Intracranial Stent Placement for Symptomatic Intracranial Stenosis as Part of a Clinical Trial Versus Outside a Clinical Trial
Background and Purpose—A high rate of postprocedure complications in the Stenting versus Aggressive Medical Therapy for Intracranial Arterial Stenosis (SAMMPRIS) trial has raised concerns whether such results are representative of intracranial stent placement in actual routine practice.
Methods—Using the Nationwide Inpatient Sample from 2008 to 2010, patients with cerebral ischemic events treated with intracranial stent as part of a clinical trial or outside the trial were identified. The composite end point (postoperative stroke, cardiac complications, and mortality) was reported.
Results—Of the 3447 patients who underwent intracranial stent placement, 223 patients (6.5%) were enrolled in a clinical trial. The rate of composite end point was higher in patients treated outside clinical trials compared with those treated within clinical trials (14.2% versus 4.5%; P=0.1). The proportion of patients discharged to home was higher in those treated in clinical trials (76.8% versus 49.6%; P=0.001).
Conclusions—Intracranial stent placement procedures outside a clinical trial have higher rates of postoperative stroke, cardiac complication, and mortality.
After the approval of the Gateway-Wingspan stent system, several prospective registries reported high rates of technical success, acceptable periprocedure complication rates, and 1-year ipsilateral stroke/death rates after intracranial stent (ICS) placement.1–3 The Stenting versus Aggressive Medical Therapy for Intracranial Arterial Stenosis (SAMMPRIS) trial4 was prematurely halted at the recommendation of the Safety Monitoring Board because of high 30-day stroke and death rate in the endovascular group (14.7%) compared with the medical group (5.8%).4
Several reports have raised concerns regarding the impact of operator experience, volume of the endovascular center, choice of the treatment device, and timing of the procedure.5,6 A recent study from 3 centers not participating in SAMMPRIS7 reported that the overall 30-day postprocedure stroke and death rate with angioplasty and ICS was 7.2% in the SAMMPRIS-eligible group and 3.3% in the SAMMPRIS-eligible, angioplasty-treated subgroup. Therefore, the question regarding generalization of the SAMMPRIS results is an essential component of medical decision making for patients with symptomatic intracranial stenosis. In our study, we compared the clinical outcomes of patients undergoing ICS placement enrolled in a trial with patients not enrolled in a trial. Because SAMMPRIS and Vitesse Intracranial Stent Study for Ischemic Therapy (VISSIT) were the only Investigational Device Exemption trials in the United States during the study time period, the results are effectively based on a comparison of the rates of postoperative complications between those treated in a trial and in general practice.
National Inpatient Sample (NIS) from 2008 to 2010 was used for our analysis. A comprehensive synopsis on NIS data is available at http://www.hcup-us.ahrq.gov.
We used the International Classification of Disease, 9th Revision, Clinical Modification (ICD-9-CM) primary diagnosis codes 433 to 437.1 to identify the patients admitted with ischemic stroke regardless of severity. ICD-9-CM procedure codes 00.62–00.65 were used to identify patients undergoing ICS placement excluding patients who received thrombolytic, endovascular treatment for stroke or intracranial aneurysm by using diagnosis-related group codes 543 and 39.71–39.79. We also used ICD-9-CM diagnostic codes V70.5 and V70.7 to identify patients enrolled in a clinical trial (CT) group. The remaining patients were placed in the outside clinical trial (OCT) group. V code identifies all patients who are enrolled in a trial and is used for Medicare reimbursements. Therefore, we expect a high level of accuracy and scrutiny for use of such code. Furthermore, among hospitals that identified themselves in the NIS database, ICS placement procedures with V code assignment were performed only in hospitals that were part of SAMMPRIS or VISSIT trials. Given the relationship between V code use and Medicare reimbursement, we performed an analysis including only Medicare patients for whom more consistent use is expected.
Study end points were neurological complication (ICD-9-CM code 997.00–997.09), cardiac complication (ICD-9-CM code 997.10), in-hospital mortality, and composite end point. Discharge status in NIS database was categorized into discharge to home and discharge to rehabilitation/nursing home.
Expanded materials and methods are provided in the online-only Data Supplement.
Of the 3447 patients who underwent ICS placement during the study period, 223 were enrolled in a CT group, whereas 3224 patients underwent ICS placement outside the clinical trial. The CT group comprised older patients compared with the OCT group (68±10 versus 65±14 years; P=0.008). Sex and medical comorbidities (hypertension, diabetes mellitus, congestive heart failure, renal failure, dyslipidemia, and atrial fibrillation) were not different between the 2 groups.
CT group patients had significantly shorter length of stay (5±4 versus 9±11 days; P<0.0001) compared with OCT group. Hospital teaching status was not different between the 2 groups (P=0.4).
The proportion of patients discharged to home was higher in the CT group (76.8% versus 49.6%; P=0.001). The proportion of patients discharged to rehabilitation/nursing home was higher in the OCT group (40.8% versus 18.7%; P=0.01). The adjusted odds ratio for discharge to rehabilitation/nursing home was 3.5 (95% confidence interval, 1.2 to 9.7; P=0.01) for the OCT group.
The rate of postoperative mortality was 9.6% in the OCT group and 4.5% in the CT group (P=0.4). The adjusted odds ratio of postoperative mortality was 2.0 (95% confidence interval, 0.2 to 14.4; P=0.5) for the OCT group. The rate of composite end point was 14.2% in the OCT group compared with 4.5% in the CT group (P=0.1). The adjusted odds ratio of composite end point was 3.4 (95% confidence interval, 0.5 to 21.5; P=0.1) for the OCT group.
In the subgroup analysis including only Medicare patients, the direction and magnitude of increased odds of the composite end point in the OCT group remained unchanged. The rate of composite end point was 14.7% in OCT group compared with 7.2% in the CT group (P=0.4). The adjusted odds ratio for composite end point for OCT group was 3.2 (95% confidence interval, 0.5 to 22.8; P=0.3).
Our study demonstrated that the composite of postprocedure mortality, neurological, and cardiac complications was 14.2% in OCT patients compared with 4.5% in CT patients. Interestingly, there were no neurological complications reported in the CT group. There may be an ascertainment bias because the methods used for neurological assessment are usually more rigorous with independent neurologist ascertainment in a clinical trial as opposed to self-ascertainment in routine practice. Similarly, lower rates of in-hospital mortality and composite end point of stroke, cardiac events, and death are observed with cervical carotid artery stent placement in patients as part of a clinical trial compared with those in actual clinical practice.8 One explanation is that patients treated within clinical trials have more favorable characteristics than those treated as part of routine practice (cherry-picking phenomenon).9
NIS data analysis has the inherent shortcomings of a large database analysis.10 The accuracy of the V code is not known; however, the code is essential for Medicare reimbursement for procedures that are not currently reimbursed as part of routine care. It is possible that some patients may have had ICS for other causes such as dissection or acute ischemic stroke treatment. Long-term outcomes cannot be assessed from the NIS database. We assume that discharge disposition has significant correlation with long-term outcome, but we do recognize that discharge to a short- or long-term care facility is not an ideal surrogate for postprocedure morbidity. Predictive value of discharge destination as a surrogate for evaluating unfavorable outcome at 3 and 12 months after stroke has been validated previously.11
Periprocedure complication rate remains high in patients treated with the currently available ICS technology outside a clinical trial. There is no evidence to support that SAMMPRIS results are not a valid representation of intracranial angioplasty and stent placement in general practice at large.
Sources of Funding
This study was performed independently of any financial support.
This article has not been published previously and is not being considered for publication elsewhere in whole or in part in any language except as an abstract.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.113.002567/-/DC1.
- Received June 19, 2013.
- Accepted August 20, 2013.
- © 2013 American Heart Association, Inc.
- Fiorella D,
- Levy EI,
- Turk AS,
- Albuquerque FC,
- Niemann DB,
- Aagaard-Kienitz B,
- et al
- Bose A,
- Hartmann M,
- Henkes H,
- Liu HM,
- Teng MM,
- Szikora I,
- et al
- Marks MP