21%
Moderate (2)=2.77%
High (ESRD, severe disability or over 2 risk factors)=5.56%

Mean complication rate across groups=2.6%

Complication rates (especially CVA) are underestimated by administrative database.

No assessment of patients by neurologist.

All hospitals in 1 region, may not be generalizable.

Good

Halm 2007⁴⁹

72%

Not reported

Asymptomatic without high comorbidity=2.69%
Asymptomatic with high comorbidity=7.13%

Large number of cases excluded due to missing data. 

Complication rates (especially CVA) are underestimated by administrative database.

No assessment of patients by neurologist.

All hospitals in 1 region, may not be generalizable.

Fair

VA-NSQIP
Samsa, 2002⁴⁶

3,231

Not reported

30-day death, CVA, MI
Overall=2.4%
1994-95=2.7%
1996-97=2.2%

Reviewer not blinded to treatment, hospital course.

Loss to follow-up not discussed, although likely very little.

No comprehensive exam by neurologist for outcome assessment.

No discussion of hospital selection.

Other complications not listed.

Generalizability low select population (white males).

Good

Horner, 2002⁴⁵

2852
(44%)

20% >75 yrs
92% White
0% Female
28% DM
10% COPD
2% CHF
% CHD NR
% HTN NR
% Smoke NR
% Stenosis NR

Stroke or death:
1.6% White
2.1% Black
2.2% Hispanic

Stroke, MI or death:
2.3% White
2.1% Black
3.2% Hispanic

Little selection within VA (VA patients are a selected subgroups of US population)

Good

Karp, 1998⁵⁰

972
(51%)

Not reported

Mortality=0.8%
Mod/Severe Stroke=1.0%
MI=0.8%
Combined (above)=2.6%
All Stroke=2.4%

Symptomatic patients:
Mortality=1.7%
Mod/Severe Stroke=2.7%
MI=1.4%
Combined (above)=5.8%
All stroke=4.7%

Found statistically significant increase in morbidity, mortality and less severe complications at hospitals performing <10 CEAs.

No comprehensive exam by neurologist for outcome assessment.

No discussion of hospital selection.

Generalizability low (all males, mostly white).

Fair

Kresowik, 2000⁵²

671
(20% '94; 40% '95-96)

Not reported

Overall=3.4%
1994=3.8%
1995-96=3.3%

Unclear when reports of outcomes were given to hospitals & surgeons.

No comprehensive evaluation, depended on medical records for outcomes.

Relied on claims database for readmissions for stoke, death occurring after discharge.

Generalizability

Good

Kresowik, 2001⁵¹

3120
(39%)

Not reported

Combined events 3.7%
Mortality 1.1%

The combined event rate by state for asymptomatic patients ranged from 2.3% to 6.7%.  Mortality ranged from 0.5% to 2.5%.  Only 2 states significantly different from the mean.

Missed nonfatal neurologic events occurring after discharge that did not result in another hospitalization.

Good

Kresowik, 2004⁴⁴

4093

Not reported

Combined events 3.8%

The combined event rate by state for asymptomatic patients ranged from 1.4% to 6.0%.  Only 3 states significantly different from the mean.

Missed nonfatal neurologic events occurring after discharge that did not result in another hospitalization.

Good