Article Text
Abstract
Background and purpose Mechanical thrombectomy (MT) for the treatment of acute ischemic stroke has been growing in popularity while the therapeutic benefit of MT has been increasingly debated. Our objective was to examine national trends in mortality following MT.
Methods We analyzed the National Inpatient Sample (2012) and the Nationwide Inpatient Sample (2008–2011) for patients with a primary diagnosis of acute ischemic stroke that received MT. Temporal trends in mortality were examined using Spearman's rank correlation. To account for confounding factors, mortality was further analyzed in binary logistic regression.
Results Hospitals performing MT comprised 8% of all hospitals treating ischemic stroke. The percentage of stroke cases treated with MT increased from 0.6% of cases in 2008 to 1.1% in 2012, totaling 16 307 MT cases in a 5 year period. Inhospital mortality decreased over the study period from 25.4% in 2008 to 16.1% in 2012 (r=−0.081, p<0.001). This finding was supported by regression analysis as each incremental year reduced the odds of mortality by 20% (OR=0.832, p<0.001). Administration of recombinant tissue plasminogen activator was associated with a decrease in the odds of mortality (OR=0.805, p<0.001).
Conclusions Utilization of MT represents a small percentage of stroke cases, although the trend is increasing. Mortality following MT has been showing a steady decline over the past 5 years. This may be a result of a learning curve, improved patient selection, and/or device improvements. Randomized trials remain essential to evaluate the potential benefit of endovascular devices and identify the most appropriate patients.
- Stroke
- Thrombectomy
- Intervention
- Economics
Statistics from Altmetric.com
Introduction
Prior to the 1996 Food and Drug Administration (FDA) approval of recombinant tissue plasminogen activator (rt-PA), therapy for acute ischemic stroke was confined to only a few highly specialized centers in Europe, the USA, and Japan. It is now common to administer intravenous rt-PA within 3 h of symptom onset to recanalize arterial occlusions, thereby restoring cerebral blood flow and improving patient outcome;1 a mildly more selective spectrum of patients can be treated between 3 and 4.5 h.2 A more focused approach to recanalization of large arterial occlusions and those deemed ineligible for intravenous rt-PA led to the development of clot busting techniques using endovascular therapy.3 This extends the treatment window and allows for intra-arterial administration of thrombolytic agents and mechanical thrombectomy (MT).4
MT was scrutinized after three neutral randomized clinical trials (RCTs) were published in the New England Journal of Medicine in 2013.5–7 However, since December 2014, the New England Journal of Medicine has published three positive RCTs that demonstrated a beneficial effect of MT for acute ischemic stroke.8–10 As stroke intervention is continually evolving, it is impossible to develop the ‘perfect’ clinical trial for such a complex disease entity that is still not fully understood. While the results of recent clinical trials are promising, it may be premature to conclude that equipoise no longer exists regarding MT.
Our objective was to examine temporal trends in mortality following MT for acute ischemic stroke with the hypothesis that mortality would decrease over time.
Materials and methods
Database characteristics
We analyzed discharge data from the National Inpatient Sample and the Nationwide Inpatient Sample (NIS), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality (Rockville, Maryland, USA). This database represents an approximately 20% stratified sample of USA non-federal hospitals. Detailed information on the design of the NIS is available at http://www.hcup-us.ahrq.gov.
Identification of patients and classification of hospitals
Patients with a primary diagnosis of acute ischemic stroke were identified in the NIS using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes (433.01, 433.11, 433.21, 433.31, 433.81, 433.91, 434.01, 434.11, 434.91, and 437.1). MT was identified with the procedure code 39.74, and the administration of rt-PA was identified by procedure code 99.10.
Case severity was determined using the All Patient Refined-Diagnosis Related Group (APR-DRG) risk of mortality. This proprietary four point ordinal scale (minor, moderate, major, extreme risk of mortality), developed by 3M Health Information Systems, has been validated to predict mortality more reliably than other severity measures using administrative datasets and has been used as a severity indicator in previous stroke studies.11–14 Any cases classified as minor were excluded from analysis (n=59). Individual comorbidity burden was determined by the Elixhauser system based on ICD-9-CM codes.15
Variables and statistical analysis
Data were analyzed using SPSS V.17 (IBM Corporation, Armonk, New York, USA). To obtain national estimates, discharge weights were applied. Hospital charges were converted to costs using the group weighted average cost to charge ratio (GAPICC) for years 2005–2011, and the cost to charge ratio (CCR) for 2012. The costs were adjusted to 2014 levels using the inflation calculator provided by the Bureau of Labor Statistics.16
Variables were compared using Kruskal–Wallis and χ2 tests, as appropriate. Temporal trends in mortality were examined using Spearman's rank correlation and the correlation coefficient, r. To account for confounding factors, mortality was analyzed in binary logistic regression using the enter method and controlled for age, gender, rt-PA administration, number of comorbidities, APR-DRG risk of mortality, and year. Comorbidity burden and age were transformed into categorical variables; comorbidity level by tertiles and age by decades. Year was kept as a continuous variable and the linearity of the logit assumption was verified using the Box–Tidwell test. ORs and their 95% CIs are reported. A probability value of 0.01 was considered statistically significant in order to nominally control for type I error.
Results
A total of 2 156 564 acute ischemic stroke cases were identified from 2008 to 2012. The percentage of stroke cases treated with MT increased, albeit very slowly (0.6% of cases in 2008 to 1.1% in 2012), totaling 16 307 MT cases in a 5 year period (table 1). Inhospital mortality significantly decreased over the study period (25.4% in 2008 to 16.1% in 2012; r=−0.081, p<0.001). When examining only those cases involving administration of rt-PA (n=8941), the trend in mortality displayed a highly linear relationship (R2=0.97, figure 1) and an overall decrease in mortality of almost 12% (26.5% in 2008 to 14.6% in 2012; r=−0.106, p<0.001).
We further examined mortality following MT in regression analysis (table 2). Each subsequent year in the study period was associated with a 20% decrease in the odds of mortality (OR=0.832, p<0.001). Administration of rt-PA and younger age were also associated with a decrease in the odds of mortality.
Discussion
The cost of healthcare continues to rise with unsustainable growth.17 Efforts must be made to continually assess the therapeutic benefit of interventions. Our study on MT is one of the largest to date in the USA and examined the most recent trends in mortality and cost using the past 5 years of data from the NIS.
MT was scrutinized in 2013 following the publication of three neutral RCTs in the New England Journal of Medicine: Interventional Management of Stroke III (IMS-III), Intra-arterial Versus Systemic Thrombolysis for Acute Ischemic Stroke (SYNTHESIS Expansion), and Mechanical Retrieval and Recanalization of Stroke Clots Using Embolectomy (MR RESCUE).5–7 Numerous questions were raised concerning the design and conduct of these trials, including a relatively long interval before intra-arterial treatment, the absence of pretreatment vascular imaging to confirm a proximal intracranial arterial occlusion, and the limited use of third generation MT devices, such as retrievable stents.8 The latter had repeatedly been shown to be superior to prior generation thrombectomy devices in early studies.18–20
In December 2014, the results of the Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands (MR CLEAN) were published and demonstrated significantly increased odds of favorable outcome (modified Rankin Scale score ≤2) for patients receiving intra-arterial treatment.8 This trial enrolled 500 patients, 233 of which were assigned to intra-arterial treatment and 190 treated with retrievable stents (81.5%). At the same time, three other RCTs were stopped early due to efficacy: the Endovascular Treatment for Small Core and Anterior Circulation Proximal Occlusion with Emphasis on Minimizing CT to Recanalization Times (ESCAPE) trial, Extending the Time for Thrombolysis in Emergency Neurological Deficits—Intra-Arterial (EXTEND-IA) trial, and Solitaire FR With the Intention For Thrombectomy as Primary Endovascular Treatment for Acute Ischemic Stroke (SWIFT PRIME).9 ,10 ,21 The mortality rate in the endovascular arms of the above trials ranged from 9% to 21% at 90 days.8–10 ,21 As the mortality data in the present study only reflected the inpatient period, the number is likely deflated in comparison with the typical 90 day evaluation point in RCTs, but the mortality rate in 2012 (16%) still compares well with recent RCTs.
It is important to note that the above RCTs did not demonstrate a significant difference in mortality between arms, with the notable exception of ESCAPE where intervention reduced mortality (10.4% vs 19.0% in the control arm).9 This may be attributable to patient selection in RCTs. A general exclusion criterion for clinical trials is the decision for comfort care measures or the enactment of a do not resuscitate (DNR) order, leading the population enrolled in clinical trials to be more aggressive with life saving measures in comparison with the overall population. We hypothesize that this may have imparted a selection bias that minimized mortality in the rt-PA arm of these clinical trials while increasing severe disability. For example, SWIFT PRIME demonstrated no difference in mortality (9.2% in the interventional arm vs 12.4% in the control arm), but the proportion of patients with a severe disability (modified Rankin Scale score=5) was more than four times higher in the control arm (13.4% vs 3%).21 The proportion of severely disabled patients in the rt-PA arm of EXTEND-IA and MR CLEAN similarly demonstrated much higher incidences than the endovascular arms (11% vs 3%, and 12% vs 6%, respectively).8 ,10
The trend for reduced mortality over time is likely a multifactorial response owing to technological developments, improved patient selection, and a learning curve, as reported during the implementation of previous endovascular operations.22 ,23 An earlier study by Hassan et al24 examined utilization and mortality trends in the NIS from 2004 to 2009 for ischemic stroke patients treated with endovascular therapy. The authors included patients receiving either MT or intra-arterial thrombolysis. They documented a decrease in mortality from the Merci era (2004–2007) to the Penumbra era (2008–2009) of 3%, but also a decrease in the number of routine discharges home, meaning that more patients were being discharged with moderate–severe disability. We noted an even more substantial reduction in mortality when considering 2008–2012, and observed a decrease of more than 9%. We also noted the number of discharges home increased from 2008 to 2012, with one additional patient discharged home per 25 treated. It is important to note that the latest generation of MT devices, stent retrievers, was not officially FDA approved until 2012 and represents the currently preferred technique for the endovascular management of acute ischemic stroke.25–27 It will remain important to study temporal trends as new years of the NIS become available.
We performed a subgroup analysis of patients receiving rt-PA (figure 1) as this population has been the focus of a large number of RCTs. It is reassuring to note that having rt-PA onboard did not increase mortality in comparison with the overall cohort. It is also important to note that the trend in reduced mortality over the study period was apparent when examining the overall cohort, and not just the limited subgroup treated with thrombolysis. Patients without rt-PA were inherently biased towards a later time to treatment due to the prompt time windows for thrombolysis. Similar outcomes may reflect an improvement in patient selection for endovascular therapy and the ability of modern imaging criteria, including perfusion based mismatch and the presence of pial collaterals, to identify patients beyond the traditional time windows that would benefit from intervention.28 ,29
Opposed to the soaring rises in cost seen in the treatment of some disease processes,30 the cost of acute ischemic stroke admission treated with MT remained relatively stable over the 5 year study period. This is reassuring but we should highlight that our analysis does not reflect the true societal cost.31–33 Long term rehabilitation and care should be considered in future studies, but these measures were beyond the scope of the anonymized database used in this study.
Due to limitations in the NIS, we cannot further investigate disease specifics known to affect stroke outcome, including size of occlusion, collateral reperfusion, and the admission National Institutes of Health Stroke Scale score. In addition, the outcomes do not include any specific neurological or functional assessment.34 We chose to analyze mortality as it is considered a ‘hard endpoint’ and not subject to coding errors or missed during chart extraction. A further limitation is that we could not evaluate the efficacy of any specific endovascular device.
Conclusions
Utilization of MT represents a small percentage of stroke cases, although the trend is increasing. Mortality following MT has been showing a steady decline over the past 5 years. This may be a result of a learning curve, improved patient selection, and/or device improvements. Randomized trials remain essential to evaluate the potential benefit of endovascular devices and identify the most appropriate patients.
References
Footnotes
Contributors EMD and MRV: study concept and design, and drafting of the manuscript. MRV: acquisition of the data. All authors: analysis and interpretation of the data, and critical revision of the manuscript for important intellectual content. EMD: guarantor.
Competing interests EMD is a physician consultant for MicroVention, Covidien Neurovascular, and Integra LifeSciences Corporation.
Ethics approval This study used the Nationwide Inpatient Sample, a de-identified patient database. Therefore, this study did not require institutional review board review in accordance with the Code of Federal Regulations, 45 CFR 46.
Provenance and peer review Not commissioned; externally peer reviewed.