Author + information
- Received April 17, 2015
- Accepted April 23, 2015
- Published online August 17, 2015.
- Suzanne V. Arnold, MD, MHA∗,†,
- Matthew R. Reynolds, MD, MSc‡,
- Kaijun Wang, PhD∗,
- Elizabeth A. Magnuson, ScD∗,†,
- Suzanne J. Baron, MD, MSc∗,†,
- Khaja M. Chinnakondepalli, MS∗,
- Michael J. Reardon, MD§,
- Peter N. Tadros, MD‖,
- George L. Zorn, MD‖,
- Brij Maini, MD¶,
- Mubashir A. Mumtaz, MD¶,
- John M. Brown, MD#,
- Robert M. Kipperman, MD#,
- David H. Adams, MD∗∗,
- Jeffrey J. Popma, MD††,
- David J. Cohen, MD, MSc∗,†∗ (, )
- CoreValve US Pivotal Trial Investigators
- ∗Saint Luke’s Mid America Heart Institute, Kansas City, Missouri
- †University of Missouri-Kansas City, Kansas City, Missouri
- ‡Harvard Clinical Research Institute, Boston, Massachusetts
- §Houston-Methodist-DeBakey Heart and Vascular Center, Houston, Texas
- ‖University of Kansas Hospital, Kansas City, Kansas
- ¶Pinnacle Health, Harrisburg, Pennsylvania
- #Morristown Memorial Hospital, Morristown, New Jersey
- ∗∗Mount Sinai Medical Center, New York, New York
- ††Beth Israel Deaconess Medical Center, Boston, Massachusetts
- ↵∗Reprint requests and correspondence:
Dr. David J. Cohen, Saint Luke’s Mid America Heart Institute, University of Missouri-Kansas City School of Medicine, 4401 Wornall Road, Kansas City, Missouri 64111.
Objectives This study sought to compare the health status outcomes for patients treated with either self-expanding transcatheter aortic valve replacement (TAVR) or surgical aortic valve replacement (AVR).
Background In patients at increased surgical risk, TAVR with a self-expanding bioprosthesis is associated with improved 1-year survival compared with AVR. However, elderly patients may be just as concerned with quality-of-life improvement as with prolonged survival as a goal of treatment.
Methods Between 2011 and 2012, 795 patients with severe aortic stenosis at increased surgical risk were randomized to TAVR or AVR in the CoreValve US Pivotal Trial. Health status was assessed at baseline, 1 month, 6 months, and 1 year using the Kansas City Cardiomyopathy Questionnaire, Medical Outcomes Study Short-Form 12 Questionnaire, and EuroQOL 5-dimension questionnaire; growth curve models were used to examine changes over time.
Results Over the 1-year follow-up period, disease-specific and generic health status improved substantially for both treatment groups. At 1 month, there was a significant interaction between the benefit of TAVR over AVR and access site. Among surviving patients eligible for iliofemoral (IF) access, there was a clinically relevant early benefit with TAVR across all disease-specific and generic health status measures. Among the non-IF cohort, however, most health status measures were similar for TAVR and AVR, although there was a trend toward early benefit with TAVR on the Short-Form 12 Questionnaire’s physical health scale. There were no consistent differences in health status between TAVR and AVR at the later time points.
Conclusions Health status improved substantially in surviving patients with increased surgical risk who were treated with either self-expanding TAVR or AVR. TAVR via the IF route was associated with better early health status compared with AVR, but there was no early health status benefit with non-IF TAVR compared with AVR. (Safety and Efficacy Study of the Medtronic CoreValve® System in the Treatment of Symptomatic Severe Aortic Stenosis in High Risk and Very High Risk Subjects Who Need Aortic Valve Replacement; NCT01240902)
Whereas patients with severe aortic stenosis previously had to choose between surgical aortic valve replacement (AVR) or medical therapy, over the last decade transcatheter aortic valve replacement (TAVR) has emerged as a viable alternative to these treatment options. Less invasive than AVR, TAVR with a balloon-expandable valve has been shown to have superior outcomes compared with medical therapy for both mortality (1) and quality of life (2). Among patients at increased risk for surgery, balloon-expandable TAVR has similar late outcomes to AVR (3), although TAVR performed via the transfemoral route did show improved early quality of life compared with AVR (4).
Recently, an alternative TAVR platform with a self-expanding bioprosthesis (CoreValve, Medtronic, Inc., Minneapolis, Minnesota) was shown to be associated with improved survival at 1 year compared with AVR in patients at increased surgical risk (5). Although improved survival is an unequivocally important benefit of the CoreValve, how this device affects patients’ symptoms, functional status, and quality of life compared with AVR is unknown. Because the self-expanding bioprosthesis differs from the balloon-expandable valve in terms of the device itself, the risk of particular complications (e.g., paravalvular regurgitation, new pacemaker), and in the device delivery (e.g., smaller sheath size, different alternative access sites), the health status outcomes of patients treated with these different devices may also be different. These outcomes are particularly important, because elderly patients with multiple comorbidities—the typical population for whom TAVR would be considered—may be more concerned with quality of life than prolonged survival (6,7). To address this gap in knowledge, we used data from the CoreValve US High Risk Pivotal Trial to compare the health status outcomes for patients with aortic stenosis who are at increased surgical risk and are treated with either self-expanding TAVR or AVR.
Patient population and study protocol
The design of the CoreValve US High Risk Pivotal Trial, including inclusion and exclusion criteria, study procedures, and follow-up protocols, was published previously (5). Briefly, the trial enrolled patients with severe, symptomatic aortic stenosis who were considered to be at increased risk for perioperative mortality with AVR. Severe aortic stenosis was defined as: 1) aortic valve area ≤0.8 cm2 or aortic valve area index ≤0.5 cm2/m2; and 2) mean aortic valve gradient >40 mm Hg or peak aortic jet velocity >4.0 m/s. Patients also had to have New York Heart Association (NYHA) functional class II or higher heart failure symptoms and be considered to be at increased surgical risk—defined as a risk of death within 30 days after surgery of ≥15%, as estimated by 2 cardiac surgeons and 1 interventional cardiologist at the investigative site. Patient risk eligibility was also confirmed by consensus among at least 2 senior cardiac surgeons and 1 interventional cardiologist who were members of the national screening committee.
The approach for the TAVR procedure (iliofemoral [IF] or noniliofemoral [NIF] [performed either via subclavian artery or direct aortic approach]) was determined using computed tomography. Patients were then randomly assigned in a 1:1 ratio to treatment with TAVR or AVR. Randomization was stratified according to investigational site and intended access site (IF or NIF). The study was approved by the institutional review board at each investigational site, and all patients provided written informed consent.
Health status measures
Disease-specific and generic health status was assessed at baseline and at 1 month, 6 months, and 1 year after enrollment using validated written questionnaires. Baseline questionnaires were administered in person after enrollment but before the implant procedure. Follow-up questionnaires were administered by mail from a central coordinating center. Surveys that were not returned by mail in a timely fashion were administered by telephone interview. Non-English speakers completed validated translations of the questionnaires.
Disease-specific health status was assessed using the Kansas City Cardiomyopathy Questionnaire (KCCQ) (8), a 23-item self-administered questionnaire that has been shown to be a reliable and valid measure of symptoms, functional status, and quality of life in patients with heart failure symptoms, including those with severe, symptomatic aortic stenosis (9). The KCCQ assesses specific health domains—physical limitation, symptoms, quality of life, social limitation, and self-efficacy—the first 4 of which are combined into an overall summary score, which was the pre-defined primary endpoint for this study. Values for all KCCQ domains and the summary score range from 0 to 100, with higher scores indicating less symptom burden and better quality of life. The KCCQ overall summary score generally correlates with NYHA functional class as follows: class I: KCCQ 75 to 100; class II: 60 to 74; class III: 45 to 59; class IV: 0 to 44 (9,10). Changes in the KCCQ of 5, 10, and 20 points correspond to small, moderate, or large clinical improvements, respectively (10).
Generic health status was evaluated with the Medical Outcomes Study Short-Form 12 (SF-12) questionnaire (11) and the EuroQOL 5-dimension questionnaire (EQ-5D) (12). Derived from the Short-Form 36, the SF-12 provides mental and physical summary scores that are scaled to overall U.S. norms of 50 and an SD of 10, with higher scores indicating better quality of life (11). The minimum clinically important difference for the SF-12 physical and mental summary scores is ∼2 points (13). The EQ-5D is a generic health status measure consisting of 5 domains (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) that can be converted to utilities using an algorithm developed for the U.S. population (14). Utilities are preference-based health status measures and range from 0 to 1, with 1 representing perfect health and 0 corresponding to the worst imaginable health state (15).
In addition to examining the disease-specific and generic health status of survivors, we also examined rates of acceptable and favorable outcomes after TAVR using definitions that combined mortality and quality of life into a single outcome (16). As previously described, an acceptable outcome was defined as the presence of all of the following at 6 months: 1) alive; 2) KCCQ overall summary score ≥45 (roughly equivalent to NYHA functional class III or better); and 3) stability or improvement in the KCCQ score from baseline to 6 months (decrease of <10 points). A favorable outcome was defined as all of the following at 1 year: 1) alive; 2) KCCQ overall summary score ≥60 (roughly equivalent to NYHA functional class I to II); and 3) stability or improvement in the KCCQ score from baseline to 6 months (decrease of <10 points) (17).
The primary analysis compared the health status of patients randomized to TAVR versus AVR on an intention-to-treat basis. As a secondary analysis, we compared treatments using a per-protocol approach, where patients were included only if they received their assigned treatment via the assigned access route. Baseline characteristics, including health status, were compared between groups using 2-sample Student t tests for continuous variables and chi-square tests for categorical variables. Mean follow-up health status scores at 1 month, 6 months, and 1 year were compared with baseline within each treatment group using paired Student t tests. Rates of acceptable and favorable outcomes at 6 months and 1 year, respectively, were compared between groups using chi-square tests.
For each of the primary and secondary health status outcomes, longitudinal random-effects growth curve models were used to examine the relative effect of TAVR versus AVR over time (18). These growth curve models incorporate all available health status data from all follow-up time points, including those for patients who subsequently died, withdrew from the study, or were lost to follow-up. The models included baseline health status, TAVR access site (IF vs. NIF), treatment assignment, and the interaction between access site and treatment group. The pre-defined analytic plan specified that if a significant interaction (p < 0.05) between treatment assignment and access site was observed on the KCCQ overall summary score at any time point, then all health status outcomes would be analyzed separately for the IF and NIF groups. The models also considered age, sex, and severe chronic obstructive pulmonary disease. The intercept and linear effects of time were estimated using both fixed and random effects. Cubic and quadratic effects of time were considered (both terms were not included in the same model to avoid over-parameterization) as well as all 2- and 3-way interactions between treatment, time, and TAVR access site. Starting with the highest order time-by-treatment interaction, variables were retained in the model if p < 0.05 using a backward stepwise selection process. Estimates of differences in mean scores between treatment groups and 95% confidence intervals were obtained at each follow-up time point from these growth curve models. All analyses were performed using SAS version 9.3 (SAS Institute, Cary, North Carolina), and all tests were 2-tailed with a nominal type 1 error rate of 5% and no adjustments for multiple comparisons.
From 2011 to 2012, 795 patients with severe, symptomatic aortic stenosis from 45 U.S. centers were randomized to either self-expanding TAVR (n = 394) or AVR (n = 401) in the CoreValve US High Risk Pivotal Trial. Health status data were available for 709 patients (89%) at baseline, which formed our analytic cohort. Overall, the patients were elderly (mean age 83 years), had a high burden of cardiac and noncardiac comorbidities, had severe aortic stenosis (average mean gradient 48 mm Hg), and had substantial functional limitations due to heart failure symptoms (67% NYHA functional class III, 19% NYHA functional class IV). Frailty indicators were common, with 9% of patients not living independently, low serum albumin (a marker of chronic disease) in 16% of patients, and an average 5-m walk time of 9.5 s (gait speed ≥6 s is considered slow and a marker of frailty).
A total of 84% of patients were eligible for IF access, whereas 16% required NIF access, which was performed either via subclavian artery or direct aortic approach. The baseline characteristics of these patients, stratified by access site (IF vs. NIF), are shown in Table 1. Patients who required NIF access generally had more cardiac comorbidities, more peripheral vascular disease, and lower body weights compared with those who were eligible for IF access. The treatment groups were well-matched, with only minor differences between groups.
Baseline health status, stratified by access site, is shown in Table 1. In the overall population, the mean KCCQ overall summary score was 46.8 points, which is generally consistent with NYHA functional class III symptoms. The mean SF-12 physical summary score was 30.9 points (∼2 SDs below population mean), and the mean SF-12 mental summary score was 47.9 points (near population mean). Mean EQ-5D utility was 0.73. Patients who required NIF access generally had higher KCCQ and SF-12 physical summary scores compared with those eligible for IF access. There were no significant differences in baseline health status between the TAVR and AVR groups in either access stratum.
Within group comparisons
Health status data were available for 59%, 75%, and 74% of patients eligible for follow-up at 1 month, 6 months, and 1 year, respectively, with slightly more missing data in the AVR arm compared with the TAVR arm, particularly at the 1-month time point (Online Table 1). Baseline characteristics were similar between those with and those missing 1-month health status data (Online Table 2), although patients missing data were more likely to have had a major stroke during the index hospitalization and were less likely to have been discharged to home compared with those patients in the analytic cohort.
The unadjusted health status scores for patients by treatment group and stratified by access site for all time points are shown in Online Table 3. The within-group comparisons of follow-up versus baseline health status are shown in Table 2 (IF group) and Table 3 (NIF group). In general, both disease-specific and generic health status improved substantially by 1 year after TAVR or AVR, regardless of access site. These improvements were evident by 1 month with IF-TAVR and by 6 months with NIF-TAVR and AVR. By 1 year, surviving patients had experienced, on average, 19 to 24 point increases in the KCCQ overall summary scores, 5 to 7 point increases in SF-12 physical summary scores, and 3 to 5 point increases in SF-12 mental summary scores. At 1 year, only the IF-TAVR group demonstrated a significant improvement in EQ-5D utilities, with an increase of 0.04 points compared with baseline.
The comparisons of health status between those patients randomized to TAVR versus AVR, according to the longitudinal growth curve models, are shown in Table 4 and Figures 1 and 2. There was a significant interaction between treatment assignment and access site for several of the key health status measures at the 1-month time point, and thus, all analyses were stratified by access site. IF-TAVR was associated with greater early improvement in health status compared with AVR, with 16.7-point higher KCCQ overall summary scores (95% confidence interval: 12.0 to 21.3, p < 0.001) at 1 month. However, at 6 months and 1 year, there were no differences between IF-TAVR and AVR in KCCQ overall summary scores. Similar trends (significantly better scores at 1 month for IF-TAVR vs. AVR with no differences at later time points) were observed for the KCCQ subscales, the SF-12 physical and mental summary scores, and the EQ-5D. For patients ineligible for IF access, there were no significant differences between TAVR and AVR for any of the health status measures at any time points, although confidence intervals for the differences were wide due to the much smaller sample size of the NIF cohort.
Rates of acceptable and favorable outcomes
An acceptable outcome after TAVR or AVR, which combines survival status with health status outcomes at 6-month follow-up, occurred in 73% of TAVR patients versus 64% of AVR patients (p = 0.022) (Table 5). This difference was confined to the IF cohort (75% vs. 63%; p = 0.005), with no difference in the rates of favorable 6-month outcomes between TAVR and AVR among patients in the NIF access stratum. At 1 year, the rates of favorable outcomes did not differ significantly between treatment groups, regardless of access site (overall population TAVR vs. AVR: 58% vs. 51%; p = 0.143).
Of the 709 randomized patients who had baseline quality-of-life data, 1 TAVR patient and 6 AVR patients did not have their assigned procedure, and 6 TAVR patients required a change in access site. As such, the per-protocol analytic population included 369 TAVR patients and 327 AVR patients who were randomized to and received their assigned treatment. There were no notable differences in either the within-group or between-group comparisons in the per-protocol analysis compared with the intention-to-treat analyses (Online Tables 4 to 6).
In this large, multicenter clinical trial of patients with severe symptomatic aortic stenosis who were at increased surgical risk, surviving patients who were treated with TAVR with a self-expanding valve or with surgical AVR had substantial improvements in health status. Among surviving patients eligible for IF access, those treated with TAVR had an early health status benefit compared with AVR, with greater improvements in heart failure symptoms, physical function, and quality of life at 1 month. However, by 6 months, there were no differences between groups, and there was little change in health status in either group between 6 months and 1 year. Among surviving patients who required NIF access, there were no significant differences between TAVR and AVR for any of the health status measures at any of the time points.
These access-specific results are particularly interesting when compared with those from Cohort A of the PARTNER (Placement of Aortic Transcatheter Valve) trial, which was a trial of balloon-expandable TAVR versus AVR in patients at increased surgical risk (4). In the PARTNER trial, there was a similarly strong interaction between access site and early health status benefits, with those patients eligible for transfemoral access receiving an early benefit with TAVR compared with AVR but no such benefit among those patients who underwent transapical TAVR. The more prolonged recovery among transapical patients was posited to reflect the greater pain associated with the lateral thoracotomy required for transapical access as compared with median sternotomy. In the CoreValve trial, patients who required NIF access were treated via either the subclavian artery or a direct aortic approach—both of which might be expected to produce less post-operative pain compared with a thoracotomy. Nonetheless, our data demonstrate that NIF-TAVR via these routes was still associated with health status and quality-of-life outcomes that were similar to those seen with surgical AVR—even at the 1-month time point.
There are several potential explanations for the apparent lack of early health status benefit with NIF-TAVR versus AVR seen in the CoreValve trial. First, it is possible that even the ministernotomy or minithoracotomy required for the direct aortic approach (the most common NIF approach used in the trial) affects recovery in elderly, debilitated, and frail patients to a similar extent as a full sternotomy. Second, there may be important differences in post-operative recovery between TAVR via the direct aortic or subclavian approach and surgical AVR that could not be measured by the health status instruments used in our study. Had we employed more sensitive approaches (such as a visual analog pain scale), it is possible that differences between groups would have then been detected. Finally, it is possible that there are important differences in early health status between NIF-TAVR and AVR that could not be detected because of the small sample size in the NIF cohort of our study. Indeed, the point estimates for the difference in SF-12 physical component scores between TAVR and AVR were similar for IF and NIF patients (and the interaction test was nonsignificant), suggesting a potential early benefit in recovery of physical function with NIF-TAVR versus AVR. Future studies will be necessary to determine how to optimize the health status recovery of patients requiring NIF access and to examine whether there are any differences among the alternative access sites.
It is also important to recognize that the health status outcomes reported in this study apply only to surviving patients. Because there was a survival advantage with TAVR compared with AVR in the CoreValve High Risk trial, it is possible that longer-term health status benefits of TAVR could have been missed due to differential attrition of the sickest patients in the AVR group. Indeed, when we integrated survival and quality-of-life outcomes into a single metric, we found that patients treated with TAVR were more likely to have an acceptable outcome at 6 months compared with patients undergoing AVR, with a similar trend in favorable outcomes at 1 year, regardless of access site. We believe that examining the results of the trial in this manner is both patient-centered and clinically relevant, because patients considering TAVR or AVR may value more than just survival. If a treatment saves lives but the quality of those lives is poor, this is unlikely to be viewed as a desirable outcome by patients or their treating physicians. Given the mortality benefit seen with TAVR in the CoreValve High-Risk trial, it is therefore encouraging to see that the rates of acceptable and favorable outcomes were improved as well with TAVR compared with AVR.
First, due to the small sheath size required for the CoreValve device, only a small proportion of patients required NIF access; as such, the comparisons of NIF-TAVR with AVR may be underpowered to detect modest yet important differences between the 2 treatments. Because many of the interactions of treatment by access site were highly significant at 1 month, however, it is clear that there are meaningful differences between IF- and NIF-TAVR in terms of early health status recovery. Second, there was a fair amount of missing health status data over follow-up, particularly in the AVR arm. We used growth curve models to analyze the health status data, which take advantage of all data available and limit the bias due to missing data. However, it is still possible that missing data could have influenced our results. Third, the trial was unblinded, which could have influenced how patients complete the health status assessments. Finally, the quality-of-life results are only reported up to 1 year; thus, the durability of these results beyond this time frame is unknown.
In a cohort of patients with severe aortic stenosis at increased surgical risk, treatment with either AVR or TAVR with the CoreValve self-expanding bioprosthesis results in substantial improvements in disease-specific and generic health status. Surviving patients treated via the IF route experienced more rapid improvement in heart failure symptoms and functional status compared with AVR, whereas patients treated via the NIF route had similar health status outcomes as AVR. Furthermore, when integrating survival with quality of life, we found that patients treated with TAVR were more likely to have an acceptable outcome compared with patients undergoing AVR. We believe that use of these combined outcomes is particularly relevant in the elderly population of patients considering TAVR or AVR, as they best represent the overarching goals of treatment from a patient perspective.
WHAT IS KNOWN? In previous studies, among patients at increased risk for surgery, balloon-expandable TAVR via a transfemoral approach showed improved early quality of life and similar late outcomes compared with surgical AVR.
WHAT IS NEW? An alternative TAVR platform with a self-expanding bioprosthesis has recently been shown to be associated with improved survival compared with AVR in patients at increased surgical risk, but how this device affects patients’ symptoms, functional status, and quality of life compared with AVR was unknown. Because the self-expanding bioprosthesis differs from the balloon-expandable valve in terms of the device, the routes of delivery, and the risk of particular complications, the health status outcomes of patients treated with these different devices may also be different. Using data from the CoreValve High Risk Pivotal trial, we found that—similar to the balloon-expandable TAVR—patients eligible for IF access had an early health status benefit compared with AVR across all disease-specific and generic health status measures, but there were no differences between groups at later time points. In contrast, there were no significant health status differences at any of the time points between TAVR and AVR for patients who required NIF access.
WHAT IS NEXT? Future studies are needed to determine how to optimize the health status recovery of patients requiring NIF access and to examine whether there are any differences among the alternative access sites (transapical, subclavian, or transaortic).
For supplemental tables, please see the online version of this paper.
The CoreValve US Pivotal trial was sponsored by Medtronic. All analyses, the preparation of the manuscript, and the decision to submit the manuscript for publication were done independently of the study sponsor. Dr. Arnold is supported by a Career Development Grant Award (K23 HL116799) from the National Heart, Lung, and Blood Institute. Dr. Reynolds has received research support from Edwards Lifesciences and Medtronic; and has received consulting income from Medtronic. Dr. Magnuson has received grant support from AstraZeneca, Eli Lilly, Daiichi Sankyo, Boston Scientific, Medtronic, and Edwards Lifesciences. Dr. Reardon has received honoraria from Medtronic for participation on a surgical advisory board. Dr. Tadros has received consulting fees from Medtronic and St. Jude Medical. Dr. Zorn has served as a consultant/proctor for Medtronic and Edwards Lifesciences. Dr. Maini has served on the Speakers Bureau and/or advisory board, served as a proctor, and performed contracted research for Medtronic, St. Jude Medical, Abbott Vascular; and has performed contracted research for Direct Flow. Dr. Mumtaz has served as a proctor, served as a consultant, and received honoraria, consulting fees, and travel reimbursements from Medtronic, Abbott, Edwards Lifesciences, and Atricure. Dr. Adams has received royalties through his institution from Medtronic for a patent related to a tricuspid valve annuloplasty ring and from Edwards Lifesciences for a patent related to degenerative valvular disease-specific annuloplasty rings. Dr. Popma has received honoraria from Boston Scientific, Abbott Vascular, and St. Jude Medical for participation on medical advisory boards; has received institutional grants from Boston Scientific, Direct Flow, and Abbott Vascular; and has received research grants from Medtronic. Dr. Cohen has received research support from Edwards Lifesciences, Medtronic, Boston Scientific, Abbott Vascular, Eli Lilly, Daiichi-Sankyo, and AstraZeneca; has received consulting income from Medtronic, Abbott Vascular, AstraZeneca, Eli Lilly, and Merck; and has received speaking honoraria from AstraZeneca. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- surgical aortic valve replacement
- EuroQOL 5-dimension questionnaire
- Kansas City Cardiomyopathy Questionnaire
- New York Heart Association
- Medical Outcomes Study Short-Form 12 Questionnaire
- transcatheter aortic valve replacement
- Received April 17, 2015.
- Accepted April 23, 2015.
- 2015 American College of Cardiology Foundation
- Reynolds M.R.,
- Magnuson E.A.,
- Lei Y.,
- et al.
- Reynolds M.R.,
- Magnuson E.A.,
- Wang K.,
- et al.
- Green C.P.,
- Porter C.B.,
- Bresnahan D.R.,
- Spertus J.A.
- Arnold S.V.,
- Spertus J.A.,
- Lei Y.,
- et al.
- Ware J.,
- Kosinski M.,
- Bjorner J.B.,
- Turner-Bowkes D.M.,
- Gandek B.,
- Maruish M.E.
- Arnold S.V.,
- Spertus J.A.,
- Lei Y.,
- et al.
- Arnold S.V.,
- Reynolds M.R.,
- Lei Y.,
- et al.