- Email: [email protected]
Meta-Analysis and Systematic Review of the Relationship between Hospital Volume and Outcome Following Carotid Endarterectomy
Eur J Vasc Endovasc Surg 33, 645e651 (2007) doi:10.1016/j.ejvs.2007.01.014, available online at http://www.sciencedirect.com on
REVIEW
Meta-Analysis and Systematic Review of the Relationship between Hospital Volume and Outcome Following Carotid Endarterectomy P.J.E. Holt,1* J.D. Poloniecki,2 I.M. Loftus1 and M.M. Thompson1 1
St George’s Vascular Institute, 4th floor, St James’ Wing, St George’s Hospital, London SW17 0QT, UK, and 2 Community Health Sciences, St George’s University of London, London SW17 0QT, UK Objectives. This study investigated the relationship between annual hospital volume and the outcomes in carotid endarterectomy and quantified critical volume threshold for this procedure. Data sources. PubMed, EMBASE and the Cochrane library were searched for all articles on the volume-outcome relationship in CEA. Review methods. Articles were included if they presented data on post-operative mortality and/or stroke rates and annual hospital volume of CEA. The review conformed to the QUOROM statement. The data were meta-analysed and a pooled effect estimate of volume on the stroke and/or mortality rates from CEA quantified, along with the critical volume threshold. Results. Twenty-five articles, encompassing 936 436 CEA, were analysed. Significant relationships between mortality rate and stroke rate and annual volume were seen. Overall, the pooled effect estimate was odds ratio 0.78 [95% confidence interval 0.64e0.92], in favour of surgery at higher volume units, with a critical volume threshold of 79 CEA per annum. Conclusions. Significantly lower mortality and stroke rates were achieved at hospitals providing a higher annual hospital volume of CEA. Hospitals wishing to provide CEA should adhere to minimum volume criteria. Keywords: Hospital volume; Mortality; Stroke; Outcome; Carotid endarterectomy.
Introduction Carotid endarterectomy is undertaken to reduce the risk of stroke in both symptomatic and asymptomatic patients with carotid artery stenosis.1e3 CEA has been shown to be superior to medical management in defined sub-groups of patients with carotid stenosis.4 With prompt surgery following a carotid territory event, highly significant absolute risk reductions in subsequent stroke may be attained5 with low numbers of patients requiring treatment to prevent a neurological event. The absolute risk reduction in stroke depends upon the natural history of the disease and the periprocedural risk of stroke and death. The lower the peri-operative event rate, the greater the absolute
risk reduction. Selected single centre series and trial participants have produced very low 30-day stroke rates, but there have been some studies that have questioned whether these results are reproducible in the community.6 It has been suggested that the number of adverse events related to CEA may be associated with the annual hospital volume undertaken. If a volumeoutcome relationship existed for CEA, then it may be expected to inform health policy, and direct health economists in the planning of vascular services. This article reviews the published evidence for the relationship between annual hospital volume and outcomes of CEA.
Methods *Corresponding author. P.J.E. Holt, St George’s Vascular Institute, 4th floor, St James’ Wing, St George’s Hospital, Tooting, London SW17 0QT, UK. E-mail address: [email protected]
PubMed, EMBASE and the Cochrane library medical databases were searched for articles investigating the
1078–5884/000645 + 07 $32.00/0 Ó 2007 Elsevier Ltd. All rights reserved.
646
P. J. E. Holt et al.
Validity assessment
Quorum Flowchart
Potentially relevant abstracts identified from searches (n=12465)
12420 excluded as abstract unsuitable
Articles retrieved for more detailed evaluation (n=45)
Having met the inclusion criteria, articles were assessed for quality and their findings. In-hospital death and post-operative stroke were the principal outcome measures. The presence, or absence, of case-mix adjustment was recorded, along with the impact of case-mix where adjusted and unadjusted data were presented. Case-mix adjusted data were used for the meta-analysis where available.
20 excluded as the paper did not address this review
Data abstraction
Potentially appropriate studies to be included in the meta-analysis (n=25)
4 excluded as data were unusable in metaanalysis Studies included in meta-analysis (n=21)
0 exclusions at this stage
Studies with usable information by outcome n=15 stroke rate n=9 death rate n=9 stroke and death
hospital-volume of CEA, and outcome from this procedure, using broad terms. The search terms employed were ‘‘carotid endarterectomy and volumeoutcome relationship,’’ ‘‘carotid endarterectomy and mortality,’’ ‘‘carotid endarterectomy and postoperative stroke,’’ ‘‘carotid endarterectomy and outcome measures,’’ ‘‘carotid endarterectomy and hospital volume,’’ ‘‘carotid endarterectomy and provider volume.’’ A final search ‘‘vascular surgery and mortality’’ was included to increase the sensitivity, though not specificity of the search.
Article selection The searches identified forty-five potentially relevant articles to which exclusion criteria were applied. Articles investigating the relationship between surgeon-operative volume and outcomes were excluded. Relevant citations were identified from scrutiny of retrieved articles in order to collect any articles missed by the searches. Eur J Vasc Endovasc Surg Vol 33, June 2007
Data abstraction was performed independently by PH. In-hospital death, post-operative stroke, or combined stroke/death rates were extracted from the original articles, along with the odds ratios and 95% confidence intervals for higher- and lower-volume hospitals. Where only annual volumes and adverse event rates were presented for the different hospital volumes, the number of affected patients was calculated, within the error of the published results. This allowed the generation of confidence intervals for these studies. The threshold value for the number of CEA repairs performed per annum between higher- and lowervolume hospitals was recorded, as stated in the original articles. Datasets were dichotomised to higher- and lower-volume categories where articles published a series of volume groupings (e.g. higher-, moderateand lower-volume hospitals). Study characteristics Analysis of the data was by meta-analysis and systematic review. Four articles were included in a systematic review, as the data could not be metaanalysed, due to inadequate published data. Quantitative data synthesis Weighted averages were calculated for fixed effect meta-analyses by the inverse variance method. The inverse variance method of meta-analysis is a widely applicable approach to meta-analysis and is based on a mathematical assumption that every study evaluated a common effect i.e. chance was the only factor effecting the results other than the effect investigated. It involved a weighted average of the effect estimates from the separate studies. The weight for each study was taken to be the inverse of the variance (one divided by the square of the standard error) of the effect estimate.7
Volume-outcome Relationships of Carotid Endarterectomy
Forest plots were used to illustrate the information from the individual studies used in the meta-analyses, in terms of odds ratios and 95% confidence intervals.8 The plots demonstrated the variation between studies and a pooled point estimate of the overall result, which represented the volume-outcome relationship in CEA.
647
and outcome of CEA. The mean death rate was 1.6% (range 0.3 to 5.2%), and the mean disabling stroke rate was 2.7% (0.23 to 6.1%). The results from 885 034 CEA were suitable for meta-analysis and were sub-divided into three groups depending on the focus of the published article. The pooled effects for each group were (odds ratio [95% confidence intervals], threshold value):
Publication bias Publication bias was assessed through separate funnel plots for each outcome variable investigated. Symmetry of the funnel plots indicated no publication bias.9 Heterogeneity Clinical heterogeneity may account for part of the effect seen in meta-analysis. Although the estimates of treatment effect at high-volume hospitals varied by chance between studies, the assessment of heterogeneity assessed whether there was more variation than expected by chance alone. By testing for heterogeneity, the validity of the combined effect estimate was assessed.9 This described whether the effects found in the individual studies were similar enough to be confident that the combined estimate was a meaningful description of the set of studies. The test of heterogeneity was not the sole determinant of model choice in meta-analysis, and clinical insight was relevant to both the investigation and interpretation of heterogeneity.10 The Q-statistic was calculated along with the degrees of freedom of the study (number of individual studies minus one) and compared using a chi-squared distribution. As a second test of heterogeneity the I-squared value was calculated. Sensitivity analysis Sensitivity analyses were performed by comparing the results obtained through both fixed and random effects meta-analyses. Furthermore, the exclusion of the largest trial in each sub-group, followed by recalculation of the meta-analyses, determined whether the results were being heavily determined by that trial.9 Results After exclusions, twenty-five articles (935 156 cases) provided information on the annual hospital volume
Stroke rate 0.84 [0.79e0.88], 72 CEA p.a. (Fig. 1A); Death rate 0.76 [0.74e0.81], 81 CEA p.a. (Fig. 1B); Combined stroke/death rate 0.73 [0.68e0.78], 84 CEA p.a. (Fig. 1C). Overall, stroke and death attributable to CEA occurred less frequently in higher-volume hospitals (0.78 [0.64e0.92]; Fig. 1D). The critical volume threshold between higher- and lower-volume hospitals was 79 CEA per annum, defined by weighting the threshold values stated in the included articles. Four articles (50 122 cases) were presented as a systematic review (Table 1). Two articles found significant improvements in outcomes with volume, and two did not. Study validity No significant heterogeneity was found in any of the sub-group analyses by either the Q-statistic or I-squared method. The funnel plots showed good symmetry indicating insignificant publication bias (Fig. 2aec). Exclusion of the largest study in each sub-group did not change the pooled effect estimate for each sub-group. This indicated that the results were not influenced by a single large trial, supporting the validity of these results. In comparing the pooled effect estimates from the three sub-group analyses, the results in terms of effect and volume threshold were consistent between the mortality, and stroke-mortality sub-groups. The stroke-only sub-group had a more conservative pooled effect estimate at a higher volume threshold compared to the other two sub-groups. Discussion The articles utilised in this review were largely based on retrospective, administrative data from the USA, and demonstrated significant relationships between the annual hospital volume and outcomes of CEA in terms of stroke and mortality. For these data, all-cause post-operative mortality was accurately quantified through discharge Eur J Vasc Endovasc Surg Vol 33, June 2007
648
P. J. E. Holt et al.
Table for Fig. 1: Sub-sections AeD Study
Threshold
Odds Ratio
95% Cases Confidence Interval
Study Weight
Birkmeyer18 Perler19 Edwards20 Wennberg6 Karp21 Roddy22 Dudley23 Dardik24 Harthun25 Segal26 Cowan27 Manheim28 Hannan29 Kempczinski13 Khuri30 Sub-total
164 50 50 21 50 50 100 50 100 30 100 100 100 50 29 81
0.88 1.14 0.80 0.85 0.64 0.54 0.78 0.85 0.40 0.93 0.67 0.66 0.73 1.22 1.06 0.77
0.80 0.79 0.61 0.77 0.45 0.30 0.59 0.30 0.24 0.59 0.51 e e e e 0.74
0.96 1.64 1.04 0.94 0.92 0.96 1.04 2.42 0.67 1.48 0.87 e e e e 0.81
479289 9918 11199 106790 10569 10211 28207 9842 14095 5657 50122 106493 28207 750 10173 881522
22.2 4.2 8.2 20.9 7.5 5.4 7.9 1.7 8.2 4.0 9.9 e e e e 100
Perler19 Edwards20 Mayo31 Roddy22 Dardik24 Harthun25 Cowan27 Kempczinski13 Khuri30 Sub-total
50 50 28 50 50 100 100 50 29 72
1.18 0.91 0.69 0.78 0.65 0.70 0.85 0.93 0.69 0.84
0.86 0.72 0.14 0.51 0.37 0.50 0.72 e e 0.79
1.61 1.13 3.40 1.20 1.14 0.97 1.02 e e 0.88
9918 11199 362 10211 9842 14095 5012 750 10173 71562
10.8 19.8 2.5 11.8 10.6 17.3 27.1 e e 100
Perler19 Edwards20 Karp21 Cebul14 Roddy22 Feasby32 Harthun25 Cowan27 Peck33 Sub-total
50 50 50 62 50 150 100 100 90 84
1.17 0.93 0.61 0.29 0.68 0.74 0.59 0.79 1.24 0.73
0.92 0.77 0.36 0.12 0.48 0.60 0.45 0.68 0.35 0.68
1.48 1.11 1.04 0.69 0.95 0.91 0.78 0.92 4.41 0.78
918 11199 10569 678 10211 14268 14095 5012 560 67510
8.6 14.2 7.1 8.5 10.3 15.5 14.6 20.1 1.2 100
79
0.78
0.64
0.92
885034
Overall total
Fig. 1. Forest plot demonstrating the volume-outcome relationship for CEA by outcome: (1A) post-operative death; (1B) post-operative stroke; (1C) combined stroke and death. Subtotals for each outcome, and an overall pooled effects estimate; (1D) are presented as odds ratios and 95% confidence intervals. Values to the left of the line of equivocal effect favour surgery at higher volume hospitals. Studies whose odds ratio is denoted by have been included on the plots to demonstrate their odds ratio only, but as no 95% confidence intervals were calculable, they were not used to calculate the pooled effect estimate.
Eur J Vasc Endovasc Surg Vol 33, June 2007
100
abstracts.11 Only two outcomes were investigated in this meta-analysis (stroke and death), therefore it is highly likely that the total complication rate was higher than presented here. Furthermore, as these studies included symptomatic and asymptomatic patients, the overall stroke and death rates were lower that expected, especially when compared to European data that are based on performing the majority of procedures on symptomatic patients.12 A further problem identified was in the coding of post-operative stroke, as patients may have been admitted to hospital due to cerebrovascular disease, or have suffered a stroke post-operatively. In the process of generating the data, either of these may be miscoded. There was significant variation in the way in which ‘‘stroke’’ was defined in the different articles ranging from a focal neurological deficit present for two or
Volume-outcome Relationships of Carotid Endarterectomy
649
Table 1. Summary of four studies for which the data were unsuitable for meta-analysis Author
Case Numbers
Threshold Value
Volume Effect Seen?
Statistically Significant
Summary
Middleton15 Pearce16
689 45744
10 expressed as change in risk with doubling of volume
N Y
N Y
Kantonen34 Fisher35
1600 2089
? 40
N Y
N Y
No difference in mortality or stroke rates. Median volume ¼ 102 CEA pa. Hospital volume, physician volume and specialty significant factors. Significant relationship for stroke, MI and death rates. No difference in mortality or stroke rates. Higher volume hospitals had lower stroke and mortality rates.
more consecutive days,13 to a deficit unresolved by discharge14 or fatal stroke.15 A number of articles specified the use of ICD-9 code (997.x) to determine the rate of post-operative stroke, with no further
Fig. 2. Funnel Plots assessing publication bias. None is demonstrated in any sub-group analysis. (2A) post-operative death; (2B) post-operative stroke; (2C) combined stroke and death.
assessment of severity, or duration. One study stated that in the presence of a code relating to stroke, it was ‘‘assumed to be post-operative’’.16 The difficulty of interpreting ‘post-operative stroke’ was supported in the sensitivity analyses, where the pooled effect estimate for this sub-group was of a lesser extent than the other sub-groups. The heterogeneous way that ‘post-operative stroke’ was interpreted suggested that the most valid measure for CEA outcome is post-operative death. Fig. 1a and b did show similarities in the impact of volume on outcome. Therefore, post-operative death may be a valid outcome surrogate for the hospital post-operative stroke-rate. The impact on the results of these confounding factors was impossible to quantify, and these differences will be difficult to rectify in the future without increasing clinician involvement in the coding process in parallel with the continued development of diagnostic coding systems, such as ICD-10. The results demonstrated that CEA performed at higher-volume hospitals had improved outcome quantified through significantly lower stroke and death rates. The key factors underlying this complex relationship related to hospital infrastructure (cardiology, specialist anaesthetists and neurology support onsite), the provision of suitable intensive care facilities where necessary, and concomitant provision of highvolume surgeons with vascular sub-specialisation.17 For CEA, lower-volume surgeons achieved results similar to higher-volume surgeons when operating in higher-volume hospitals. This supported the concept that hospital infrastructure was as a major component of the volume-outcome relationship,14 and acted independently to a relationship between surgeon operative volume and outcome. Death and stroke rate for CEA are lowest for high-volume vascular specialist surgeons, operating in high-volume hospitals.14 It has been suggested that differences in case-mix between higher- and lower-volume hospitals may underlie the volume-outcome relationship observed, Eur J Vasc Endovasc Surg Vol 33, June 2007
650
P. J. E. Holt et al.
with higher-volume hospitals taking on a larger number of asymptomatic cases, and lower-risk patients. However, this was not supported by articles in this study, with significant crude results remaining significant after case-mix adjustment, whether those results showed a volume-outcome relationship or not.18 The studies utilised in this analysis varied in the way the data were presented, with some presenting an in-hospital death or stroke rate, and others a combined stroke/death rate. In order to more easily assess the data for CEA, it would be advisable that all future studies make an assessment of both primary outcome measures for this procedure, and present a combined stroke/death rate. Further risk adjustment would be made possible through the routine inclusion of physiological parameters, including a pre-operative neurological scoring system and functional assessment, whether the patient was symptomatic or asymptomatic, and the degree of carotid artery stenosis. The evidence presented here demonstrated that the relationship between hospital volume and outcome was consistent and reproducible. We suggest that all healthcare systems should establish volume criteria for CEA. The volume threshold will need individual assessment in each system, as different policies regarding the proportion of asymptomatic carotid artery stenoses operated will have an effect on the threshold.
Conclusion CEA is the gold standard therapy in the prevention of neurological symptoms secondary to carotid artery stenosis. This study demonstrated that adverse outcome from CEA was reduced as the annual hospital volume of surgery increased. Volume criteria should be established in every healthcare system for CEA, to reduce the incidence of these events.
References 1 Group ECSTC. Randomised trial of endarterectomy for recently symptomatic carotid stenosis: final results of the MRC European Carotid Surgery Trial (ECST). Lancet 1998 May 9;351(9113):1379e 1387. 2 Study ECftACA. Endarterectomy for asymptomatic carotid artery stenosis. JAMA 1995 May 10;273(18):1421e1428. 3 FERGUSON GG, ELIASZIW M, BARR HW, CLAGETT GP, BARNES RW, WALLACE MC et al. The North American Symptomatic Carotid Endarterectomy Trial: surgical results in 1415 patients. Stroke 1999 Sep;30(9):1751e1758. 4 ROTHWELL PM. ACST: which subgroups will benefit most from carotid endarterectomy? Lancet 2004 Sep 25eOct 1;364(9440): 1122e1123 [author reply 5e6]. 5 ROTHWELL PM, ELIASZIW M, GUTNIKOV SA, WARLOW CP, BARNETT HJ. Endarterectomy for symptomatic carotid stenosis
Eur J Vasc Endovasc Surg Vol 33, June 2007
6
7 8 9 10 11
12 13 14 15
16
17
18 19
20
21
22 23 24
25
26
in relation to clinical subgroups and timing of surgery. Lancet 2004 Mar 20;363(9413):915e924. WENNBERG DE, LUCAS FL, BIRKMEYER JD, BREDENBERG CE, FISHER ES. Variation in carotid endarterectomy mortality in the Medicare population: trial hospitals, volume, and patient characteristics. JAMA 1998 Apr 22e29;279(16):1278e1281. Collaboration C. Revman. [cited 2006 20th December 2006]; Available from: http://www.cc-ims.net/RevMan/ivnotes.htm. LEWIS S, CLARKE M. Forest plots: trying to see the wood and the trees. BMJ 2001 Jun 16;322(7300):1479e1480. Collaboration C. openlearning. [cited 2006 20th December]; Available from: www.cochrane-net.org/openlearning. HARDY RJ, THOMPSON SG. Detecting and describing heterogeneity in meta-analysis. Stat Med 1998 Apr 30;17(8):841e856. WEINGART SN, IEZZONI LI, DAVIS RB, PALMER RH, CAHALANE M, HAMEL MB et al. Use of administrative data to find substandard care: validation of the complications screening program. Med Care 2000 Aug;38(8):796e806. PANAYIOTOPOULOS YP, PADAYACHEE S, COLCHESTER AC, TAYLOR PR. The changing indications for carotid endarterectomy in the UK. Int J Clin Pract 1997 Sep;51(6):375e377. KEMPCZINSKI RF, BROTT TG, LABUTTA RJ. The influence of surgical specialty and caseload on the results of carotid endarterectomy. J Vasc Surg 1986 Jun;3(6):911e916. CEBUL RD, SNOW RJ, PINE R, HERTZER NR, NORRIS DG. Indications, outcomes, and provider volumes for carotid endarterectomy. JAMA 1998 Apr 22e29;279(16):1282e1287. MIDDLETON S, DONNELLY N. Outcomes of carotid endarterectomy: how does the Australian state of New South Wales compare with international benchmarks? J Vasc Surg 2002 Jul;36(1):62e69. PEARCE WH, PARKER MA, FEINGLASS J, UJIKI M, MANHEIM LM. The importance of surgeon volume and training in outcomes for vascular surgical procedures. J Vasc Surg 1999 May;29(5):768e776. [discussion 77e78]. HOLT PJE, POLONIECKI JD, LOFTUS IM, GERRARD D, THOMPSON MM. Meta-analysis and systematic review of the relationship between volume and outcome in abdominal aortic aneurysm surgery. Br J Surg 2007;94(4):395e403. BIRKMEYER JD, SIEWERS AE, FINLAYSON EV, STUKEL TA, LUCAS FL, BATISTA I et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002 Apr 11;346(15):1128e1137. PERLER BA, DARDIK A, BURLEYSON GP, GORDON TA, WILLIAMS GM. Influence of age and hospital volume on the results of carotid endarterectomy: a statewide analysis of 9918 cases. J Vasc Surg 1998 Jan;27(1):25e31 [discussion 3]. EDWARDS WH, MORRIS Jr JA, JENKINS JM, BASS SM, MACKENZIE EJ. Evaluating quality, cost-effective health care. Vascular database predicated on hospital discharge abstracts. Ann Surg 1991 May; 213(5):433e438 [discussion 8e9]. KARP HR, FLANDERS WD, SHIPP CC, TAYLOR B, MARTIN D. Carotid endarterectomy among Medicare beneficiaries: a statewide evaluation of appropriateness and outcome. Stroke 1998 Jan;29(1): 46e52. RODDY SP, O’DONNELL Jr TF, WILSON AL, ESTES JM, MACKEY WC. The Balanced Budget Act: potential implications for the practice of vascular surgery. J Vasc Surg 2000 Feb;31(2):227e236. DUDLEY RA, JOHANSEN KL, BRAND R, RENNIE DJ, MILSTEIN A. Selective referral to high-volume hospitals: estimating potentially avoidable deaths. JAMA 2000 Mar 1;283(9):1159e1166. DARDIK A, BOWMAN HM, GORDON TA, HSIEH G, PERLER BA. Impact of race on the outcome of carotid endarterectomy: a population-based analysis of 9,842 recent elective procedures. Ann Surg 2000 Nov;232(5):704e709. HARTHUN NL, KONGABLE GL, BAGLIONI AJ, MEAKEM TD, KRON IL. Examination of sex as an independent risk factor for adverse events after carotid endarterectomy. J Vasc Surg 2005 Feb;41(2): 223e230. SEGAL HE, RUMMEL L, WU B. The utility of PRO data on surgical volume: the example of carotid endarterectomy. QRB Qual Rev Bull 1993 May;19(5):152e157.
Volume-outcome Relationships of Carotid Endarterectomy
27 COWAN Jr JA, DIMICK JB, THOMPSON BG, STANLEY JC, UPCHURCH Jr GR. Surgeon volume as an indicator of outcomes after carotid endarterectomy: an effect independent of specialty practice and hospital volume. J Am Coll Surg 2002 Dec;195(6):814e821. 28 MANHEIM LM, SOHN MW, FEINGLASS J, UJIKI M, PARKER MA, PEARCE WH. Hospital vascular surgery volume and procedure mortality rates in California, 1982e1994. J Vasc Surg 1998 Jul; 28(1):45e56 [discussion 8]. 29 HANNAN EL, POPP AJ, TRANMER B, FUESTEL P, WALDMAN J, SHAH D. Relationship between provider volume and mortality for carotid endarterectomies in New York state. Stroke 1998 Nov;29(11): 2292e2297. 30 KHURI SF, DALEY J, HENDERSON W, HUR K, HOSSAIN M, SOYBEL D et al. Relation of surgical volume to outcome in eight common operations: results from the VA National Surgical Quality Improvement Program. Ann Surg 1999 Sep;230(3):414e429 [discussion 29e32]. 31 MAYO SW, ELDRUP-JORGENSEN J, LUCAS FL, WENNBERG DE, BREDENBERG CE. Carotid endarterectomy after NASCET and ACAS: a statewide study. North American Symptomatic Carotid
32 33 34
35
651
Endarterectomy Trial. Asymptomatic Carotid Artery Stenosis Study. J Vasc Surg 1998 Jun;27(6):1017e1022 [discussion 22e3]. FEASBY TE, QUAN H, GHALI WA. Hospital and surgeon determinants of carotid endarterectomy outcomes. Arch Neurol 2002 Dec;59(12):1877e1881. PECK C, PECK J, PECK A. Comparison of carotid endarterectomy at high- and low-volume hospitals. Am J Surg 2001 May;181(5): 450e453. KANTONEN I, LEPANTALO M, SALENIUS JP, MATZKE S, LUTHER M, YLONEN K. Influence of surgical experience on the results of carotid surgery. The Finnvasc Study Group. Eur J Vasc Endovasc Surg 1998 Feb;15(2):155e160. FISHER ES, MALENKA DJ, SOLOMON NA, BUBOLZ TA, WHALEY FS, WENNBERG JE. Risk of carotid endarterectomy in the elderly. Am J Public Health 1989 Dec;79(12):1617e1620.
Accepted 21 January 2007 Available online 30 March 2007
Eur J Vasc Endovasc Surg Vol 33, June 2007
REVIEW
Meta-Analysis and Systematic Review of the Relationship between Hospital Volume and Outcome Following Carotid Endarterectomy P.J.E. Holt,1* J.D. Poloniecki,2 I.M. Loftus1 and M.M. Thompson1 1
St George’s Vascular Institute, 4th floor, St James’ Wing, St George’s Hospital, London SW17 0QT, UK, and 2 Community Health Sciences, St George’s University of London, London SW17 0QT, UK Objectives. This study investigated the relationship between annual hospital volume and the outcomes in carotid endarterectomy and quantified critical volume threshold for this procedure. Data sources. PubMed, EMBASE and the Cochrane library were searched for all articles on the volume-outcome relationship in CEA. Review methods. Articles were included if they presented data on post-operative mortality and/or stroke rates and annual hospital volume of CEA. The review conformed to the QUOROM statement. The data were meta-analysed and a pooled effect estimate of volume on the stroke and/or mortality rates from CEA quantified, along with the critical volume threshold. Results. Twenty-five articles, encompassing 936 436 CEA, were analysed. Significant relationships between mortality rate and stroke rate and annual volume were seen. Overall, the pooled effect estimate was odds ratio 0.78 [95% confidence interval 0.64e0.92], in favour of surgery at higher volume units, with a critical volume threshold of 79 CEA per annum. Conclusions. Significantly lower mortality and stroke rates were achieved at hospitals providing a higher annual hospital volume of CEA. Hospitals wishing to provide CEA should adhere to minimum volume criteria. Keywords: Hospital volume; Mortality; Stroke; Outcome; Carotid endarterectomy.
Introduction Carotid endarterectomy is undertaken to reduce the risk of stroke in both symptomatic and asymptomatic patients with carotid artery stenosis.1e3 CEA has been shown to be superior to medical management in defined sub-groups of patients with carotid stenosis.4 With prompt surgery following a carotid territory event, highly significant absolute risk reductions in subsequent stroke may be attained5 with low numbers of patients requiring treatment to prevent a neurological event. The absolute risk reduction in stroke depends upon the natural history of the disease and the periprocedural risk of stroke and death. The lower the peri-operative event rate, the greater the absolute
risk reduction. Selected single centre series and trial participants have produced very low 30-day stroke rates, but there have been some studies that have questioned whether these results are reproducible in the community.6 It has been suggested that the number of adverse events related to CEA may be associated with the annual hospital volume undertaken. If a volumeoutcome relationship existed for CEA, then it may be expected to inform health policy, and direct health economists in the planning of vascular services. This article reviews the published evidence for the relationship between annual hospital volume and outcomes of CEA.
Methods *Corresponding author. P.J.E. Holt, St George’s Vascular Institute, 4th floor, St James’ Wing, St George’s Hospital, Tooting, London SW17 0QT, UK. E-mail address: [email protected]
PubMed, EMBASE and the Cochrane library medical databases were searched for articles investigating the
1078–5884/000645 + 07 $32.00/0 Ó 2007 Elsevier Ltd. All rights reserved.
646
P. J. E. Holt et al.
Validity assessment
Quorum Flowchart
Potentially relevant abstracts identified from searches (n=12465)
12420 excluded as abstract unsuitable
Articles retrieved for more detailed evaluation (n=45)
Having met the inclusion criteria, articles were assessed for quality and their findings. In-hospital death and post-operative stroke were the principal outcome measures. The presence, or absence, of case-mix adjustment was recorded, along with the impact of case-mix where adjusted and unadjusted data were presented. Case-mix adjusted data were used for the meta-analysis where available.
20 excluded as the paper did not address this review
Data abstraction
Potentially appropriate studies to be included in the meta-analysis (n=25)
4 excluded as data were unusable in metaanalysis Studies included in meta-analysis (n=21)
0 exclusions at this stage
Studies with usable information by outcome n=15 stroke rate n=9 death rate n=9 stroke and death
hospital-volume of CEA, and outcome from this procedure, using broad terms. The search terms employed were ‘‘carotid endarterectomy and volumeoutcome relationship,’’ ‘‘carotid endarterectomy and mortality,’’ ‘‘carotid endarterectomy and postoperative stroke,’’ ‘‘carotid endarterectomy and outcome measures,’’ ‘‘carotid endarterectomy and hospital volume,’’ ‘‘carotid endarterectomy and provider volume.’’ A final search ‘‘vascular surgery and mortality’’ was included to increase the sensitivity, though not specificity of the search.
Article selection The searches identified forty-five potentially relevant articles to which exclusion criteria were applied. Articles investigating the relationship between surgeon-operative volume and outcomes were excluded. Relevant citations were identified from scrutiny of retrieved articles in order to collect any articles missed by the searches. Eur J Vasc Endovasc Surg Vol 33, June 2007
Data abstraction was performed independently by PH. In-hospital death, post-operative stroke, or combined stroke/death rates were extracted from the original articles, along with the odds ratios and 95% confidence intervals for higher- and lower-volume hospitals. Where only annual volumes and adverse event rates were presented for the different hospital volumes, the number of affected patients was calculated, within the error of the published results. This allowed the generation of confidence intervals for these studies. The threshold value for the number of CEA repairs performed per annum between higher- and lowervolume hospitals was recorded, as stated in the original articles. Datasets were dichotomised to higher- and lower-volume categories where articles published a series of volume groupings (e.g. higher-, moderateand lower-volume hospitals). Study characteristics Analysis of the data was by meta-analysis and systematic review. Four articles were included in a systematic review, as the data could not be metaanalysed, due to inadequate published data. Quantitative data synthesis Weighted averages were calculated for fixed effect meta-analyses by the inverse variance method. The inverse variance method of meta-analysis is a widely applicable approach to meta-analysis and is based on a mathematical assumption that every study evaluated a common effect i.e. chance was the only factor effecting the results other than the effect investigated. It involved a weighted average of the effect estimates from the separate studies. The weight for each study was taken to be the inverse of the variance (one divided by the square of the standard error) of the effect estimate.7
Volume-outcome Relationships of Carotid Endarterectomy
Forest plots were used to illustrate the information from the individual studies used in the meta-analyses, in terms of odds ratios and 95% confidence intervals.8 The plots demonstrated the variation between studies and a pooled point estimate of the overall result, which represented the volume-outcome relationship in CEA.
647
and outcome of CEA. The mean death rate was 1.6% (range 0.3 to 5.2%), and the mean disabling stroke rate was 2.7% (0.23 to 6.1%). The results from 885 034 CEA were suitable for meta-analysis and were sub-divided into three groups depending on the focus of the published article. The pooled effects for each group were (odds ratio [95% confidence intervals], threshold value):
Publication bias Publication bias was assessed through separate funnel plots for each outcome variable investigated. Symmetry of the funnel plots indicated no publication bias.9 Heterogeneity Clinical heterogeneity may account for part of the effect seen in meta-analysis. Although the estimates of treatment effect at high-volume hospitals varied by chance between studies, the assessment of heterogeneity assessed whether there was more variation than expected by chance alone. By testing for heterogeneity, the validity of the combined effect estimate was assessed.9 This described whether the effects found in the individual studies were similar enough to be confident that the combined estimate was a meaningful description of the set of studies. The test of heterogeneity was not the sole determinant of model choice in meta-analysis, and clinical insight was relevant to both the investigation and interpretation of heterogeneity.10 The Q-statistic was calculated along with the degrees of freedom of the study (number of individual studies minus one) and compared using a chi-squared distribution. As a second test of heterogeneity the I-squared value was calculated. Sensitivity analysis Sensitivity analyses were performed by comparing the results obtained through both fixed and random effects meta-analyses. Furthermore, the exclusion of the largest trial in each sub-group, followed by recalculation of the meta-analyses, determined whether the results were being heavily determined by that trial.9 Results After exclusions, twenty-five articles (935 156 cases) provided information on the annual hospital volume
Stroke rate 0.84 [0.79e0.88], 72 CEA p.a. (Fig. 1A); Death rate 0.76 [0.74e0.81], 81 CEA p.a. (Fig. 1B); Combined stroke/death rate 0.73 [0.68e0.78], 84 CEA p.a. (Fig. 1C). Overall, stroke and death attributable to CEA occurred less frequently in higher-volume hospitals (0.78 [0.64e0.92]; Fig. 1D). The critical volume threshold between higher- and lower-volume hospitals was 79 CEA per annum, defined by weighting the threshold values stated in the included articles. Four articles (50 122 cases) were presented as a systematic review (Table 1). Two articles found significant improvements in outcomes with volume, and two did not. Study validity No significant heterogeneity was found in any of the sub-group analyses by either the Q-statistic or I-squared method. The funnel plots showed good symmetry indicating insignificant publication bias (Fig. 2aec). Exclusion of the largest study in each sub-group did not change the pooled effect estimate for each sub-group. This indicated that the results were not influenced by a single large trial, supporting the validity of these results. In comparing the pooled effect estimates from the three sub-group analyses, the results in terms of effect and volume threshold were consistent between the mortality, and stroke-mortality sub-groups. The stroke-only sub-group had a more conservative pooled effect estimate at a higher volume threshold compared to the other two sub-groups. Discussion The articles utilised in this review were largely based on retrospective, administrative data from the USA, and demonstrated significant relationships between the annual hospital volume and outcomes of CEA in terms of stroke and mortality. For these data, all-cause post-operative mortality was accurately quantified through discharge Eur J Vasc Endovasc Surg Vol 33, June 2007
648
P. J. E. Holt et al.
Table for Fig. 1: Sub-sections AeD Study
Threshold
Odds Ratio
95% Cases Confidence Interval
Study Weight
Birkmeyer18 Perler19 Edwards20 Wennberg6 Karp21 Roddy22 Dudley23 Dardik24 Harthun25 Segal26 Cowan27 Manheim28 Hannan29 Kempczinski13 Khuri30 Sub-total
164 50 50 21 50 50 100 50 100 30 100 100 100 50 29 81
0.88 1.14 0.80 0.85 0.64 0.54 0.78 0.85 0.40 0.93 0.67 0.66 0.73 1.22 1.06 0.77
0.80 0.79 0.61 0.77 0.45 0.30 0.59 0.30 0.24 0.59 0.51 e e e e 0.74
0.96 1.64 1.04 0.94 0.92 0.96 1.04 2.42 0.67 1.48 0.87 e e e e 0.81
479289 9918 11199 106790 10569 10211 28207 9842 14095 5657 50122 106493 28207 750 10173 881522
22.2 4.2 8.2 20.9 7.5 5.4 7.9 1.7 8.2 4.0 9.9 e e e e 100
Perler19 Edwards20 Mayo31 Roddy22 Dardik24 Harthun25 Cowan27 Kempczinski13 Khuri30 Sub-total
50 50 28 50 50 100 100 50 29 72
1.18 0.91 0.69 0.78 0.65 0.70 0.85 0.93 0.69 0.84
0.86 0.72 0.14 0.51 0.37 0.50 0.72 e e 0.79
1.61 1.13 3.40 1.20 1.14 0.97 1.02 e e 0.88
9918 11199 362 10211 9842 14095 5012 750 10173 71562
10.8 19.8 2.5 11.8 10.6 17.3 27.1 e e 100
Perler19 Edwards20 Karp21 Cebul14 Roddy22 Feasby32 Harthun25 Cowan27 Peck33 Sub-total
50 50 50 62 50 150 100 100 90 84
1.17 0.93 0.61 0.29 0.68 0.74 0.59 0.79 1.24 0.73
0.92 0.77 0.36 0.12 0.48 0.60 0.45 0.68 0.35 0.68
1.48 1.11 1.04 0.69 0.95 0.91 0.78 0.92 4.41 0.78
918 11199 10569 678 10211 14268 14095 5012 560 67510
8.6 14.2 7.1 8.5 10.3 15.5 14.6 20.1 1.2 100
79
0.78
0.64
0.92
885034
Overall total
Fig. 1. Forest plot demonstrating the volume-outcome relationship for CEA by outcome: (1A) post-operative death; (1B) post-operative stroke; (1C) combined stroke and death. Subtotals for each outcome, and an overall pooled effects estimate; (1D) are presented as odds ratios and 95% confidence intervals. Values to the left of the line of equivocal effect favour surgery at higher volume hospitals. Studies whose odds ratio is denoted by have been included on the plots to demonstrate their odds ratio only, but as no 95% confidence intervals were calculable, they were not used to calculate the pooled effect estimate.
Eur J Vasc Endovasc Surg Vol 33, June 2007
100
abstracts.11 Only two outcomes were investigated in this meta-analysis (stroke and death), therefore it is highly likely that the total complication rate was higher than presented here. Furthermore, as these studies included symptomatic and asymptomatic patients, the overall stroke and death rates were lower that expected, especially when compared to European data that are based on performing the majority of procedures on symptomatic patients.12 A further problem identified was in the coding of post-operative stroke, as patients may have been admitted to hospital due to cerebrovascular disease, or have suffered a stroke post-operatively. In the process of generating the data, either of these may be miscoded. There was significant variation in the way in which ‘‘stroke’’ was defined in the different articles ranging from a focal neurological deficit present for two or
Volume-outcome Relationships of Carotid Endarterectomy
649
Table 1. Summary of four studies for which the data were unsuitable for meta-analysis Author
Case Numbers
Threshold Value
Volume Effect Seen?
Statistically Significant
Summary
Middleton15 Pearce16
689 45744
10 expressed as change in risk with doubling of volume
N Y
N Y
Kantonen34 Fisher35
1600 2089
? 40
N Y
N Y
No difference in mortality or stroke rates. Median volume ¼ 102 CEA pa. Hospital volume, physician volume and specialty significant factors. Significant relationship for stroke, MI and death rates. No difference in mortality or stroke rates. Higher volume hospitals had lower stroke and mortality rates.
more consecutive days,13 to a deficit unresolved by discharge14 or fatal stroke.15 A number of articles specified the use of ICD-9 code (997.x) to determine the rate of post-operative stroke, with no further
Fig. 2. Funnel Plots assessing publication bias. None is demonstrated in any sub-group analysis. (2A) post-operative death; (2B) post-operative stroke; (2C) combined stroke and death.
assessment of severity, or duration. One study stated that in the presence of a code relating to stroke, it was ‘‘assumed to be post-operative’’.16 The difficulty of interpreting ‘post-operative stroke’ was supported in the sensitivity analyses, where the pooled effect estimate for this sub-group was of a lesser extent than the other sub-groups. The heterogeneous way that ‘post-operative stroke’ was interpreted suggested that the most valid measure for CEA outcome is post-operative death. Fig. 1a and b did show similarities in the impact of volume on outcome. Therefore, post-operative death may be a valid outcome surrogate for the hospital post-operative stroke-rate. The impact on the results of these confounding factors was impossible to quantify, and these differences will be difficult to rectify in the future without increasing clinician involvement in the coding process in parallel with the continued development of diagnostic coding systems, such as ICD-10. The results demonstrated that CEA performed at higher-volume hospitals had improved outcome quantified through significantly lower stroke and death rates. The key factors underlying this complex relationship related to hospital infrastructure (cardiology, specialist anaesthetists and neurology support onsite), the provision of suitable intensive care facilities where necessary, and concomitant provision of highvolume surgeons with vascular sub-specialisation.17 For CEA, lower-volume surgeons achieved results similar to higher-volume surgeons when operating in higher-volume hospitals. This supported the concept that hospital infrastructure was as a major component of the volume-outcome relationship,14 and acted independently to a relationship between surgeon operative volume and outcome. Death and stroke rate for CEA are lowest for high-volume vascular specialist surgeons, operating in high-volume hospitals.14 It has been suggested that differences in case-mix between higher- and lower-volume hospitals may underlie the volume-outcome relationship observed, Eur J Vasc Endovasc Surg Vol 33, June 2007
650
P. J. E. Holt et al.
with higher-volume hospitals taking on a larger number of asymptomatic cases, and lower-risk patients. However, this was not supported by articles in this study, with significant crude results remaining significant after case-mix adjustment, whether those results showed a volume-outcome relationship or not.18 The studies utilised in this analysis varied in the way the data were presented, with some presenting an in-hospital death or stroke rate, and others a combined stroke/death rate. In order to more easily assess the data for CEA, it would be advisable that all future studies make an assessment of both primary outcome measures for this procedure, and present a combined stroke/death rate. Further risk adjustment would be made possible through the routine inclusion of physiological parameters, including a pre-operative neurological scoring system and functional assessment, whether the patient was symptomatic or asymptomatic, and the degree of carotid artery stenosis. The evidence presented here demonstrated that the relationship between hospital volume and outcome was consistent and reproducible. We suggest that all healthcare systems should establish volume criteria for CEA. The volume threshold will need individual assessment in each system, as different policies regarding the proportion of asymptomatic carotid artery stenoses operated will have an effect on the threshold.
Conclusion CEA is the gold standard therapy in the prevention of neurological symptoms secondary to carotid artery stenosis. This study demonstrated that adverse outcome from CEA was reduced as the annual hospital volume of surgery increased. Volume criteria should be established in every healthcare system for CEA, to reduce the incidence of these events.
References 1 Group ECSTC. Randomised trial of endarterectomy for recently symptomatic carotid stenosis: final results of the MRC European Carotid Surgery Trial (ECST). Lancet 1998 May 9;351(9113):1379e 1387. 2 Study ECftACA. Endarterectomy for asymptomatic carotid artery stenosis. JAMA 1995 May 10;273(18):1421e1428. 3 FERGUSON GG, ELIASZIW M, BARR HW, CLAGETT GP, BARNES RW, WALLACE MC et al. The North American Symptomatic Carotid Endarterectomy Trial: surgical results in 1415 patients. Stroke 1999 Sep;30(9):1751e1758. 4 ROTHWELL PM. ACST: which subgroups will benefit most from carotid endarterectomy? Lancet 2004 Sep 25eOct 1;364(9440): 1122e1123 [author reply 5e6]. 5 ROTHWELL PM, ELIASZIW M, GUTNIKOV SA, WARLOW CP, BARNETT HJ. Endarterectomy for symptomatic carotid stenosis
Eur J Vasc Endovasc Surg Vol 33, June 2007
6
7 8 9 10 11
12 13 14 15
16
17
18 19
20
21
22 23 24
25
26
in relation to clinical subgroups and timing of surgery. Lancet 2004 Mar 20;363(9413):915e924. WENNBERG DE, LUCAS FL, BIRKMEYER JD, BREDENBERG CE, FISHER ES. Variation in carotid endarterectomy mortality in the Medicare population: trial hospitals, volume, and patient characteristics. JAMA 1998 Apr 22e29;279(16):1278e1281. Collaboration C. Revman. [cited 2006 20th December 2006]; Available from: http://www.cc-ims.net/RevMan/ivnotes.htm. LEWIS S, CLARKE M. Forest plots: trying to see the wood and the trees. BMJ 2001 Jun 16;322(7300):1479e1480. Collaboration C. openlearning. [cited 2006 20th December]; Available from: www.cochrane-net.org/openlearning. HARDY RJ, THOMPSON SG. Detecting and describing heterogeneity in meta-analysis. Stat Med 1998 Apr 30;17(8):841e856. WEINGART SN, IEZZONI LI, DAVIS RB, PALMER RH, CAHALANE M, HAMEL MB et al. Use of administrative data to find substandard care: validation of the complications screening program. Med Care 2000 Aug;38(8):796e806. PANAYIOTOPOULOS YP, PADAYACHEE S, COLCHESTER AC, TAYLOR PR. The changing indications for carotid endarterectomy in the UK. Int J Clin Pract 1997 Sep;51(6):375e377. KEMPCZINSKI RF, BROTT TG, LABUTTA RJ. The influence of surgical specialty and caseload on the results of carotid endarterectomy. J Vasc Surg 1986 Jun;3(6):911e916. CEBUL RD, SNOW RJ, PINE R, HERTZER NR, NORRIS DG. Indications, outcomes, and provider volumes for carotid endarterectomy. JAMA 1998 Apr 22e29;279(16):1282e1287. MIDDLETON S, DONNELLY N. Outcomes of carotid endarterectomy: how does the Australian state of New South Wales compare with international benchmarks? J Vasc Surg 2002 Jul;36(1):62e69. PEARCE WH, PARKER MA, FEINGLASS J, UJIKI M, MANHEIM LM. The importance of surgeon volume and training in outcomes for vascular surgical procedures. J Vasc Surg 1999 May;29(5):768e776. [discussion 77e78]. HOLT PJE, POLONIECKI JD, LOFTUS IM, GERRARD D, THOMPSON MM. Meta-analysis and systematic review of the relationship between volume and outcome in abdominal aortic aneurysm surgery. Br J Surg 2007;94(4):395e403. BIRKMEYER JD, SIEWERS AE, FINLAYSON EV, STUKEL TA, LUCAS FL, BATISTA I et al. Hospital volume and surgical mortality in the United States. N Engl J Med 2002 Apr 11;346(15):1128e1137. PERLER BA, DARDIK A, BURLEYSON GP, GORDON TA, WILLIAMS GM. Influence of age and hospital volume on the results of carotid endarterectomy: a statewide analysis of 9918 cases. J Vasc Surg 1998 Jan;27(1):25e31 [discussion 3]. EDWARDS WH, MORRIS Jr JA, JENKINS JM, BASS SM, MACKENZIE EJ. Evaluating quality, cost-effective health care. Vascular database predicated on hospital discharge abstracts. Ann Surg 1991 May; 213(5):433e438 [discussion 8e9]. KARP HR, FLANDERS WD, SHIPP CC, TAYLOR B, MARTIN D. Carotid endarterectomy among Medicare beneficiaries: a statewide evaluation of appropriateness and outcome. Stroke 1998 Jan;29(1): 46e52. RODDY SP, O’DONNELL Jr TF, WILSON AL, ESTES JM, MACKEY WC. The Balanced Budget Act: potential implications for the practice of vascular surgery. J Vasc Surg 2000 Feb;31(2):227e236. DUDLEY RA, JOHANSEN KL, BRAND R, RENNIE DJ, MILSTEIN A. Selective referral to high-volume hospitals: estimating potentially avoidable deaths. JAMA 2000 Mar 1;283(9):1159e1166. DARDIK A, BOWMAN HM, GORDON TA, HSIEH G, PERLER BA. Impact of race on the outcome of carotid endarterectomy: a population-based analysis of 9,842 recent elective procedures. Ann Surg 2000 Nov;232(5):704e709. HARTHUN NL, KONGABLE GL, BAGLIONI AJ, MEAKEM TD, KRON IL. Examination of sex as an independent risk factor for adverse events after carotid endarterectomy. J Vasc Surg 2005 Feb;41(2): 223e230. SEGAL HE, RUMMEL L, WU B. The utility of PRO data on surgical volume: the example of carotid endarterectomy. QRB Qual Rev Bull 1993 May;19(5):152e157.
Volume-outcome Relationships of Carotid Endarterectomy
27 COWAN Jr JA, DIMICK JB, THOMPSON BG, STANLEY JC, UPCHURCH Jr GR. Surgeon volume as an indicator of outcomes after carotid endarterectomy: an effect independent of specialty practice and hospital volume. J Am Coll Surg 2002 Dec;195(6):814e821. 28 MANHEIM LM, SOHN MW, FEINGLASS J, UJIKI M, PARKER MA, PEARCE WH. Hospital vascular surgery volume and procedure mortality rates in California, 1982e1994. J Vasc Surg 1998 Jul; 28(1):45e56 [discussion 8]. 29 HANNAN EL, POPP AJ, TRANMER B, FUESTEL P, WALDMAN J, SHAH D. Relationship between provider volume and mortality for carotid endarterectomies in New York state. Stroke 1998 Nov;29(11): 2292e2297. 30 KHURI SF, DALEY J, HENDERSON W, HUR K, HOSSAIN M, SOYBEL D et al. Relation of surgical volume to outcome in eight common operations: results from the VA National Surgical Quality Improvement Program. Ann Surg 1999 Sep;230(3):414e429 [discussion 29e32]. 31 MAYO SW, ELDRUP-JORGENSEN J, LUCAS FL, WENNBERG DE, BREDENBERG CE. Carotid endarterectomy after NASCET and ACAS: a statewide study. North American Symptomatic Carotid
32 33 34
35
651
Endarterectomy Trial. Asymptomatic Carotid Artery Stenosis Study. J Vasc Surg 1998 Jun;27(6):1017e1022 [discussion 22e3]. FEASBY TE, QUAN H, GHALI WA. Hospital and surgeon determinants of carotid endarterectomy outcomes. Arch Neurol 2002 Dec;59(12):1877e1881. PECK C, PECK J, PECK A. Comparison of carotid endarterectomy at high- and low-volume hospitals. Am J Surg 2001 May;181(5): 450e453. KANTONEN I, LEPANTALO M, SALENIUS JP, MATZKE S, LUTHER M, YLONEN K. Influence of surgical experience on the results of carotid surgery. The Finnvasc Study Group. Eur J Vasc Endovasc Surg 1998 Feb;15(2):155e160. FISHER ES, MALENKA DJ, SOLOMON NA, BUBOLZ TA, WHALEY FS, WENNBERG JE. Risk of carotid endarterectomy in the elderly. Am J Public Health 1989 Dec;79(12):1617e1620.
Accepted 21 January 2007 Available online 30 March 2007
Eur J Vasc Endovasc Surg Vol 33, June 2007