Short-term Outcomes of and Ventral Hernia A Meta-analysis ORIGINAL ARTICLE Philip P. Goodney, MD; Christian M. Birkmeyer, MS; John D. Birkmeyer, MD Background: Although laparoscopic repair of ventral hernia has become increasingly popular, its outcomes relative to open repair have not been well characterized. For this reason, we performed a meta-analysis of studies comparing open and laparoscopic ventral (including incisional) hernia repair. Hypothesis: ventral hernia repair results in better short-term outcomes than open ventral hernia repair. Data Sources: Structured MEDLINE search for published studies. One unpublished study was also identified. Study Selection: Studies were selected on the basis of study design (comparison of laparoscopic and open ventral hernia repair). The 3 main outcome measures were perioperative complications, operative time, and length of hospital stay. Of 83 potential studies identified by abstract review, 8 (10%) met the inclusion criteria. Data Extraction: Two reviewers assessed each article to determine eligibility for inclusion and, where appropriate, abstracted information on patient characteristics and main outcome measures. Data Synthesis: Across 8 studies, 390 patients underwent open repair and 322 underwent laparoscopic repair. Perioperative complications were less than half as likely to occur in patients undergoing laparoscopic repair (14% vs 27%; P=.03; odds ratio, 0.42; 95% confidence interval, 0.29-0.68). Average length of stay was shorter in the laparoscopic group (2.0 vs 4.0 days; P=.02). No statistically significant difference in operative times was noted between laparoscopic and open repair (99 vs 96 minutes; P=.38). Conclusions: ventral hernia repair offers lower complication rates and shorter length of stay than open repair. However, randomized controlled trials and studies with long-term follow-up are needed to confirm these findings and to assess long-term rates of hernia recurrence. Arch Surg. 2002;137:1161-1165 From the VA Outcomes Group, Department of Veterans Affairs Medical Center, White River Junction, Vt (Drs Goodney and J. D. Birkmeyer); the Department of Surgery, Dartmouth-Hitchcock Medical Center, Lebanon, NH (Drs Goodney and J. D. Birkmeyer and Mr C. M. Birkmeyer); and the Center for the Evaluative Clinical Sciences, Dartmouth Medical School, Hanover, NH (Drs Goodney and J. D. Birkmeyer and Mr C. M. Birkmeyer). APPROXIMATELY 90 000 ventral hernias are repaired yearly in the United States, 1 including incisional, epigastric, and spigelian defects. Although open repair, preferably with mesh, 2-7 has long been the standard approach, laparoscopic repair is becoming increasingly popular among surgeons and patients following the development of minimally invasive techniques. Several observational studies 8-16 have raised the possibility that laparoscopic ventral hernia repair may be associated with fewer complications, decreased length of postoperative hospital stay, and lower recurrence rates. Although numerous studies have described outcomes of laparoscopic ventral hernia repair, there remains uncertainty about the relative outcomes of laparoscopic and open ventral hernia repair. Most studies 8-15 have been case series, lacking control groups. Studies 17-19 with control groups have been relatively small, single-center series. To better understand the outcomes of See Invited Critique at end of article these 2 techniques, we performed a metaanalysis of studies evaluating both laparoscopic and open ventral hernia repair. MATERIALS AND METHODS STUDY SELECTION AND DATA ABSTRACTION Studies were selected from MEDLINE using the strategy described in Figure 1. We searched with the medical subject headings terms ventral hernia and umbilical hernia, and then we combined these studies with those with the medical subject headings term laparoscopy. In addition, we hand searched references of 1161
included articles for other relevant studies. One unpublished study was identified. 20 After excluding non-english articles, we were left with 83 studies. We then systematically reviewed abstracts. Studies without explicit comparison of laparoscopic and open ventral hernia repair were excluded. Eight studies met the inclusion criteria. Two reviewers (P.P.G. and C.M.B.) independently reviewed the 8 studies and extracted information about the study design, sample size, patient characteristics, hernia characteristics, and outcomes. Rate of complications was the primary outcome measure. In addition, we abstracted information on operative time and length of hospital stay. STATISTICAL ANALYSIS We compared the odds of developing complications for patients treated laparoscopically and those treated with the open approach. Complication rates were weighted inversely with the variance and event rates across studies, related to the overall sample size. Because most studies did not provide confidence intervals, we conservatively calculated these using the Fisher exact method. 21 In primary analysis, we aggregated these results across studies using the Mantel-Haenszel method 22 and used a fixed-effects model to determine confidence intervals. We also analyzed the data using a random-effects model. However, because this approach changed the point estimates minimally, we present only the former. Study uniformity was assessed using the test of homogeneity. We calculated the mean operative time and length of hospital stay from each study, weighted by the number of patients in each study. The unpaired t test was then used to determine significance between the weighted averages. All tests of significance are at the 5% level, and all P values are 2-tailed. All calculations were performed using Stata (Stata Corp, College Station, Tex). RESULTS Ventral Hernia n = 1134 Nonpublished Studies Identified (n = 1) Combined (or) n = 1657 Umbilical Hernia n = 563 Combined (and) n = 100 Potentials n = 83 Eligible Studies n = 8 Laparoscopy n = 19 221 Figure 1. MEDLINE search strategy (OVID). Exclusion Criterion: Not in English (n = 18) Exclusion Criterion: Did Not Explicitly Compare With (n = 75) PATIENT AND STUDY CHARACTERISTICS We included 1 randomized controlled trial 19 and 7 cohort studies 17,18,20,23-26 (Table 1). One cohort study 17 used historical controls (open repair cases from an earlier period). Patient characteristics, such as demographics, body mass index, comorbidities, and previous attempt at hernia repair, were abstracted when available. However, only information on patient age and sex was consistently available in most studies. Average patient age ranged from 46 to 60 years (Table 2). Other than the study by Holzman et al, 26 patient age was similar in the laparoscopic and open groups. Similar trends were noted for patient sex, with the exception of the study by Robbins et al. 18 Last, patients undergoing laparoscopic repair were more likely to have undergone previous (failed) hernia repair in all studies in which that information was available; this difference was statistically significant in 2 of 5 studies. The operative technique used for open repair varied across studies (Table 1). Whereas 7 of 8 studies used Table 1. Characteristics of the 8 Studies Included in the Meta-analysis of Studies Comparing With Ventral Hernia * Study and Year Study Design Patients, No. Mesh Used in All Patients Technique Mesh Position Relative to Fascia Outcomes Assessed Randomized controlled trial 30 30 Yes Not assessed Complications, OR time, LOS Retrospective cohort study 14 14 Yes Underlay Complications, OR time, LOS DeMaria et al, 23 2000 Prospective cohort study 21 18 Yes Onlay Complications, OR time, recurrence, LOS, cost Retrospective cohort study 21 16 No Not assessed Complications, OR time, recurrence, LOS, cost Prospective cohort study 65 71 Yes Onlay Complications, OR time, cost Retrospective cohort study 56 49 Yes Inlay Complications, OR time, recurrence, LOS Retrospective cohort study 79 174 Yes Not assessed Complications, OR time, recurrence, LOS Robbins et al, 18 2001 Prospective cohort study 36 18 Yes Underlay Complications Total 322 390 *OR indicates operating room; LOS, length of stay. Onlay indicates mesh placed in a prefascial position; inlay, mesh sewn to edges of fascial defect, with overlap; and underlay, mesh placed retromuscular to the rectus sheath. Historical controls. 1162
Table 2. Patient Characteristics Source, y Age, Mean, y Men, % Previous, % 58 55 27 40.........*............... DeMaria et al, 23 2000 46 47 38 39 52 17 52 60...... 38 25 54 56 53 56 40 21 59 59 53 47 29 18 51 56 44 45 46 29 Robbins et al, 18 2001 46 49 19 44...... *Ellipses indicate not available. P.05 as reported by the authors or calculated from study data, when available. Table 3. Summary of Complication Rates, vs Ventral Hernia * Complication Rate, % Source, y Weight, % OR (95% CI) 14 14 4 1.00 (0.15-6.72) DeMaria et al, 23 2000 19 50 9 0.24 (0.06-0.95) 14 31 7 0.37 (0.08-1.71) 5 15 11 0.27 (0.08-0.95) 18 37 24 0.37 (0.16-0.91) 19 26 45 0.65 (0.34-1.25) Summary measure 14 27 100 0.42 (0.29-0.68) *Mantel-Haenszel test of homogeneity: 2 5 = 3.69; P =.59. OR indicates odds ratio; CI, confidence interval. mesh in all open repairs, some open procedures in one study 26 involved primary repair with nonabsorbable sutures. The position of the mesh placement was either not noted explicitly 19,24,26 or varied from onlay 20,23 to inlay 25 to underlay. 17,18 Onlay was defined as placed anteriorly to the fascia, inlay was defined as sewn to the edges of the fascial defect, and underlay was defined as placed retromuscular to the rectus sheath. 17,18 COMPLICATIONS Of the 6 studies included in the complications summary measure, 5 (83%) reported trends toward decreased risks of complications with laparoscopy (Table 3 and Figure 2). Three of these reductions were statistically significant. The only study showing no benefit with laparoscopic repair was the smallest study, with only 14 patients in each arm. 17 In evaluating complication rates, 2 studies were excluded from meta-analysis. The first excluded study 18 recorded only wound complications, excluding any other type of complication, such as pulmonary embolism or pneumonia. They reported wound complication rates of 28% in the open group and 16% in the laparoscopic group. The other excluded study 19 reported only total number of complications, not number of patients with complications. They identified 2 complications in 30 laparoscopic patients compared with 15 in 30 open patients. In pooled analysis, the summary odds ratio was 0.42 (95% confidence interval, 0.29-0.68; P=.03) for risk of complications with laparoscopic relative to open repair. In other Robbins et al, 18 2001 DeMaria et al, 23 2000 Less Risk for Cases 0 0.25 0.50 0.75 1.00 1.25 1.50 1.75 2.00 Odds Ratio words, patients undergoing laparoscopic ventral hernia repair were 58% less likely to experience a complication as those undergoing open repair. Our test of homogeneity (5 df) yielded a P=.59, demonstrating that the outcomes from these studies were consistent enough for aggregation. LENGTH OF STAY More Risk for Cases (2.59) (6.71) Figure 2. Odds of perioperative complications for laparoscopic vs open ventral hernia repair. Error bars represent 95% confidence intervals. Carbajo et al 19 assessed total complications (rather than rate of complications) and Robbins et al 18 included only wound complications. These 2 studies were excluded from the summary measure calculation. Seven studies reported shorter postoperative hospital stays for patients undergoing laparoscopic repair (Figure 3). Three studies 19,23,25 reported statistically significant re- 1163
Shorter for Cases Longer for Cases Shorter for Cases (P>.05) Longer for Cases DeMaria et al, 23 2000 (P<.02) 10 8 6 4 2 0 2 4 6 8 10 Length of Stay, d ( vs ) Figure 3. Comparison of average hospital lengths of stay for laparoscopic vs open ventral hernia repair. 60 45 30 15 0 15 30 45 60 Operative Time, min ( vs ) Figure 4. Comparison of average operative times for laparoscopic vs open ventral hernia repair. ductions in length of stay, 3 studies 17,24,26 did not assess statistical significance, and the final study 20 found a statistically nonsignificant reduction. One study 18 did not report data on length of stay. In pooled analysis, average length of stay was shorter in the laparoscopic group (2.0 vs 4.0 days; P=.02). The study showing the largest reduction in length of stay was the only randomized controlled trial 19 in our analysis. OPERATIVE TIME Six studies compared average operating room times in the 2 groups (Figure 4). Four 17,20,25,26 of the 6 studies noted longer operative time (range, 17-46 minutes longer) with laparoscopic repair. The 2 remaining studies 19,24 found average operating room times 24 and 29 minutes shorter with laparoscopy. In pooled analysis, we found no statistically significant difference in operative times between the laparoscopic and open groups (99 vs 96 minutes; P=.38). COMMENT This study examined the current surgical literature comparing laparoscopic and open ventral hernia repair. Eight studies, with a total of 712 patients, were identified in the meta-analysis. Compared with open repair, laparoscopic surgery was found to have lower risks of complication, longer operative times, and shorter length of hospital stay. This study has several limitations. First, it is difficult to rule out unmeasured differences in case-mix as an explanation for the findings, particularly since 7 of the 8 studies reviewed had observational designs. However, there is little reason to believe that differences in case-mix explain the findings. Measured patient characteristics did not imply that sicker patients were undergoing open repair. In fact, more patients in the laparoscopic group had undergone previous (failed) attempts at hernia repair. Also, the only randomized controlled trial 19 in our analysis, in which patient characteristics should have been evenly distributed between the 2 groups, found similar results to our pooled analysis. Therefore, we believe it is unlikely that our findings can be attributed to unmeasured differences in case-mix. Second, we did not assess postoperative pain, another potentially important outcome measure. Minimally invasive procedures are often assumed by patients and surgeons to be less painful. However, this assumption has not been well tested in the literature; only one study 23 in our analysis used a pain assessment scale to show a small decrease in pain with laparoscopic repair. Our own clinical observations suggest that patients often have considerable discomfort after laparoscopic repair of abdominal wall hernias. More studies assessing this outcome rigorously are needed. Third, the nature of postoperative complications may differ substantially between the 2 techniques. With open repair, most complications tend to be wound related, only infrequently requiring reoperation and rarely causing permanent disability. In contrast, the risk of unrecognized enterotomy in laparoscopic repair is significant (2%-4%), 8,9,14 and delay in diagnosis can result in intraabdominal contamination and life-threatening sepsis. Therefore, although complications are less frequent in laparoscopic repair, their sequelae may be more severe. Furthermore, several studies used different definitions of what constituted a complication, making comparison across studies difficult. Future trials should use consistent definitions and scales to account for the differences in complications between laparoscopic and open repair. Fourth, there was significant variability in operative technique, particularly in open ventral hernia, across the 8 studies included in our analysis. Seven studies used mesh in all open repairs. However, the mesh location varied across studies, with 2 studies using an onlay technique, 20,23 1 an inlay technique, 25 and 2 a retromuscular technique. 17,18 Despite this variation, we find little reason to believe that mesh location would affect shortterm complications; mesh location in open repair is much more likely to affect recurrence rate. 3 Finally, we did not assess hernia recurrence rates, which is one of the most important outcomes of ventral hernia repair. Considerable uncertainty exists surrounding recurrence rates in open and laparoscopic ventral hernia repair. The 3-year cumulative recurrence rate in a large randomized controlled trial 2 of open ventral hernia repair was 24%. This study used retromuscular mesh placement with a 2- to 3-cm overlap, considerably less overlap than that described by Stoppa 4 and Rives. 3 Uncontrolled studies 3,27-30 using the Stoppa/Rives technique have reported recurrence rates as low as 2% to 6%. In laparoscopic repair, retrospective studies 9,31,32 have reported recurrence rates as low as 3% to 4%. Of the 8 studies included in our meta-analysis, only 3 23-25 contained data regarding recurrence rates for the laparoscopic technique (range, 1164
3%-13%). These data were short term (usually 2 years) and often were not evaluated by independent examiners or objective measures (eg, imaging). Given uncertainty about long-term recurrence rates after laparoscopic repair, trials with long-term follow-up are needed to compare the durability of open and laparoscopic repair. In conclusion, laparoscopic ventral hernia repair offers lower complication rates and shorter length of hospital stay compared with traditional open repair. However, randomized controlled trials are necessary to confirm these findings and to provide information on long-term recurrence rates. This study was supported by a Career Development Award from the Veterans Administration Health Services Research and Development Service, Washington, DC (Dr J. D. Birkmeyer). Corresponding author: Philip P. Goodney MD, VA Outcomes Group (111B), Department of Veteran Affairs Medical Center, 215 N Main St, White River Junction, VT 05009 (e-mail: philip.goodney@hitchcock.org). REFERENCES 1. Toy FK, Bailey RW, Carey S, et al. Prospective, multicenter study of laparoscopic ventral hernioplasty: preliminary results. Surg Endosc. 1998;12:955-959. 2. Luijendijk RW, Hop WC, van den Tol MP, et al. A comparison of suture repair with mesh repair for incisional hernia. N Engl J Med. 2000;343:392-398. 3. Incisional hernia: the problem and the cure. J Am Coll Surg. 1999;188:429-447. 4. Stoppa RE. The treatment of complicated groin and incisional hernias. World J Surg. 1989;13:545-554. 5. Amid PK, Lichtenstein IL. Retromuscular alloplasty of large scar hernias: a simple staple attachment technique [in German]. Chirurg. 1996;67:648-652. 6. Amid PK, Shulman AG, Lichtenstein IL. A simple stapling technique for prosthetic repair of massive incisional hernias. Am Surg. 1994;60:934-937. 7. Wantz GE. Incisional hernioplasty with Mersilene. Surg Gynecol Obstet. 1991; 172:129-137. 8. Heniford BT, Park A, Ramshaw BJ, Voeller G. ventral and incisional hernia repair in 407 patients. J Am Coll Surg. 2000;190:645-650. 9. Heniford BT, Ramshaw BJ. ventral hernia repair: a report of 100 consecutive cases. Surg Endosc. 2000;14:419-423. 10. Sanders LM, Flint LM, Ferrara JJ. Initial experience with laparoscopic repair of incisional hernias. Am J Surg. 1999;177:227-231. 11. Amid PK. repair of anterior abdominal wall herniation using composite mesh. Am J Surg. 1996;171:542-543. 12. Costanza MJ, Heniford BT, Arca MJ, Mayes JT, Gagner M. repair of recurrent ventral hernias. Am Surg. 1998;64:1121-1125; discussion, 1126-1127. 13. Farrakha M. treatment of ventral hernia: a bilayer repair. Surg Endosc. 2000;14:1156-1158. 14. Kyzer S, Alis M, Aloni Y, Charuzi I. repair of postoperation ventral hernia: early postoperation results. Surg Endosc. 1999;13:928-931. 15. LeBlanc KA, Booth WV, Whitaker JM, Bellanger DE. incisional and ventral herniorrhaphy in 100 patients. Am J Surg. 2000;180:193-197. 16. Larson GM. Ventral hernia repair by the laparoscopic approach. Surg Clin North Am. 2000;80:1329-1340. 17. Chari R, Chari V, Eisenstat M, Chung R. A case controlled study of laparoscopic incisional hernia repair. Surg Endosc. 2000;14:117-119. 18. Robbins SB, Pofahl WE, Gonzalez RP. ventral hernia repair reduces wound complications. Am Surg. 2001;67:896-900. 19. Carbajo MA, Martin del Olmo JC, Blanco JI, et al. treatment vs open surgery in the solution of major incisional and abdominal wall hernias with mesh. Surg Endosc. 1999;13:250-252. 20. McGreevy JM, Goodney PP, Laycock WS, Birkmeyer CM, Birkmeyer JD. A prospective study of complication rates after laparoscopic and open ventral hernia repair. Surg Endosc. In press. 21. D Agostino R, Chase W, Belanger A. The appropriateness of some common procedures for testing the equality of two independent binomial populations. Am Stat. 1988;42:198-202. 22. Mantel N, Haenszel M. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst. 1959;22:719-748. 23. DeMaria EJ, Moss JM, Sugerman HJ. intraperitoneal polytetrafluoroethylene (PTFE) prosthetic patch repair of ventral hernia: prospective comparison to open prefascial polypropylene mesh repair. Surg Endosc. 2000;14: 326-329. 24. Ramshaw BJ, Esartia P, Schwab J, et al. Comparison of laparoscopic and open ventral herniorrhaphy. Am Surg. 1999;65:827-831; discussion, 831-832. 25. Park A, Birch DW, Lovrics P. and open incisional hernia repair: a comparison study. Surgery. 1998;124:816-821; discussion, 821-822. 26. Holzman MD, Purut CM, Reintgen K, Eubanks S, Pappas TN. ventral and incisional hernioplasty. Surg Endosc. 1997;11:32-35. 27. Korenkov M. Classification and surgical treatment of incisional hernia: results of an experts meeting. Langenbecks Arch Surg. 2001;386:65-73. 28. Avisse C, Palot JP, Flament JB. Treatment of inguinal hernia by the Jean Rives technique: replacement of the fascia traversalis by a Dacron prosthesis: apropos of the reports of G.E. Wantz and E.P. Pelissier: session of 12 May 1993 [in French]. Chirurgie. 1993;119:362-365. 29. Rives J, Fortesa L, Drouard F, Hibon J, Flament JB. Subperitoneal abdominal approach in the treatment of inguinal hernia [in French]. Ann Chirurg. 1978;32: 245-253. 30. Schumpelick V, Conze J, Klinge U. Preperitoneal mesh-plasty in incisional hernia repair: a comparative retrospective study of 272 operated incisional hernias. Chirurg. 1996;67:1028-1035. 31. Reitter DR, Paulsen JK, Debord JR, Estes NC. Five-year experience with the fourbefore laparoscopic ventral hernia repair. Am Surg. 2000;66:465-468; discussion, 468-469. 32. Chowbey PK, Sharma A, Khullar R, Mann V, Baijal M, Vashistha A. ventral hernia repair. J Laparoendosc Surg. 2000;10:79-84. Invited Critique H aving demonstrated the feasibility of laparoscopic ventral hernia repair, surgeons are performing this procedure before its efficacy has been established by randomized controlled trials. In this meta-analysis, the authors evaluated the existing literature to determine whether superior outcomes have been demonstrated for laparoscopic ventral hernia repair. They found only 8 studies that compared open ventral hernia repair with the laparoscopic approach. Of these, only half were prospective. The outcomes that the authors focused on, presumably because they were the end points that they were able to measure in most studies, were perioperative complications, operative time, and length of hospital stay. They concluded that complication rates were lower and length of stay was shorter in the laparoscopic group and that operative times were not significantly different between the 2 groups. It is surprising that these studies did not adequately address the differences in postoperative pain or hernia recurrence rates between the 2 groups. Because a decrease in postoperative pain is a strong argument for the use of minimally invasive technology in other procedures, one would assume that this outcome would be carefully assessed in these comparative analyses. Shorter length of stay may be a surrogate for diminished postoperative pain. However, because well-described tools are available to study pain, this variable should have been addressed. Hernia recurrence, one of the most important outcomes to measure when determining efficacy of repair, requires longer follow-up to be adequately addressed. Nevertheless, the fact that only 3 of the 8 studies even looked at this outcome is unexpected. Considerable controversy still exists regarding the superiority of laparoscopic inguinal hernia repair over the tension-free anterior approach. I suspect that this same controversy will persist for laparoscopic ventral hernia repair. This study illustrates the need for large, well-designed studies with the power to demonstrate the long-term benefits of the laparoscopic approach. Diana Marie Weber, MD Washington, DC 1165