A Comparison of Open vs. Laparoscopic Abdominal Rectopexy for Full-Thickness Rectal Prolapse: A Meta-Analysis Sanjay Purkayastha, M.R.C.S., Paris Tekkis, M.D., F.R.C.S., Thanos Athanasiou, Ph.D., F.E.C.T.S., Omer Aziz, M.R.C.S., Paraskeva Paraskevas, Ph.D., F.R.C.S., Paul Ziprin, M.D., F.R.C.S., Ara Darzi, M.D., F.R.C.S., K.B.E. Department of Surgical Oncology and Technology, Imperial College, London, St. Mary’s Hospital, London, United Kingdom PURPOSE: Using meta-analytical techniques, this study was designed to compare open and laparoscopic abdominal procedures used to treat full-thickness rectal prolapse in adults. METHODS: Comparative studies published between 1995 and 2003, cited in the literature of open abdominal rectopexy vs. laparoscopic abdominal rectopexy, were used. The primary end points were recurrence and morbidity, and the secondary end points assessed were operative time and length of hospital stay. A random effect model was used to aggregate the studies reporting these outcomes, and heterogeneity was assessed. RESULTS: Six studies, consisting of a total of 195 patients (98 open and 97 laparoscopic) were included. Analysis of the data suggested that there is no significant difference in recurrence and morbidity between laparoscopic abdominal rectopexy and open abdominal rectopexy. Length of stay was significantly reduced in the laparoscopic group by 3.5 days (95 percent confidence interval, 3.1–4; P < 0.01), whereas the operative time was significantly longer in this group, by approximately 60 minutes (60.38 minutes; 95 percent confidence interval, 49– 71.8). CONCLUSIONS: Laparoscopic abdominal rectopexy is a safe and feasible procedure, which may compare equally with the open technique with regards to recurrence and morbidity and favorably with length of stay. However large-scale randomized trials, with comparative, sound methodology are still needed to ascertain detailed outcome
Correspondence to: Paris Tekkis, M.D., F.R.C.S., Department of Surgical Oncology and Technology, St. Mary’s Hospital, 10th Floor QEQM Building, Praed Street, W2 1NY, London, United Kingdom, e-mail:
[email protected] Dis Colon Rectum 2005; 48: 1930–1940 DOI: 10.1007/s10350-005-0077-x © The American Society of Colon and Rectal Surgeons Published online: 20 June 2005
measures accurately. [Key words: Rectal prolapse; Rectopexy; Laparoscopy; Meta-analysis]
F
ull-thickness rectal prolapse (FTRP) is the complete eversion of the rectum through the anal canal. FTRP may occur at any age, with an overall incidence of 4 per 1,000 population1; however, it is more common in females, especially in the elderly population, with a female to male distribution ranging from 10 to 6:1.2 The etiology of this condition still has not been elicited, but there are theories evaluated, including functional problems, such as constipation,3 anatomic problems, such as caliber differences between the rectum and sigmoid,4 and problems related to childbirth.5 Patients usually complain of an anal lump, which might require manual reduction. Bleeding, incontinence, and mucous per rectum also are common complaints, which are all distressing for the individual concerned. These described symptoms are similar to those of prolapsing hemorrhoids and large external skin tags, both of which must be excluded promptly as the differential diagnosis. There are many procedures described for the treatment of FTRP6; however, generally these can be divided into perineal and abdominal approaches. It is recognized that although in elderly, frail patients, the perineal approach has acceptable results, might avoid the necessity for a general anesthetic, and has a short hospital stay, abdominal rectopexy has a lower recurrence rate.7,8
1930
Vol. 48, No. 10
LAP VS. OPEN RECTOPEXY META-ANALYSIS
Laparoscopic approaches for the surgical treatment of rectal prolapse necessitate an abdominal rectopexy, with or without resection of the sigmoid colon. Therefore, this study has evaluated trials comparing open abdominal rectopexy (OAR) vs. laparoscopic abdominal rectopexy (LAR). The authors have previously used meta-analytic techniques to evaluate successfully the impact of minimally invasive procedures in routine practice.9,10 A previous Cochrane review11 has reported results on two comparative studies evaluating short-term and long-term outcomes in patients undergoing LAR or OAR, only one of which was a randomized, controlled trial. This study uses meta-analytical techniques to compare laparoscopic vs. open rectopexy in adults, with regards to morbidity, recurrence, operative time, and length of hospital stay. Other end points, such as opiate usage and return of bowel function, also were considered. The specific goals of our study were the following: 1. Are there significant differences in the abovementioned end points between the open and laparoscopic approach? 2. Is there significant heterogeneity in the estimates of the outcomes of interest between the studies comparing OAR and LAR? 3. Assessment of potential publication bias between the studies. 4. Assessment of potential discrepancy between randomized and observational evidence.
METHODS PUBMED was used as the search engine for a MEDLINE search of the literature for studies comparing OAR and LAR for FTRP. The following text searches and search headings were used: “rectopexy and comparative study,” “rectal prolapse and comparative study,” “rectopexy and laparoscopy,” “rectal prolapse and laparoscopy,” “laparoscopic vs. open rectopexy.” The “related articles” function was used to broaden the search, and all abstracts, studies, and citations scanned were reviewed. The Cochrane database was searched using the terms: “rectal prolapse” and “rectopexy.” The date of the most recent search was September 2004. Mesh search headings were “rectal prolapse and surgery.”
Data Extraction Two reviewers (SP and PT) independently performed the search and reviewed and extracted the
1931
following data from each study according to prespecified protocol: first author, year of publication, study demographics, study design, number of patients operated on with each technique, and end point data (operative time, length of hospital stay, adverse events, mortality, recurrence, opiate usage, and return of bowel function). The data were tabulated and entered into the relevant format in Review Manager 4.2 (The Cochrane Collaboration, Software Update, Oxford) in use for analysis.
Inclusion Criteria Comparative studies with complete data for LAR and OAR were used, including both randomized and nonrandomized data on the outcomes discussed below. Analysis was performed only on patients undergoing abdominal rectopexy for the treatment of FTRP.
Exclusion Criteria Noncomparative studies were excluded. All the studies excluded cases from their data if malignancy was involved. Because of variability between studies with regards to data reporting, this study focused on outcomes that had common definitions within the studies.
Outcomes of Interest and Definitions The primary outcomes evaluated were 1) recurrence of FTRP anytime during the follow-up period, and 2) morbidity. Secondary outcomes were 1) operative time, and 2) length of in-hospital stay. These were the only meta-analyzable outcomes because of numbers and methodology of data reporting between the studies. Other outcomes also were described in the comparative studies. These included opiate requirements in the early postoperative period, intraoperative blood loss (ml), operative mortality, constipation, and fecal incontinence rates.
Statistical Analysis For studies that presented continuous data as mean and range values, the standard deviation (SD) was calculated using statistical algorithms and checked using “bootstrap” resampling techniques. Thus, all continuous data were standardized for analysis. The effect measures estimated were odds ratio (OR) for dichotomous data and weighted mean difference for continuous data, both with 95 percent confidence intervals (CI). Odds ratio <1 favored the treatment group, and the point estimate of the odds ratio was
PURKAYASTHA ET AL
1932
Dis Colon Rectum, October 2005
considered statistically significant at the P < 0.05 level if the 95 percent confidence interval did not include the value 1. Aggregation of the overall rates of the outcomes of interest was performed with the MantelHaenszel chi-squared test. Studies that contained a zero in one cell for the number of events of interest in one of the two groups resulted in problems with the computation of ratio measurement and a value of 0.5 was added in both groups from that particular study. Three strategies were used to quantitatively assess heterogeneity. First, data was reanalyzed using both random and fixed effect models. Second, graphic exploration with funnel plots was used to evaluate publication bias.12,13 Third, sensitivity analysis was undertaken using subgroup analysis. To do this, the following variables were evaluated: 1) all studies, 2) prospective studies, 3) randomized studies, 4) excluding studies with resection rectopexy data, 5) excluding studies causing significant funnel plot asymmetry. In a “fixed effect” model, it is assumed that there is no heterogeneity in treatment effect between studies, whereas in a “random effect” model, it is assumed that there is variation between studies and the calculated odds ratios have a more conservative value.12,14 Although meta-analysis of randomized, controlled trials is preferable, random effect models can be used to aggregate results of studies with different methodology, focusing not only in the calculation of the overall effect, but in explaining heterogeneity and for performing sensitivity analysis. In surgical research, meta-analysis using the random effect model is preferable particularly because patients who are operated on in different centers have varying risk profiles and selection criteria for each surgical technique. Trials that represented significant asymmetry were excluded for particular specific end points. Data for the chosen outcomes were entered into Review Manager Version 4.2 (The Cochrane Collaboration, Software Update) in dichotomous or continuous formats for use in metaanalysis allowing the estimation of odds ratios and weighted mean difference as the respective summary statistics. Subsequently, confidence intervals, pooled estimates, and tests for heterogeneity were calculated, and visual evaluation of possible publication bias was performed by the use of funnel plots.
comparing LAR vs. OAR, and reported the incidence of operative time, length of stay, morbidity, and recurrence.15–20 Two of the studies were by the same author18,19; however, they were conducted during two completely separate time periods and hence both studies were included. These included two retrospective, two prospective, nonrandomized, one prospective, randomized study, and one prospective study with a retrospective control group. A combined total of 195 patients, of which 97 (49.7 percent) underwent LAR and 98 (50.3 percent) OAR, were included in the final analysis. On review of the data extraction, there was 100 percent agreement between the two reviewers. The demographic characteristics, matching, and exclusion criteria of these studies are demonstrated in Table 1. The age of patients ranged from 22 to 93 years. The majority of patients were female (91 percent). Four studies were matched for age, gender, and operative technique. The comparison of outcomes measured in the respective studies is shown in Table 2.
RESULTS
Five of the studies reported the incidence of recurrence; however, two studies reported “zero” in their recurrence rates18,20 in both the LAR and OAR groups and were excluded from the analysis. The three remaining studies were compared. Meta-analysis
Selected Studies Six comparative studies were published between 1995 and 2003 that matched the inclusion criteria,
Primary and Secondary Outcomes All six studies reported the incidence of morbidity, operative time, and length of stay, and five reported the incidence of recurrence. The results from metaanalysis of the studies with regards to morbidity, length of stay and operative time are summarized later (Table 3). Figure 1 demonstrates the outcome for meta-analysis for recurrence. Figures 2A and B represent the meta-analytic outcome for the overall morbidity and the corresponding funnel plot when all the studies were considered. Figures 3 and 4 represent results from meta-analysis of the continuous variables (length of stay and operative time), respectively. Other outcomes that were not meta-analyzable (Table 2) because of difference in data set measurement or lack of data across the studies were biochemical stress markers, time to first bowel movement, opiate requirements, blood loss measurements, time to toleration of oral fluids, time to toleration of normal diet, and constipation rates.
Recurrence
21 20 10
Prospective, nonrandomized Prospective, randomized Retrospective
24 19 8
10 13 26
Open
22–93 NVG 49–85
44–63 22–76 22–88
Age Range (yr)
5/39 NVG 0/18
NVG 1/22 3/49
Male/ Female Ratio 1,2 None 1,2,3,5,6,7, 8,9,10,11 1 1,2,3 1,2,3,4
Matching
A,B,C,D,E NVG
A A NVG
Exclusion Criteria
Stapled mesh Stapled mesh Resection and sutured
Resection only Sutured mesh Sutured only
Type of Rectopexy
NA 24 12
27 31 12
Follow-up (mo)
√ √ √ √ √ √
√ √ √ √ √ √
√
√
√
√
√
√
Length of Stay
√
√
√
√
√
Recurrence
√
√
√
Time to First Bowel Movement/ Flatus
√
√
Opiate Requirements
√
√
√
Blood Loss Measurements
√
Time to Toleration of Oral Fluids
√
Time to Toleration of Normal Diet
√
√
Constipation
√
√
√
Fecal Incontinence
√
√
√
√
√
√
Mortality
LAP VS. OPEN RECTOPEXY META-ANALYSIS
√ = Measured Outcome
Baker et al. (1995)15 Boccasanta et al. (1999)16 Kairaluoma et al. (2003)17 Solomon and Eyers (1996)18 Solomon et al. (2002)19 Xynos et al. (1999)20
Author (yr)
Morbidity
Operative Time
Biochemical Stress Markers
Table 2. Outcomes Measured by Included Studies
NVG = no values given; NA = not available. Matching: 1 = age; 2 = gender; 3 = surgical technique, 4 = parity; 5 = body mass index; 6 = ASA (American Society of Anesthesiologists) score; 7 = no. of previous operations; 8 = presence of solitary rectal ulcer; 9 = diverticular disease; 10 = incontinence; 11 = constipation. Exclusion criteria: A = malignancy; B = concomitant gynecologic procedure necessary; C = previous rectopexy performed; D = large irreducible prolapse; E = intractable constipation.
8 10 26
Retrospective Prospective, nonrandomized Prospective, nonrandomized
Baker et al. (1995)15 Boccasanta et al. (1999)16 Kairaluoma et al. (2003) no resection group17 Solomon and Eyers (1996)18 Solomon et al. (2002)19 Xynos et al.(1999)20
Laparoscopic
Study Type
Authors (yr)
Cases
Table 1. Comparison of Demographics and Methodology of Included Studies
Vol. 48, No. 10 1933
Random Effect Model
−2.97 (−4.30– −1.63) Not applicable Not applicable
−4.58 (−7.72– −1.44) 1.21 0.55
−2.92 (−4.11– −1.73) 3.62 0.06
−3.89 (−4.96– −2.81) 3.95 0.27
0.5 (0.25–0.97)
0.41 (0.17–1.02)
0.49 (0.22–1.12)
0.39 (0.06–2.41)
0.49 (0.22–1.12)
0.5 (0.25–0.97)
−2.87 (−3.98– −1.76) 0.66 0.72
Random Effect Model
−2.87 (−3.98– −1.76) 3.95 0.27
Fixed Effect Model 0.56 (0.24–1.29)c
Random Effect Model
Morbidity
−3.54 −3.54 0.55 (−4.01– −3.08)b (−4.01– −3.08)b (0.29–1.06)c 5.44c 0.66d 0.24c 0.72d
Fixed Effect Model
Length of Stay
1.07 (0.31–3.69)
0.3 (0.01–7.85)
0.91 (0.26–3.20)
0.91 (0.26–3.20)d
Fixed Effect Model
1.1 (0.3–4.02)
0.3 (0.01–7.85)
0.94 (0.26–3.44)
0.94 (0.26–3.44)d
Random Effect Model
Recurrence
a
PURKAYASTHA ET AL
Data are numbers with 95% confidence intervals in parentheses unless otherwise indicated. Baker et al. (1995)15 were excluded from analysis because of significant heterogeneity. b Boccasanta et al. (1999)16 were excluded from analysis because of significant heterogeneity. c Xynos et al. (1999)20 were excluded from analysis because of significant heterogeneity. d Baker et al. (1995),15 Solomon and Eyers (1996),18 Xynos et al. (1999)20 were excluded from analysis because no values in study or values equal to zero.
Overall Weighted mean difference or 60.08 60.38 odds ratio (49.94–70.22)a (48.99–71.76)a Heterogeneity (chi-squared) 4.98a 2.87b P value of heterogeneity 0.29a 0.58b Prospective Weighted mean difference or 61.83 62.56 odds ratio (50.84–72.82) (49.27–79.85) Heterogeneity (chi-squared) 4.32 1.13 P value of heterogeneity 0.23 0.57 Randomized Weighted mean difference or 58.23 58.51 odds ratio (43.69–72.77) (41.96–75.06) Heterogeneity (chi-squared) 1.28 1.06 P value of heterogeneity 0.26 0.3 Excluding studies with resection rectopexy Weighted mean difference or 1.86 2.66 odds ratio (1.45–2.27) (1.07–4.25) Heterogeneity (chi-squared) 39.16 16.17 P value of heterogeneity <0.01 <0.01
Fixed Effect Model
Operative Time
Table 3. Results of Sensitivity Analysis for Primary Outcomes
1934 Dis Colon Rectum, October 2005
Vol. 48, No. 10
LAP VS. OPEN RECTOPEXY META-ANALYSIS
1935
Figure 1. Forrest plot for analysis of comparison of morbidity for included studies. OR = odds ratio; CI = confidence interval; Chi2 = chi-squared; df = degrees of freedom; I2 = I-squared value; Z = Z-value.
showed no significant difference in the recurrence rate between LAR and OAR (OR, 0.94; 95 percent CI, 0.26–3.44; chi-squared = 0.66; P = 0.72) using random effect modeling. This finding was consistent in the sensitivity analysis for the prospective studies and the exclusion of resection studies subgroup, with no significant heterogeneity identified. It was not possible to calculate values for solely randomized studies because of the zero cell effect; there was only one study remaining for analysis.19 The mean follow-up time for the studies ranged from 12 to 31 months. There was no longer-term data for recurrence presented, which may be a relevant factor, especially for the younger patients. More detailed data on preoperative and postoperative constipation and fecal incontinence rates compared with recurrence rates would allow a better understanding of the association with recurrence.
Morbidity Although all six studies reported data on overall morbidity, only two17,19 described major and minor morbidities separately, therefore, separate subgroup analysis was not performed for such small data sets. One study caused significant funnel plot asymmetry20 (this was an outlier) and was excluded from the overall morbidity meta-analysis. The results demonstrated that there was no significant difference in the overall morbidity between LAR and OAR (OR, 0.56; 95 percent CI, 0.24–1.29, chi-squared = 5.44; P = 0.24) using random effect modeling. This finding was the same when comparing prospective and randomized studies only. The only prospective, randomized study reported increased adverse events in the OAR group (6 of 20 vs. 14 of 19; P < 0.01).19 Another study reported no significant difference in morbidity between the
two groups but did report two port site hernias.17 A third study also noted port site complications of the LAR group,18 notably two port site hematomas, associated with poor technique and one case of pseudomembranous colitis. In the same study, the open group had one case of wound dehiscence. Other complications were those usually associated with major abdominal surgery (1 case of deep vein thrombosis, 1 case of pneumonia, and 3 cases of prolonged postoperative ileus). There was one postoperative death in the OAR group; however, the cause was not stated.
Length of Hospital Stay All six studies reported data on length of stay. Evaluating the data also showed that one study demonstrated significant heterogeneity16 and was therefore excluded from the overall meta-analysis for length of in-hospital stay. The present study demonstrated a significant difference in the overall length of stay between LAR and OAR, favoring the laparoscopic group by 3.54 days (95 percent CI, 3.08–4.01 days; chi-squared = 2.87; P = 0.58) using random effect modeling. Similar results were obtained when only prospective or randomized studies were compared (Table 3). The excluded study16 also showed a significant reduction for in-hospital stay in the LAR group (6.8 vs. 15.1 days; P < 0.05). When sensitivity analysis was performed, excluding the resection rectopexy studies, there was still a significant difference in the length of stay; however, heterogeneity also significantly increased.
Operative Time All six studies reported data on operative time. One study demonstrated significant heterogeneity and was
1936
PURKAYASTHA ET AL
Dis Colon Rectum, October 2005
Figure 2. A. Forrest plot for analysis of comparison of overall morbidity for included studies. B. Funnel plot of studies used for comparison of overall morbidity. OR = odds ratio; CI = confidence intervals; Chi2 = chi-squared; df = degrees of freedom; I2 = I-squared value; Z = Z-value; SE = standard error.
excluded for the operative time meta-analysis.15 Using the random effect model, our study demonstrated a significant difference in the operative time between LAR and OAR, favoring the open group by 60.38 minutes (95 percent CI, 48.99–71.76; chi-squared = 4.98; P = 0.29). Similar results were obtained when prospective and randomized studies were compared. Interestingly, the randomized control trial, which was a follow-up study to an initial prospective trial with a retrospective control group (all cases performed by one or two surgeons only), noted a reduction in the
approximate average laparoscopic operating time from 70 to 50 minutes. This might be the result of advances in the surgical learning curve, which with more cases would mean less time differences between the OAR and LAR groups. The study that was excluded because of heterogeneity also reported greater operative time in the LAR group (177.1 ± 23 minutes vs. 86.5 ± 8.6 minutes). It must be noted that the studies included did not have precise definitions for the operative time. When sensitivity analysis was performed, excluding the resection rectopexy studies,
Vol. 48, No. 10
LAP VS. OPEN RECTOPEXY META-ANALYSIS
1937
Figure 3. Forrest plot for analysis of comparison of length of stay for included studies. WMD = weighted mean difference; CI = confidence interval; Chi2 = chi-squared; df = degrees of freedom; I2 = I-squared value; Z = Z-value. Figure 4. Forrest plot for analysis of comparison of operative time for included studies. WMD = weighted mean difference; CI = confidence interval; Chi2 = chi-squared; df = degrees of freedom; I2 = I-squared value; Z = Z-value.
there was still a significant difference in the operative time; however, heterogeneity also significantly increased.
Publication Bias Although publication bias may be problematic with multiple studies of small sample size, graphic exploration of the results with funnel plots of the primary and secondary outcomes did not demonstrate any evidence of publication bias. An example of a “funnel plot” of the studies used in this meta-analysis is shown in Figure 2B, which evaluates publication bias (outlier detection) for postoperative morbidity. This is a graph of the treatment effects estimated from individual studies plotted on the horizontal axis (OR), against a measure of study size, plotted on the vertical axis (SE[logOR]). The funnel plot resembles a symmetrical inverted funnel (95 percent CI), inside which all the studies are included in the meta-analysis. The name “funnel plot” is based on the fact that precision in the estimation of the underlying treatment effect will increase as the sample size of the component studies increases. Similar plots were constructed to assess publication bias with regards to the other end points.
Other Outcomes of Interest Table 2 demonstrates other outcomes measured in the six comparative studies assessed. These outcomes were not meta-analyzable because of lack of overlap in the reporting between studies or because of measurement differences in the way the data were collected and/or represented. Although mortality was measured in all six studies, zero values were present in all the studies, therefore, this data was not analyzed. Three studies described data on blood loss: one measured this as reduction in the postoperative hemoglobin (Hb),19 one as blood loss collected in milliliters,15 and the third stated no difference but gave no numeric values.20 The two studies that documented numeric values suggested that blood loss was less in their laparoscopic groups (LAR, 184.4 ± 31 ml; OAR, 285 ± 35 ml15; LAR, 1.2 percent Hb reduction postoperative; OAR, 1.9 percent Hb reduction postoperative; P = 0.0219). Opiate requirements were recorded in two studies18,19: one for a period of two days, and the other for five days. Both these studies suggested that less opiate analgesia was required by the laparoscopic groups, suggesting that the procedures were less painful, although no formal pain
PURKAYASTHA ET AL
1938
scores were demonstrated. Time taken for patients to tolerate oral fluid and normal diet were recorded in one study only,19 which suggested shorter times for laparoscopic cases (16 vs. 2 cases tolerated solid diet by Day 2; P < 0.01). Constipation rates were recorded in two studies,16,17 and although both studies measured preoperative and postoperative constipation rates, there was no data available for comparison of the open group in the second study.17 The same two studies16,17 evaluated fecal incontinence rates as did a third,20 but again significant difference in the collection of the measurements meant that no real evaluation could be made of such potentially comparable data.
Cost Only two studies have assessed cost issues.16,21 The first showed that although the mean cost of the surgical materials used was higher in the LAR group (US $927 vs. $231), the reduced overall hospital stay in the LAR group meant that the overall mean costs were actually less in the LAR group: $1,873 ($1,073–3,873) compared with $3,369 ($2,569–4,569). This infers that although periprocedural cost is greater for LAR, the overall cost is less, perhaps because of lower morbidity. The second study had similar outcomes and concluded that laparoscopic rectopexy was associated with an overall mean cost saving of £357 (95 percent CI, £64–592; P = 0.042), or approximately $607 per patient.
Sample Size Considerations The overall incidence of postoperative complications between studies in the OAR groups was 39 of 92 (approximately 42 percent) and the incidence of recurrence was 5 of 55 (9 percent). To rule out a 20 percent relative risk reduction (from 42 to 34 percent) for postoperative morbidity and for recurrence a 50 percent reduction (from 9 to 5 percent) with a 5 percent significance level and 80 percent power, calculations showed that a traditional, randomized, controlled trial would require approximately 700 patients in each arm. Thus, it is evident that even with combining the data from these trials a large-sized clinical trial is still needed.
DISCUSSION The results of this meta-analysis of six studies (retrospective, prospective, and randomized) suggest that
Dis Colon Rectum, October 2005
although the operative time is greater, the length of hospital stay is reduced in patients undergoing LAR compared with OAR. These results were statistically significant. There was no difference in recurrence rates or morbidity (the primary outcomes) between the two techniques. The use of meta-analytic techniques allowed the inclusion of a total of 195 patients, of which 97 (49.7 percent) underwent LAR and 98 (50.3 percent) underwent OAR. A sample of this size is substantial because to accumulate such data in a randomized, controlled trial setting would take considerable time and cost. Of note is the fact that subgroup analysis of the only randomized trial revealed a significant difference between LAR and OAR for length of stay and operative time. However, these findings should be interpreted with caution because of the small numbers of patients enrolled in the studies and because of inherent weaknesses of the observational studies included. The results for the meta-analysis of operative time show that LAR takes significantly longer compared with OAR (Fig. 4). This is not surprising considering the increased complexity and instrumentation during laparoscopic compared with open procedure, not to mention the set-up time involved and the fact that most surgeons may still be in their learning curve for such procedures. This is a limitation of laparoscopic surgery that would automatically be assumed for most laparoscopic vs. open operations. Ultimately the benefit of laparoscopic surgery is unlikely to be the length of the operation, but the quicker healing of smaller operative wounds and earlier recovery. Reducing surgical trauma and allowing abdominal rectopexy procedures on frailer individuals will enable provision of procedures that have a lower recurrence rate and low morbidity. Length of hospital stay results from the studies included in this meta-analysis show that LAR significantly reduces length of stay compared with OAR (Fig. 3). These results may be because mobilization after LAR is quicker, thereby facilitating recovery and subsequent discharge from hospital. Longer LOS in the OAR group may be related to wound pain, infection, and ileus formation, all of which impair a patient’s mobility after an operation. It is worth mentioning that the advantage of the procedure (LAR) with regards to LOS may be subject to bias. Surgeons may have discharged patients from the hospital sooner than patients who had a conventional approach. For patients, the time taken to return to normal activity is an important issue, reducing the chances of hospital
Vol. 48, No. 10
LAP VS. OPEN RECTOPEXY META-ANALYSIS
admission-related complications, such as hospital acquired pneumonia, and reducing the long-term psychologic effects of hospitalization. Heterogeneity is a measure of the diversity between different studies focusing on a similar subject. This may be because of methodologic (differences between study design or case mix) or statistical (when the true effects being evaluated differ between studies, and may be detectable if the variation between the results of the studies is above that expected by chance22). As previously mentioned, one meta-analysis has been used in the past to assess the most favorable operative methods for the treatment of rectal prolapse.11 However, this work did not focus purely on laparoscopic vs. open comparisons and only included one randomized, prospective trial (without taking into account possible discrepancies evident from the observational studies to explain the differences between studies).19 It also has been reported that there may be conflicting results from such meta-analysis and subsequent randomized trials.23 Although the trials used in this analysis were relatively small, that does not mean that their results should be ignored. Moreover, small studies can challenge conventional but untested therapeutic knowledge. Also, because the patient number in any trial is the number of events, rather than the number of study patients, some small trials are so definitively positive that they are sufficient to identify the best therapy. Small trials, even when individually inconclusive, can serve as the basis for convincingly conclusive overviews and meta-analyses that carry, and deserve, greater credibility than a single large trial of similar size to their sum.24 Another limitation of this study is that it does not attempt to evaluate the different laparoscopic surgical techniques and methods for reducing the prolapse itself. There has been discussion in the literature as to the strength of small meta-analyses.25
CONCLUSIONS Our analysis further highlights the need for high quality, multicenter, randomized trials. Further work focusing on the high-risk groups, such as elderly females and those with common comorbidities, is necessary. This will allow a better understanding of the outcomes in the type of patients who commonly present with rectal prolapse. These studies should incor-
1939
porate structured and standardized methodology measuring larger numbers of specific, comparative end points, which will be needed to further the discussion on this subject.
REFERENCES 1. Flowers LK. Rectal prolapse. In: E-medicine 2002. Available at: http://www.emedicine.com/emerg/topic496. htm. Accessed September 2004. 2. Senapati A. Rectal prolapse. In: Phillips RK, ed. Colorectal surgery. London: WB Saunders, 2001:251–71. 3. Eu KW, Seow-Choen F. Functional problems in adult rectal prolapse and controversies in surgical treatment. Br J Surg 1997;84:904–11. 4. Stelzner F. Etiology and therapy of rectal prolapse. Experiences with 308 cases 1956-1991 [in German]. Chirurg 1994;65:533–45. 5. Karasick S, Spettell CM. Defecography: does parity play a role in the development of rectal prolapse? Eur Radiol 1999;9:450–3. 6. Wassef R, Rothenberger DA, Goldberg SM. Rectal prolapse. Curr Probl Surg 1986;23:397–451. 7. Azimuddin K, Khubchandani IT, Rosen L, Stasik JJ, Riether RD, Reed JF 3rd. Rectal prolapse: a search for the “best” operation. Am Surg 2001;67:622–7. 8. Kim DS, Tsang CB, Wong WD, Lowry AC, Goldberg SM, Madoff RD. Complete rectal prolapse: evolution of management and results. Dis Colon Rectum 1999;42: 460–9. 9. Athanasiou T, Aziz O, Skapinakis P, et al. Leg wound infection after coronary artery bypass grafting: a metaanalysis comparing minimally invasive vs. conventional vein harvesting. Ann Thorac Surg 2003;76:2141–6. 10. Athanasiou T, Aziz O, Mangoush O, et al. Do off-pump techniques reduce the incidence of postoperative atrial fibrillation in elderly patients undergoing coronary artery bypass grafting? Ann Thorac Surg 2004;77:1567–74. 11. Bachoo P, Brazzelli M, Grant A. Surgery for complete rectal prolapse in adults. Cochrane Database Syst Rev 2000(2):CD001758. 12. Sterne JA, Egger M, Smith G. Investigating and dealing with publication and other biases. In: Egger M, Smith G, Altman DG, eds. Systematic reviews and healthcare: meta-analysis in context. 2nd ed. London: BMJ Publishing Group, 2001:189–208. 13. Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphic test. BMJ 1997;315:629–34. 14. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–88. 15. Baker R, Senagore AJ, Luchtefeld MA. Laparoscopicassisted vs. open resection. Rectopexy offers excellent results. Dis Colon Rectum 1995;38:199–201. 16. Boccasanta P, Venturi M, Reitano MC, et al. Lapa-
1940
17.
18.
19.
20.
PURKAYASTHA ET AL
rotomic vs. laparoscopic rectopexy in complete rectal prolapse. Dig Surg 1999;16:415–9. Kairaluoma MV, Viljakka MT, Kellokumpu IH. Open vs. laparoscopic surgery for rectal prolapse: a casecontrolled study assessing short-term outcome. Dis Colon Rectum 2003;46:353–60. Solomon MJ, Eyers AA. Laparoscopic rectopexy using mesh fixation with a spiked chromium staple. Dis Colon Rectum 1996;39:279–84. Solomon MJ, Young CJ, Eyers AA, Roberts RA. Randomized clinical trial of laparoscopic vs. open abdominal rectopexy for rectal prolapse. Br J Surg 2002;89:35–9. Xynos E, Chrysos E, Tsiaoussis J, Epanomeritakis E, Vassilakis JS. Resection rectopexy for rectal prolapse.
21.
22. 23. 24. 25.
Dis Colon Rectum, October 2005
The laparoscopic approach. Surg Endosc 1999;13: 862–4. Salkeld G, Bagia M, Solomon M. Economic impact of laparoscopic vs. open abdominal rectopexy. Br J Surg 2004;91:1188–91. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21:1539–58. Lau J, Ioannidis JP, Schmid CH. Summing up evidence: one answer is not always enough. Lancet 1998;351:123–7. Sackett DL, Cook DJ. Can we learn anything from small trials? Ann N Y Acad Sci 1993;703:25–32. Flather MD, Farkouh ME, Pogue JM, Yusuf S. Strengths and limitations of meta-analysis: larger studies may be more reliable. Control Clin Trials 1997;18:568–79.