Retroperitoneal versus transperitoneal approach for elective open abdominal aortic aneurysm repair
Abstract
Background
There has been extensive debate in the surgical literature regarding the optimum surgical access approach to the infrarenal abdominal aorta during an operation to repair an abdominal aortic aneurysm. The published trials comparing retroperitoneal (RP) and transperitoneal (TP) aortic surgery show conflicting results. This is an update of the review first published in 2016.
Objectives
To assess the effectiveness and safety of the retroperitoneal versus transperitoneal approach for elective open abdominal aortic aneurysm repair on mortality, complications, hospital stay and blood loss.
Search methods
The Cochrane Vascular Information Specialist searched the Cochrane Vascular Specialised Register, CENTRAL, MEDLINE, Embase and CINAHL databases and the World Health Organization International Clinical Trials Registry Platform and the ClinicalTrials.gov trials registers to 30 November 2020. The review authors searched the Chinese Biomedical Literature Database and handsearched reference lists of relevant articles to identify additional trials.
Selection criteria
We included randomized controlled trials (RCTs) that assessed the RP approach versus the TP approach for elective open abdominal aortic aneurysm (AAA) repair. There were no restrictions on language or publication status.
Data collection and analysis
Two review authors independently extracted data from the included trials. We resolved any disagreements through discussion with a third review author. Two review authors independently assessed the risk of bias in included trials with the Cochrane risk of bias tool. For dichotomous outcomes, we calculated the odds ratio (OR) with the corresponding 95% confidence interval (CI). For continuous data, we calculated a pooled estimate of treatment effect by calculating the mean difference (MD) and standard deviation (SD) with corresponding 95% CIs. We pooled data using a fixed‐effect model, unless we identified heterogeneity, in which case we used a random‐effects model. We used GRADE to assess the overall certainty of the evidence. We evaluated the outcomes of mortality, complications, intensive care unit (ICU) stay, hospital stay, blood loss, aortic cross‐clamp time and operating time.
Main results
We identified no new studies from the updated searches. After reassessment, we included one study which had previously been excluded. Five RCTs with a combined total of 152 participants are included. The overall certainty of the evidence ranged from low to very low because of the low methodological quality of the included trials (unclear random sequence generation method and allocation concealment, and no blinding of outcome assessors), small sample sizes, small number of events, high heterogeneity and inconsistency between the included trials, no power calculations and relatively short follow‐up.
There was no evidence of a difference between the RP approach and the TP approach regarding mortality (odds ratio (OR) 0.32, 95% CI 0.01 to 8.25; 3 studies, 110 participants; very low‐certainty evidence). Similarly, there was no evidence of a difference in complications such as hematoma (OR 0.90, 95% CI 0.13 to 6.48; 2 studies, 75 participants; very low‐certainty evidence), abdominal wall hernia (OR 10.76, 95% CI 0.55 to 211.78; 1 study, 48 participants; very low‐certainty evidence), or chronic wound pain (OR 2.20, 95% CI 0.36 to 13.34; 1 study, 48 participants; very low‐certainty evidence) between the RP and TP approaches in participants undergoing elective open AAA repair. The RP approach may reduce ICU stay (mean difference (MD) ‐19.02 hours, 95% CI ‐30.83 to ‐7.21; 3 studies, 106 participants; low‐certainty evidence); hospital stay (MD ‐3.30 days, 95% CI ‐4.85 to‐1.75; 5 studies, 152 participants; low‐certainty evidence); and blood loss (MD ‐504.87 mL, 95% CI ‐779.19 to ‐230.56; 4 studies, 129 participants; very low‐certainty evidence). There was no evidence of a difference between the RP approach and the TP approach regarding aortic cross‐clamp time (MD 0.69 min, 95% CI ‐7.23 to 8.60; 4 studies, 129 participants; very low‐certainty evidence) or operating time (MD ‐15.94 min, 95% CI ‐34.76 to 2.88; 4 studies, 129 participants; very low‐certainty evidence).
Authors' conclusions
Very low‐certainty evidence from five small RCTs showed no clear evidence of a difference between the RP approach and the TP approach for elective open AAA repair in terms of mortality, or for rates of complications including hematoma (very low‐certainty evidence), abdominal wall hernia (very low‐certainty evidence), or chronic wound pain (very low‐certainty evidence). However, a shorter intensive care unit (ICU) stay and shorter hospital stay was probably indicated following the RP approach compared to the TP approach (both low‐certainty evidence). A possible reduction in blood loss was also shown after the RP approach (very low‐certainty evidence). There is no clear difference between the RP approach and TP approach in aortic cross‐clamp time or operating time. Further well‐designed, large‐scale RCTs assessing the RP approach versus TP approach for elective open AAA repair are required.
PICOs
Plain language summary
Different surgical approaches to access the infrarenal abdominal aorta during an operation to repair an abdominal aortic aneurysm
Background
There has been a lot of debate in the surgical literature about the best way to surgically access the infrarenal abdominal aorta during an operation to repair an abdominal aortic aneurysm (AAA: a ballooning of an artery (blood vessel) which occurs in the major artery in the abdomen (aorta)). Two approaches are commonly used: the retroperitoneal (RP) approach and the transperitoneal (TP) approach. Both approaches appear to have advantages and disadvantages. Many trials comparing RP and TP aortic surgery have been published, with conflicting results. The aim of this Cochrane Review is to assess the effectiveness and safety of the RP versus TP approach for planned surgical open AAA repair, taking into account mortality, complications, hospital stay and blood loss. This is an update of the review originally published in 2016.
Key results
We found no new studies from the updated searches. After reassessment, we included one study which had previously been excluded. Therefore, the review includes five small randomized controlled trials (RCTs), including 152 participants. The evidence in this Cochrane Review is current to 30 November 2020. There were no clear differences between RP and TP for the outcome of death. Similarly, there was no clear evidence that RP might increase complications such as hematoma (swelling of clotted blood), chronic wound pain and abdominal wall hernia compared with TP, but there were variations between the included trials. We found that RP might result in shorter hospital stay and intensive care unit (ICU) stay and less blood loss compared with TP. There were no clear differences between the two approaches for operating time and aortic cross‐clamp time (length of time that a surgical instrument, used to clamp the aorta and separate the circulation from the outflow of the heart, is used).
Reliability of the evidence
Four of the five included trials had methodological weaknesses ‐ such as unclear randomization methods, and no reporting of blinding of the people assessing the outcome ‐ which compromised the value of their results. In addition, the included trials only included a small number of people, few outcomes were reported, participants were followed up for a relatively short time, and there were inconsistencies between the included trials, resulting in evidence of very low to low certainty. More large‐scale RCTs of the RP approach versus the TP approach for planned surgical open AAA repair are needed.
Authors' conclusions
Summary of findings
| Retroperitoneal (RP) approach compared with transperitoneal (TP) approach for AAA repair | ||||||
| Participant or population: people with AAA Settings: hospital Intervention: RP approach Comparison: TP approach | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with TP approach | Risk with RP approach | |||||
| Mortality Follow‐up: up to 30 days | Study population | OR 0.32 (0.01 to 8.25) | 110 | ⊕⊝⊝⊝ | No deaths reported in two of the three RCTs in this comparison. There were no clear differences in mortality detected between the groups. | |
| 17 per 1000 | 5 per 1000 | |||||
| Complications: hematoma Follow‐up: 30 days post operation | Study population | OR 0.90 (0.13 to 6.48) | 75 | ⊕⊝⊝⊝ | There was no clear difference in hematoma detection between the groups. | |
| 28 per 1000 | 25 per 1000 | |||||
| Medium risk population | ||||||
| 42 per 1000 | 38 per 1000 | |||||
| Complications: abdominal wall hernia Follow‐up: 30 days post operation | Study population | OR 10.76 (0.55 to 211.78) | 48 | ⊕⊝⊝⊝ | No abdominal wall hernia complications reported in the TP approach group. There were no clear differences in abdominal wall hernia detected between the groups. | |
| See comment | See comment | |||||
| Complications: chronic wound pain Follow‐up: 30 days post operation | Study population | OR 2.20 (0.36 to 13.34) | 48 | ⊕⊝⊝⊝ | There were no clear differences in chronic wound pain detected between the groups. | |
| 83 per 1000 | 167 per 1000 | |||||
| ICU stay (hrs) recorded at end of ICU stay | The mean ICU stay ranged from | The mean ICU stay was 19.02 hrs lower (30.83 lower to 7.21 lower) | ‐ | 106 | ⊕⊕⊝⊝ | There may be a shorter ICU stay for participants in the RP approach group. |
| Hospital stay (days) recorded at end of hospital stay | The mean hospital stay ranged from | The mean hospital stay was 3.3 days lower (4.85 lower to 1.75 lower) | ‐ | 152 | ⊕⊕⊝⊝ | There may be a shorter hospital stay for participants in the RP approach group. |
| Blood loss (mL) recorded after operation | The mean blood loss ranged from | The mean blood loss was 504.87 mL lower (779.19 lower to 230.56 lower) | ‐ | 129 | ⊕⊝⊝⊝ | There may be reduced blood loss in the RP approach group compared to the TP approach group. |
| *The risk in the intervention group (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 | ||||||
| aWe downgraded the evidence by one level as random sequence generation was unclear in three of the four included trials (Darling 1992; Komori 1997; Laohapensang 2005a), and therefore these trials are at unclear risk of selection bias. In addition, blinding of outcome assessors was unclear in the included trials and therefore these trials are at risk of detection bias. | ||||||
| Retroperitoneal (RP) approach compared with transperitoneal (TP) approach for AAA repair | ||||||
| Participant or population: people with AAA Settings: hospital Intervention: RP approach Comparison: TP approach | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with TP approach | Risk with RP approach | |||||
| Aortic cross‐clamp time (minutes) recorded after operation | The mean aortic cross‐clamp time ranged from 56 to 68 minutes | The mean aortic cross‐clamp time was 0.69 minutes higher (7.23 lower to 8.60 higher) | ‐ | 129 | ⊕⊝⊝⊝ | There was no clear difference in aortic cross‐clamp time between the groups. |
| Operating time (minutes) recorded after operation | The mean operating time ranged from 160 to 258.1 minutes | The mean operating time was 15.94 minutes lower (34.76 lower to 2.88 higher) | ‐ | 129 | ⊕⊝⊝⊝ | There was no clear difference in operating time between the groups. |
| *The risk in the intervention group (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 | ||||||
| aWe downgraded the evidence by one level as random sequence generation was unclear in three of the four included trials (Darling 1992; Komori 1997; Laohapensang 2005a), and therefore these trials are at unclear risk of selection bias. In addition, blinding of outcome assessors was unclear in the included trials and therefore these trials are at risk of detection bias. | ||||||
Background
Description of the condition
Abdominal aortic aneurysm (AAA) is an abnormal enlargement in the diameter of a person's aorta. They are often asymptomatic until they rupture. National Institute for Health and Care Excellence (NICE) guidelines define AAA as an enlargement either 1.5 times the size of the normal aorta or a diameter greater than 3 cm (NICE 2020). The development of an AAA is multifactorial. Some of the established risk factors for the onset of AAA include age, male sex, white race, family history, atherosclerotic disease, and cigarette smoking, with the latter being considered the primary modifiable risk factor (Ullery 2018). Due to the reduction in tobacco smoking prevalence over time, population‐based studies have shown that the prevalence and incidence rates of AAA also declined (Sampson 2014). Reported AAA prevalence rates in 65‐year‐old men who attended the UK national AAA screening programme was 1.3% (Benson 2016). Untreated AAAs are likely to increase in size and rupture, eventually causing massive internal bleeding. Rupture of the abdominal aorta is the most serious complication, which presents as a surgical emergency. About one third of people with a ruptured abdominal aorta do not even reach hospital alive, giving an 80% overall mortality rate (Reimerink 2013). Data from the Global Burden of Disease showed that an average annual death rate of 2.8 per 100,000 (more than 20 million) is attributed to aortic aneurysm rupture (Golledge 2017; Sampson 2014). Regarding treatment for aortic aneurysm, elective open and endovascular surgical repairs are initially indicated in people with AAA to prevent death from rupture.
Description of the intervention
Dubost 1952 performed the first successful excision of an AAA via the retroperitoneal (RP) route in 1951. However, the RP approach, as reported by Oudot 1951, received little exposure. In the following years, most surgeons preferred to use the transperitoneal (TP) approach or transabdominal aortic replacement for open infrarenal AAA repair (Creech 1966). The RP approach was not forgotten. Rob 1963 wrote a detailed description of the RP approach, including its advantages, such as easier postoperative course, and its disadvantages, such as limited exposure. Williams 1980 reported an extended RP approach, which offers a better exposure, not only of the infrarenal aorta, but also of the pararenal and suprarenal aorta. Endovascular repair is now the standard method used to treat AAA, as described by a number of reviews and meta‐analyses comparing endovascular versus open surgery (Antoniou 2020; Bulder 2019), although these reviews do not address the different approaches for open surgery. The 30‐day mortality for elective endovascular aneurysm repair is lower than for open repair, but long‐term mortality has been shown to be similar (Jetty 2010). Because endovascular repair is associated with higher costs, open surgical repair is still an important method for treating people with AAAs.
How the intervention might work
The TP approach is most familiar to surgeons. It allows for easy access to the infrarenal aorta and iliac vessels, and at the same time, permits the surgical evaluation of the whole intra‐abdominal cavity to deal with concomitant surgical disease, such as colon carcinoma (Buck 2016). The inferior mesenteric artery could also be repaired, which can be incorporated in an infrarenal aortic graft. An equal TP aortic approach to the abdominal aorta can be attained through midline and transverse abdominal incisions. Lacy 1994 reported no statistically significant differences in morbidity through transverse and midline abdominal incisions, and therefore suggested that the type of incision used can be left to the surgeon's preference. However, the TP approach usually involves intestinal manipulation, mesenteric traction and blood contamination of the peritoneal cavity ‐ all of which may lead to impaired intestinal motility (Arya 2009). In order to avoid these complications, many doctors prefer to use the RP approach. Compared with the TP approach, the RP approach does not require opening the whole intra‐abdominal cavity. However, it is time‐consuming and would not be better for emergency cases (Nevelsteen 2005). Although one study reported that respiratory function after aortic aneurysm repair was similar between the two groups (Volta 2003), other studies have reported that people who had AAA repair using the RP approach had fewer postoperative respiratory complications, reduced incidence of intestinal obstruction, reduced intubation time, and decreased hospital stay and costs (Helsby 1975; Leather 1989; Taheri 1983). Moreover, several similar prospective randomized studies of the two approaches for aortic surgery have been performed, with conflicting results (Cambria 1990; Sieunarine 1997; Taheri 1983).
Why it is important to do this review
This Cochrane Review analyses the available evidence to assess the advantages and disadvantages of the RP versus TP approach for open AAA repair. We hope that this will help inform decision‐making for healthcare professionals and their patients.
Objectives
To assess the effectiveness and safety of the retroperitoneal versus transperitoneal approach for elective open abdominal aortic aneurysm repair on mortality, complications, hospital stay and blood loss.
Methods
Criteria for considering studies for this review
Types of studies
We included randomized controlled trials (RCTs). There was no restriction on language or publication status. We included all trials that met the inclusion criteria even if the trials did not report all of the pre‐specified outcomes of the review.
Types of participants
We included people who received elective open surgery for abdominal aortic aneurysm (AAA) (including with juxtarenal or pararenal aneurysm).
We excluded people with previous aortic repair (including previous laparotomy or previous endovascular stent graft) undergoing a redo aortic procedure or people undergoing emergency or urgent repair. Individuals who underwent aortoiliac or aortobifemoral bypasses for obstructive aortoiliac disease were not eligible for inclusion.
Types of interventions
We included RCTs comparing the retroperitoneal (RP) approach versus the transperitoneal (TP) approach for elective open AAA repair.
Types of outcome measures
Primary outcomes
-
Mortality: we analyzed in‐hospital mortality, 30‐day and late mortality separately.
-
Complications: we included hematoma, abdominal wall hernia and chronic wound pain.
Secondary outcomes
-
Intensive care unit (ICU) or high dependency unit (HDU) stay: all participants are normally initially managed in the HDU or the ICU and transferred to the vascular surgery ward when deemed appropriate. We defined ICU or HDU stay as the time of participant stay in the ICU or HDU.
-
Hospital stay: we defined hospital stay as from the day of operation to the day when the participant left hospital.
-
Blood loss: we defined blood loss as the total amount of blood obtained from suction, cell saver and weighed swabs.
-
Aortic cross‐clamp time
-
Operating time: we defined operating times as from the start of incision to the time of closure.
Search methods for identification of studies
Electronic searches
For this update, the Cochrane Vascular Information Specialist conducted systematic searches of the following databases for relevant trials without language, publication year or publication status restrictions.
-
Cochrane Vascular Specialised Register via the Cochrane Register of Studies (CRS‐Web searched on 30 November 2020).
-
Cochrane Central Register of Controlled Trials (CENTRAL) via the Cochrane Register of Studies Online (CRSO 2020, Issue 10).
-
MEDLINE (Ovid MEDLINE Epub Ahead of Print, In‐Process & Other Non‐Indexed Citations, Ovid MEDLINE Daily and Ovid MEDLINE) (searched from 11 May 2015 to 30 November 2020).
-
Embase Ovid (searched from 11 May 2015 to 30 November 2020).
-
CINAHL Ebsco (Cumulative Index to Nursing and Allied Health Literature; searched from 11 May 2015 to 30 November 2020).
The Information Specialist modelled search strategies for other databases on the search strategy designed for CENTRAL. Where appropriate, the search strategies were combined with adaptations of the highly sensitive search strategy designed by the Cochrane Collaboration for identifying randomized controlled trials and controlled clinical trials (as described in the Cochrane Handbook for Systematic Reviews of Interventions Chapter 6, Lefebvre 2011, hereafter referred to as the Cochrane Handbook). Search strategies for major databases are provided in Appendix 1.
The Information Specialist searched the following trials registries on 30 November 2020.
-
World Health Organization (WHO) International Clinical Trials Registry Platform (ICTRP) (apps.who.int/trialsearch/).
-
ClinicalTrials.gov (clinicaltrials.gov/).
Authors' searches
We searched the China BioMedical (CBM) Literature database (searched from inception to 30 November 2020) using the search strategies shown in Appendix 1. We based the CBM search on the search terms used in the MEDLINE search strategy.
Searching other resources
We checked the reference lists of retrieved articles and narrative and systematic reviews to find additional potentially relevant studies for inclusion.
Data collection and analysis
Selection of studies
For the update of this review, two authors (FM and FC) initially screened abstracts and titles using Covidence software, following the pre‐determined eligibility criteria to discard studies that were not applicable. If we could not decide whether or not the articles satisfied the inclusion criteria based on the abstracts, we obtained the full‐text articles of the studies. If there were two or more publications relating to one trial, we selected one report as the source of the study results (primary report) after checking the relevant information. If baseline information and additional results data for a trial were reported separately in multiple publications, we documented them all and extracted outcome data and underlying information of interest separately. We resolved any disagreements through discussion and, if necessary, by involving a third review author (BM).
Data extraction and management
Two review authors (BZ and KYH) independently extracted data using a data extraction form designed by Cochrane Vascular. We resolved any disagreements through discussion with a third review author (BM). We sought additional information from the authors of included trials if required. Where trial authors presented median and interquartile range (IQR) values and we were unable to obtain additional information from the trial authors, we assumed the median to be the mean and calculated the standard deviations (SDs) from the IQR using the formula IQR/1.35, in accordance with Higgins 2011. We did this for Arya 2009 and the outcomes of ICU or HDU stay, hospital stay, blood loss, aortic cross‐clamp time and operating time.
Assessment of risk of bias in included studies
For this update, two review authors (FM and LZ) independently reviewed or evaluated the risk of bias for each included trial using the Cochrane Collaboration's tool for assessing risk of bias (Higgins 2011), and based on the following domains.
-
Random sequence generation (selection bias).
-
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).
-
Other sources of bias.
We evaluated each criterion as high risk of bias, low risk of bias, or unclear risk of bias, in accordance with Higgins 2011. We resolved discrepancies through discussion until we reached consensus.
Regarding blinding, this cannot be implemented for surgeons in a trial about surgical operation. In addition, we believe that unblinded participants will not affect the measurement outcomes of our review, such as mortality, ICU stay, hospital stay, blood loss, aortic cross‐clamp time and operating time. Therefore, we mainly evaluated the implementation of blinding for outcome assessors in the subjective outcomes in our review, such as hematoma, abdominal wall hernia and chronic wound pain.
Measures of treatment effect
For dichotomous outcomes, we calculated a pooled estimate of the treatment effect for each outcome across trials as odds ratio (OR) with 95% confidence intervals (CIs). For continuous data, we calculated a pooled estimate of treatment effect by calculating the mean difference (MD) and standard deviation (SD) with corresponding 95% CIs. We used Review Manager software to calculate the measures of treatment effect (Review Manager).
Unit of analysis issues
We did not intend to include non‐standard designs, such as cross‐over trials and cluster‐RCTs, in this Cochrane Review. We considered each participant as an individual unit of analysis.
Dealing with missing data
We contacted the authors of the included trials via email for clarification regarding any missing data. We planned to undertake sensitivity analyses to assess the impact of missing data on the quality of the included trials when necessary (see the Sensitivity analysis section).
Assessment of heterogeneity
We used the Chi² test on N‐1 degrees of freedom with a significance level of P less than 0.05, and the I² statistic to examine the heterogeneity among trials. A guide to interpretation is as follows, as described in the Cochrane Handbook (Higgins 2011). I² statistic values of 25%, 50% and 75% correspond to low, moderate and high levels of heterogeneity, respectively. If the I² statistic estimate is greater than 50%, we regarded the level of heterogeneity among trials as moderate or high, and we investigated the potential reasons for heterogeneity. We considered clinical, methodological and statistical heterogeneity. For clinical heterogeneity, we studied the participants, interventions and outcomes. For methodological and statistical heterogeneity, we conducted sensitivity analyses as described below (Sensitivity analysis). We presented results separately and attempted to report the reasons if heterogeneity persisted.
Assessment of reporting biases
We planned to construct a funnel plot to investigate the likelihood of potential publication bias if we had included more than 10 trials in a meta‐analysis (Higgins 2011). As only five studies were included in this review, we could not undertake an assessment of reporting bias.
Data synthesis
We analyzed the data using Review Manager. We used a fixed‐effect model when there was no or low heterogeneity and a random‐effects model if there was high heterogeneity in the meta‐analysis because of the likely heterogeneous population. When the I² statistic values were greater than 75%, we used a random‐effects model. For levels smaller than the cut‐off point (I² statistic greater than 75%), we used a fixed‐effect model.
Subgroup analysis and investigation of heterogeneity
We were unable to carry out any subgroup analysis because of the difference in reporting of outcomes. If there were an adequate number of trials to merit subgroup analysis, we intended to perform subgroup analysis according to the following subgroups.
-
Age (≥ 65 years old versus < 65 years old).
-
Gender (female versus male).
-
Diabetes (with diabetes versus without diabetes).
-
Duration of treatment.
Sensitivity analysis
We undertook sensitivity analyses by analyzing the following categories of trials separately.
-
Trials with and without adequate randomization and concealment of treatment allocation.
-
Trials with and without intention‐to‐treat (ITT) analysis.
-
Trials with a dropout rate of more than 20% and less than 20%.
For sensitivity analyses, we described both the main effects within strata and a coefficient (and 95% CIs) describing the interaction between them.
Summary of findings and assessment of the certainty of the evidence
We created a summary of findings table for the comparison of the RP approach versus the TP approach for elective open AAA repair, using GRADEpro GDT software (GRADEpro GDT). We reported the following outcomes: mortality, complications (hematoma, abdominal wall hernia and chronic wound pain), ICU stay, hospital stay and blood loss (summary of findings Table 1). We created a second summary of findings table for the additional secondary outcomes of aortic cross‐clamp time and operating time (summary of findings Table 2). We downgraded the evidence from high certainty to moderate, low or very low certainty for serious or very serious study limitations (risk of bias), indirectness and inconsistency of evidence, imprecision of effect estimates or potential publication bias using the guidance developed by the Cochrane Handbook (Higgins 2011), and the GRADE working group (Atkins 2004).
Results
Description of studies
Results of the search
We identified no new studies for inclusion from the updated searches. We excluded two new studies (Deery 2019; Teixeira 2016), and two studies are awaiting classification (Arya 2010; Malek 1996). Upon reassessment, we included one study which was previously excluded (Volta 2003). One study which was previously listed as an additional publication to an included study was reassessed as a different study and subsequently excluded (Laohapensang 2005b). See Figure 1 for details of the search results.
Study flow diagram
Included studies
See Characteristics of included studies for further details on the included studies.
We identified no new studies from the updated searches for this review update. Volta 2003 was previously assessed as excluded (outcomes of interest not reported), but reassessed as an included study for this update as it met our inclusion criteria. A total of five RCTS met the inclusion criteria (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003). These five RCTs included a total of 152 participants and were conducted in at least three different countries: Darling 1992 in the USA, Komori 1997 in Japan and Laohapensang 2005a in Thailand, while Arya 2009 and Volta 2003 did not state in which country the study was conducted. Of the five included trials, Arya 2009 and Volta 2003 did not report the AAA size. All five included trials compared the retroperitoneal (RP) and transperitoneal (TP) approach for repairing open abdominal aortic aneurysm (AAA).
In Arya 2009, participants in Group I were repaired via the RP approach, while participants in Groups II and III were repaired via the TP approach, with the bowel packed within the peritoneal cavity or exteriorized in a bowel bag, respectively. In Komori 1997, Group I comprised people without previous laparotomy who were treated by the TP approach, while Groups II and III comprised people with previous laparotomy, treated by RP and TP approach, respectively. However, only the participants of Groups II and III were randomly assigned. The other three included trials, Darling 1992, Laohapensang 2005a and Volta 2003, divided participants into two groups (TP group versus RP group).
Excluded studies
We excluded two new studies (Deery 2019; Teixeira 2016). One study which was previously assessed as an additional publication to included study Laohapensang 2005a, was reassessed as a different study and subsequently excluded for not having an eligible intervention (Laohapensang 2005b). Therefore, we excluded a total of six studies (Cambria 1990; Deery 2019; Laohapensang 2005b; Sicard 1995; Sieunarine 1997; Teixeira 2016). For three studies, we could not stratify or obtain AAA participant data from the study authors (Cambria 1990; Sicard 1995; Sieunarine 1997). Two studies did not have an eligible study design (Deery 2019; Teixeira 2016). See Characteristics of excluded studies.
Studies awaiting classification
We assessed two studies as awaiting classification (Arya 2010; Malek 1996). See Characteristics of studies awaiting classification.
Risk of bias in included studies
We assessed the risk of bias for each included study using Cochrane's risk of bias tool. See Figure 2 and Figure 3.
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies
Risk of bias summary: review authors' judgements about each risk of bias item for each included study
Allocation
All included trials were randomized controlled trials (RCTs) but in four trials, no details were available on the method of allocation other than a statement that the trial was 'randomized' (Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003). Four trials had unclear selection bias for random sequence generation because they did not adequately describe their generation method or it was unclear if the method would be truly random (Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003).
Only Arya 2009 described allocation concealment and used an envelope system so was at low risk of selection bias. The other included trials did not mention allocation concealment.
Blinding
Blinding can not be implemented for surgeons in a trial about surgical operation. In addition, unblinded participants also will not affect the outcomes of our review so we have assessed the studies as unclear risk of performance bias. We also evaluated the implementation of blinding for outcome assessors, especially for the subjective outcomes. Unfortunately, it is unclear whether the outcome assessors were blinded, based on the reporting of the included trials, so we judged all five studies to be at unclear risk of detection bias (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003).
Incomplete outcome data
Two included trials reported their incomplete outcome data (Arya 2009; Darling 1992), and three included trials had no losses to follow‐up and withdrawals (Komori 1997; Laohapensang 2005a; Volta 2003). All included trials therefore had low risk of attrition bias.
Selective reporting
According to the reports, all included trials presented all the outcomes that they had planned to present. However, we failed to find the protocols of the included trials. Therefore, we are unsure if there was selective reporting in the included trials so all trials were at unclear risk of reporting bias (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003).
Other potential sources of bias
None of the five included trials had power calculations, and the small sample size can lead to a reduction of the testing efficiency. Therefore, we judged all studies as having unclear risk of other bias.
Effects of interventions
See: Summary of findings 1 Retroperitoneal approach compared with transperitoneal approach for elective open AAA repair; Summary of findings 2 Retroperitoneal approach compared with transperitoneal approach for elective open AAA repair (additional secondary outcomes)
For Arya 2009, Group I participants had repair through the retroperitoneal (RP) approach. Group II and Group III participants had their aneurysms repaired through a midline transperitoneal (TP) approach, with the small bowel packed within the peritoneal cavity and the small intestine exteriorized into a plastic bowel bag, separately. We split the study data into two for pair‐wise comparisons. We avoided double‐counting in the overall comparison by halving the Group 1 sample size (Higgins 2011).
Primary outcomes
Mortality
We intended to present and analyze data by early and late mortality. However, the included trials did not report these mortality details and therefore we reported morality as a single variable.
Three trials, including 110 participants, evaluated mortality after treatment with the RP approach versus the TP approach (Arya 2009; Darling 1992; Laohapensang 2005a). We found no clear difference between the treatment groups (odds ratio (OR) 0.32, 95% CI 0.01 to 8.25; P = 0.49, I² = 0%; very low‐certainty evidence; Analysis 1.1).
Complications
Two trials, including 75 participants, evaluated hematoma (Darling 1992; Laohapensang 2005a). We found no clear difference between the treatment groups (OR 0.90, 95% CI 0.13 to 6.48; P = 0.92, I² = 14%; very low‐certainty evidence; Analysis 1.2).
One trial, Laohapensang 2005a, including 48 participants, evaluated abdominal wall hernia. We found no clear difference between the treatment groups (OR 10.76, 95% CI 0.55 to 211.78; P = 0.12; very low‐certainty evidence; Analysis 1.2).
This trial also evaluated chronic wound pain. We found no clear difference between the treatment groups (OR 2.20, 95% CI 0.36 to 13.34; P = 0.39; very low‐certainty evidence; Analysis 1.2).
Secondary outcomes
Intensive care unit or high dependency unit stay
Three trials, including 106 participants, evaluated intensive care unit (ICU) or high dependency unit (HDU) stay (Arya 2009; Laohapensang 2005a; Volta 2003). Results showed a shorter ICU or HDU stay following the RP approach compared to the TP approach (MD ‐19.02 hours, 95% CI ‐30.83 to ‐7.21; P = 0.002, I² = 0%; low‐certainty evidence; Analysis 1.3).
Hospital stay
Five trials, including 152 participants, evaluated hospital stay (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003). The results showed a shorter hospital stay following the RP approach compared to the TP approach (MD ‐3.3 days, 95% CI ‐4.85 to ‐1.75; P < 0.001; low‐certainty evidence). There was high heterogeneity (I² = 79%), so we used a random‐effects model (Analysis 1.4).
Blood loss
Four trials, including 129 participants, evaluated blood loss (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a). Results showed lower blood loss for the RP approach compared to the TP approach (MD ‐504.87 mL, 95% CI ‐779.19 to ‐230.56; P < 0.001; very low‐certainty evidence; Analysis 1.5). There was high heterogeneity (I² = 76%), so we used a random‐effects model (Analysis 1.5).
Aortic cross‐clamp time
Four trials, including 129 participants, evaluated aortic cross‐clamp time (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a). Results showed no clear difference between the treatment groups (MD 0.69 min, 95% CI ‐7.23 to 8.60; P = 0.86; very low‐certainty evidence). There was high heterogeneity (I² = 81%), so we used a random‐effects model (Analysis 1.6).
Operating time
Four trials, including 129 participants, evaluated operating time (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a). Results showed no clear difference between the treatment groups (MD ‐15.94 min, 95% CI ‐34.76 to 2.88; P = 0.10; very low‐certainty evidence). We found high heterogeneity (I² = 81%), so we used a random‐effects model (Analysis 1.7).
Sensitivity analysis
We planned to conduct sensitivity analyses by analyzing these categories of trials separately:
-
trials with and without adequate randomization and concealment of treatment allocation;
-
trials with and without intention‐to‐treat (ITT) analysis;
-
trials with a dropout rate of more than 20% and less than 20%.
No trials reported ITT analysis or a dropout rate of more than 20%. Therefore, we did not perform a sensitivity analysis for these categories of trials.
Of the five included RCTs, only one trial mentioned adequate randomization and allocation concealment (Arya 2009). Therefore, we carried out sensitivity analyses by excluding the other four included trials (Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003). Regarding mortality, Arya 2009 did not report any cases of death. Therefore, a test for overall effects was no longer applicable. For the outcome 'complications', we could not perform a sensitivity analysis as none of the trials included in this comparison had adequate randomization and allocation concealment. For hospital stay (days) and blood loss (mL), there were no longer differences between the techniques (hospital stay: MD ‐0.72 days, 95% CI ‐2.93 to 1.50; P = 0.53; Analysis 2.3; blood loss: MD ‐231.78 mL, 95% CI ‐1142.47 to 678.90; P = 0.62; Analysis 2.4). We did not observe any changes for ICU stay (hours) (Analysis 2.2); aortic cross‐clamping (Analysis 2.5); or operating time (Analysis 2.6).
Discussion
Summary of main results
Five RCTs which compared participants who underwent the retroperitoneal (RP) and transperitoneal (TP) approach for elective open abdominal aortic aneurysm (AAA) repair were identified and included in meta‐analyses (Arya 2009; Darling 1992; Komori 1997; Laohapensang 2005a; Volta 2003). Our analyses did not show any clear difference between the RP approach and the TP approach regarding mortality (very low‐certainty evidence), or any clear evidence of a difference for rates of complications such as hematoma (very low‐certainty evidence), abdominal wall hernia (very low‐certainty evidence), or chronic wound pain (very low‐certainty evidence). However, a shorter intensive care unit (ICU) stay and shorter hospital stay was probably indicated following the RP approach compared to the TP approach (both low‐certainty evidence). A possible reduction in blood loss was also shown after the RP approach compared to the TP approach (very low‐certainty evidence). There were no clear differences between the RP approach and the TP approach regarding aortic cross‐clamp time or operating time (both very low‐certainty evidence). See summary of findings Table 1 and summary of findings Table 2.
Overall completeness and applicability of evidence
Given the thorough search strategy and clear inclusion and exclusion criteria, this review provides a comprehensive overview of the current evidence comparing the RP and TP approach in people undergoing AAA repair. We found five RCTs that investigated this topic. While most studies appropriately reported the outcomes of interests, the data were insufficient to perform pre‐specified subgroup analyses exploring whether the approach effect differed importantly according to key participant characteristics. Notably, the time points of mortality and the calculation of sample size reported in the included studies were generally poorly described. Between studies, moreover, there was a degree of clinical heterogeneity among individual surgeons and units. Therefore, we have limited confidence in generalizing these results to other situations.
We found that no further studies have been done since the latest Arya study was published in 2009 (Arya 2009). With advances in minimally‐invasive techniques for the treatment of aneurysmal aortoiliac disease, open procedures have become fewer and more technically demanding (Schermerhorn 2012). These procedures have been progressively reserved for younger people and those unable to comply with long‐term surveillance (Twine 2012). Future research regarding the management of AAA should focus on time points for assessing outcomes and key participant characteristics, as data regarding for example age, gender and diabetes are lacking. Such research can facilitate a meta‐analysis and create much clearer clinical guidance and prognostic information.
Quality of the evidence
Methodological weaknesses in the five included RCTs compromised the value of their results. The trial authors described each of their respective trials as randomized. However, only Arya 2009 reported their random sequence generation thoroughly. The other included trials had unclear selection bias for random sequence generation because they did not adequately describe their generation method, or it was unclear if the method would be truly random. There was no reporting of blinding of outcome assessors, leading to unclear detection bias. The sample sizes of the included trials were small, with two studies including fewer than 30 participants (Darling 1992; Komori 1997). In addition, the number of events was small, leading to imprecision in the effect estimates. Heterogeneity and inconsistency between the included RCTs was high for most outcomes. Due to the small number of included trials, we were unable to assess the reasons for inconsistency and heterogeneity via subgroup analyses. None of the included RCTs provided prior power calculations and all reported primary outcomes after a relatively short follow‐up period. Moreover, none of the included trials reported whether the qualification of the surgeon was required to ensure the relative consistency of operation quality among different surgeons. Thus, overall, the certainty of the available evidence was low to very low. See summary of findings Table 1 and summary of findings Table 2.
Potential biases in the review process
The Cochrane Vascular Information Specialist carried out a comprehensive search of the literature. To further expand the search, we checked the reference lists of relevant articles for additional studies. We contacted study authors to obtain data that were missing in the original publications or abstracts. Two review authors independently screened the titles and abstracts of references identified by the searches. However, this Cochrane Review included published data only. As a result, selective biases may exist in our review. In future updates of this Cochrane Review, we will attempt to identify any additional studies by further searching the grey literature. Some study results, such as the secondary outcomes of ICU or HDU stay, hospital stay, blood loss, aortic cross‐clamp time and operating time reported by Arya 2009, were not reported as mean and SD values but as median and IQR values. Assuming normal distribution, we took the median value to be the mean and calculated the SD according to Higgins 2011. Therefore, one should consider the bias of pooled effect.
Agreements and disagreements with other studies or reviews
The results of this Cochrane Review are consistent with other meta‐analyses, which showed that the RP approach could reduce the duration of ICU stay, but that it did not reduce the incidence of mortality and operation time compared with the TP approach in people undergoing elective open AAA repair (Twine 2013). A large retrospective cohort study by the Society for Vascular Surgery Vascular Quality Initiative (SVS‐VQI) found that people who underwent the TP approach experienced higher rates of repair‐related re‐interventions and readmissions (Deery 2019). Additional studies show that the RP approach could reduce the operative time (Borkon 2010). Additional studies also show that the RP approach has some postoperative advantages, including a faster recovery, quicker return to bowel function, fewer pulmonary complications, less pain and, potentially lower costs compared with the TP approach (Arko 2001; Ballard 2000; Hioki 2002; Johnson 1986; Leather 1989).
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies
Risk of bias summary: review authors' judgements about each risk of bias item for each included study
Comparison 1: Retroperitoneal versus transperitoneal approach, Outcome 1: Mortality
Comparison 1: Retroperitoneal versus transperitoneal approach, Outcome 2: Complications
Comparison 1: Retroperitoneal versus transperitoneal approach, Outcome 3: ICU stay (hrs)
Comparison 1: Retroperitoneal versus transperitoneal approach, Outcome 4: Hospital stay (days)
Comparison 1: Retroperitoneal versus transperitoneal approach, Outcome 5: Blood loss (mL)
Comparison 1: Retroperitoneal versus transperitoneal approach, Outcome 6: Aortic cross‐clamp time (mins)
Comparison 1: Retroperitoneal versus transperitoneal approach, Outcome 7: Operating time (mins)
Comparison 2: Sensitivity analysis of retroperitoneal versus transperitoneal approach, Outcome 1: Mortality
Comparison 2: Sensitivity analysis of retroperitoneal versus transperitoneal approach, Outcome 2: ICU stay (hrs)
Comparison 2: Sensitivity analysis of retroperitoneal versus transperitoneal approach, Outcome 3: Hospital stay (days)
Comparison 2: Sensitivity analysis of retroperitoneal versus transperitoneal approach, Outcome 4: Blood loss (mL)
Comparison 2: Sensitivity analysis of retroperitoneal versus transperitoneal approach, Outcome 5: Aortic cross‐clamp time (mins)
Comparison 2: Sensitivity analysis of retroperitoneal versus transperitoneal approach, Outcome 6: Operating time (mins)
| Retroperitoneal (RP) approach compared with transperitoneal (TP) approach for AAA repair | ||||||
| Participant or population: people with AAA Settings: hospital Intervention: RP approach Comparison: TP approach | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with TP approach | Risk with RP approach | |||||
| Mortality Follow‐up: up to 30 days | Study population | OR 0.32 (0.01 to 8.25) | 110 | ⊕⊝⊝⊝ | No deaths reported in two of the three RCTs in this comparison. There were no clear differences in mortality detected between the groups. | |
| 17 per 1000 | 5 per 1000 | |||||
| Complications: hematoma Follow‐up: 30 days post operation | Study population | OR 0.90 (0.13 to 6.48) | 75 | ⊕⊝⊝⊝ | There was no clear difference in hematoma detection between the groups. | |
| 28 per 1000 | 25 per 1000 | |||||
| Medium risk population | ||||||
| 42 per 1000 | 38 per 1000 | |||||
| Complications: abdominal wall hernia Follow‐up: 30 days post operation | Study population | OR 10.76 (0.55 to 211.78) | 48 | ⊕⊝⊝⊝ | No abdominal wall hernia complications reported in the TP approach group. There were no clear differences in abdominal wall hernia detected between the groups. | |
| See comment | See comment | |||||
| Complications: chronic wound pain Follow‐up: 30 days post operation | Study population | OR 2.20 (0.36 to 13.34) | 48 | ⊕⊝⊝⊝ | There were no clear differences in chronic wound pain detected between the groups. | |
| 83 per 1000 | 167 per 1000 | |||||
| ICU stay (hrs) recorded at end of ICU stay | The mean ICU stay ranged from | The mean ICU stay was 19.02 hrs lower (30.83 lower to 7.21 lower) | ‐ | 106 | ⊕⊕⊝⊝ | There may be a shorter ICU stay for participants in the RP approach group. |
| Hospital stay (days) recorded at end of hospital stay | The mean hospital stay ranged from | The mean hospital stay was 3.3 days lower (4.85 lower to 1.75 lower) | ‐ | 152 | ⊕⊕⊝⊝ | There may be a shorter hospital stay for participants in the RP approach group. |
| Blood loss (mL) recorded after operation | The mean blood loss ranged from | The mean blood loss was 504.87 mL lower (779.19 lower to 230.56 lower) | ‐ | 129 | ⊕⊝⊝⊝ | There may be reduced blood loss in the RP approach group compared to the TP approach group. |
| *The risk in the intervention group (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 | ||||||
| aWe downgraded the evidence by one level as random sequence generation was unclear in three of the four included trials (Darling 1992; Komori 1997; Laohapensang 2005a), and therefore these trials are at unclear risk of selection bias. In addition, blinding of outcome assessors was unclear in the included trials and therefore these trials are at risk of detection bias. | ||||||
| Retroperitoneal (RP) approach compared with transperitoneal (TP) approach for AAA repair | ||||||
| Participant or population: people with AAA Settings: hospital Intervention: RP approach Comparison: TP approach | ||||||
| Outcomes | Anticipated absolute effects* (95% CI) | Relative effect | No. of participants | Certainty of the evidence | Comments | |
|---|---|---|---|---|---|---|
| Risk with TP approach | Risk with RP approach | |||||
| Aortic cross‐clamp time (minutes) recorded after operation | The mean aortic cross‐clamp time ranged from 56 to 68 minutes | The mean aortic cross‐clamp time was 0.69 minutes higher (7.23 lower to 8.60 higher) | ‐ | 129 | ⊕⊝⊝⊝ | There was no clear difference in aortic cross‐clamp time between the groups. |
| Operating time (minutes) recorded after operation | The mean operating time ranged from 160 to 258.1 minutes | The mean operating time was 15.94 minutes lower (34.76 lower to 2.88 higher) | ‐ | 129 | ⊕⊝⊝⊝ | There was no clear difference in operating time between the groups. |
| *The risk in the intervention group (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 | ||||||
| aWe downgraded the evidence by one level as random sequence generation was unclear in three of the four included trials (Darling 1992; Komori 1997; Laohapensang 2005a), and therefore these trials are at unclear risk of selection bias. In addition, blinding of outcome assessors was unclear in the included trials and therefore these trials are at risk of detection bias. | ||||||
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 1.1 Mortality Show forest plot | 3 | 110 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.32 [0.01, 8.25] |
| 1.2 Complications Show forest plot | 2 | Odds Ratio (M‐H, Fixed, 95% CI) | Subtotals only | |
| 1.2.1 hematoma | 2 | 75 | Odds Ratio (M‐H, Fixed, 95% CI) | 0.90 [0.13, 6.48] |
| 1.2.2 abdominal wall hernia | 1 | 48 | Odds Ratio (M‐H, Fixed, 95% CI) | 10.76 [0.55, 211.78] |
| 1.2.3 chronic wound pain | 1 | 48 | Odds Ratio (M‐H, Fixed, 95% CI) | 2.20 [0.36, 13.34] |
| 1.3 ICU stay (hrs) Show forest plot | 3 | 106 | Mean Difference (IV, Fixed, 95% CI) | ‐19.02 [‐30.83, ‐7.21] |
| 1.4 Hospital stay (days) Show forest plot | 5 | 152 | Mean Difference (IV, Random, 95% CI) | ‐3.30 [‐4.85, ‐1.75] |
| 1.5 Blood loss (mL) Show forest plot | 4 | 129 | Mean Difference (IV, Random, 95% CI) | ‐504.87 [‐779.19, ‐230.56] |
| 1.6 Aortic cross‐clamp time (mins) Show forest plot | 4 | 129 | Mean Difference (IV, Random, 95% CI) | 0.69 [‐7.23, 8.60] |
| 1.7 Operating time (mins) Show forest plot | 4 | 129 | Mean Difference (IV, Random, 95% CI) | ‐15.94 [‐34.76, 2.88] |
| Outcome or subgroup title | No. of studies | No. of participants | Statistical method | Effect size |
| 2.1 Mortality Show forest plot | 1 | 35 | Odds Ratio (M‐H, Fixed, 95% CI) | Not estimable |
| 2.2 ICU stay (hrs) Show forest plot | 1 | 35 | Mean Difference (IV, Fixed, 95% CI) | ‐28.59 [‐47.70, ‐9.48] |
| 2.3 Hospital stay (days) Show forest plot | 1 | 35 | Mean Difference (IV, Fixed, 95% CI) | ‐0.72 [‐2.93, 1.50] |
| 2.4 Blood loss (mL) Show forest plot | 1 | 35 | Mean Difference (IV, Fixed, 95% CI) | ‐231.78 [‐1142.47, 678.90] |
| 2.5 Aortic cross‐clamp time (mins) Show forest plot | 1 | 35 | Mean Difference (IV, Fixed, 95% CI) | ‐3.73 [‐19.09, 11.63] |
| 2.6 Operating time (mins) Show forest plot | 1 | 35 | Mean Difference (IV, Fixed, 95% CI) | 5.87 [‐27.99, 39.73] |
