J P Pell, R Slack, M Dennis*, G Welch+
Greater Glasgow NHS Board, *Western General Hospital Edinburgh +Southern General Hospital Glasgow
e-mail: jill.pell@gghb.scot.nhs.uk
SMJ 2003 49(1): 53-56
Abstract
Background and aims: The effectiveness of carotid endarterectomy (CEA) depends upon selection of patients with a high absolute risk of stroke, avoidance of delays from initial presentation to surgery and provision of surgery with a low perioperative risk of stroke. We evaluated these aspects of practice in Scottish hospitals and, by encouraging adherence to national (SIGN) guidelines, attempted to improve the effectiveness of CEA. Methods: Prospective independent survey of patient selection, pre-operative delays and surgical performance for all CEAs performed in Scottish National Health Service hospitals over two 13 month periods from September 1997 and February 1999 respectively. Results: Thirteen hospitals performed 485 CEAs in the first period and 392 in the second, equating to an overall annual rate of 79 per million population. During both periods at least 95% of patients reported symptoms of carotid territory ischaemia, but the proportion with stenoses greater than > 70% rose from 89% to 97% (p<0.0001). The delays between referral, surgical consultation and subsequent surgery fell significantly but remained unacceptably long. The proportion seen by the surgeon within two weeks of referral rose from 36% to 43% (p=0.05) and the proportion operated on within one month thereafter rose from 35% to 49% (p<0.0001). The perioperative combined major stroke and death rate was 3% in both periods.
Conclusion: We demonstrated significant changes in practice, in line with national guidelines, which would be expected to improve the effectiveness of our national programme of CEA. There is further scope for improving performance, particularly in relation to pre-operative delays.
Key words (MeSH terms) : Endarterectomy, carotid; Surgical procedure, vascular; Audit, medical
Introduction
In a National Health Service (NHS) with limited resources, including limited capacity for CEA, it is important to ensure that the maximum benefit is accrued from the resources allocated. The results of two large randomised trials suggest that performing CEA on nine symptomatic patients with carotid stenoses greater than 70% should prevent one stroke over the subsequent three years, provided the operations are performed soon after the onset of symptoms and with a peri-operative combined major stroke and death rate less than 7%.1-3 Among patients with stenoses of greater than 70% who have not had recent symptoms of carotid territory ischaemia around 50 operations may require to be performed for a similar net benefit.4 Application of additional clinical and angiographic selection criteria may reduce further the number of operations needed to prevent one stroke in both symptomatic and asymptomatic patients.5 The Scottish Intercollegiate Guidelines Network (SIGN) published evidence-based guidelines aimed at helping clinicians identify those patients who could gain most from CEA, maximising the benefits of surgery in these patients by reducing delays, and therefore pre-operative strokes, and encouraging surgical practices which minimise perioperative complications (www.sign.ac.uk).6 We undertook a survey of CEA services in Scottish NHShospitals, in order to monitor and encourage adherence to the e guidelines and change practice where required. The survey was overseen by the Scottish Vascular Audit Group which included all vascular surgeons in Scotland as well as representatives from interventional radiology, stroke medicine and public health.
Methods
Four dedicated audit nurses collected data prospectively on all patients who had a CEA in a Scottish NHS hospital between 1 September 1997 and 30 September 1998 to ascertain baseline performance. Following completion of the first survey we attempted to improve practice by disseminating individual and aggregated survey results and reminders of the principal SIGN recommendations (Figure 1) to participating surgeons, trust managers and directors of public health. Guidelines on the use and reporting of investigations were also disseminated to all vascular radiologists. The same four audit nurses completed a second survey of all patients having CEA between 1 February 1999 and 29 February 2000. In accordance with the SIGN recommendations, the audit nurses were not involved in the clinical management of the patients. They extracted data from the clinical records during the hospital admission for CEA, including dates and sources of referral, symptoms, dates of surgical consultation and surgery, preoperative assessments and investigations and perioperative care. Where patients underwent both noninvasive imaging and angiography, the result of the latter was used to grade their carotid stenosis. A perioperative major stroke was defined as a stroke occurring during the index admission either during or after surgery, which resulted in disabling symptoms which were still present at the time of discharge. A minor stroke was defined as one not associated with disabling symptoms at discharge. The nurses did not personally assess patients to determine whether a perioperative stroke had occurred. Missing data were sought directly from the attending surgeon. The data were collated and analysed centrally in Greater Glasgow Health Board. The statistical significance of differences between the two periods was tested using ÷2, ÷2 for trend and Mann Whitney U (MWU) tests for categorical, ordinal and continuous data respectively.
Results
CEA was performed in 13 of the 33 acute NHS hospitals in Scotland. Four hundred and eighty five CEAs were performed in the first period and 392 in the second; equivalent to an overall annual rate of 79 operations per million population per year. The number of operations performed per hospital ranged from 4 to 99 in the first period and from 3 to 76 in the second. In both periods only three hospitals performed at least 50 operations. The number of surgeons performing CEA fell from 30 in the first period to 29 in the second. The number of operations per surgeon ranged from 1 to 49 in the first period and from 1 to 43 in the second. In the first period, 20(67%) surgeons performed at least 10 operations compared with 17(59%) in the second.
Patient selection
The demographic characteristics of patients did not differ significantly between periods (Table I). Their overall median age was 68 years (IQR 60-73) and 510(58%) patients were male. At least 95% of patients in each period reported relevant focal symptoms (Table I). Very few patients had surgery for asymptomatic stenoses in either period. Modifiable risk factors and comorbid conditions were common (Table II). Between periods there was an increase in non-smokers and decrease in current smokers (x2 trend, p<0.01). Knowledge of cholesterol status increased from 67% to 84% (p<0.0001). In both periods two-thirds of those with known hypercholesterolaemia were taking lipidlowering drugs. The proportion of patients operated on with a stenosis of greater than 70% increased from 89% in the first period to 97% in the second (p<0.0001)(Table III).
Referrals, delays and pr pre- e- operative investigations
One hundred and seventy (19%) patients were referred directly by their general practitioner to the operating surgeon. Among the remainder, the most common sources of referral were general physicians (25%), neurologists (11%) and stroke physicians (10%). Overall, 155(18%) patients were assessed pre-operatively by a neurologist. The median waiting time from referral to first consultation was 21 days (IQR 8-33) in the first period and 18 days (IQR 7-31) in the second (MWU, p=0.06) (Table IV). In the first period 167(36%) were seen within two weeks compared with 163(43%) in the second (x2 p=0.05) (Table IV). Waiting times varied by referral source. Among referrals from secondary care the percentage seen within two weeks rose significantly (p<0.01) whereas there was a non-significant fall in the proportion seen within two weeks among patients referred by their general practitioners.
Overall, the number of patients operated on within one month of their first consultation with the surgeon increased from 169(35%) to 189(49%) (p<0.0001) (Table IV). The median time delays in the first and second periods were 42 days (IQR 20-76) and 30 days (IQR 13-55) respectively (MWU, p<0.0001). In the first period the waiting time was significantly greater for general practitioner referrals than secondary care referrals (p<0.01). However, in both groups, there were significant improvements between the first and second period (p<0.05 and p<0.01 respectively).
Pre-operative duplex scanning was performed on 471 (97%) patients in the first period and 387(99%) in the second (p>0.05). The number of patients having preoperative conventional angiography fell from 181 (37%) to 93(24%) (p<0.0001). Thirty two (7%) patients in the first period and 37(9%) in the second underwent magnetic resonance (MR) or computerised tomography (CT) angiography (p>0.05). In the first period the degree of stenosis could not be categorised in 34(7%) patients. In 22(5%) cases this was due to the stenosis being recorded as a range which spanned a critical value, such as 50%-80% (Table III). The number of such patients fell to only 5(1%) in the second period (p<0.01).
Surgical and anaesthetic practice
The number of patients who received solely general anaesthesia fell from 430(89%) to 302(79%). The number of operations performed solely under local anaesthesia remained low at 14(3%) and 10(3%) respectively, while the numbers in which a combination was used increased from 40(8%) to 71(19%) (÷2 trend, p<0.0001). Overall, 497(57%) patients had shunts inserted and 700(80%) received patches with no significant statistical differences between the periods. In the first period, 73(15%) patients had some form of perioperative monitoring compared with 81(21%) in the second (p<0.05).
Outcomes
In-hospital events were infrequent and there were no significant changes between periods. Overall, 17(2%) patients required postoperative ventilation and 28(3%) reoperation. Sixty five (7%) developed new hypertension, eight (1%) a new cardiac arrhythmia and eight (1%) suffered myocardial infarctions. Five (1%) patients developed a wound infection and 21(2%) required drainage of a haematoma. While in hospital, 22(3%) patients suffered major strokes, six (1%) minor strokes and 14(2%) transient ischaemic attacks. Ten deaths occurred, seven of which were attributed to major stroke. The combined major stroke and death rate was 3%. The median post-operative length of stay was 3 days (IQR 2-4) in both periods.
Discussion
We are not aware of any previous attempt to prospectively survey and alter the practice of CEA on a national basis. Countries such as the USA7 and Canada8 report annual CEA rates per head of population in excess of 1000 and 350 per million respectively. Our much lower rate is less likely to reflect a lower incidence of carotid stenosis than a igher threshold for surgery due, in part, to fewer resources. Relatively few patients in Scotland had CEA for asymptomatic carotid stenosis or for mild or moderate symptomatic stenoses. The SIGN guidelines recommend that asymptomatic patients should be recruited to ongoing trials.6 In our study, 22 (3%) operations were performed on asymptomatic patients, but only three were undertaken within the context of a randomised trial.
Routine hospital activity data suggested that the number of CEA operations was increasing steadily in Scotland prior to our study ([M Bain], unpublished data, 2000). However, within our study, the number of operations performed was lower during the second period than the first. This may, in part, have been due to the recruitment of some patients to the non-interventional arms of trials. However, it can also be attributed, in part, to a reduction in operations on patients with only moderate stenoses, in whom CEA is not indicated.. .. . We were not able to access data on CEA operations performed in private hospitals. However, in Scotland, the vast majority of operations are performed in NHS hospitals and this proportion is unlikely to have changed over the course of this study.
Our data suggest that by issuing guidelines we were able to improve the selection process so that the limited number of operations were focused on individuals at highest risk of stroke. The long delays between initial referral and surgery will inevitably lead to some patients having strokes during this period which may be fatal or may result in the patient becoming unfit for surgery, thereby reducing the effectiveness of our CEA service at a population level. Since recruitment to our survey was restricted to patients who actually underwent surgery we were unable to identify such patients. Because the risk of stroke in untreated patients falls with time, the net benefit from surgery in those who survive to undergo surgery also decreases as waiting times increase.2 In the published trials patients received surgery well within six months of recruitment whereas in our study one-third of patients waited more than six months from the onset of symptoms to surgery. Therefore, although our survey demonstrated reductions in pre-operative delays, further improvements are required. Specialised neurovascular outpatient clinics, which offer an early specialist assessment, same day carotid Duplex and close links with vascular surgeons, are being set up in many hospitals and will hopefully improve the service further. In our study, patients underwent a variety of pre-operative investigations including plain, digital subtraction and MR angiography, and Duplex scanning. The SIGN guideline recommends Duplex scans for all patients considered for CEA followed by conventional angiography in those in whom a severe stenosis is detected. In our survey, the majority of patients received only Duplex, or other non-invasive, imaging. Conventional angiography was not used in all patients in whom non-invasive investigations suggested a stenosis of more than 70% and the proportion undergoing angiography fell between rounds. Published trials were based on contrast angiography and the results may, therefore, not be totally applicable to current practice.1-3
No randomised trials comparing local and general anaesthesia have been completed, although the GALA multicentre trial is in progress (www.dcn.ed.ac.uk/gala). A review of 17 non-randomised trials suggests that, in comparison with general anaesthesia, local anaesthesia may halve the risk of perioperative stroke, myocardial infarction and death and reduce hospital stay.9 In our study, the percentage of patients operated on using a combination of general and local anaesthesia increased from 8% to 19% but the proportion in whom local anaesthesia was used in isolation remained static at 3%. In our study the perioperative combined major stroke and death rate was 3% in both periods. This is the same as in a previous UK audit,10 and well within the figures reported by trials.1-3 Therefore, net benefits similar to those demonstrated in the trials might be expected. However, our study collected only in-hospital complications rather than 30 day event rates. The median post-operative length of stay in our study was three days. Also, although the data were collected by an independent review of clinical records, the information contained in these was recorded predominantly by the operating surgeon which may result in an under-recording of adverse events.11
A number of studies have suggested a possible inverse association between the volume of throughput for individual surgeons and surgical complications.12-16 Between five and 10 operations per annum have been suggested as minimum acceptable numbers.12,14-16 A number of studies have also demonstrated an association with hospital volumes.15,17-19 The published SIGN guideline recommends a minimum of 50 operations per annum in hospitals,6 primarily because higher volumes facilitate the organisation of services, including provision of specialist neurological, radiological and anaesthetic input, and more precise estimation of complication rates. However the potential benefits of increasing volumes by reducing the numbers of hospitals performing CEA need to be balanced against poorer patient access, particularly in more rural areas where patients already have to travel considerable distances.
Ideally, future surveys should track patients prospectively from initial diagnosis through to the postoperative period and include independent validation of the accuracy of diagnosis and completeness of identification of complications. Despite the methodological limitations of our survey, we have demonstrated significant changes in practice in line with national guidelines which would be expected to improve the effectiveness of our national programme of CEA. There is further scope to improve performance, particularly in relation to pre-operative delays. However, given our lower rate of CEA compared with some other countries, further reductions in delays may be difficult to achieve within existing levels of funding.
ACKNOWLEDGEMENTS: The authors are grateful to the research nurses who collected the data: A Finnie, M Imrie, J Seaton and J Thorburn, and to the other vascular surgeons who participated: S Boom, I McMillan and G Stewart (Ayr), R Brookes and R Scott (Monklands), *A Bradbury, R Chalmers, A Jenkins, J Murie, *CV Ruckley and J Sathianathan (Edinburgh Royal Infirmary), D Byrne, AJ MacKay, *R Quin and P Rogers (Gartnavel General), M Calvert and P Teenan (Stobhill), H Campbell and D Reid (Law), *G Cooper, S Cross and J Engeset (Aberdeen Royal Infirmary), D Gilmour, P Leiberman and J Pollock (Glasgow Royal Infirmary), G Griffiths, P McCollum, S Naidu, *P Stonebridge and T Varma (Ninewells), R Holdsworth (Stirling Royal Infirmary), J McCormick (Dumfries & Galloway Royal Infirmary) and J Reidy (Inverclyde Royal Hospital). The authors are grateful to the Clinical Resources Audit Group of the Scottish Executive for providing the funding to finance this study.
*Members of Scottish Carotid Endarterectomy Audit Steering Group
REFERENCES
1 European Carotid Surgery Trialists’ Collaborative Group. MRC European Carotid Surgery Trial: interim results for symptomatic patients with severe (70-99%) or with mild (0-29%) carotid stenosis. Lancet 1991;337:1235-43.
2 European Carotid Surgery Trialists’ Collaborative Group. Randomised trial of endarterectomy for recently symptomatic carotid stenosis: final results of the MRC European Carotid Surgery Trial. Lancet 1998;351:1379-87.
3 North American Symptomatic Carotid Endarterectomy Trial Collaborators. Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. New Engl J Med 1991;325:445-53.
4 Hobson RW, Weiss DG, Fields WS, Goldstone J, Moore WS, Towne JB, Wright CB. Efficacy of carotid endarterectomy for asymptomatic carotid stenosis. The Veterans Affairs Cooperative Study Group.N Engl J Med 1993;328:221-7.
5 Rothwell PM, Slattery J, Warlow CP. Clinical and angiographic predictors of stroke and death from carotid endarterctomy: systematic review. Br Med J 1997;315:1571-7.
6 Scottish Intercollegiate Guideline Network. Management of patients with stroke. II. Management of carotid stenosis and carotid endarterectomy. Volume 14. Royal College of Physicians, Edinburgh: May 1997.
7 Morasch MD, Parker MA, Feinglass J, Manheim LM, Pearce WH. Carotid endarterectomy: characterization of recent increases in procedure rates.J Vasc Surg 2000;31:901-9.
8 Simunovic M, To T, Johnston KW, Naylor CD. Trends and variations in the use of vascular surgery in Ontario. Can J Cardiol 1996 12;249-53.
9 Tangkanakul C, Counsell CE, Warlow CP. Local versus general anaesthesia in carotid endarterectomy: a systematic review of the evidence. Eur Vasc Endovasc Surg 1997;13:491-9.
10 McCollum PT, da Silva A, Ridler BD, de Cossart L. Carotid endarterectomy n the UK and Ireland: audit of 30 day outcome. The Audit Committee for the Vascular Surgical Society. Eur J Vasc Endo Soc 1997;14:386-91.
11 Rothwell P, Warlow C. Is self-audit reliable. Lancet 1995;346:1623.
12 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 Surg 1998;15:155-60.
13 Pearce WH, Parker MA, Feinglass J, Ujiki M, Mannheim LM. The importance of surgeon volume and training in outcomes for vascular surgical procedures. J Vasc Surg 1999;29:768-76.
14 Kucey DS, Bowyer B, Iron K, Austin P, Anderson G, Tu JV. Determinants of outcome after carotid endarterectomy. J Vasc Surg 1998;28:1051-8.
15 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;29:2292-7.
16 Ruby ST, Robinson D, Lynch JT, Mark H. Outcome analysis of carotid endarterectomy in Connecticut: the impact of volume and specialty. Ann Vasc Surg 1996;10:22-6.
17 Mannheim LM, Sohn MW, Feinglass J, Ujiki M, Parker MA, Pearce WH. Hospital vascular surgery volume and procedure mortality rates in California, 1982-1994. J Vasc Surg 1998;28:45-56.
18 Cebul RD, Snow RJ, Pine R, Hertzer NR, Norris DG. Indications, outcomes and provider volumes for carotid endarterectomy. J Am Med Assoc 1998;279:1282-7.
19 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;27:25-31.