Cirugía de derivación para la isquemia crónica de extremidades inferiores
Resumen
Antecedentes
La cirugía de derivación es uno de los principales tratamientos para los pacientes con isquemia crítica de las extremidades inferiores (ICEI). Esta es la segunda actualización de la revisión publicada por primera vez en 2000.
Objetivos
Evaluar los efectos de la cirugía de derivación en pacientes con isquemia crónica de las extremidades inferiores.
Métodos de búsqueda
Para esta actualización, el Grupo Cochrane Vascular (Cochrane Vascular Group) realizó búsquedas en su registro de ensayos (última búsqueda en octubre de 2016) y en el Registro Cochrane Central de Ensayos Controlados (Cochrane Central Register of Controlled Trials, CENTRAL) en The Cochrane Library (última búsqueda Número 9, 2016).
Criterios de selección
Se seleccionaron ensayos controlados aleatorizados de cirugía de derivación versus control o cualquier otro tratamiento. Los parámetros de resultados primarios se definieron como complicaciones postoperatorias no trombóticas tempranas, mortalidad en la intervención, mejoría clínica, amputación, permeabilidad primaria y mortalidad en el seguimiento.
Obtención y análisis de los datos
Para esta actualización, dos autores de la revisión extrajeron los datos y evaluaron la calidad de los ensayos. Se analizaron los datos mediante el odds ratio (OR) e intervalos de confianza (IC) del 95%. Se aplicaron modelos de efectos fijos o aleatorios.
Resultados principales
Se seleccionaron 11 ensayos que informaron un total de 1486 participantes. Seis ensayos compararon la cirugía de bypass con la angioplastia transluminal percutánea (ATP), y uno con la endarterectomía remota, la tromboendarterectomía, la trombólisis, el ejercicio y la estimulación de la médula espinal. La calidad de la evidencia de los resultados más importantes de la cirugía de derivación frente a la ATP fue alta, excepto por la mejora clínica y la permeabilidad primaria. Se consideró que la calidad de la evidencia de la mejora clínica fue baja, debido a la heterogeneidad entre los estudios y al hecho de que se trataba de una evaluación subjetiva de los resultados y, por tanto, con riesgo de sesgo de detección. Se consideró que la calidad de la evidencia de la permeabilidad primaria fue moderada debido a la heterogeneidad entre los estudios. Para las restantes comparaciones, la evidencia fue limitada. Para varios resultados, los IC fueron amplios.
La comparación de la cirugía de derivación con la ATP reveló un posible aumento de las complicaciones no trombóticas tempranas tras la intervención (OR 1,29; IC del 95%: 0,96 a 1,73; seis estudios; 1015 participantes) con la cirugía de derivación, pero la cirugía de derivación se asoció con mayores tasas de éxito técnico (OR 2,26; IC del 95%: 1,49 a 3,44; cinco estudios; 913 participantes). Los análisis según la diferente gravedad clínica de la enfermedad (claudicación intermitente (CI) o ICEI) revelaron que las complicaciones perintervención ocurrieron más frecuentemente en los participantes con ICEI que se sometieron a una cirugía de derivación que en los que se sometieron a ATP (OR 1,57; IC del 95%: 1,09 a 2,24). No se identificaron diferencias en la mortalidad durante la intervención (OR 1,67; IC del 95%: 0,66 a 4,19; cinco estudios; 913 participantes). La tasa de permeabilidad primaria al año fue mayor después de la cirugía de derivación que después de la ATP (OR 1,94; IC del 95%: 1,20 a 3,14; cuatro estudios; 300 participantes), pero esta diferencia no se demostró a los cuatro años (OR 1,15; IC del 95%: 0,74 a 1,78; dos estudios; 363 participantes). No se identificaron diferencias en la mejoría clínica (OR 0,65; IC del 95%: 0,03 a 14,52; dos estudios; 154 participantes), las tasas de amputación (OR 1,24; IC del 95%: 0,82 a 1,87; cinco estudios; 752 participantes), las tasas de reintervención (OR 0,76; IC del 95%: 0,42 a 1,37; tres estudios; 256 participantes), ni en la mortalidad dentro del período de seguimiento (OR 0,94; IC del 95%: 0,71 a 1,25; cinco estudios; 961 participantes) entre el tratamiento quirúrgico y el endovascular. No se comunicaron diferencias en los parámetros de resultados subjetivos, indicados por la calidad de vida y el bienestar físico y psicosocial. Se informó de que la estancia en el hospital para el procedimiento inicial fue más prolongada en los participantes sometidos a una cirugía de derivación que en los tratados con ATP.
En el único estudio (116 participantes) que comparó la cirugía de derivación con la endarterectomía remota de la arteria femoral superficial, la frecuencia de complicaciones tempranas no trombóticas tras la intervención fue similar en los grupos de tratamiento (OR 1,11; IC del 95%: 0,53 a 2,34). No se registró mortalidad en los 30 días siguientes al tratamiento inicial ni durante la estancia en el hospital en ninguno de los dos grupos. No se identificaron diferencias en cuanto a la permeabilidad (OR 1,66; IC del 95%: 0,79 a 3,46), la amputación (OR 1,70; IC del 95%: 0,27 a 10,58) y las tasas de mortalidad dentro del período de seguimiento (OR 1,66; IC del 95%: 0,61 a 4,48). No se dispuso de información sobre la mejora clínica.
No se identificaron diferencias en las complicaciones graves (OR 0,66; IC del 95%: 0,34 a 1,31) ni la mortalidad (OR 2,09; IC del 95%: 0,67 a 6,44) dentro de los 30 días de tratamiento entre la cirugía y la trombólisis (un estudio, 237 participantes) para la isquemia crónica de las extremidades inferiores. La tasa de amputación fue menor después de la cirugía de derivación (OR 0,10, IC del 95%: 0,01 a 0,80). No se encontraron diferencias en la mortalidad tardía (OR 1,56; IC del 95%: 0,71 a 3,44). No se informaron datos sobre las tasas de permeabilidad y la mejora clínica.
El éxito técnico que resultó en la restauración del flujo sanguíneo fue mayor después de la cirugía de derivación que la tromboendarterectomía para la enfermedad oclusiva aortoilíaca (un estudio, 43 participantes) (OR 0,01, IC del 95%: 0 a 0,17). La mortalidad durante la intervención (OR 0,33; IC del 95%: 0,01 a 8,65), la mortalidad en el seguimiento (OR 3,29; IC del 95%: 0,13 a 85,44) y las tasas de amputación (OR 0,47; IC del 95%: 0,08 a 2,91) no difirieron entre los tratamientos. No se informaron las tasas de mejora clínica y de permeabilidad.
La comparación de la cirugía y el ejercicio (un estudio, 75 participantes) no identificó diferencias en las complicaciones tempranas tras la intervención (OR 7,45; IC del 95%: 0,40 a 137,76) ni en la mortalidad (OR 1,55; IC del 95%: 0,06 a 39,31). No se informaron los resultados primarios restantes. No hubo diferencias en el tiempo máximo de caminata entre el ejercicio y la cirugía (1,66 min, IC del 95%: ‐1,23 a 4,55).
En cuanto a las comparaciones de la cirugía de derivación con la estimulación de la médula espinal para la ICEI, no hubo diferencias en las tasas de amputación después de 12 meses de seguimiento (OR 4,00; IC del 95%: 0,25 a 63,95; un estudio, 12 participantes). No se informaron los parámetros de resultados primarios restantes.
Conclusiones de los autores
Hay evidencia limitada de alta calidad sobre la eficacia de la cirugía de derivación en comparación con otros tratamientos; ningún estudio comparó la derivación con un tratamiento médico óptimo. Nuestro análisis ha demostrado que la ATP se asocia con una disminución de las complicaciones durante la intervención en los participantes tratados por ICEI y con una estancia hospitalaria más corta en comparación con la cirugía de derivación. El tratamiento quirúrgico parece conferir mejores tasas de permeabilidad hasta un año. El tratamiento endovascular puede ser aconsejable en pacientes con una comorbilidad significativa, lo que los convierte en candidatos quirúrgicos de alto riesgo. No se pueden extraer conclusiones sólidas sobre las comparaciones de la cirugía de derivación con otros tratamientos debido a la escasez de evidencia disponible. Es necesario realizar más ensayos grandes para evaluar el impacto de la ubicación anatómica y la extensión de la enfermedad, así como la gravedad clínica.
PICOs
Resumen en términos sencillos
Cirugía de derivación para la isquemia crónica de extremidades inferiores
Antecedentes
El síntoma más común de la isquemia crónica de las extremidades inferiores (flujo sanguíneo insuficiente a las piernas) es la claudicación, un dolor por calambres causado por un suministro insuficiente de sangre al músculo afectado. A menudo afecta al músculo de la pantorrilla y habitualmente se desencadena por el ejercicio y se alivia con el reposo. La restricción más grave de la irrigación sanguínea puede producir un dolor en reposo, úlceras de la pierna o gangrena. Estas afecciones, y la claudicación severa, pueden requerir una cirugía de bypass (o derivación) u otros tratamientos para mejorar el flujo de sangre a la pierna.
Resultados clave
Esta revisión de once ensayos con un total de 1486 participantes (actualizada hasta octubre de 2016) identificó seis ensayos que comparaban la cirugía de derivación con la angioplastia (estiramiento con balón y/o stent de la arteria estrechada u ocluida), y uno con la endarterectomía a distancia (una combinación de extracción de placa y stent), otro con la tromboendarterectomía (extracción de la placa y el coágulo), otro con la trombólisis (disolución del coágulo), otro con el ejercicio y otro con la estimulación de la médula espinal. En esta revisión, no se encontró evidencia que favorezcan la cirugía de derivación sobre la angioplastia en cuanto al efecto sobre la muerte, la mejoría de los síntomas, la tasa de amputación, la necesidad de un procedimiento adicional o la mortalidad a largo plazo. Las complicaciones del procedimiento se produjeron con mayor frecuencia en los pacientes con isquemia grave en las piernas (dolor en reposo, úlceras o gangrena) que se sometieron a una cirugía de derivación que en los que se sometieron a una angioplastia. Hubo evidencia de que la cirugía de derivación tenía éxito técnico más a menudo, se asociaba con una estancia hospitalaria más larga y que una tasa más alta del injerto de bypass permanecía abierto (permeable) un año después del procedimiento en comparación con la angioplastia; esta diferencia de permeabilidad a favor de la cirugía desaparecía después de cuatro años. Tampoco hubo evidencia claras que favoreciera la cirugía de derivación (bypass) en comparación con otros tratamientos, como lo indicaban las complicaciones de los procedimientos y las muertes, la mejoría clínica, la permeabilidad de los vasos sanguíneos y la mortalidad a largo plazo. Las comparaciones de la cirugía de bypass con la trombólisis mostraron menos amputaciones en los pacientes sometidos a cirugía de derivación, mientras que para el resto de las comparaciones la tasa de amputaciones fue similar.
Calidad de la evidencia
En general, la calidad de la evidencia fue alta para todos los resultados más importantes desde el punto de vista clínico, excepto dos. Se consideró que la calidad de la evidencia de la mejora clínica era baja, ya que se trataba de un resultado subjetivo con riesgo de sesgo, ya que a los evaluadores de resultados no se les ocultaron los tratamientos del estudio y porque había diferencias en los resultados entre los estudios. La calidad de la evidencia de la permeabilidad del injerto de bypass fue moderada debido a las diferencias de resultados entre los estudios. Se necesitan más estudios, con un gran número de participantes, para investigar la eficacia de la cirugía de derivación para la isquemia crónica de as extremidades inferiores.
Authors' conclusions
Summary of findings
| Bypass surgery compared with angioplasty for chronic lower limb ischaemia | ||||||
| Patient or population: Individuals with peripheral arterial disease Settings: Hospital Intervention: Bypass surgery Comparison: Percutaneous transluminal angioplasty | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Angioplasty | Bypass surgery | |||||
| Early postoperative non‐thrombotic complications Follow up: 30 days | Medium risk population1 | OR 1.29 (0.96 to 1.73) | 1015 | ⊕⊕⊕⊕ | ||
| 245 per 1000 | 295 per 1000 | |||||
| Procedural mortality Follow up: 30 days | Medium risk population1 | OR 1.67 (0.66 to 4.19) | 913 | ⊕⊕⊕⊕ | Three studies reported no cases of procedural mortality | |
| 15 per 1000 | 25 per 1000 | |||||
| Clinical improvement Follow up: 23‐48 months | Medium risk population1 | OR 0.65 (0.03 to 14.52) | 154 | ⊕⊕⊝⊝ | Estimate effect based on two studies | |
| 800 per 1000 | 722 per 1000 | |||||
| Amputation Follow up: 12‐48 months | Medium risk population1 | OR 1.24 (0.82 to 1.87) | 752 | ⊕⊕⊕⊕ | ||
| 126 per 1000 | 152 per 1000 | |||||
| Primary patency Follow up: 12 months | Medium risk population1 | OR 1.94 (1.20 to 3.14) | 300 | ⊕⊕⊕⊝ | ||
| 583 per 1000 | 731 per 1000 | |||||
| Primary patency Follow up: 4 years | Medium risk population1 | OR 1.15 (0.74 to 1.78) | 363 | ⊕⊕⊕⊕ | Estimate effect based on two studies | |
| 633 per 1000 | 665 per 1000 | |||||
| Mortality within follow‐up Follow up: 12‐48 months | Medium risk population1 | OR 0.94 (0.71 to 1.25) | 961 | ⊕⊕⊕⊕ | ||
| 371 per 1000 | 357 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 Based on studies including both claudication and critical ischaemia participants; the assumed risk was calculated by the mean number of events in the control groups of the selected studies for each outcome. | ||||||
Background
Description of the condition
Even though peripheral arterial disease (PAD) may present with acute limb ischaemia, chronic lower limb ischaemia is the most common presentation of PAD and results from an atherosclerotic process affecting the lower extremity arteries causing a reduction in the blood supply to the leg. Intermittent claudication (IC) is the most common presenting symptom for patients with PAD. It is thought to be produced by an inadequate supply of oxygen to the calf, thigh, or buttock muscles during exercise, resulting in anaerobic metabolism and pain. In its more severe manifestations, PAD may lead to critical limb ischaemia (CLI), which is characterised by intractable rest pain, ischaemic ulceration, or gangrene. Patients with CLI are at significant risk of developing irreversible ischaemic damage to the leg or foot if no appropriate treatment is undertaken, and this may result in amputation of the limb (Norgren 2007). In the National Health and Nutrition Examination Survey (NHANES), the overall prevalence of symptomatic or asymptomatic PAD in individuals aged 40 years or older was 4.3%, with a dramatic increase with age, rising from 0.9% in those younger than 50 years to 14.5% in those 70 years or older (Selvin 2004). As well as having a detrimental impact on functional capacity and quality of life, PAD indicates a more widespread systemic atherosclerotic disease affecting arterial trees in different organ systems, such as the coronary and cerebral circulation.
Description of the intervention
The importance of identifying and appropriately treating patients with PAD lies in both the management of ischaemia, to relieve symptoms or prevent amputation, or both, and the control of atherosclerotic risk factors, life style modifications and optimal medical treatment to mitigate the cardiovascular and cerebrovascular risk. Treatments for PAD range from conservative measures, such as management of cardiovascular risk factors with antiplatelets, statins, and exercise regimens, to interventional therapies, including surgical and endovascular arterial reconstruction.
Surgical bypass of the diseased arterial segment is one of the main treatments for the patient with life‐limiting claudication or CLI. The type of bypass procedure in the lower limbs depends on the extent of disease and involves reconstructions of the aorto‐iliac segment or infra‐inguinal arterial segment or both. The first successful femoro‐popliteal bypass operation was performed in 1950 by William Holden using a section of the patient's own vein; this is called autogenous vein graft (Holden 1950). Since then, a number of synthetic materials have been developed, including Dacron and expanded polytetrafluoroethylene (PTFE), a whole range of collagen tubes derived from human umbilical vein, and bovine carotid artery. Cadaveric homografts have also been used. Autogenous vein is considered the preferred conduit for infra‐inguinal bypass.
Apart from a surgical bypass, endoluminal procedures have been developed for the treatment of PAD. Endovascular techniques for the treatment of patients with lower extremity ischaemia include balloon angioplasty, insertion of stents and stent‐grafts, plaque debulking procedures, thrombolysis, and percutaneous thrombectomy (Tepe 2006). The range of new adjunct or alternative endovascular treatments and techniques is consistent with the constant advent of technological developments and bioengineering.
How the intervention might work
The primary goals of interventional treatment of chronic lower limb ischaemia are to relieve ischaemic pain, heal ischaemic ulcers, prevent limb loss, and improve patient's functional capacity and quality of life. In order to achieve these outcomes, some patients will ultimately require a surgical or endovascular revascularization procedure. Bypass procedures have the advantages of technical success, satisfactory anatomical patency and clinical durability, whereas proponents of endovascular therapies emphasise the minimally invasive nature of the procedures with subsequent reduced morbidity and mortality, enhanced recovery, and improved resource utilisation. A plethora of clinical research provides supporting evidence for the relative merits of approaches and techniques of lower limb revascularization.
Why it is important to do this review
The aging population, the rising prevalence of diabetes in western societies, and continued tobacco abuse worldwide is likely to result in a wide spread increase of PAD and increase in the number of revascularization procedures in the foreseeable future, with resultant socioeconomic implications and consumption of health care resources. As with many surgical interventions, bypass surgery was introduced without formal evaluation. Nowadays, however, patients and doctors are expected to make informed decisions based on evidence from randomised controlled trials, and it is important that the evidence comparing surgery with other treatment modalities is readily available (Antoniou 2013a). This review summarizes all previous trials of bypass surgery and highlights the advantages and disadvantages of surgery compared with other treatments. Furthermore, it identifies areas for future research. The review provides comparisons of bypass surgery to other treatments for symptomatic PAD, but does not assess the effect of any treatment on the natural history of PAD, either claudication or CLI.
Objectives
To assess the effects of bypass surgery in patients with chronic lower limb ischaemia.
Methods
Criteria for considering studies for this review
Types of studies
All randomised controlled trials of bypass surgery versus control (no treatment) or any other regimen were eligible for the review. Possible comparisons included endovascular intervention, exercise therapy, and medical treatment. Any method of randomisation was eligible and differences in quality were taken into account in the analysis. Trials that were not analysed on an intention‐to‐treat basis were included provided all randomised participant were accounted for.
Types of participants
Trials in individuals with lower limb ischaemia due to atherosclerotic disease, in which disease was confirmed by objective testing, were eligible for the review (Fontaine stages II, III, and IV) (Fontaine 1954). Trials of individuals with chronic ischaemia were included, defined as the presence of symptoms for more than 14 days. The severity of symptoms did not affect inclusion in the review, but this factor was taken into account in the analysis.
Types of interventions
Any surgical bypass procedure for the treatment of chronic lower limb ischaemia was included, irrespective of the approach, route, or type of graft employed. This was, therefore, likely to focus on individuals undergoing femoro‐popliteal bypass surgery, but other routes such as aorto‐iliac segment surgery were also included, if performed to treat lower limb ischaemia.
Types of outcome measures
Primary outcomes
The primary outcome measures were divided into early perioperative or peri‐interventional outcomes and follow‐up outcomes, as follows:
Early perioperative or peri‐interventional outcomes
-
early postoperative non‐thrombotic complications
-
procedural mortality
Follow‐up outcomes
-
clinical improvement (defined as improvement in Rutherford category) (Rutherford 1997)
-
amputation
-
primary patency (vessel or graft patency following initial procedure with no further intervention)
-
mortality
Secondary outcomes
Similarly, the secondary outcome measures were divided into early perioperative or peri‐interventional outcomes and follow‐up outcomes, as follows:
Early peri‐operative or peri‐interventional outcomes
-
technical success (defined as technical accomplishment of the intended intervention)
Follow‐up outcomes
-
assisted primary patency (patency not lost but maintained with prophylactic intervention)
-
secondary patency (restored patency after occlusion)
-
vessel or graft occlusion
-
reinterventions
-
walking distance (time to onset of pain and maximal walking distance)
-
success in ulcer healing assessed by complete healing
Subjective measures included:
-
quality of life scores as reported in the included studies
-
use of resources (such as length of hospital stay)
Technical success, clinical improvement, vessel or graft patency, and reinterventions are additional outcomes to those included in the initial review. The selected outcome parameters were thought to provide valuable additional information related to the comparative effectiveness of bypass surgery for the treatment of PAD.
Search methods for identification of studies
Electronic searches
For this update, the Cochrane Vascular Clinical Information Specialist (CIS) searched the following databases for relevant trials:
The Cochrane Vascular Specialised Register (October 2016);
The Cochrane Central Register of Controlled Trials (CENTRAL (2016, Issue 9)) via The Cochrane Register of Studies Online.
See Appendix 1 for details of the search strategy used to search CENTRAL.
The Cochrane Vascular Specialised Register is maintained by the CIS and is constructed from weekly electronic searches of MEDLINE Ovid, Embase Ovid, CINAHL, AMED, and through handsearching relevant journals. The full list of the databases, journals, and conference proceedings which have been searched, as well as the search strategies used are described in the Specialised Register section of the Cochrane Vascular module in the Cochrane Library (www.cochranelibrary.com).
In addition, the CIS searched the following trial registries (October 2016) for details of ongoing and unpublished studies;
-
ClinicalTrials.gov (www.clinicaltrials.gov)
-
World Health Organization International Clinical Trials Registry Platform (www.who.int/trialsearch)
-
ISRCTN Register (www.isrctn.com/).
Searching other resources
The reference lists of relevant articles retrieved by electronic searches were searched for additional citations.
Data collection and analysis
Selection of studies
For the present update, eligibility assessment of the reports provided by the Cochrane Vascular CIS was performed independently by two review authors (GAA and GSG). Disagreements were discussed with a third review author (FT), who acted as an adjudicator in the event of disagreement. We contacted the principal investigators of trials that were potentially included but terminated early and no published results could be found, to check availability of additional information.
Data extraction and management
For this update, two review authors (GSG and SAA) independently extracted data using a prespecified data collection form based on the Cochrane Vascular data extraction template. Disagreements were resolved by discussion with the contact author (GAA).
Assessment of risk of bias in included studies
The Cochrane Collaboration's risk of bias tool was applied to assess the risk of bias of the selected trials according to Higgins 2011. This tool evaluates six main domains: random sequence generation and allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other sources of bias. We completed a risk of bias table for each eligible study. For each individual domain, we classified studies into low, unclear, or high risk of bias. We considered blinding and incomplete outcome data separately for each outcome.Two review authors (RRM, JDS) independently assessed the methodological quality of the trials. The contact author (GAA) acted as an adjudicator in the event of disagreement. FT performed the risk of bias assessment of an article written in the Italian language and discussed the results with the contact author (GAA).
Measures of treatment effect
Analysis of dichotomous variables, such as mortality and the occurrence of postoperative complications, was carried out using the odds ratio (OR) with 95% confidence interval (CI) as the summary statistic. The total numbers of participants and numbers of events for each outcome parameter were entered into RevMan 5.3 to calculate the effect measure. Patency rates were transformed into a dichotomous outcome for specific time periods. For continuous variables, such as length of hospital stay, we aimed to calculate mean differences (MDs) using means and standard deviations (SD). If different scales were used in the different studies, the results were standardised, where possible, and then combined to form a standardised mean difference (SMD). Where these data were unavailable, we reported medians and interquartile range, but did not include these results in a meta‐analysis.
Unit of analysis issues
We did not identify any cluster‐randomised or cross‐over trials; therefore, no special issues with regard to analyses of studies with non‐standard designs existed. Each participant was counted as the unit of analysis for the defined outcome measures (e.g. primary patency).
Dealing with missing data
We planned to contact authors of selected studies to clarify any missing or unclear outcome data. Quantitative analyses were performed on an intention‐to‐treat basis where possible.
Assessment of heterogeneity
We anticipated that there might be considerable heterogeneity among the studies because of differences in severity of chronic lower limb ischaemia, anatomical level of disease, and methods of surgical or other treatments applied. In‐between study heterogeneity was examined with the combination of the Cochrane Q (Chi2) test and the I2 statistic. Important heterogeneity (Chi2 P < 0.05 and I2 ≥ 75%) was investigated, where possible, by subgroup analyses.
Assessment of reporting biases
For each study, the effect by the inverse of its standard error was plotted. If 10 or more studies were included in any single meta‐analysis, we planned to assess publication bias both visually evaluating the symmetry of such funnel plots and using the Egger’s regression intercept.
Data synthesis
Pooled ORs with 95% CIs were calculated using the Mantel–Haenszel fixed‐effect model, unless evidence of between study heterogeneity (Chi2 P < 0.05 and I2 ≥ 75%) existed, in which case random‐effects models of DerSimonian and Laird were applied.
Subgroup analysis and investigation of heterogeneity
Where sufficient information was available, we planned to investigate the following subgroups, which could account for heterogeneity among studies: individuals undergoing arterial reconstruction at different anatomical levels (e.g. aorto‐iliac or infra‐inguinal reconstruction) and individuals with different disease severity (e.g. IC or CLI).
Sensitivity analysis
We prespecified several additional analyses to assess the robustness of our results; we tested the effect of removing one study at a time on the pooled effect measure. We also undertook analyses to explore the contribution of risk of bias by excluding the trials that were found to be at high risk of bias in one or more domains.
Summary of findings table
We constructed a table compiling and summarizing the best evidence of relevant outcomes for the comparison of bypass surgery with PTA. We considered study populations consisting of individuals with disease severity ranging form claudication to severe limb ischaemia. We selected the most important and clinically relevant outcomes (both desirable and undesirable) that were thought to be essential for decision‐making for the summary of findings Table for the main comparison. We calculated assumed control intervention risks by the mean number of events in the control groups of the selected studies for each outcome. We used the system developed by the Grades of Recommendation, Assessment, Development and Evaluation Working Group (GRADE working group) for grading the quality of evidence as high, moderate, low and very low, based on within‐study risk of bias, directness of evidence, heterogeneity, precision of effects estimates, and risk of population bias (GRADE 2004).
Results
Description of studies
See Figure 1.
Study flow diagram.
Results of the search
The search of CENTRAL and the Specialist Register identified 4734 reports, after duplicates were removed. Irrelevant reports were discarded and we assessed the full text of 40 articles for eligibility. Of these, 12 additional studies (14 reports) were excluded and the reasons for exclusion are provided in the Characteristics of excluded studies table, seven reports were additional publications of a previously included study (BASIL study), and nine (12 reports) were ongoing trials. Three new studies (seven reports) were identified, which along with the eight studies included in the previous version of this review made a total of 11 studies included in the qualitative synthesis and meta‐analysis.
Included studies
Three additional studies were included in this update (Lepantalo 2009; McQuade 2010; REVAS Trial). There were also seven additional publications added for one study (BASIL study). Eleven studies reporting a total of 1486 participants fulfilled our inclusion criteria and were selected for analysis (BAESIC study; BASIL study; Gaspard 1972; Guarnera 1994; Holm 1991; Lepantalo 2009; Lundgren 1989; McQuade 2010; REVAS Trial; STILE Trial; Veterans Study). They are summarized in the Characteristics of included studies table. Approximately half of these studies (five trials) were published in 2000s; the first trial investigating the effects of bypass surgery in individuals with chronic lower limb ischaemia was published in the early 1970s (Gaspard 1972). The number of participants in the included trials ranged from 12 to 452. The largest trial is the BASIL study, which assigned individuals with severe lower limb ischaemia secondary to infra‐inguinal arterial disease to receive bypass surgery or balloon angioplasty. The eleven included trials were conducted in six different countries (seven trials in Europe and four in North America).
There was also some variation in the types of participant included in the eleven trials. Most trials included both men and women, except the Veterans Study which involved men only. Seven trials included individuals with a range of disease severity (both IC and CLI), but two were restricted to individuals with claudication only (BAESIC study; Lundgren 1989) and two included only those with CLI (BASIL study; Guarnera 1994). The proportion of claudicants in the trials with mixed groups ranged from 34% in the STILE Trial to 89% in the Lepantalo 2009 trial. In addition, the STILE Trial included a mixture of individuals with native artery disease and individuals with existing grafts, but only the subset with native artery disease has been included in this review.
There were no trials that compared bypass surgery with a placebo, no intervention, or medical management. Six trials compared bypass surgery with PTA (BAESIC study; BASIL study; Holm 1991; Lepantalo 2009; McQuade 2010; Veterans Study). In the remaining trials, bypass surgery was compared with: remote endarterectomy of the superficial femoral artery (REVAS Trial); thromboendarterectomy (Gaspard 1972); thrombolysis (STILE Trial); spinal cord stimulation (Guarnera 1994); and exercise, in which the control group performed dynamic leg exercises that were supervised by a physiotherapist (Lundgren 1989).
The type of bypass procedure performed in each trial was similar in most studies. Vein grafts were generally used for distal reconstructions, and synthetic prostheses for aorto‐iliac or ilio‐femoral bypasses and some femoro‐popliteal bypasses above the knee. In the REVAS Trial, the type of graft for the femoro‐popliteal bypass above the knee was either prosthetic (PTFE) or vein (great saphenous vein), and separate analyses were performed for the two types of graft. There was also some variation in the types of surgery performed. In four trials, unfortunately, a number of participant underwent endarterectomy rather than bypass surgery (BASIL study; Holm 1991; Lundgren 1989; STILE Trial) and these different groups were not separated in the analysis. Furthermore, the type of endovascular procedure varied among trials comparing effects of bypass surgery with those of endovascular treatment for chronic lower limb ischaemia. Four trials (BAESIC study; BASIL study, Holm 1991; Veterans Study) used PTA without stenting or with stenting at the discretion of the treating physician, whereas two trials (Lepantalo 2009; McQuade 2010) used an endograft (covered stent) in all their endovascular procedures.
Nine of the eleven trials included mortality and procedural (or technical) success as outcome measures (BAESIC study; BASIL study; Gaspard 1972; Holm 1991;Lepantalo 2009; McQuade 2010; REVAS Trial; STILE Trial; Veterans Study), and most of these also reported complications, patency rates, and need for amputation. Three trials included subjective measures (BASIL study; Guarnera 1994; Veterans Study); one included only treadmill testing and measures of lower limb blood flow (Lundgren 1989).
Excluded studies
For this update, an additional 12 studies were excluded (ABC 2010; CLEVER study; Djoric 2011; Gavrilenko 2008; IRONIC Trial; Matyas 2008; Nordanstig 2011; PROOF 2007; Stanisic 2009; TECCO Trial; Tiek 2009; Tiek 2012). This made a total of 23 excluded studies (ABC 2010; CLEVER study; de Donato 2002; Devine 2004; Djoric 2011; Gavrilenko 2008; Gelin 2001; Hamsho 1999; IRONIC Trial; Jensen 2007; Linhart 1991; Matyas 2008; McCollum 2003; Mohammadi 2007; Nordanstig 2011; Panneton 2004; PROOF 2007; Stanisic 2009; Taft 2004; TECCO Trial; Tiek 2009; Tiek 2012; Vukobratov 2006) These studies are summarised in the Characteristics of excluded studies table. Most of the excluded studies compared different techniques of bypass surgery or different types of bypass grafts (de Donato 2002; Devine 2004; Hamsho 1999; Gavrilenko 2008; Jensen 2007; Matyas 2008; McCollum 2003; Mohammadi 2007; Panneton 2004; Stanisic 2009; Tiek 2012; Vukobratov 2006). Three studies did not have a bypass group (CLEVER study; Djoric 2011; Tiek 2009), and in another study there is no mention of randomisation and the two different treatment options (surgery and medical therapy) were not compared in the analysis (Linhart 1991). Two of the excluded trials (PROOF 2007; ABC 2010) potentially fulfilled the inclusion criteria, as they were randomised controlled trials comparing bypass surgery with plaque excision (Silverhawk Plaque Excision) or angioplasty for the treatment of participants with CLI and IC, respectively. Unfortunately, both trials were terminated and no published or presented results could be found. The principal investigators either confirmed the absence of published results (ABC 2010) or did not respond to our request (PROOF 2007). Another four trials (Gelin 2001; IRONIC Trial; Nordanstig 2011; Taft 2004) comparing invasive with non‐invasive treatment for lower limb ischaemia were excluded because the enrolled participants were randomised to any invasive treatment (including surgical or endovascular) rather than bypass surgery. The CLEVER study is a randomised controlled trial comparing optimal medical management, stent placement, supervised exercise rehabilitation, and combined stenting with supervised exercise rehabilitation for aorto‐iliac occlusive disease in individuals suffering from IC. It was excluded from our review and analysis because bypass surgery was not included in the treatment arms. Furthermore, even though the TECCO Trial compared surgery with endovascular treatment for common femoral artery disease, a minority of participants underwent bypass surgery in the surgical treatment arm and therefore, this study was excluded.
Ongoing studies
Nine ongoing trials were identified through searches of clinical trials databases. BASIL 2 is a multi‐centre randomised controlled trial conducted in the UK comparing the clinical and cost effectiveness of a "vein bypass first" with an "endovascular first" revascularization strategy for severe limb ischaemia due to infra‐geniculate arterial disease. Best endovascular treatment involves balloon angioplasty and possibly the use of stents. Participant recruitment started in May 2014 and the anticipated end date is October 2019. The BEST‐CLI trial is a pragmatic, multicentre, open label, randomised trial that compares best endovascular therapy with best open surgical treatment in individuals with CLI eligible for both treatments. This trial is funded by the National Lung Heart and Blood Institute of the National Institutes of Health and aims to enrol 2100 participants with CLI at 120 sites in North America . Participant recruitment started in August 2014 and the anticipated end date is December 2018. FINNPTX is a Finnish multicentre randomised clinical trial comparing paclitaxel‐eluting stent with femoro‐popliteal bypass using PTFE graft for the treatment of long superficial femoral artery occlusion in individuals with life‐limiting IC or CLI. The trial commenced in October 2011 and is estimated to be completed in 2017 with an enrolment of 400 participants. ROBUST is a single‐centre randomised clinical trial comparing bypass surgery with angioplasty and stenting for TASC II B and C lesions of the superficial femoral artery. It was launched in 2009 and aims to enrol 320 individuals with IC that does not respond to medical management or with CLI. SUPERB is a randomised controlled trial comparing heparin‐bonded endoluminal with surgical femoro‐popliteal bypass in individuals with symptomatic PAD. This trial, which is currently recruiting participants in the Netherlands, commenced in October 2010 and the estimated date of completion is December 2019. The ZILVERPASS trial is another randomised controlled trial comparing the Cook Zilver PTX drug‐eluting stent with bypass surgery for the treatment of femoro‐popliteal TASC C and D lesions in individuals with symptomatic PAD. This study is being conducted in Belgium, commenced in August 2014, and is anticipated to enrol 220 participants by November 2017. The Optimized Strategy for Diabetic Patients with Critical Limb Ischaemia study (NCT01171703) randomises diabetic individuals with chronic long occlusion of the superficial femoral artery to receive a femoro‐popliteal PTFE bypass above the knee or stenting. Another ongoing randomised controlled trial (ISRCTN18315574) compares bypass surgery with ipsilateral great saphenous vein with percutaneous transluminal angioplasty with stent placement in individuals with IC or CLI and TASC C/D femoropopliteal disease. The NCT02580084 trial is the only trial comparing aorto‐femoral bypass with hybrid intervention consisting of common femoral endarterectomy and iliac balloon angioplasty and stenting. The study is being conducted in Russia and the estimated completion date is August 2020.
Risk of bias in included studies
Our risk of bias assessments for each included study are summarized in Figure 2 and as percentages across all studies in Figure 3. Details and reasons for each assessment are listed in the Characteristics of included studies table.
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Allocation
In three trials, the allocation sequence was adequately generated (BAESIC study; BASIL study; REVAS Trial). A computerised random‐number generator or a central telephone number in a permuted‐block sequence was applied to generate the allocation sequence. A sequential treatment assignment, with balancing for prognostic factors, which was applied in two trials (Holm 1991; Lundgren 1989), was not considered an appropriate method of sequence generation. Unfortunately, the rest of the selected trials provided insufficient information about the sequence generation process to permit judgement (Gaspard 1972; Guarnera 1994; Lepantalo 2009; McQuade 2010; STILE Trial; Veterans Study).
Appropriate methods of allocation concealment were used in five trials (BAESIC study; BASIL study; REVAS Trial; STILE Trial; Veterans Study). Either a central computerised allocation or sealed envelopes were used to conceal allocation.
Blinding
Inevitably, in trials of a surgical intervention, blinding was not possible, but as there were no comparison groups that received no treatment this may be less significant. None of the reports stated that those taking measurements were blinded to the treatment group and, therefore, it must be assumed that they were not. However, in several trials there was a set protocol for follow‐up assessment, and objective measures for assessment of main outcomes, such as Duplex ultrasound or angiographic imaging for the assessment of patency, were used. Therefore, we judged that the outcome and the outcome measurement were not likely to be influenced by lack of blinding.
Incomplete outcome data
There were either no or minimal losses to follow up in most trials. No issues with incompleteness of data were identified for six of the trials (BAESIC study; BASIL study; Gaspard 1972; Holm 1991; Lundgren 1989; REVAS Trial). Missing outcome data were balanced in numbers across intervention groups (BAESIC study; Holm 1991; Lundgren 1989) or the proportion of missing outcomes compared with observed event risk was not enough to have a clinically relevant impact on the intervention effect estimate (BASIL study; REVAS Trial). The STILE Trial reported a transparent process of dealing with missing or incomplete data and was therefore considered to be of low risk of bias. In the McQuade 2010 trial, six (12%) and 15 participants (30%) were lost to follow‐up in the stent and bypass group, respectively, and this trial was therefore considered to be high risk of attrition bias because of the imbalance in numbers across intervention groups. The remaining studies provided insufficient information to permit judgment.
Selective reporting
Reporting bias was judged to be present in three trials (Gaspard 1972; Lepantalo 2009; McQuade 2010). In the Gaspard 1972 trial, the outcome measures were not clearly defined and the study failed to include key outcomes. In the Lepantalo 2009 trial, even though costs were prespecified as a secondary outcome parameter, they were not reported. Insufficient information was available concerning whether amputation and mortality were prespecified outcomes in the McQuade 2010 trial. For the remaining trials, either no issues with regard to reporting bias existed or insufficient information to permit judgment was provided.
Other potential sources of bias
Two trials, the BAESIC study and the Lepantalo 2009 trial, were terminated prematurely because of recruitment issues and lack of benefit of endoluminal stent‐graft placement in the superficial femoral artery over bypass surgery in the Lepantalo 2009 trial. Furthermore, there might be a risk of bias in relation to participant compliance with exercise treatment in the Lundgren 1989 trial, but insufficient evidence that this problem would introduce bias was available. The 237 participants with native artery disease in the STILE Trial were a subset of a larger trial of 393 participants, which included both native artery and graft disease, and this may have biased the results. Reinterventions affecting assisted primary patency are potentially subject to intervention use bias, unless the criteria for reintervention are prespecified and applied equally to both interventions. Of the trials providing data for assisted primary patency (Lepantalo 2009; REVAS Trial), the REVAS Trial only defined criteria for reintervention in both groups, whereas the Lepantalo 2009 trial did not provide specific reintervention criteria to maintain primary patency and is, therefore, subject to reintervention bias.
Effects of interventions
See: Summary of findings for the main comparison
Bypass surgery compared with angioplasty
Six trials compared bypass surgery with PTA (BAESIC study; BASIL study; Holm 1991; Lepantalo 2009; McQuade 2010; Veterans Study). Participant groups in these trials included participants with IC, CLI, or both. The follow‐up period for each trial varied from 12 months (Holm 1991; Lepantalo 2009), 23 months (BAESIC study), 36 months (BASIL study), 48 months (McQuade 2010) to 49 months (Veterans Study).
Early postoperative non‐thrombotic complications
Early postoperative non‐thrombotic complications were reported in all trials comparing bypass surgery with angioplasty for chronic lower limb ischaemia (BAESIC study; BASIL study; Holm 1991; Lepantalo 2009; McQuade 2010; Veterans Study). Early complications occurred either within 30 days of the index treatment or during the initial hospital stay, and were local or systemic. Most trials reported major complications significantly affecting the participant's postoperative course or requiring intervention (e.g. surgical treatment of groin infection). Even though early non‐thrombotic complications occurred more frequently in participants undergoing bypass surgery, the difference did not reach statistical significance (OR 1.29, 95% CI 0.96 to 1.73; six studies; 1015 participants; Analysis 1.1). See Figure 4. Heterogeneity among the trials was I2 = 55%.
Forest plot of comparison: 1 Bypass surgery compared with angioplasty (PTA), outcome: 1.1 Early postoperative non‐thrombotic complications ‐ by symptoms at time of intervention.
We performed separate meta‐analyses for participants treated for CLI and those treated for claudication. No significant difference in the frequency of early postoperative non‐thrombotic complications between surgery and angioplasty in participants with claudication was found (OR 2.15, 95% CI 0.50 to 9.21), whereas in those with CLI, complications occurred more frequently in the bypass group (OR 1.57, 95% CI 1.09 to 2.24; test for subgroup differences P = 0.06; Analysis 1.1). Furthermore, the risk of complications in the surgery and angioplasty group was similar when separate meta‐analyses were performed for participants with iliac disease (OR 0.62, 95% CI 0.24 to 1.58) and those treated for femoro‐popliteal disease (OR 1.34, 95% CI 0.97 to 1.86; test for subgroup differences P = 0.13; Analysis 1.2).
Sensitivity analysis removing one study at a time showed an effect in favour of angioplasty when the Veterans Study was removed (OR 1.55, 95% CI 1.12 to 2.15; I2 = 30%). Sensitivity analysis excluding the trials that were found to be at high risk of bias in one or more domains (Holm 1991; Lepantalo 2009; McQuade 2010) found no difference between the groups (OR 1.17, 95% CI 0.43 to 3.19; I2 = 75%).
Procedural mortality
Mortality occurring within 30 days of treatment or during the hospital stay for the index procedure was reported in five trials (BAESIC study; BASIL study; Lepantalo 2009; McQuade 2010; Veterans Study). No significant difference in procedural mortality between surgical and endovascular treatment for chronic lower limb ischaemia was identified (OR 1.67, 95% CI 0.66 to 4.19; 913 participants; Analysis 1.3). The heterogeneity among the studies was I2 = 0%. Three studies reported no cases of procedural mortality (BAESIC study; Lepantalo 2009; McQuade 2010).
Repeating the analysis after removing one study at a time and after excluding the high risk of bias trials showed no difference between treatments.
Clinical improvement
Improvement in the clinical grade of the Rutherford classification was reported in two trials (BAESIC study; McQuade 2010). Our analyses revealed similar clinical improvement after bypass surgery and PTA (OR 0.65, 95% CI 0.03 to 14.52; 154 participants; Analysis 1.4). Heterogeneity was I2 = 75%.
Sensitivity analysis removing one study at a time showed no difference between the treatment groups.
Amputation
Five trials reported the numbers of participants who had an amputation of the treated limb during the follow‐up period (BAESIC study; BASIL study; Holm 1991; Lepantalo 2009; McQuade 2010). Participants treated with bypass surgery had a similar rate of progression to amputation of the treated limb to participants treated with PTA (OR 1.24, 95% CI 0.82 to 1.87; 752 participants; Analysis 1.5). Heterogeneity among the included studies was I2 = 31%.
We created separate meta‐analysis models to involve trials reporting outcome data for participants with claudication and those with CLI. Such analyses found no difference in amputation rates between surgery and angioplasty in participants treated for claudication (OR 0.40, 95% CI 0.04 to 4.02) and those treated for critical ischaemia (OR 1.14, 95% CI 0.73 to 1.77) (test for subgroup differences P = 0.15; Analysis 1.5).
Sensitivity analysis removing one study at a time showed an effect in favour of angioplasty when the BASIL study was excluded (OR 2.91, 95% CI 1.13, 7.48; I2 = 0%). Repeating the analysis after excluding trials at high risk of bias (Holm 1991; Lepantalo 2009; McQuade 2010) revealed no difference between the treatment groups (OR 0.96, 95% CI 0.60 to 1.52; I2 = 0%).
Primary patency
Primary patency rates were reported in five trials (BAESIC study; Holm 1991; Lepantalo 2009; McQuade 2010; Veterans Study). The primary patency at one year was found to be significantly higher in participants treated with bypass surgery than those receiving endovascular treatment (OR 1.94, 95% CI 1.20 to 3.14; four studies; 300 participants; Analysis 1.6). See Figure 5. The statistical heterogeneity among the studies was I2 = 71%. However, two of four trials (Holm 1991; McQuade 2010) found no effect and the other two (BAESIC study; Lepantalo 2009), which were relatively small trials, found a benefit. When applying a random‐effects model, there was no longer a combined benefit in favour of bypass surgery (OR 2.47, 95% CI 0.92 to 6.61). Sensitivity analysis removing one study at a time showed no difference when the BAESIC study (OR 1.53, 95% CI 0.90 to 2.61; I2 = 66%) and the Lepantalo 2009 trial were removed (OR 1.58, 95% CI 0.94 to 2.67; I2 = 67%). Repeating the analysis after excluding trials that were found to be at high risk of bias (Holm 1991; Lepantalo 2009; McQuade 2010) showed a difference in favour of bypass surgery (OR 6.54, 95% CI 1.79 to 23.84).
Forest plot of comparison: 1 Bypass surgery compared with angioplasty (PTA), outcome: 1.6 Primary patency at 1 year.
At four years, the primary patency was not found to be different between surgery and angioplasty (OR 1.15, 95% CI 0.74 to 1.78; two studies; 363 participants; Analysis 1.7). See Figure 6. The statistical heterogeneity was I2 = 0%. The Veterans Study provided specific four‐year patency information for participants with claudication and critical ischaemia, as well as for participants with iliac and femoro‐popliteal disease. Meta‐analyses found no significant differences in primary patency at four years between surgical and endovascular treatment in participants with claudication (OR 1.44, 95% CI 0.77 to 2.69) or critical ischaemia (OR 0.95, 95% CI 0.37 to 2.43; test for subgroup differences P = 0.63; Analysis 1.7), and in participants with femoro‐popliteal disease (OR 0.91, 95% CI 0.41 to 2.01) or iliac disease (OR 1.57, 95% CI 0.78 to 3.14; test for subgroup differences P = 0.31; Analysis 1.8). Sensitivity analysis removing one study at a time revealed no difference between the treatment groups.
Forest plot of comparison: 1 Bypass surgery compared with angioplasty (PTA), outcome: 1.7 Primary patency at 4 years ‐ by symptoms at time of intervention.
Mortality within follow‐up
Five out of the six trials reported mortality of the study populations within the follow‐up period (BASIL study; Holm 1991; Lepantalo 2009; McQuade 2010; Veterans Study). No significant difference in mortality between the treatment modalities was identified (OR 0.94, 95% CI 0.71 to 1.25; 961 participants; Analysis 1.9). Between‐study heterogeneity was I2 = 0%.
When separate analyses for participants with claudication and those with CLI were performed, no differences in mortality were identified (OR 0.41, 95% CI 0.02 to 10.55; and OR 0.77, 95% CI 0.54 to 1.11, respectively; test for subgroup differences P = 0.16 Analysis 1.9).
Sensitivity analysis removing one study at a time revealed no significant difference between the groups. Similarly, repeating the analysis without the trials that were found to be at high risk of bias (Holm 1991; Lepantalo 2009; McQuade 2010) showed no difference in mortality within follow‐up (OR 0.95, 95% CI 0.70 to 1.29; I2 = 73%).
Technical success
Technical success rates were reported in five trials (BAESIC study; BASIL study; Lepantalo 2009; McQuade 2010; Veterans Study). Technical success was either inconsistently defined by the study authors or a clear definition was not provided. Bypass surgery was found to be associated with a higher technical success rate than PTA (OR 2.26, 95% CI 1.49 to 3.44; 913 participants; Analysis 1.10). The statistical heterogeneity was I2 = 65%.
Sensitivity analysis removing one study at a time confirmed higher technical success with bypass surgery. However, excluding the high risk of bias trials (Lepantalo 2009; McQuade 2010) showed no significant difference between treatments (OR 5.91, 95% CI 0.51 to 69.01; I2 = 75%).
Assisted primary patency
Assisted primary patency rates were provided by one trial only (Lepantalo 2009). Assisted primary patency at one year was found to be significantly higher after bypass surgery than PTA (OR 8.71, 95% CI 1.64 to 46.31; 44 participants; Analysis 1.11). However, this result should be cautiously interpreted because of the very wide CI.
Secondary patency
Secondary patency rates were reported in three trials (Holm 1991; Lepantalo 2009; McQuade 2010). Pooled analysis revealed that the secondary patency at one year was similar in the bypass surgery and the angioplasty group (OR 1.28, 95% CI 0.71 to 2.34; 246 participants; Analysis 1.12). The statistical heterogeneity was I2 = 74%. Repeating the analysis after removing one study at a time showed no difference between treatments. All three trials were found to be of high risk of bias.
Similar to primary patency, no difference in secondary patency rates at four years between the treatment groups was identified (OR 0.90, 95% CI 0.37 to 2.19; one study; 100 participants; Analysis 1.13).
Vessel or graft occlusion
Two studies reported vessel or graft occlusion within the follow‐up period (BAESIC study; McQuade 2010). Even though the incidence of vessel or graft occlusion was higher in the angioplasty group, no statistically significant difference between the treatment groups was found (OR 0.56, 0.27 to 1.15; 154 participants; Analysis 1.14). The statistical heterogeneity was I2 = 46%.
Repeating the analysis after removing one study at a time revealed no difference between the treatment groups.
Reinterventions within follow‐up
Three trials reported numbers of reinterventions within the follow‐up period (BAESIC study; Holm 1991; McQuade 2010). Bypass surgery was associated with a lower reintervention rate, but the difference between surgery and angioplasty was not statistically significant (OR 0.76, 95% CI 0.42 to 1.37; 256 participants; Analysis 1.15). Heterogeneity among the trials was I2 = 0%.
Sensitivity analysis removing one study at a time revealed no difference between the treatment groups. Similarly, repeating the analysis after excluding the trials that were found to be at high risk of bias (Holm 1991; McQuade 2010) showed no difference between the treatments (OR 1.33, 95% CI 0.37 to 4.82).
Walking distance
Not reported.
Ulcer healing
Not reported.
Subjective measures
In the Veterans Study, the Sickness Impact Profile (SIP) was used to evaluate health status. This instrument provides a score of physical and psychosocial well‐being, measured on an interval scale from zero (no impairment) to 100 (maximum impairment), with an average score of 5.2 in healthy controls. Mean SIP scores after 12 months were 10.6 in the surgery group and 10.8 in the angioplasty group compared with 15.8 and 15.6 at baseline, respectively. Both groups showed a significant improvement in scores compared with baseline, without any significant differences between the two groups at any point in the follow‐up. In the BASIL study, self reported health‐related quality of life (HRQOL) was measured using the Vascular Quality of Life Questionaire (VascuQol), the EuroQoL (EQ‐5D) health outcome measure, and the Short Form 36 (SF‐36) up to three years from randomisation. These generic measures were recorded at baseline and at three, six, 12, 24, and 36 months after randomisation. No significant differences in HRQOL indices from baseline scores were found in either treatment group. There were no subjective measures of health status reported in the BAESIC study; Holm 1991; Lepantalo 2009; and McQuade 2010 trials.
Use of resources
The BASIL study reported that over three years, the use of inpatient hospital services was broadly similar in the treatment arms, as measured by the number of hospital admissions and total days in the hospital. Over three years, both groups had an average of three hospital stays. Furthermore, by three years, there was an insignificant difference in the mean length of hospital stay between the two groups, with 60 days (16 to 82 days) for the bypass group and 57 days (8 to 73 days) for the angioplasty group. Three more studies mention resource utilisation (Holm 1991; Lepantalo 2009; McQuade 2010), where the length of stay in hospital is reported. In the Holm 1991 trial, the post‐treatment stay was significantly shorter in the angioplasty group. The median lengths of stay for the IC and CLI groups, respectively, were 8.6 days and 15.0 days for the surgery group, and 2.6 and 5.0 days for the angioplasty group. In the Lepantalo 2009 trial, the hospital stay was longer for participants assigned bypass surgery (mean 4.5 days, range 2 to 10 days) than for participants randomised to endovascular treatment (mean 1.7 days, range 0 to 7 days). Similarly, in the McQuade 2010 trial, the length of hospital stay was significantly longer in the bypass group than that in the endovascular treatment group (mean 3.1 days, SD 1.8 days versus mean 0.9 days, SD 0.8 days; P < 0.01). Unfortunately, data from the Holm 1991 and Lepantalo 2009 trials were not available in a form suitable for inclusion in a statistical meta‐analysis.
Bypass surgery compared with remote endarterectomy
Only one trial with 116 participants (REVAS Trial) compared bypass surgery with remote endarterectomy for the treatment of lower limb ischaemia. This trial enrolled participants with severe claudication or CLI treated with supra‐geniculate bypass surgery with long saphenous vein or PTFE graft, or remote endarterectomy of the superficial femoral artery. The median duration of follow‐up for the whole study population was 37 months.
Early postoperative non‐thrombotic complications
The frequency of early postoperative non‐thrombotic complications was similar in the treatment groups (OR 1.11, 95% CI 0.53 to 2.34; Analysis 2.1).
Procedural mortality
No mortality within 30 days of the index treatment or during stay in hospital was recorded in either group.
Clinical improvement
Not reported.
Amputation
Three participants in the bypass group (55 participants) and two participants in the remote endarterectomy group (61 participants) progressed to major amputation of the treated limb, and the difference between the treatment groups was insignificant (OR 1.70, 95% CI 0.27 to 10.58; Analysis 2.2).
Primary patency
The primary patency rate at three years was similar after bypass surgery and remote endarterectomy (OR 1.66, 95% CI 0.79 to 3.46; Analysis 2.3).
Mortality within follow‐up
No differences in late mortality during the follow‐up period between the treatment arms were found (OR 1.66, 95% CI 0.61 to 4.48; Analysis 2.4).
Technical success
The technical success rate was higher in the bypass surgery group, but the difference between bypass surgery and remote endarterectomy was not significant (OR 10.81, 95% CI 0.58 to 200.08; Analysis 2.5). However, this result should be interpreted with caution because of the very wide CI.
Assisted primary patency
No differences in assisted primary patency at three years between treatments were identified (OR 1.35, 95% CI 0.63 to 2.93; Analysis 2.6).
Secondary patency
The secondary patency rate at three years was similar in the bypass and remote endarterectomy group (OR 1.21, 95% CI 0.54 to 2.69; Analysis 2.7).
Vessel or graft occlusion
Similar episodes of graft or native vessel occlusion occurred in the bypass group and the remote endarterectomy group within the follow‐up period (OR 0.97, 95% CI 0.43 to 2.19; Analysis 2.8).
Reinterventions within follow‐up
Not reported.
Walking distance
Not reported.
Ulcer healing
Not reported.
Subjective measures
Not reported.
Use of resources
The only information about resource utilisation provided by the REVAS Trial was the length of stay in hospital for the index procedure. Participants with chronic lower limb ischaemia treated with bypass surgery stayed in hospital for a significantly longer period (median 6 days, range 3 to 28 days) than those undergoing remote endarterectomy (median 4 days, range 1 to 21; P = 0.004).
Bypass surgery compared with thrombolysis
Bypass surgery was compared with thrombolysis in the STILE Trial only (237 participants). As mentioned in the Characteristics of included studies table, these results were included with the proviso that only 86% of the surgery group had a bypass procedure and 20% of the participants had acute rather than chronic ischaemia.
Early postoperative non‐thrombotic complications
In the Weaver 1996 report of the STILE Trial for results in native arteries, complications were combined into a measure of "major morbidity". This category included: life‐threatening haemorrhage requiring resuscitation; perioperative complications, for example myocardial infarction or stroke; renal failure requiring dialysis; serious anaesthesia‐related complications; vascular complications, for example, dissection; and postoperative wound complications. No significant difference in major morbidity within 30 days of treatment between surgery and thrombolysis was identified (OR 0.66, 95% CI 0.34 to 1.31; Analysis 3.1). Data were not provided for major morbidity excluding those participants with acute ischaemia or separated according to symptoms at presentation. Analysis by site of lesion also showed no significant difference between surgery and thrombolysis within 30 days of intervention (test for subgroup differences P = 0.19; Analysis 3.1).
Procedural mortality
There was no significant difference in 30‐day mortality between the surgery group and the thrombolysis group (OR 2.09, 95% CI 0.67 to 6.44; Analysis 3.2). No separate data for participants treated for claudication or CLI were provided. The site of the lesion also did not significantly affect mortality (test for subgroup differences P = 0.50; Analysis 3.2).
Clinical improvement
Not reported.
Amputation
After one year, there were significantly fewer amputations in the surgery group compared with the thrombolysis group (OR 0.10, 95% CI 0.01 to 0.80; Analysis 3.3). The significantly lower amputation rate in the surgery group at one year persisted when those with acute symptoms were excluded from the analysis. This significant difference was also present in the group of participants presenting with CLI (OR 0.06, 95% CI 0 to 1.02), but not in the group presenting with claudication (OR 0.44, 95% CI 0.02 to 11.12; test for subgroup differences P = 0.36; Analysis 3.3). Amputation rates at one year were also affected by the site of the lesion: there were significantly fewer amputations in those participants with femoro‐popliteal occlusions who received surgery compared with thrombolysis (OR 0.05, 95% CI 0 to 0.83), but there was no difference in the group with ilio‐femoral occlusions (OR 0.65, 95% CI 0.03 to 16.46; test for subgroup differences P = 0.24; Analysis 3.4). Analysis in the original article demonstrated that both diabetes and critical ischaemia were significant prognostic factors for amputation (P = 0.03). Diabetes as a risk factor alone did not reach statistical significance (OR 1.97, 95% CI 0.99 to 3.93; Weaver 1996).
Primary patency
Not reported.
Mortality within follow‐up
There was no significant difference in mortality at one year between the surgery group and the thrombolysis group (OR 1.56, 95% CI 0.71 to 3.44; Analysis 3.5). Mortality rates also did not differ significantly between surgery and thrombolysis when the analysis was performed excluding those participants with acute limb ischaemia. When the data were split by symptoms at the time of intervention (IC and CLI), there were also no significant differences between surgery and thrombolysis, although mortality tended to be less in the CLI group treated with thrombolysis (test for subgroup differences P = 0.53; Analysis 3.5). The site of the lesion also did not significantly affect mortality (OR 1.45, 95% CI 0.66 to 3.18) (test for subgroup differences P = 0.17; Analysis 3.6).
Technical success
Not reported.
Assisted primary patency
Not reported.
Secondary patency
Not reported.
Vessel or graft occlusion
Not reported.
Reinterventions within follow‐up
Not reported.
Walking distance
Not reported.
Ulcer healing
Not reported.
Subjective measures
Not reported.
Use of resources
Not reported.
Bypass surgery compared with thromboendarterectomy
Bypass surgery was compared with thromboendarterectomy in the Gaspard 1972 trial only. The results from this trial were unfortunately limited because it included only 43 participants, the follow‐up period was short (approximately six weeks), and relatively few outcome measures were included.
Early postoperative non‐thrombotic complications
The only complication reported was blood loss during surgery, which was reported by the study authors to be significantly greater in the thromboendarterectomy group (an average of 3.6 units per participant were required compared with 2.7 units required in the bypass group).
Procedural mortality
One participant in the thromboendarterectomy group died during hospital stay for the index procedure, whereas no in‐hospital death was recorded in the bypass group (OR 0.33, 95% CI 0.01 to 8.65; Analysis 4.1).
Clinical improvement
Not reported.
Amputation
Six weeks after intervention, there were two amputations in the bypass surgery group and four in the thromboendarterectomy group; this difference was not statistically significant (OR 0.47, 95% CI 0.08 to 2.91; Analysis 4.2).
Primary patency
Not reported.
Mortality within follow‐up
During follow up, there was one death in the bypass group due to a disrupted aortic suture line, whereas no death in the thromboendarterectomy group was recorded (OR 3.29, 95% CI 0.13 to 85.44; Analysis 4.3). Caution is required when interpreting this result because of the wide CI.
Technical success
Technical success was achieved in all participants in the bypass group, whereas the thromboendarterectomy was unsuccessful in six participants (OR 0.01, 95% CI 0 to 0.17; Analysis 4.4).
Assisted primary patency
Not reported.
Secondary patency
Not reported.
Vessel or graft occlusion
Not reported.
Reinterventions within follow‐up
Not reported.
Walking distance
Not reported.
Ulcer healing
Not reported.
Subjective measures
Not reported.
Use of resources
This was reported as the average time to complete the procedure. The time was an hour longer in the thromboendarterectomy group (5.8 hours compared with 4.6 hours), but no statistical tests were reported.
Bypass surgery compared with exercise
Bypass surgery was compared with exercise in only one trial (Lundgren 1989). This was a small trial enrolling only 75 participants, with a relatively short follow‐up period (eight to nine months). Twenty five participants were randomised to surgery, 25 participants to exercise, and another 25 participants were randomised to combined treatment with surgery and exercise therapy.
Early postoperative non‐thrombotic complications
Comparisons of the complication rates in the surgery and exercise group produced very wide CI (OR 7.45, 95% CI 0.40 to 137.76; Analysis 5.1). In the total of 50 participants randomised to surgery (surgery alone and surgery plus exercise), three developed a wound haematoma, two developed a myocardial infarction, and one suffered a pulmonary embolus. There were no direct complications of exercise.
Procedural mortality
One participant died in the surgery group and, similar to the comparison of early postoperative non‐thrombotic complications, comparisons of procedural mortality produced a wide CI (OR 1.55, 95% CI 0.06 to 39.31; Analysis 5.2).
Clinical improvement
Not reported.
Amputation
No amputations were reported in either group.
Primary patency
Not reported.
Mortality within follow‐up
Within the follow‐up period, there were two deaths in the surgery group, but this result should be cautiously interpreted because of the very wide CI (OR 2.63, 95% CI 0.12 to 56.86; Analysis 5.3).
Technical success
Not reported.
Assisted primary patency
Not reported.
Secondary patency
Not reported.
Vessel or graft occlusion
Not reported.
Reinterventions within follow‐up
The requirement for further intervention did not differ significantly between the two groups (OR 2.19, 95% CI 0.43 to 11.19; Analysis 5.4). In the total of 50 participants randomised to surgery, three participants required thrombectomy and five required a second reconstruction. In two of the participants randomised to exercise, limb‐threatening ischaemia developed and bypass surgery was performed.
Walking distance
There was no significant difference in maximal walking time between the exercise and surgery group at the end of the trial (1.66 min, 95% CI ‐1.23 to 4.55; Analysis 5.5), although improvement was slightly less in the exercise group (150% compared with 173%). There was no significant difference between the surgery group and a third group receiving both surgery and exercise.
Ulcer healing
Not relevant (the trial included only participants with IC).
Subjective measures
Not reported.
Use of resources
Not reported.
Bypass surgery compared with spinal cord stimulation
Bypass surgery was compared with spinal cord stimulation in one small trial of 12 participants in which very few outcome measures were reported (Guarnera 1994). This small sample size did not allow the demonstration of any statistically significant differences.
Early postoperative non‐thrombotic complications
Not reported.
Procedural mortality
No procedural deaths were reported in either group.
Clinical improvement
Clinical improvement defined as improvement in Rutherford category was not reported. However, therapeutic success was reported as good or fair where either complete or evident pain regression and trophic ulcer healing were obtained; otherwise, the result was considered poor. At 12 months, the results were poor in 60% of the bypass group and in 28% of the cord stimulation group; this difference was not statistically significant (OR 3.75, 95% CI 0.33 to 42.47; Analysis 6.1). However, this result should be cautiously interpreted because of the very wide CI.
Amputation
There was no significant difference in amputation rates between surgery and spinal cord stimulation after 12 months of follow‐up (OR 4.00, 95% CI 0.25 to 63.95; Analysis 6.2). Again, the result should be interpreted with caution because of the very wide CI.
Primary patency
Not reported.
Mortality within follow‐up
Not reported.
Technical success
Not reported.
Assisted primary patency
Not reported.
Secondary patency
Not reported.
Vessel or graft occlusion
Not reported.
Reinterventions within follow‐up
Not reported.
Walking distance
Not reported.
Ulcer healing
Ulcer healing was included in the overall measure of therapeutic success (see above), but was not reported separately.
Subjective measures
Not reported.
Use of resources
Not reported.
Discussion
Summary of main results
No studies comparing bypass surgery with no intervention or medical treatment were identified. In the trials selected in this review, bypass surgery was the de facto "gold standard" for the management of chronic lower extremity ischaemia of sufficient symptomatic severity to require treatment. The review provides comparisons of bypass surgery to other treatments for symptomatic PAD, but does not assess the effect of any treatment on the natural history of PAD, either claudication or CLI.
Comparisons of bypass surgery with angioplasty is the area in which most evidence is currently available. The main results are outlined in the summary of findings Table for the main comparison. Six of the eleven randomised trials reported comparisons of bypass surgery with endovascular therapy for chronic lower limb ischaemia in a total of 1015 participants (BAESIC study; BASIL study; Holm 1991; Lepantalo 2009; McQuade 2010; Veterans Study). In terms of the peri‐interventional outcomes, pooled analysis revealed that bypass surgery was associated with greater technical success than PTA. However, when interpreting this finding, one should take into account that technical success was either inconsistently defined among the trials included in the meta‐analysis or not defined at all. Furthermore, it was not possible to assess haemodynamic parameters as a measure of technical success, and no distinction was made between procedural success measures, such as improved blood flow to the foot and increased ankle brachial index (ABI), and technical success of merely accomplishing the intended intervention. Even though a trend towards reduced early postinterventional non‐thrombotic complications and procedural mortality in the endovascular treatment group was demonstrated, such differences did not reach statistical significance. Regarding the follow‐up outcomes, an interesting finding of our analysis is that the primary patency rate at one year was higher after bypass surgery, but no difference was found at four years of follow up. Furthermore, the assisted primary patency was higher in the bypass group, but this finding should be cautiously interpreted in the presence of a wide CI. One should also take into account that primary patency is a measure of the effectiveness of the procedure itself, while assisted primary patency is a function of the postprocedure surveillance process and willingness to intervene and is, therefore, subject to surveillance, observer and intervention bias. Modern objective performance comparisons consider the number and types of interventions needed to maintain patency, but data on this were not available. No differences were identified in the rest of the clinical outcome parameters for which data were available for meta‐analyses, including clinical improvement, amputation and reintervention rates, and late mortality. Similar results were found when subgroup analyses, comprising of participants with similar clinical severity of disease and anatomical level of reconstruction, were performed, except that bypass surgery was associated with a higher early non‐thrombotic complication rate compared with angioplasty in participants with CLI but not in those with claudication. No difference in subjective outcome parameters, indicated by quality of life and physical and psychosocial well‐being, was reported. The hospital stay for the index procedure was reported to be longer in participants undergoing bypass surgery than those treated with angioplasty. It is unknown, however, why participants undergoing bypass surgery stayed in the hospital for a longer period, and whether there were more minor amputations, debridement of pedal wounds or gangrene in one group versus the other, or the difference was entirely due to recovery from the index revascularization procedure itself.
Limited literature information was identified regarding comparisons of bypass surgery with other treatment modalities for chronic lower limb ischaemia. The recent REVAS Trial compared outcomes of bypass surgery and remote endarterectomy for femoro‐popliteal arterial disease. In this trial cohort of 116 patients, femoro‐popliteal bypass above the knee demonstrated similar outcomes with remote endarterectomy of the superficial femoral artery, expressed by technical success, perioperative morbidity and mortality, vessel or graft patency, progression to amputation, and late mortality.
The only clinical trial evaluating the comparative effectiveness of bypass surgery and thrombolysis for the treatment of chronic lower limb ischaemia is the STILE Trial. The results of this trial suggest some benefit of surgery compared with thrombolysis; the amputation rate was lower in the bypass group, particularly in patients with CLI and femoro‐popliteal lesions. Furthermore, the frequency of ongoing or recurrent ischaemia was lower after bypass surgery than after thrombolysis. There were no differences in the mortality rate or complications of the intervention between the two groups.
One trial only compared bypass surgery with thromboendarterectomy in a total of 43 participants with chronic limb ischaemia (Gaspard 1972). Despite the small size of the trial, a difference in restoration of blood flow and blood loss in favour of bypass surgery was found. Mortality and amputation rates did not differ significantly between the treatment groups. Unfortunately, information on the rest of the clinical outcome parameters was not provided.
Surgery was compared with exercise in a small trial (Lundgren 1989), and these results may, therefore, also have limited relevance. The only clear statistically significant difference was in the ABI, which was higher in the surgery group following intervention.
Bypass was compared with spinal cord stimulation in one trial of only 12 participants (Guarnera 1994). No differences were reported between the two interventions. Due to the small sample size of the study, analyses produced very wide CIs, so the finding of no difference should be interpreted with caution.
Overall completeness and applicability of evidence
Our search of bibliographic databases identified limited clinical research validating bypass surgery for the treatment of chronic lower limb ischaemia. The eleven trials included in this review varied in size, quality, and in the intervention served as control. There were no trials in which bypass surgery was compared with a placebo or no intervention, undoubtedly for ethical reasons. Furthermore, no trials comparing surgery with medical treatment were identified. Most existing evidence provides comparative information about bypass surgery versus endovascular treatment for chronic lower limb ischaemia (six trials). For the comparisons of bypass surgery with other treatment modalities, the evidence is limited, being provided by only one trial in each comparison. Most trials reported important outcome parameters, such as mortality, morbidity, patency and amputation rates, but other clinical information, such as walking distance in claudicants and ulcer healing in patients with CLI, was inadequately reported. Furthermore, amputation free survival and major adverse limb events, which are important outcomes and part of modern recommendations for studies of comparative effectiveness, were not reported in the studies included in review. Similarly, limited information was provided regarding subjective measures, such as quality of life and resource utilization.
Unfortunately, the existing results of bypass surgery versus PTA do not provide an overall clear picture favouring one treatment over the other. This may be because the existing trials were too small, because the effects really are similar, or because differences will appear only in defined subgroups of patients. It must also be remembered that these trials include only a small subset of patients with lower limb ischaemia, as those patients with multilevel disease requiring extensive or hybrid surgical and endovascular arterial reconstruction (Antoniou 2009) or those requiring tibial or pedal artery bypass may not be eligible for angioplasty. Extrapolation of our findings to the entire population suffering from PAD should be judiciously performed, because a great proportion of patients presenting with chronic lower limb ischaemia may be unsuitable for intervention due to the presence of comorbid conditions and, therefore, treated conservatively. Furthermore, these trials only included patients deemed suitable for either surgery or PTA. Patients with extensive lesions could have been excluded from receiving angioplasty and, therefore, participating in the trial. It is possible that these patients with more severe lesions receive more benefits from a surgical approach. This remains to be investigated. The long‐term (< 5 years) effects or bypass surgery in comparison to endovascular therapy remain unknown. The BASIL study found that for those patients who survived for at least two years after randomisation, a bypass‐first revascularization strategy was associated with a significant increase in overall survival and a trend towards improved amputation‐free survival compared to a balloon angioplasty‐first revascularization strategy. The trials also did not give detailed descriptions of how the angioplasties were performed and whether stents were used as an adjunct. A systematic review and meta‐analysis supports the use of primary stenting as opposed to balloon angioplasty alone, mainly for long lesions, as a first‐line endovascular treatment for symptomatic disease in the femoro‐popliteal segment (Acin 2012). None of the trials mentioned subintimal angioplasty, which may have produced different results (Chang 2013). Lastly, balloon angioplasty and stent technology is constantly evolving. Application of modern endovascular therapies, such as drug‐eluting balloons and stents and bioabsorbable stents for the treatment of anatomically "difficult" lesions, may reveal subtle differences in outcomes between treatments (Antoniou 2014). Furthermore, the introduction of endovascular tools, such as atherectomy, chronic total occlusion and re‐entry devices, may play an important role in improving procedural success and limb salvage.
All the evidence comparing bypass surgery with thrombolysis came from the STILE Trial. This was a large high quality trial, but was included in the review with some reservations (see Characteristics of included studies table). Some of these problems were mitigated by the randomisation method, which ensured that those with native artery disease were balanced across the interventions, and by the analysis, which was performed after excluding those patients with acute ischaemia. The third problem, relating to the use of endarterectomy in 14% of the surgery group, could not be addressed.
The rest of the trials comparing bypass surgery with remote endarterectomy (REVAS Trial), thromboendarterectomy (Gaspard 1972), exercise (Lundgren 1989), and spinal cord stimulation (Guarnera 1994) were relatively small, with a short follow‐up period; therefore, the results must be viewed with caution. Lack of statistical significance may result from the small sample size and not from any true absence of difference. The trials Gaspard 1972 and Guarnera 1994 are old studies, involving extensive aorto‐iliac endarterectomy and spinal cord stimulation, respectively, procedures which are currently rarely used in clinical practice.
Quality of the evidence
See summary of findings Table for the main comparison
The majority of the evidence investigating effects of bypass surgery for chronic lower limb ischaemia derives from randomised trials of surgical versus endovascular treatment for PAD affecting the lower limbs. Six such trials were identified, reporting a total of 1015 participants. Most of these studies had an adequate design and were executed well. No significant methodological constraints were identified. The available information is limited, however, by the fact that most of these studies recruited patients with a wide range of disease severity and/or anatomical location and extent of disease. According to current guidelines, the extent of atherosclerotic disease has a great impact on decision making and the selection of type of treatment (Norgren 2007). We attempted to circumvent such limitations by performing subgroup analyses, but the numbers of participants included in such analyses were relatively small. Furthermore, the type of endovascular treatment varied among studies, ranging from balloon angioplasty with bare stent placement at the discretion of the treating physician to routine use of covered stents. Despite the heterogenous nature of study populations and treatments, the trials consistently reported no great differences in main outcomes between surgical and endovascular treatment of chronic lower limb ischaemia. No solid conclusions can be drawn regarding comparisons of bypass surgery with other treatments because of the paucity of available evidence; only one study for each comparison was identified. When assessing bypass surgery versus PTA, we judged the quality of the evidence to be high for all primary outcomes except for clinical improvement and primary patency. We judged the quality of the evidence for clinical improvement to be low due to due to heterogeneity between the studies and the fact this was a subjective outcome assessment and therefore at risk of detection bias. We judged the quality of the evidence for primary patency moderate due to heterogeneity between the studies. Furthermore, the CI for this outcome and for several outcomes in the comparisons of bypass surgery with other treatments was large, which might be due to either a lack of studies, small participant numbers, or low number of events for some of the outcomes. For the comparisons of bypass surgery with other treatment modalities, the evidence is also limited by being provided by only one trial in each comparison.
Potential biases in the review process
We were unable to undertake funnel plots or to assess publication bias because we identified fewer than 10 studies for any outcome. We used participants as the unit of analysis, but one study (McQuade 2010) used limbs and this unit was used in the analysis. Sensitivity analysis excluding this study revealed no effect on the outcomes.
Agreements and disagreements with other studies or reviews
A related systematic review and meta‐analysis of surgical versus endovascular reconstruction of femoro‐popliteal arterial disease was recently conducted by our evidence synthesis research group (Antoniou 2013b). This review identified four randomised trials and six observational studies comprising a total of 2817 patients. Pooled analysis revealed that endovascular treatment was accompanied by lower 30‐day morbidity (OR 2.93, 95% CI 1.34 to 6.41) and higher technical failure than bypass surgery (OR 0.10, 95% CI 0.05 to 0.22). Similar to the results of the present review, analyses of follow‐up outcomes demonstrated higher primary patency rates in the surgical treatment arm one (OR 2.42, 95% CI 1.37 to 4.28), two (OR 2.03, 95% CI 1.20 to 3.45), and three years following intervention (OR 1.48, 95% CI 1.12 to 1.97), with this difference favouring surgery disappearing at four years (OR 1.09, 95% CI 0.74 to 1.60). Limb loss rates within two and three years of intervention was found to be higher in the endovascular group; however, at the end of the fourth year, the benefit in limb salvage in favour bypass surgery was eliminated. The study concluded that an endovascular‐first approach may be advisable in patients with significant comorbidity, whereas for fit patients with a longer‐term perspective, a bypass procedure may be offered as a first line interventional treatment.
A systematic review of nine studies (3071 subjects) investigating the comparative effectiveness of bypass surgery versus endovascular treatment for severe or CLI found no difference in mortality (OR 0.72, 95% CI 0.44 to 1.16) or amputation (OR 1.2, 95% CI 0.87 to 1.65), but higher primary patency (OR 2.5, 95% CI 1.25 to 4.99) and assisted primary patency (OR 3.39, 95% CI 1.53 to 7.51) (Abu Dabrh 2016). Another recent meta‐analysis found that angioplasty was not inferior to bypass surgery in patients with CLI, as indicated by amputation‐free survival, revascularisation, leg amputation, and overall mortality (Fu 2015).
No systematic reviews and analyses of bypass surgery versus other treatment modalities for chronic lower limb ischaemia were identified. A systematic review of the literature undertaken by our research group (Antoniou 2008) demonstrated that remote endarterectomy of the superficial femoral artery had acceptable outcomes, as indicated by technical success, procedure‐related complications, and patency rates. However, it was limited by the fact that it included single‐arm observational studies only and no comparisons with other therapeutic modalities, such as bypass surgery, were performed, as no related information was available. A recent systematic review in patients with acute lower limb ischaemia (< 14 days) found that thrombolysis may be associated with a higher risk of ongoing limb ischaemia and haemorrhagic complications, including stroke, than surgery (Berridge 2013).
Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Forest plot of comparison: 1 Bypass surgery compared with angioplasty (PTA), outcome: 1.1 Early postoperative non‐thrombotic complications ‐ by symptoms at time of intervention.
Forest plot of comparison: 1 Bypass surgery compared with angioplasty (PTA), outcome: 1.6 Primary patency at 1 year.
Forest plot of comparison: 1 Bypass surgery compared with angioplasty (PTA), outcome: 1.7 Primary patency at 4 years ‐ by symptoms at time of intervention.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 1 Early postoperative non‐thrombotic complications ‐ by symptoms at time of intervention.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 2 Early postoperative non‐thrombotic complications ‐ by site of intervention.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 3 Procedural mortality.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 4 Clinical improvement.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 5 Amputation ‐ by symptoms at time of intervention.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 6 Primary patency at 1 year.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 7 Primary patency at 4 years ‐ by symptoms at time of intervention.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 8 Primary patency at 4 years ‐ by site of intervention.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 9 Mortality within follow‐up ‐ by symptoms at time of intervention.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 10 Technical success.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 11 Assisted primary patency at 1 year.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 12 Secondary patency at 1 year.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 13 Secondary patency at 4 years.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 14 Vessel or graft occlusion.
Comparison 1 Bypass surgery compared with angioplasty (PTA), Outcome 15 Reinterventions.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 1 Early postoperative non‐thrombotic complications.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 2 Amputation.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 3 Primary patency at 3 years.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 4 Mortality within follow‐up.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 5 Technical success.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 6 Assisted primary patency at 3 years.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 7 Secondary patency at 3 years.
Comparison 2 Bypass surgery compared with remote endarterectomy, Outcome 8 Vessel or graft occlusion.
Comparison 3 Bypass surgery compared with thrombolysis, Outcome 1 Early postoperative non‐thrombotic complications ‐ by site of intervention.
Comparison 3 Bypass surgery compared with thrombolysis, Outcome 2 Procedural mortality ‐ by site of intervention.
Comparison 3 Bypass surgery compared with thrombolysis, Outcome 3 Amputation ‐ by symptoms at time of intervention.
Comparison 3 Bypass surgery compared with thrombolysis, Outcome 4 Amputation ‐ by site of intervention.
Comparison 3 Bypass surgery compared with thrombolysis, Outcome 5 Mortality within follow‐up ‐ by symptoms at time of intervention.
Comparison 3 Bypass surgery compared with thrombolysis, Outcome 6 Mortality within follow‐up ‐ by site of intervention.
Comparison 4 Bypass surgery compared with thromboendarterectomy (TE), Outcome 1 Procedural mortality.
Comparison 4 Bypass surgery compared with thromboendarterectomy (TE), Outcome 2 Amputation.
Comparison 4 Bypass surgery compared with thromboendarterectomy (TE), Outcome 3 Mortality within follow‐up.
Comparison 4 Bypass surgery compared with thromboendarterectomy (TE), Outcome 4 Technical success.
Comparison 5 Bypass surgery compared with exercise, Outcome 1 Early postoperative non‐thrombotic complications.
Comparison 5 Bypass surgery compared with exercise, Outcome 2 Procedural mortality.
Comparison 5 Bypass surgery compared with exercise, Outcome 3 Mortality within follow‐up.
Comparison 5 Bypass surgery compared with exercise, Outcome 4 Reintervention within follow‐up.
Comparison 5 Bypass surgery compared with exercise, Outcome 5 Maximal walking time (minutes).
Comparison 6 Bypass surgery compared with spinal cord stimulation (SCS), Outcome 1 Therapeutic success: poor result of intervention.
Comparison 6 Bypass surgery compared with spinal cord stimulation (SCS), Outcome 2 Amputation.
| Bypass surgery compared with angioplasty for chronic lower limb ischaemia | ||||||
| Patient or population: Individuals with peripheral arterial disease Settings: Hospital Intervention: Bypass surgery Comparison: Percutaneous transluminal angioplasty | ||||||
| Outcomes | Illustrative comparative risks* (95% CI) | Relative effect | No of Participants | Quality of the evidence | Comments | |
| Assumed risk | Corresponding risk | |||||
| Angioplasty | Bypass surgery | |||||
| Early postoperative non‐thrombotic complications Follow up: 30 days | Medium risk population1 | OR 1.29 (0.96 to 1.73) | 1015 | ⊕⊕⊕⊕ | ||
| 245 per 1000 | 295 per 1000 | |||||
| Procedural mortality Follow up: 30 days | Medium risk population1 | OR 1.67 (0.66 to 4.19) | 913 | ⊕⊕⊕⊕ | Three studies reported no cases of procedural mortality | |
| 15 per 1000 | 25 per 1000 | |||||
| Clinical improvement Follow up: 23‐48 months | Medium risk population1 | OR 0.65 (0.03 to 14.52) | 154 | ⊕⊕⊝⊝ | Estimate effect based on two studies | |
| 800 per 1000 | 722 per 1000 | |||||
| Amputation Follow up: 12‐48 months | Medium risk population1 | OR 1.24 (0.82 to 1.87) | 752 | ⊕⊕⊕⊕ | ||
| 126 per 1000 | 152 per 1000 | |||||
| Primary patency Follow up: 12 months | Medium risk population1 | OR 1.94 (1.20 to 3.14) | 300 | ⊕⊕⊕⊝ | ||
| 583 per 1000 | 731 per 1000 | |||||
| Primary patency Follow up: 4 years | Medium risk population1 | OR 1.15 (0.74 to 1.78) | 363 | ⊕⊕⊕⊕ | Estimate effect based on two studies | |
| 633 per 1000 | 665 per 1000 | |||||
| Mortality within follow‐up Follow up: 12‐48 months | Medium risk population1 | OR 0.94 (0.71 to 1.25) | 961 | ⊕⊕⊕⊕ | ||
| 371 per 1000 | 357 per 1000 | |||||
| *The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). | ||||||
| GRADE Working Group grades of evidence | ||||||
| 1 Based on studies including both claudication and critical ischaemia participants; the assumed risk was calculated by the mean number of events in the control groups of the selected studies for each outcome. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Early postoperative non‐thrombotic complications ‐ by symptoms at time of intervention Show forest plot | 6 | 1015 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.29 [0.96, 1.73] |
| 1.1 Intermittent claudication | 2 | 95 | Odds Ratio (M‐H, Fixed, 95% CI) | 2.15 [0.50, 9.21] |
| 1.2 Critical limb ischaemia | 2 | 513 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.57 [1.09, 2.24] |
| 1.3 Combined intermittent claudication and critical limb ischaemia | 3 | 407 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.71 [0.40, 1.28] |
| 2 Early postoperative non‐thrombotic complications ‐ by site of intervention Show forest plot | 5 | 905 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.23 [0.90, 1.67] |
| 2.1 Femoro‐popliteal disease | 5 | 748 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.34 [0.97, 1.86] |
| 2.2 Iliac disease | 1 | 157 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.62 [0.24, 1.58] |
| 3 Procedural mortality Show forest plot | 5 | 913 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.67 [0.66, 4.19] |
| 4 Clinical improvement Show forest plot | 2 | 154 | Odds Ratio (M‐H, Random, 95% CI) | 0.65 [0.03, 14.52] |
| 5 Amputation ‐ by symptoms at time of intervention Show forest plot | 5 | 752 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.24 [0.82, 1.87] |
| 5.1 Intermittent claudication | 2 | 95 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.40 [0.04, 4.02] |
| 5.2 Critical limb ischaemia | 2 | 513 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.14 [0.73, 1.77] |
| 5.3 Combined intermittent claudication and critical limb ischaemia | 2 | 144 | Odds Ratio (M‐H, Fixed, 95% CI) | 5.59 [0.94, 33.33] |
| 6 Primary patency at 1 year Show forest plot | 4 | 300 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.94 [1.20, 3.14] |
| 7 Primary patency at 4 years ‐ by symptoms at time of intervention Show forest plot | 2 | 363 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.15 [0.74, 1.78] |
| 7.1 Intermittent claudication | 1 | 191 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.44 [0.77, 2.69] |
| 7.2 Critical limb ischaemia | 1 | 72 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.95 [0.37, 2.43] |
| 7.3 Combined intermittent claudication and critical limb ischaemia | 1 | 100 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.92 [0.41, 2.04] |
| 8 Primary patency at 4 years ‐ by site of intervention Show forest plot | 1 | 263 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.24 [0.74, 2.08] |
| 8.1 Femoro‐popliteal disease | 1 | 100 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.91 [0.41, 2.01] |
| 8.2 Iliac disease | 1 | 163 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.57 [0.78, 3.14] |
| 9 Mortality within follow‐up ‐ by symptoms at time of intervention Show forest plot | 5 | 961 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.94 [0.71, 1.25] |
| 9.1 Intermittent claudication | 1 | 41 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.41 [0.02, 10.55] |
| 9.2 Critical limb ischaemia | 2 | 513 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.77 [0.54, 1.11] |
| 9.3 Combined intermittent claudication and critical limb ischaemia | 3 | 407 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.34 [0.84, 2.16] |
| 10 Technical success Show forest plot | 5 | 913 | Odds Ratio (M‐H, Fixed, 95% CI) | 2.26 [1.49, 3.44] |
| 11 Assisted primary patency at 1 year Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 12 Secondary patency at 1 year Show forest plot | 3 | 246 | Odds Ratio (M‐H, Random, 95% CI) | 1.61 [0.34, 7.50] |
| 13 Secondary patency at 4 years Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 14 Vessel or graft occlusion Show forest plot | 2 | 154 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.56 [0.27, 1.15] |
| 15 Reinterventions Show forest plot | 3 | 256 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.76 [0.42, 1.37] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Early postoperative non‐thrombotic complications Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 2 Amputation Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 3 Primary patency at 3 years Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 4 Mortality within follow‐up Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 5 Technical success Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 6 Assisted primary patency at 3 years Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 7 Secondary patency at 3 years Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 8 Vessel or graft occlusion Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Early postoperative non‐thrombotic complications ‐ by site of intervention Show forest plot | 1 | 237 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.66 [0.34, 1.31] |
| 1.1 Femoro‐popliteal disease | 1 | 168 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.84 [0.40, 1.76] |
| 1.2 Ilio‐femoral disease | 1 | 69 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.19 [0.02, 1.58] |
| 2 Procedural mortality ‐ by site of intervention Show forest plot | 1 | 237 | Odds Ratio (M‐H, Fixed, 95% CI) | 2.09 [0.67, 6.44] |
| 2.1 Femoro‐popliteal disease | 1 | 168 | Odds Ratio (M‐H, Fixed, 95% CI) | 2.59 [0.70, 9.55] |
| 2.2 Ilio‐femoral disease | 1 | 69 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.0 [0.09, 11.64] |
| 3 Amputation ‐ by symptoms at time of intervention Show forest plot | 1 | 236 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.10 [0.01, 0.80] |
| 3.1 Intermittent claudication | 1 | 80 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.44 [0.02, 11.12] |
| 3.2 Critical limb ischaemia | 1 | 156 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.06 [0.00, 1.02] |
| 4 Amputation ‐ by site of intervention Show forest plot | 1 | 237 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.10 [0.01, 0.72] |
| 4.1 Femoro‐popliteal disease | 1 | 168 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.05 [0.00, 0.83] |
| 4.2 Ilio‐femoral disease | 1 | 69 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.65 [0.03, 16.46] |
| 5 Mortality within follow‐up ‐ by symptoms at time of intervention Show forest plot | 1 | 236 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.56 [0.71, 3.44] |
| 5.1 Intermittent claudication | 1 | 80 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.02 [0.21, 4.87] |
| 5.2 Critical limb ischaemia | 1 | 156 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.83 [0.73, 4.56] |
| 6 Mortality within follow‐up ‐ by site of intervention Show forest plot | 1 | 237 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.45 [0.66, 3.18] |
| 6.1 Femoro‐popliteal disease | 1 | 168 | Odds Ratio (M‐H, Fixed, 95% CI) | 1.95 [0.80, 4.73] |
| 6.2 Ilio‐femoral disease | 1 | 69 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.37 [0.04, 3.39] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Procedural mortality Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 2 Amputation Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 3 Mortality within follow‐up Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 4 Technical success Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Early postoperative non‐thrombotic complications Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 2 Procedural mortality Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 3 Mortality within follow‐up Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 4 Reintervention within follow‐up Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 5 Maximal walking time (minutes) Show forest plot | 1 | Mean Difference (IV, Fixed, 95% CI) | Totals not selected | |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1 Therapeutic success: poor result of intervention Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
| 2 Amputation Show forest plot | 1 | Odds Ratio (M‐H, Fixed, 95% CI) | Totals not selected | |
