| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Evidence-based Practice |
1 From the Division of Abdominal Imaging and Intervention, Massachusetts General Hospital, Boston (M.M.M.); Department of Radiology, Beth Israel Deaconess Medical Center, Boston, Mass (A.M.M.); Department of Radiology, University of Chicago, Ill (P.M.M.); and Departments of Radiology (D.E.M.) and Vascular Surgery (S.J.S.), St Vincent’s University Hospital, Elm Park, Dublin 4, Ireland. Received January 21, 2002; revision requested March 19; final revision received March 4, 2003; accepted March 18. Address correspondence to (e-mail: d.malone@st-vincents.ie).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
MATERIALS AND METHODS: A focused clinical question formed the basis of a literature search. Evidence-based criteria were used to appraise and assign a "level of evidence" to retrieved articles. The following data were determined from the best studies: systemic, local, and/or vascular complications; graft failure rates; blood loss; mortality; length of intensive care and/or hospital stay; mid- and long-term outcomes; cost of endovascular repair versus that of surgery; and eligibility for endovascular repair. Absolute risk reductions and/or increases and numbers needed to treat or harm were calculated.
RESULTS: The best current evidence came from 22 studies, which showed that there is slight, if any, difference between mortality rates of endovascular repair and surgery. Hospital and/or intensive care stay is shorter, blood loss less, and systemic complications fewer (numbers needed to treat, two to 12) with endovascular repair. Some authors reported a significant increase in local and/or vascular complications with endovascular repair (numbers needed to harm, two to six). Graft failure is significantly more common with endovascular repair (numbers needed to harm, four), and substantive adjunctive interventions are needed. Endovascular repair is more expensive than surgery.
CONCLUSION: Elective endovascular repair has short-term benefits compared with surgery. There is slight, if any, difference in mortality. Endovascular repair costs more than surgery. At follow-up, surgical grafts performed better.
© RSNA, 2003
Index terms: Aneurysm, abdominal • Aneurysm, surgery, 89.1267 • Aorta, grafts and prostheses, 89.1267 • Aorta, interventional procedures, 89.1267 • Technology assessment
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
The purpose of the present study was to use evidence-based techniques to compare elective surgery with endovascular aneurysm repair by means of stent placement.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Focused Clinical Question
We used the PICO ("patient, intervention, comparison intervention, and outcomes of interest") format of evidence-based question formulation. The question would be written in text form as follows: "How do patients with abdominal aortic aneurysms treated with endovascular repair compare with those treated with surgery for morbidity, mortality, intensive care stay, hospital stay, graft failure, cost, and effect on a radiology department?"
Literature Search
The literature search was performed in four stages by three authors (M.M.M., A.M.N., D.E.M.). First, to address the focused clinical question, a computer-based search was performed to obtain the best available current evidence. The Figure shows how the focused question can be used to link concepts in a search strategy for comprehensive retrieval. Individual component terms, such as medical subject headings, title words, and author names can be linked by the Boolean operator "or" if they belong to the same column of concepts and by the Boolean operator "and" if they belong to different columns. The U.S. National Library of Medicine’s MEDLINE database was accessed by using Advanced PubMed, Grateful Med, and Knowledge Finder search engines on several occasions between January 2000 and December 2001. The search covered the period from January 1990 to December 2001. References were retrieved by using the following words and medical subject headings alone and in combination: aortic aneurysm, abdominal; aortic aneurysm; aortic aneurysm abdominal, surgery; abdominal aneurysm, vascular surgery procedures; aortic aneurysm, endovascular stent; endovascular stent; abdominal aorta stenting; abdominal aorta stent; abdominal aorta stenting; insertion abdominal aortic stent; abdominal aorta prosthesis; and aortic aneurysm abdominal prosthesis. The results of this search were reviewed by three authors (M.M.M., A.M.N., D.E.M.), who worked in a group setting.
|
|
Finally, in February 2003, during the last stage of the manuscript review process, publications in the Evidence-based Practice section of Radiology that addressed aspects of our question were included in our evaluation. A repeat computerized search of the MEDLINE database was performed, which was limited to randomized controlled trials published during 2002 in which endovascular aneurysm repair and surgery were compared. At this time, the Cochrane Library, which incorporates the Cochrane Database of Systematic Reviews, the Cochrane Central Register of Controlled Trials, the Database of Abstracts of Reviews of Effectiveness, the Cochrane Peripheral Vascular Diseases Collaborative Review, the Health Technology Assessment Database, and the U.K. National Health Service Economic Evaluation Database, was also searched for randomized controlled trials in which endovascular aneurysm repair and surgery were compared.
Evidence-based review methodology teaches that the findings of many original studies are documented but not published. Those with inconclusive or negative findings are more likely to be rejected for publication. A publication bias toward positive conclusions may result. It may therefore be important to trace unpublished work, as systematic reviewers do, to get a complete picture of work that has been done on a subject (5). This is termed the "gray" literature. (5). The gray literature (eg, theses and unpublished data) was not searched during the present review.
Critical Appraisal of Retrieved Literature
Levels of evidence and grades of recommendations.—Radiology reports in the literature vary in their degree of validity (6). The CEBM provides a table to grade the strength of the source publication type to assist in the search for the best evidence (4). The sections dealing with therapeutic benefit and harm are presented in Table 1. These grades range from systematic reviews of randomized controlled trials (optimal data) to expert opinion, unsupported by explicit clinical trials (relatively weak data). The CEBM table can be used to rank abstracts rapidly according to the quality of study design. This concept of a "hierarchy of evidence" is one of the key evidence-based concepts. It permits the reviewer to analyze in detail only the articles with methodology that is least subject to bias. The underlying assumption is that if a well-designed and a poorly designed study yield different conclusions, the conclusion from the study with the best design will be closest to the truth. It is important to stress that the practice of evidence-based medicine involves seeking the best current evidence to guide decisions while being aware of the degree of uncertainty surrounding that evidence. It is an evolution in the tools that are used to practice scientific medicine (7). It does not mean discarding of any evidence short of randomized controlled trials, provided that it is the best available evidence and its limitations are recognized. Evidence-based medicine should not be regarded as a justification for therapeutic nihilism because no randomized controlled trials or meta-analyses are available (8).
The retrieved abstracts were separated into two groups: those with the highest level of evidence retrieved (in this case, level 2b: individual cohort studies) and those assigned lower levels of evidence during abstract review. The latter were discarded. Level 2b publications were appraised critically in detail.
Detailed critical appraisal.—This consisted of two phases: (a) reading the original article carefully, while using explicit evidence-based criteria (9–15) to assess the Materials and Methods section for validity of evidence (freedom from bias introduced by weaknesses in study design), and (b) retrieving raw data from the Results section to use for evidence-based calculations of benefit and harm (16) (numerical or statistical strength of evidence).
Studies appraised.—Eleven cohort studies (level 2b evidence, Table 1) and one meta-analysis (level 2a–2b evidence) provided the best current evidence on clinical performance of endovascular aneurysm repair (Table 1), and another 10 studies addressed outcome, cost, and cost-effectiveness. The data were extracted from each study jointly by two authors to find the answers to the components of the focused clinical question.
Assessment of Evidence Validity from Materials and Methods
Standard evidence-based questions were used to assess each retrieved study for methodologic flaws that would potentially weaken its results and/or conclusions. The domains under consideration were therapeutic benefit and harm, review-type articles, and economic analysis (Appendix).
Therapeutic studies.—The following questions were used to evaluate study design in the nonrandomized unblinded therapeutic studies under consideration (9–12): Were the two groups (treatment and control) similar at the start of the study? Were they treated equally apart from the therapeutic intervention? Were outcome assessments applied equally to both groups? Were all clinically relevant outcomes reported? Was follow-up of the study population sufficiently long and complete?
Review articles.—The validity criteria for review articles were the following (13): Did the overview address a focused clinical question? Were the criteria used to select articles for inclusion appropriate? Is it unlikely that important relevant studies were missed? Was the validity of the included studies appraised? Were the assessments of the included studies reproducible? Were the results similar from study to study?
Economic analysis.—By using the CEBM tables of levels of evidence and grades of recommendations, retrieved articles were assigned a level of evidence where appropriate (4). The validity criteria (14,15) were the following: (a) Did the analysis provide a full economic comparison of health care strategies? (i) Was a broad enough viewpoint adopted? (ii) Were all relevant clinical strategies compared? (b) Were costs and outcomes properly measured and evaluated? (i) Was clinical effectiveness established? (ii) Were costs measured accurately? (iii) Were data on costs and outcomes appropriately integrated? (c) Was appropriate allowance made for uncertainties in analysis? (d) Are estimates of cost and outcome related to the baseline risk in the treatment population? (e) Do the results consider incremental costs and effects of each strategy? (i) How much does allowance for uncertainty change the results? (ii) Do incremental costs and effects differ between subgroups?
Assessment of Strength of Evidence from Results: Evidence-based Analysis
Raw data published in the Results sections of the validated articles were used to establish event rates for the surgical repair group (control event rate) and the endovascular repair group (experimental event rate). These data were subjected to evidence-based analysis by using a specially designed spreadsheet (16). The rates or risk of adverse events in the endovascular repair and surgery groups were calculated. The absolute risk reduction or increase was determined with 95% CI estimates. The number needed to treat or harm was also calculated where appropriate (16–19). Adverse events included mortality, local and/or vascular complications, medical or systemic complications, and endoleak. Short-term morbidity and complications were defined as events that occurred within 30 days of the procedure. Other factors that were used to compare endovascular repair and surgery included intraoperative blood loss (milliliters), length of stay in an intensive care setting (days), and length of hospital stay (days). The length of follow-up and survival and the percentage of patients who required additional procedures were also evaluated. Any study from which no usable raw data could be extracted was discarded.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
Table 2 shows the number of articles retrieved by means of an Advanced PubMed search with each major search term. By using the validity criteria described previously, the best current evidence on the clinical performance of endovascular repair versus surgery was found in 11 studies that achieved level 2 evidence (20–30). Nine level 2 studies were retrieved initially by using the Advanced PubMed search engine (20–24,26–29). Two additional level 2 studies were retrieved by using the related searches option (25,30). The numbers of articles retrieved by using the related searches tool for the 11 level 2 studies are shown in Table 3. The manual searches of the RSNA Index to Imaging Literature, Journal of Vascular Surgery, European Journal of Vascular and Endovascular Surgery, and the Annals of Surgery did not reveal any additional level 2 or stronger articles. One article, with "Systematic Review of Short-Term Results" in the title, was published in the Evidence-based Practice section of Radiology in 2002 (31). According to evidence-based criteria in the United Kingdom (32), that study was not truly a systematic review, since it did not address a focused clinical question, include a fully comprehensive literature search, or involve assessment of the validity of the included studies. It included heterogeneous studies and is difficult to classify according to CEBM criteria. A meta-analysis was performed (31); therefore, this article is referred to in our evaluation as a meta-analysis.
|
|
Eligibility rate among patients referred for endovascular stent placement was addressed in four articles (20,37–39).
Nine articles addressed varying aspects of the cost or cost-effectiveness of elective endovascular aneurysm repair versus conventional surgery and the economic factors associated with introducing an endovascular repair program (28,40–48).
Blood Loss, Intensive Care and Hospital Stay, and Endoleaks
In the comparison of endovascular repair and surgery, there was less blood loss with endovascular repair, and intensive care and hospital stays were shorter (Table 4). The reported incidence of endoleaks varied between 5.6% and 36% (Table 4). In the meta-analysis (31), the pooled estimate of early endoleak after endovascular repair was 20% (95% CI: 15%, 24%), there was significantly less blood loss with endovascular repair (P = .003), and patients who underwent endovascular repair had significantly shorter intensive care and hospital stays (P = .04 and P = .02, respectively).
|
|
2 test as appropriate. This yielded a P value of .03. By using a basic formula (50) that incorporates the expected event rate in the control group (0.04) and the odds ratio (0.55), we calculated a number needed to treat of 57 for endovascular repair from the data presented by Adriaensen et al (31). These results will be considered in detail in the Discussion. In the other seven studies from which data could be extracted (22,24,25,27–30), the 95% CIs cross zero. This indicates that either there was inadequate sample size or there is a true lack of difference in mortality.
Short-Term Morbidity
The reporting of short-term complications was homogeneous in the 11 retrieved articles. The homogeneity in the reporting of short-term complications is at least partially explained by the fact that most level 2 articles that deal with endovascular repair have been written by a small number of authors from leading institutions in the field, and these articles are published in three journals: Journal of Vascular Surgery, Annals of Surgery, and European Journal of Vascular and Endovascular Surgery. Complications were considered either local and/or vascular or systemic. Blood loss, hospital stay, and intensive care stay were also compared.
Local and/or vascular complications.—These results are shown in Table 5. There are 11 studies in all. No raw data were available from two studies (20,30). Evidence-based analysis of the raw data from six studies yielded 95% CIs that crossed zero, which indicated no difference or a result that was not statistically significant (21,24–26,28,39). The remaining three studies yielded 95% CIs that did not cross zero, which suggested statistically significant results (22,23,27). These three studies yielded numbers needed to harm of two to six. That is, the statistically significant data suggest that endovascular repair was associated with more local and/or vascular complications than was surgery and that two to six patients needed to be treated with endovascular aneurysm repair before a local and/or vascular complication that would not have been encountered with surgery was encountered.
The meta-analysis (31) results indicated that there was no significant difference in the likelihood of experiencing a local and/or vascular complication after endovascular repair or surgery. Differences in study selection between this evidence-based evaluation and the meta-analysis are explored in the Discussion.
Systemic complications.—These results are shown in Table 5. Of the 11 studies, no data were available in two (20,30). In five studies, the 95% CIs crossed zero, which indicated no difference or a result that was not statistically significant (22,24–26,29). In four studies, the 95% CIs did not cross zero, which suggested statistically significant results (21,23,27,28). These four studies, with statistically significant results, yielded numbers needed to treat from two to 12. That is, the statistically significant data suggest that endovascular aneurysm repair was associated with fewer systemic complications than was surgery, and two to 12 patients needed to be treated with endovascular repair before a systemic complication that would have been encountered with surgery was avoided. The meta-analysis (31) results also indicated that there was a significantly lower likelihood of experiencing a systemic and/or remote complication after endovascular repair (P < .001).
Mid- and Long-Term Outcomes
The data for mid- and long-term outcomes after endovascular repair are reported in Table 6 (20,25,27,29–31,34–36). May et al (22) performed a Kaplan-Meier analysis that suggested 3-year graft success probability was 71% (95% CI: 58%, 81%). The incidence of graft failure was higher with endovascular repair than with surgery (absolute risk increase, 0.23; 95% CI: 0.15, 0.31; number needed to harm, four) (22). The necessity for secondary surgical or endovascular procedures in the endovascular aneurysm repair group ranged from 7.5% to 23% (Table 6). The cumulative 2-year survival for endovascular repair is comparable to that of surgery (29,30). Moore et al (29) reported a 5-year survival rate of 65%, which is comparable to a 72% cumulative survival rate for surgery. Although endovascular graft repair is less invasive than surgery and sometimes effective in the long term, it is often not a definitive procedure. The pattern of complications emerging from reported long-term outcomes mandate long-term surveillance and prospective studies to prove the effectiveness of endovascular graft repair (51).
|
Three articles addressed costs of follow-up and repeat intervention (42,47). Birch et al (42) reported that during follow-up, costs for endovascular aneurysm repair are higher than those for surgery because of the greater requirement for radiologic imaging studies. Total costs in the surgery group were 65% of costs in the endovascular repair group. After decision analysis (with a mathematic model based on reported average resource use and their hospital accounting system), Patel et al (47) suggested that endovascular repair may be cost-effective if the mortality rate was less than 1.2% when surgical mortality rate was higher than 1.7%. The model suggested that endovascular repair would only become cost-effective if it led to a large decrease in the combined mortality and long-term morbidity rates associated with surgery; this does not appear to be the case (Tables 5, 6). Bosch et al (48) used a decision-analysis model to study the cost-effectiveness of elective endovascular repair. They found that in certain circumstances, endovascular repair was a cost-effective alternative to surgery for elective repair of infrarenal abdominal aortic aneurysms. The benefits and cost-effectiveness of endovascular aneurysm repair were highly dependant on uncertain outcomes, particularly endovascular graft performance as measured according to long-term failure and rupture rates. The clinical effectiveness data for the model were derived from published literature and included the 30-day mortality rates from the meta-analysis by Adriaensen et al (31). These rates are considered further in the Discussion.
Effect of Endovascular Repair on Radiology Departments
The effect of the establishment of an endovascular aneurysm repair program on the workload of a radiology department is not widely addressed in the literature. When compared with surgery, additional imaging examinations are needed (42). Secondary endovascular procedures are required in 7.5% to 23% of patients who undergo endovascular aneurysm repair (Table 6). These are not usually required for patients who undergo surgery.
Eligibility Rates for Endovascular Repair
Although we had not considered eligibility rates when we designed our focused clinical question, review of the level 2b articles showed that this is a clinically important issue for investigators of endovascular repair who work at the efficacy level and for radiologists who may be planning to start an endovascular aneurysm repair program at the effectiveness level (ordinary practice). Treiman et al (38) initially suggested that the eligibility rate among patients referred for endovascular aneurysm repair was about 20%. Subsequently, Sarkar et al (37) suggested that the true rate was closer to 27%. Brewster et al (23) documented an eligibility rate that was nearer to 50%. In their most recent article, Zarins et al (20) analyzed eligibility rates over a 3-year period. They suggested that the current eligibility rates are approaching 63% and found that eligibility rates had increased over a 3-year period from 45% in the first 1
years of the study period to 63% in the second 1-year period. The overall eligibility rate over the 3-year period was 54%. The most frequent reason for ineligibility of patients with abdominal aortic aneurysms for endovascular aneurysm repair was involvement of the proximal neck (20).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
The results of this review raised many interesting issues.
Eligibility Rates
The eligibility rates among patients referred for endovascular repair appear to be changing from an initial rate of 20% (38) to the most recent assessment of almost 63% (20). Zarins et al (20) investigated the factors that have led to increased eligibility. They found that these were a result of increased workload in patients considered unsuitable for surgery because of comorbidity. Technical improvements in endovascular repair stent design may further increase eligibility rates.
Perioperative Morbidity
Endovascular aneurysm repair has definite benefits over surgery in the early perioperative period: less blood loss, shorter intensive care stay, and fewer systemic complications (Table 4). On the other hand, tabulation of data from the best observational studies suggests that endovascular repair may result in significantly increased local and/or vascular complications (Table 5). The rates of endoleak vary from 5.6% to 36% (Table 4). The meta-analysis (31) showed no significant difference in local and/or vascular complications between endovascular aneurysm repair and surgery. Factors relevant to this finding are discussed in the next section. The rates of early local and vascular complications may reduce with time and improved technology, particularly in the area of more low-profile delivery systems.
Early (30-day) Mortality
This proved to be one of the more difficult issues to evaluate. The level 2b study results are shown in Table 5. The results of one trial (21) suggested that endovascular repair was associated with a statistically significant increase in 30-day mortality (number needed to harm, 38). In that article (21), Zarins et al state that five patients who received stent-grafts died, and no patients who underwent surgery died. The authors report results of a multicenter trial of the AneuRx stent-graft (Medtronic, Sunnyvale, Calif). This includes the first 190 patients in the United States who were treated with this graft and incorporates the learning curve for 12 centers. Three of the deaths occurred in the first 40 procedures (phase I stent graft feasibility trial) and two in the next 150 patients (phase II clinical trial). The mortality difference between phases I and II was statistically significant (P < .05). When the phases were pooled for comparison with the surgery group, the P value for the mortality difference between surgery and endovascular aneurysm repair was higher than .05 (not significant). The authors considered that the difference was perhaps related to improvements in technical aspects of the procedure as the trial proceeded (21). This result does not prove that endovascular repair has a (clinically significant) higher 30-day mortality rate compared with that of surgery. No other significant differences were found in level 2b studies.
In their meta-analysis, Adriaensen et al (31) found that endovascular repair resulted in a statistically significant lower 30-day mortality rate than that of surgery (P = .03). We noted that the point estimates of mortality for endovascular repair and surgery were overlapped by both sets of 95% CIs, which suggests either that there is no difference or that the difference is not statistically significant according to 95% CI measurement (ie, sample size too small). It is not clear how clinically significant the mortality difference is, since we are dealing with small percentage differences in outcomes and a marginal level of statistical significance. Evidence-based teaching instructs that the magnitude of the treatment effect should be estimated by calculating the number needed to treat in these circumstances. The number needed to treat was 57. That is, 57 patients would have to undergo endovascular repair rather than surgery before one extra patient would be alive at 30 days postoperatively. This may prove to be a relatively small clinical effect, when the mid- and long-term mortality and/or morbidity of endovascular repair are considered.
One of our validity questions (13) was whether the estimates of treatment effect were similar from study to study—that is, were the results of the studies homogeneous? We tabulated the individual mortality rates for endovascular repair and surgery from the studies pooled for meta-analysis by Adriaensen et al (Table 7) (31).
|
Table 7 shows that the mortality rate for surgery reported by Treharne et al (53) is an immediately obvious outlier within the selection. This article, published from a tertiary referral center, was based on the evaluation of endovascular repair and surgical procedures performed from December 1994 to November 1997 by using a physiologic scoring system to compare expected and observed morbidity and mortality rates. Although the mean physiologic scores were similar for both groups, the "operative severity score" was significantly greater in the surgery group (26.3 vs 19.7, P < .001). The authors concluded that the increased mortality rate for surgery was a result of the operative severity. Adriaensen et al (31) point out that their meta-analysis is based on observational studies. Oxman et al (55) noted that in observational studies, physicians may systematically select patients with a good prognosis to undergo experimental therapy, a pattern of practice that may be consistent over time and geographic setting. In these circumstances, subsequent randomized controlled trials often show no difference (55). The study by Treharne et al (53) was undertaken in the early days of endovascular repair (between 1994 and 1997). It is conceivable that during this period, in that center, simpler aneurysms were treated with endovascular repair—a selection bias that influenced the final result.
The discrepancy between the surgical mortality rate of Treharne et al (53) and that of the other centers is large. The surgical complexity is well documented. The random-effects model of meta-analysis gives smaller studies proportionately greater weight in the pooled estimate than in the fixed effects model (56). If the smaller studies are further from the null result, the random-effects model will tend to produce larger estimates of effect than will the fixed-effects model (56). These points raise the possibility that the meta-analysis (31) may have underestimated the difference between endovascular aneurysm repair and surgery for local and/or vascular complications and overestimated the difference in 30-day mortality. Evidence-based methodology (32,57) suggests that once the data had been pooled with data from other studies for a meta-analysis, it would have been reasonable to perform a sensitivity analysis and repeat the meta-analysis after removal of the data of Treharne et al (53). If results remained consistent despite this manipulation, one could be more confident about their reliability (57).
Overall, for practical purposes, the present evaluation with evidence-based methods did not show a convincing clinically significant difference between endovascular repair and surgery for 30-day mortality.
Mid- and Long-term Mortality Rates
Cumulative patient survival for endovascular repair and surgery appears comparable at 2, 3, and 5 years. The necessity for patients who undergo endovascular repair to undergo lifelong CT follow-up and secondary and endovascular procedures that are not necessary in patients who undergo surgery is cause for concern (51,58). This obviously results in quality of life issues for the patient and economic issues for health care systems.
Cost
Most authors who have addressed the question of cost suggest that endovascular repair is more expensive to perform than surgery (40–44). This is mainly due to the cost of the prosthesis. Some of these costs are at least partially offset by reduced intensive care stay and operating room time, as well as reduced hospital stay and the reduced necessity for blood transfusion. There are limited data on costs of follow-up imaging and repeat intervention. These are expected to be higher with endovascular repair. Current evidence suggests that the cost of long-term follow-up after endovascular repair is likely to be greater than that of surgery, when the cost of follow-up with conventional contrast material–enhanced CT and the additional costs of secondary and endovascular procedures are included. Endovascular repair may, however, be cost-effective under certain conditions (48).
Effect on Radiology Departments
The effect of the establishment of an endovascular repair program on the workload of a radiology department may include the recruitment of additional interventional vascular radiologists who will need to be available in the operating room for long periods of time. There will be increased interventional radiology workload in the treatment of early local vascular complications and also potentially with long-term graft failure. In addition, the necessity for follow-up CT will increase imaging workload. How this is viewed from country to country will vary, depending on local funding, staffing, and reimbursement policies.
In summary, current best evidence (from nonrandomized controlled trials performed at the efficacy level) shows that endovascular aneurysm repair has short-term benefits compared with surgery, but there is slight, if any, difference in mortality; surgery costs less, and surgical grafts currently perform better at mid- and long-term follow-up. Decision-analysis models suggest that endovascular repair may be cost-effective if certain parameters are met. Cost-effectiveness is highly dependent on (a) endovascular aneurysm repair producing a substantial decrease in morbidity and mortality rates and (b) the long-term performance of endovascular repair grafts (48).
What Next?
Since evidence-based methodology was used for this review, it should constitute level 2a evidence (Table 1). According to the CEBM hierarchy of evidence (4), only a multicenter randomized controlled trial or a systematic review of randomized controlled trials would constitute a higher level of evidence. Randomized controlled trials are in progress in the United Kingdom and the Netherlands, and results will determine future best practice. There are two trials underway in England: Endovascular aneurysm repair trial I involves the comparison of surgery and endovascular repair in patients considered fit for both surgery and endovascular repair. Endovascular aneurysm repair trial II involves the comparison of endovascular repair and best medical management against best medical treatment alone in patients considered unfit for surgery. The inclusion criteria for the trials are that patients must have an abdominal aortic aneurysm 5.5 cm or larger in diameter and be 60 years of age or older. An endovascular aneurysm repair registry has been in place in Europe since 1996 (36). This "EUROSTAR" registry constitutes a large case series, and the results equate to level 4 evidence according to CEBM criteria (4).
Results of the present review suggest that, for centers currently not participating in randomized controlled trials, surgery is still the elective treatment of choice for patients with low-risk abdominal aortic aneurysms and long life expectancies. The emerging data regarding mid- and long-term outcome of endovascular aneurysm repair suggest that the use of endovascular repair technology should at present reside in relatively few experienced centers, ideally those whose patients compose part of a trial population. The future of endovascular aneurysm repair must be viewed with caution at present (39,59). Our evidence-based policy decision has therefore been not to start an endovascular aneurysm repair program in our hospital until and/or unless the results of randomized controlled trials show conclusive benefit for specific subsets of patients with abdominal aortic aneurysms. In the meantime, we will consider referring selected patients to centers involved in the above randomized controlled trials on a case-by-case basis.
Evidence-based Techniques: Caveats for Potential Users
Performance of our study allowed us the opportunity to assess whether evidence-based techniques could be used by radiologists in the environment of a relatively small yet busy radiology department. We found that these techniques could be used, but we identified several barriers to evidence-based radiology practice.
Lack of "secondary" radiology literature.—Within the evidence-based paradigm, original research publications are termed "primary" literature, and summaries of primary literature that have been filtered for quality and relevance by using evidence-based methods are described as "secondary" literature. Our article would, for example, constitute secondary literature on endovascular aneurysm repair. When compared with the published and Web-based secondary literature databases available to internal medicine specialists (4,60–64), there is little secondary literature available for radiologists. The routine practice of evidence-based radiology will be greatly assisted if the authoritative style of evidence-based systematic review (as opposed to the traditional authoritarian "expert opinion" review) becomes more fundamental to the radiology literature, as it has to the internal medicine literature (7,13,60).
Problems within the secondary literature.—The terminology of the newer style of review and/or overview is used variably. For example, the study by Adriaensen et al (31) is termed a "systematic review" but does not meet the design criteria for a systematic review described by British workers (32). It is hoped that the explicit evidence-based methodology (13) will give improved consistency and reliability of results compared with traditional "expert" reviews. Meta-analysis is promoted as a mathematic method that produces a high level of evidence. No method is perfect, and however detailed the literature search or mathematic meta-analysis, inclusion and exclusion of studies in different reviews may produce differing results (65). Standardization of terminology and, as much as possible, study selection methodology defined by those working at the efficacy level would be helpful to practitioners of evidence-based medicine at the effectiveness level.
Lack of training in computerized literature searching.—Because of the lack of secondary literature, radiologists must search the primary literature by using a search engine. The annual McMaster university course "How to Teach Evidence-based Clinical Practice" and the Oxford University and CEBM course "The Practice of Evidence-based Medicine" teach the use of subscription-based search engines and software for literature searching. These are more sophisticated options than PubMed but are currently less widely available. We used PubMed because it provides free Web-based access to MEDLINE data, and we suspect that this is the search engine most radiologists who attempt to practice evidence-based radiology will use. There is a need for radiologists to develop, learn, and teach effective searching strategies and to develop databases of filtered and validated information, as has been done in other specialities. Postgraduate training in storage and retrieval of literature and performance of repeat literature searches will be needed if evidence-based radiology is to become widespread practice.
Lack of training in evidence-based critical appraisal.—One author in our group (D.E.M.) attended the McMaster and Oxford University evidence-based courses mentioned previously and acted as a facilitator for the rest of the group. This was a genuine "problem-based learning" exercise for us all, however. It was our first application of the "Users’ Guides to the Medical Literature" (9–11,13–15,61,66,68) to an interventional radiology problem. The explicitness of the evidence-based approach and our lack of prior training made it a slow process. The "Users’ Guides" for appraisal of overviews, studies on the evaluation of procedural benefit and harm, and economic analyses are, nevertheless, useful for radiologists and can be accessed easily on the Internet (61,66).
|
APPENDIX |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
To effectively carry out the third step, literature is classified as belonging to one of several "domains." Examples of domains are diagnosis, therapeutic benefit, therapeutic harm, reviews, clinical guidelines, prognosis, economic analysis, and qualitative research. The difference between evidence-based and "traditional" literature analysis is that the evidence-based researchers have described, in the peer-reviewed literature, an explicit process of appraisal for literature in each domain (66). Factors that will act as sources of bias in study design have been identified, and mathematic analyses that will give the reader a clear idea of the strength, statistical significance, and possible clinical importance of the results are described. Analyses of procedural benefit used in the present review are the absolute risk reduction and number needed to treat. Analyses of procedural harm are the absolute risk increase and the number needed to harm. These are calculated from control event rates and experimental event rates. The absolute risk reduction is the arithmetic difference in risk between the two groups (control event rate minus experimental event rate). If this is a positive number, the control group has a higher risk of this complication. If it is negative, the experimental group has a higher risk (absolute risk increase). Absolute risk data may be difficult for us to remember, especially when the numbers are less than 1. On the other hand, the inverse of absolute risk reduction (1/absolute risk reduction) is a whole number and has the useful property of telling us the numbers of patients that we need to treat with the experimental therapy for the duration of the trial to prevent one additional bad outcome (number needed to treat). Similarly, the inverse of absolute risk increase (1/absolute risk increase) is also a whole number and tells us the number of patients that we need to treat with experimental therapy for the duration of the trial to encounter one additional bad outcome (number needed to harm) (16,18,70).
In addition to being located in the original literature (71), these principles are now widely available in textbook form (67,68) and on the Internet (4,61,66,72,73). The applicability of evidence-based principles to radiology was recently discussed in detail in a special review in Radiology (69). Interested readers are referred to these resources.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviation: CEBM = Centre for Evidence-Based Medicine
Author contributions: Guarantor of integrity of entire study, D.E.M.; study concepts, D.E.M., S.J.S.; study design, D.E.M., M.M.M.; literature research, M.M.M., A.M.M.; data acquisition, M.M.M., A.M.M.; data analysis/interpretation, A.M.M., D.E.M., P.M.M., M.M.M.; statistical analysis, M.M.M., P.M.M.; manuscript preparation, D.E.M., M.M.M., P.M.M.; manuscript definition of intellectual content and editing, D.E.M., M.M.M., P.M.M., S.J.S.; manuscript revision/review and final version approval, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS DISCUSSIONAPPENDIXREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  E. J. R. van Beek and D. E. Malone Evidence-based Practice in Radiology Education: Why and How Should We Teach It? Radiology, June 1, 2007; 243(3): 633 - 640. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. M. Maher, P. A. Hodnett, and M. K. Kalra Evidence-based Practice in Radiology: Steps 3 and 4--Appraise and Apply Interventional Radiology Literature Radiology, March 1, 2007; 242(3): 658 - 670. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Rakita, A. Newatia, J. J. Hines, D. N. Siegel, and B. Friedman Spectrum of CT Findings in Rupture and Impending Rupture of Abdominal Aortic Aneurysms RadioGraphics, March 1, 2007; 27(2): 497 - 507. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. Dodd Evidence-based Practice in Radiology: Steps 3 and 4--Appraise and Apply Diagnostic Radiology Literature Radiology, February 1, 2007; 242(2): 342 - 354. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. E. Malone Evidence-based Practice in Radiology: An Introduction to the Series Radiology, January 1, 2007; 242(1): 12 - 14. [Full Text] [PDF]
|
|
|
|
|
|
M. Staunton Evidence-based Radiology: Steps 1 and 2--Asking Answerable Questions and Searching for Evidence Radiology, January 1, 2007; 242(1): 23 - 31. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Blankensteijn, S. E.C.A. de Jong, M. Prinssen, A. C. van der Ham, J. Buth, S. M.M. van Sterkenburg, H. J.M. Verhagen, E. Buskens, D. E. Grobbee, and the Dutch Randomized Endovascular Aneurysm Managem Two-Year Outcomes after Conventional or Endovascular Repair of Abdominal Aortic Aneurysms N. Engl. J. Med., June 9, 2005; 352(23): 2398 - 2405. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. N. Ghansah and J. T. Murphy Complications of Major Aortic and Lower Extremity Vascular Surgery Seminars in Cardiothoracic and Vascular Anesthesia, December 1, 2004; 8(4): 335 - 361. [Abstract] [PDF]
|
|
|
|
|
|
M. Prinssen, E. L.G. Verhoeven, J. Buth, P. W.M. Cuypers, M. R.H.M. van Sambeek, R. Balm, E. Buskens, D. E. Grobbee, J. D. Blankensteijn, and the Dutch Randomized Endovascular Aneurysm Managem A Randomized Trial Comparing Conventional and Endovascular Repair of Abdominal Aortic Aneurysms N. Engl. J. Med., October 14, 2004; 351(16): 1607 - 1618. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. T. van Heek, D. T. Ubbink, D. A. Legemate, D. E. Malone, M. M. Maher, and P. M. MacEneaney Systematic Review of Studies on Elective Endovascular Repair versus Conventional Surgery for Abdominal Aortic Aneurysms * Dr Malone and colleagues respond: Radiology, September 1, 2004; 232(3): 933 - 936. [Full Text] [PDF]
|
|
|
|
|
|
H. O. Stolberg Treatment of Abdominal Aortic Aneurysms: Review with Evidence-based Methods Radiology, September 1, 2003; 228(3): 614 - 616. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
