The incidence of ACL tears in North America is 0.38 per 1000 per annum.33 In the United States in 1996, doctors performed ACL repairs on more than 100,000 patients across ambulatory (72,000) and inpatient (35,300) set-tings.4 Therefore, ACL reconstruction has generated considerable interest and expectancy in restoring function and facilitating return to preinjury level of activity.15
The bone-patellar tendon-bone autograft,26 considered the criterion standard, remains the preferred graft for primary reconstruction.29 However, with the increased use of hamstring tendons as graft material for ACL reconstruction, surgeons have questioned the relative benefit of bone-patellar tendon-bone versus hamstring grafts with regard to functional results. In a previous meta-analysis,6 we showed there is strong evidence hamstring grafts yield less morbidity but only marginal evidence bone-patellar tendon-bone grafts offer better stability. However, the ideal graft choice for restoring function and facilitating return to preinjury level of activity remains unknown.
Although some authors have reviewed the results of both procedures with regard to function,8,39 a meta-analysis allows for quantitative comparison. A few meta-analyses have compared hamstring graft reconstruction with bone-patellar tendon-bone graft reconstruction with respect to knee function.16,17,19,41 However, the small number of trials included or the inclusion of nonrandomized controlled trials have limited their conclusions.
Therefore, we performed a meta-analysis to provide quantitative data to ascertain whether bone-patellar tendon-bone grafts or hamstring grafts provided superior knee function as determined by final overall International Knee Documentation Committee knee evaluation and return to preinjury level of activity. We hypothesized bone- patellar tendon-bone graft reconstructions would yield better final overall International Knee Documentation Committee scores than hamstring graft reconstructions. We also asked whether return to preinjury level of activity would differ between the two grafts and whether specific factors influenced treatment outcome.
MATERIALS AND METHODS
We searched MEDLINE through PubMed, Webspirs, Science Citation Index, Current Contents databases, and the Cochrane Central Register of Controlled Trials supplemented by crossbib-liographic checks of reference lists of published trials. The following search terms were entered and modified according to the requirement of each database: (randomized OR controlled clinical trial OR random allocation) AND (ACL OR anterior cruciate ligament OR knee instability OR ligament injur*) AND (rehabilitation OR reconstruction OR treatment OR surgery). We had no restrictions on date of publication, language, or publication status. In addition, we sent a copy of selected studies to all authors of these studies, general and specific orthopaedic journals, and organizations with an interest in the topic to ask if they knew of any other published or unpublished trials. The initial search performed on March 14, 2005, was updated to March 15, 2006, by electronic search of the same databases and by searching Embase from 1996.
We selected trials that were randomized or quasirandomized (ie, the allocation was not at random but based on an alternating sequence, date of birth, day of surgery), providing that groups were constituted during the same period of time, included bone-patellar tendon-bone and hamstring graft reconstructions without augmentation in the comparison, had a mean followup of at least 2 years, and had one or more primary outcomes related to function: final overall International Knee Documentation Committee score and return to preinjury level of activity.
Two of the authors (DJB, CT) independently abstracted data on study design, setting, population, condition of interest, interventions and cointerventions, outcomes, and the reported meth-odologic quality of the studies by using standardized forms. Disagreements were resolved by discussion and, if necessary, with the help of other authors (SK, RN). The following questions were used to assess the quality of randomized controlled trials: Was the randomization process described and appropriate? Was the withdrawal and dropout rate described and acceptable (less than 20%)? Were cointerventions comparable? Was the assessment independent? Were the main outcomes analyzed according to the intention-to-treat principle? One point was given for each positive answer. Reports were rated as poor for a total of 2 or less, fair for a total of 3 or 4, and good for a total of 5. All authors of the included studies were contacted to retrieve unpublished data (through recorded data) and clarify possible overlap of patients. When overlap of data occurred between reports of the same trial, only the report with the longest followup was considered. Therefore, patients were included only once in the analysis.
Fourteen trials were included in the analysis. The search strategy generated 1703 references. Twenty-one reports of 14 different trials were relevant on title, abstract, and final complete retrieval (Fig 1).1-3,5,9-14,20,22,25,28,31,32,35-38,40 Three trials had three treatment groups.5,28,36 Groups that were of no interest were excluded from the analysis5 and others were pooled.28,36 One thousand two hundred sixty-three patients were analyzed (649 in the bone-patellar tendon-bone group and 614 in the hamstring group) (Table 1). In the hamstring group, a five-strand hamstring graft was used in one trial,32 a four-strand hamstring graft in eight trials,2,3,10,13,20,22,31,38 a four- or three-strand hamstring graft in two trials,9,28 and a two-strand hamstring graft in three trials.5,36,37 The use of arthroscopy was referred to in all trials. Preconditioning of the graft was reported in one trial,13 cycling in three trials,2,10,36 securing under tension in seven tri-als,2,5,9,10,28,31,32 and flexion degree of the knee when fixing the graft in 11 trials.2,3,5,9,10,13,22,28,31,32,36 Type of femoral and tibial fixation varied considerably among trials (Table 1). The postoperative rehabilitation program varied among trials but was described or reported similarly for both groups in all trials. The quality of the report was poor for six trials20,22,31,32,36,38 (scoring 0, 1, or 2) and fair for eight trials2,3,5,9,10,13,28,37 (scoring 3 or 4) (Table 2). Details of the design of each trial, inclusion and exclusion criteria, and associated disorders at the time of surgery may be found in the original reports.
We entered eligible trials into RevMan software (version 4.2.8 [July 25, 2005]; The Cochrane Collaboration) and sorted them according to inclusion and exclusion criteria. The final overall International Knee Documentation Committee score (Class A versus other classes and Class A or B versus Class C or D) and return to preinjury level of activity (yes versus no) were treated as binary variables. The International Knee Documentation Committee knee evaluation uses a multidimensional scale including the following categories: subjective assessment, symptoms, range of motion, and ligament examination.20 Each category consists of one or more items rated normal, nearly normal, abnormal, or severely abnormal according to established guidelines. The worst rating of any item in a given category determines the overall category rating, and the worst rating of the four categories determines the final overall rating of the knee, which is designated as Class A (normal), Class B (nearly normal), Class C (abnormal), or Class D (severely abnormal). We used RevMan Analysis Software (version 1.0.3 [July 25, 2005]; The Cochrane Collaboration) with a random effects model to analyze data. Heterogeneity was tested between trials with Cochrane's test statistic with p < 0.1 indicating significant heterogeneity. In case of significant heterogeneity, influential trials were identified as the most contributing to the chi square heterogeneity test statistics. These trials were searched thoroughly to look for differential factors that might explain the heterogeneity. To look for variation in the effect of treatment based on study quality (score less than 3 versus score 3 or higher), randomization status (qua-sirandomized versus randomized), number of strands used in the hamstring group (less than 4 versus 4 or higher), and length of followup (36 months or less versus more than 36 months), we performed quantitative interaction tests.
The difference in the final overall International Knee Documentation Committee score between bone-patellar tendon-bone and hamstring grafts was similar. The final overall International Knee Documentation Committee score was available for 1125 patients in 12 trials.2,3,5,9,10,13,20,22,28,32,36,37 The final overall International Knee Documentation Committee score was rated Class A in 178 of 544 patients in the hamstring group (33%) versus 239 of 581 patients in the bone-patellar tendon-bone group (41%). The relative risk of final overall International Knee Documentation Committee Class A was 0.90 (95% confidence interval, 0.79-1.03; p = 0.13) in favor of bone-patellar tendon-bone grafts; the test for heterogeneity was not significant (p = 0.98) (Table 3; Fig 2). The final overall International Knee Documentation Committee score was rated Class A or B in 422 of 544 patients in the hamstring group (78%) versus 462 of 581 patients in the bone-patellar tendon-bone group (80%). The relative risk of final overall International Knee Documentation Committee Class A or B was 1.01 (95% confidence interval, 0.92-1.11; p = 0.83) in favor of hamstring grafts; the test for heterogeneity was significant (p = 0.06) (Table 4). Three influential trials were identified. 13,37
We found no difference in return to preinjury level of activity between bone-patellar tendon-bone and hamstring grafts. Return to preinjury level of activity information was available for 392 patients in five trials.13,31,32,36,38 In the hamstring group, 119 of 178 patients (67%) were able to return to their preinjury level of activity versus 162 of 214 patients in the bone-patellar tendon-bone group (76%). The relative risk of returning to preinjury level of activity was 0.94 (95% confidence interval, 0.85-1.05; p = 0.28) in favor of bone-patellar tendon-bone grafts; the test for heterogeneity was not significant (p = 0.81) (Table 5).
Quantitative interaction tests on the effect of treatment based on study quality, randomization status, number of strands, and length of followup were non significant in the two groups. In trials using a four- or five-strand hamstring graft, the final overall International Knee Documentation Committee score remained in favor of patients with bone-patellar tendon-bone graft reconstructions but the magnitude of the treatment effect was smaller (Table 6).
There is fair evidence that patients reconstructed with hamstring grafts report less morbidity than those reconstructed with bone-patellar tendon-bone grafts.6 The improvement of stability with bone-patellar tendon-bone grafts compared with four-strand hamstring grafts remains of questionable importance for most patients.6 However, functional results between the two types of reconstruction remain unclear, and we compared the two grafts with regard to final overall International Knee Documentation Committee score and return to preinjury level of activity.
Our study has several limitations. First, aggregate patient data meta-analysis is prone to bias.30 However, the methodology for this review was rigorous and followed the QUOROM statement to limit the risk for bias.34 The search was thorough and authors were contacted to retrieve additional information and limit publication bias. Second, heterogeneity is a concern when pooling data from different trials and may hamper the overall conclusion. However, heterogeneity was found for only one outcome (International Knee Documentation Committee Class A or B versus Class C or D) despite the conservative level accepted for significance owing to the limited power of the test. Nonetheless, three influential trials were iden-tified.5,13,37 Anderson et al5 reported a final overall International Knee Documentation Committee score in favor of bone-patellar tendon-bone grafts, whereas Röpke et al37 and Feller and Webster13 reported in favor of hamstring grafts. The only difference we found that separated the first trial from the latter ones was the gender ratio; the latter two trials included comparatively more males than the first trial. However, the complex construction of the final overall International Knee Documentation Committee score makes it difficult to infer any legitimate explanations. Finally, other factors such as the inclusion of qua-sirandomized trials, the poor quality of the reports, and the different lengths of followup of these trials may have induced biases. However, the quantitative interactions tests did not find a variation in the treatment effect based on randomization status, study quality, or length of followup and the results were consistent in each subgroup analysis.
No difference was found in the final overall International Knee Documentation Committee class between groups. The International Knee Documentation Committee evaluation uses a multidimensional scale including the following groups: subjective assessment, symptoms, range of motion, and ligament examination.21 Regression analysis suggests the majority of the variability in the final overall International Knee Documentation Committee score is accounted for by symptoms and laxity.24 Therefore, hamstring graft reconstructions may compensate with less morbidity (symptoms) for what they lack in stability (ligament examination) when compared with bone-patellar tendon-bone graft reconstructions, and comparisons between groups may not be sensitive enough to differentiate between the groups. Moreover, floor and ceiling effects are likely when using a four-class scale. It has been reported the final overall International Knee Documentation Committee score is not sensitive enough to differentiate between Classes A (normal) and B (nearly normal), and this may hide a difference between groups when comparing Class A with other classes.24 However, when comparing Class A or B with Class C or D, we did not find a difference. The new subjective knee evaluation uses a scale ranging from 0 to 100 and may overcome this draw-back.23
We found similar percentages of patients with hamstring graft reconstructions (67%) and with bone-patellar tendon-bone graft reconstructions (76%) who were able to return to their preinjury level of activity. Although bone-patellar tendon-bone graft reconstruction is considered the criterion standard for those who are adamant about returning to full activity, recovery may be hampered by anterior knee pain and extension deficit.6 Moreover, reasons other than the reconstructed knee may prevent patients from returning to their preinjury level of activity.2,13 Some patients may be afraid to resume sports at the same level of activity that generated the trauma, some may prefer not to resume the same sports, and some may reduce their activity because they are not required to practice sports as often after finishing college.2,38
At last followup, only 33% and 41% of the patients, respectively, having reconstructions with hamstring grafts and bone-patellar tendon-bone grafts were reported as normal according to their final overall International Knee Documentation Committee scores. Consequently, more than 60% of the patients (708 of 1125 for the two reconstructions) will not make a full recovery (final overall International Knee Documentation Committee score Class A) after their reconstruction. The aim of ACL reconstruction is to restore functional stability without compromising other joint function.7 The pivot shift test is positive (>1+) for 19% of patients who had reconstruction with bone-patellar tendon-bone grafts and anterior knee pain is reported by 22% of these patients; comparative figures for patients who had reconstruction with hamstring grafts are 24% and 13%, respectively.6 Extension deficit is reported by 9% and 6% after bone-patellar tendon-bone and hamstring reconstructions, respectively.6 However, little is known about the distribution of these patients. Those reporting anterior knee symptoms or extension deficit may be different from those who have not reached complete stability. Therefore, the number of patients for whom the objective of the reconstruction has been fully completed may be lower than thought. Additionally, the final overall International Knee Documentation Committee score is a very conservative scale because the worst rating of any item in a given group determines the overall group rating; as a result, only patients with an absolutely normal knee will be rated Class A. Involving patients in their own care is crucial in completing their expectations.27 However, standard evaluation tools such as the International Knee Documentation Committee are not designed to elicit patients' needs and therefore are inappropriate for taking full account of what the patient sees as a successful operation. There is a need for developing new innovative methodologic approaches sensitive to differences in patient preferences.18
We found no difference in the final overall International Knee Documentation Committee score and in the number of patients returning to full activity after hamstring graft and bone-patellar tendon-bone graft reconstructions. We emphasize the data suggest no more than 60% of the patients will make a full recovery after their reconstruction and patients should be informed accordingly. Asymptomatic functional stability is the primary objective of ACL reconstruction, and efforts should be made to attain this objective fully.
We thank the following authors who gave additional information on published and unpublished trials and helped with the literature search: S. Abdulrazik, L. Ejerhed, K. Eriksson, J. Feller, A. Harilainen, K. Jansson, J. Kartus, G. Laxdal, M. Röpke, and K. Webster. We also thank the following experts who helped in the literature search: A. Anderson, D. Brand, E. Eriksson, M. Hantes, M. Marcacci, H. Roos, R. Smith, and J. Webb. Their contribution to this work does not necessarily mean they agree with the content of this study.
1. Aglietti P, Buzzi R, Zaccherotti G, De Biase P. Patellar tendon versus doubled semitendinosus and gracilis tendons for anterior cruciate ligament reconstruction. Am J Sports Med
2. Aglietti P, Giron F, Buzzi R, Biddau F, Sasso F. Anterior cruciate ligament reconstruction: bone-patellar tendon-bone compared with double semitendinosus and gracilis tendon grafts-a prospective, randomized clinical trial. J Bone Joint Surg Am
3. Aglietti P, Zaccherotti G, Buzzi R, De Biase P. A comparison between patellar tendon and double semitendinosus/gracilis tendon for anterior cruciate ligament reconstruction: a minimum five-year follow-up. J Sports Traumatol Relat Res
5. Anderson AF, Snyder RB, Lipscomb AB Jr. Anterior cruciate ligament reconstruction: a prospective randomized study of three surgical methods. Am J Sports Med
6. Biau DJ, Tournoux C, Katsahian S, Schranz PJ, Nizard RS. Bone-patellar tendon-bone autografts versus hamstring autografts for reconstruction of anterior cruciate ligament: meta-analysis. BMJ
8. Dopirak RM, Adamany DC, Steensen RN. A comparison of autogenous patellar tendon and hamstring tendon grafts for anterior cruciate ligament reconstruction. Orthopedics
. 2004;27:837-842; quiz 843-844.
9. Ejerhed L, Kartus J, Sernert N, Kohler K, Karlsson J. Patellar tendon or semitendinosus tendon autografts for anterior cruciate ligament reconstruction? A prospective randomized study with a two-year follow-up. Am J Sports Med
10. Eriksson K, Anderberg P, Hamberg P, Lofgren AC, Bredenberg M, Westman I, Wredmark T. A comparison of quadruple semitendinosus and patellar tendon grafts in reconstruction of the anterior cruciate ligament. J Bone Joint Surg Br
11. Eriksson K, Anderberg P, Hamberg P, Olerud P, Wredmark T. There are differences in early morbidity after ACL reconstruction when comparing patellar tendon and semitendinosus tendon graft: a prospective randomized study of 107 patients. Scand J Med Sci Sports
12. Feller JA, Webster KE. A prospective randomised comparison of patellar tendon and hamstring tendon ACLR with 3-year follow-up. A Sports Medicine Odissey: challenges, controversies and change
13. Feller JA, Webster KE. A randomized comparison of patellar tendon and hamstring tendon anterior cruciate ligament reconstruction. Am J Sports Med
14. Feller JA, Webster KE, Gavin B. Early post-operative morbidity following anterior cruciate ligament reconstruction: patellar tendon versus hamstring graft. Knee Surg Sports Traumatol Arthrosc
15. Fithian DC, Paxton LW, Goltz DH. Fate of the anterior cruciate ligament-injured knee. Orthop Clin North Am
. 2002;33:621-636. v.
16. Forster MC, Forster IW. Patellar tendon or four-strand hamstring? A systematic review of autografts for anterior cruciate ligament reconstruction. Knee
17. Freedman KB, D'Amato MJ, Nedeff DD, Kaz A, Bach BR Jr. Arthroscopic anterior cruciate ligament reconstruction: a metaanalysis comparing patellar tendon and hamstring tendon autografts. Am J Sports Med
18. Frewer LJ, Salter B, Lambert N. Understanding patients' preferences for treatment: the need for innovative methodologies. Qual Health Care
. 2001;10 (Suppl 1):i50-i54.
19. Goldblatt JP, Fitzsimmons SE, Balk E, Richmond JC. Reconstruction of the anterior cruciate ligament: meta-analysis of patellar tendon versus hamstring tendon autograft. Arthroscopy
20. Harilainen A, Linko E, Sandelin J. Randomized prospective study of ACL reconstruction with interference screw fixation in patellar tendon autografts versus femoral metal plate suspension and tibial post fixation in hamstring tendon autografts: 5-year clinical and radiological follow-up results. Knee Surg Sports Traumatol Arthrosc
21. Hefti F, Muller W, Jakob RP, Staubli HU. Evaluation of knee ligament injuries with the IKDC form. Knee Surg Sports Traumatol Arthrosc
22. Ibrahim SA, Al-Kussary IM, Al-Misfer AR, Al-Mutairi HQ, Ghafar SA, El Noor TA. Clinical evaluation of arthroscopically assisted anterior cruciate ligament reconstruction: patellar tendon versus gracilis and semitendinosus autograft. Arthroscopy
23. Irrgang JJ, Anderson AF, Boland AL, Harner CD, Kurosaka M, Neyret P, Richmond JC, Shelborne KD. Development and validation of the international knee documentation committee subjective knee form. Am J Sports Med
24. Irrgang JJ, Ho H, Harner CD, Fu FH. Use of the International Knee Documentation Committee guidelines to assess outcome following anterior cruciate ligament reconstruction. Knee Surg Sports Traumatol Arthrosc
25. Jansson KA, Linko E, Sandelin J, Harilainen A. A prospective randomized study of patellar versus hamstring tendon autografts for anterior cruciate ligament reconstruction. Am J Sports Med
26. Jones KG. Reconstruction of the anterior cruciate ligament using the central one-third of the patellar ligament. J Bone Joint Surg Am
27. Kravitz RL, Melnikow J. Engaging patients in medical decision making. BMJ
28. Laxdal G, Kartus J, Hansson L, Heidvall M, Ejerhed L, Karlsson J. A prospective randomized comparison of bone-patellar tendon-bone and hamstring grafts for anterior cruciate ligament reconstruction. Arthroscopy
29. Lipscomb AB, Johnston RK, Snyder RB. The technique of cruciate ligament reconstruction. Am J Sports Med
30. Lyman GH, Kuderer NM. The strengths and limitations of meta-analyses based on aggregate data. BMC Med Res Methodol
[serial online]. 2005;5:Article No. 14. Available at: http://www.biomedcentral.com/1471-2288/5/14
31. Marder RA, Raskind JR, Carroll M. Prospective evaluation of arthroscopically assisted anterior cruciate ligament reconstruction: patellar tendon versus semitendinosus and gracilis tendons. Am J Sports Med
32. Matsumoto A, Yoshiya S, Muratsu H, Yagi M, Iwasaki Y, Kurosaka M, Kuroda R. A comparison of bone-patellar tendon-bone and bone-hamstring tendon-bone autografts for anterior cruciate ligament reconstruction. Am J Sports Med
33. Miyasaka KC, Daniel DM, Shore ML, Hirsham P. The incidence of knee ligament injuries in the general population. Am J Knee Surg
34. Moher D, Cook DJ, Eastwood S, Olkin I, Rennie D, Stroup DF. Improving the quality of reports of meta-analyses of randomised controlled trials: the QUOROM statement. Quality of Reporting of Meta-analyses. Lancet
35. O' Neill DB. Arthroscopically assisted reconstruction of the anterior cruciate ligament: a prospective randomized analysis of three techniques. J Bone Joint Surg Am
36. O'Neill DB. Arthroscopically assisted reconstruction of the anterior cruciate ligament: a follow-up report. J Bone Joint Surg Am
37. Ropke M, Becker R, Urbach D, Nebelung W. Semitendinosus tendon vs. patellar ligament. Results of a prospective randomized study after anterior cruciate ligament reconstruction. [in German]. Unfallchirurg
38. Shaieb MD, Kan DM, Chang SK, Marumoto JM, Richardson AB. A prospective randomized comparison of patellar tendon versus semitendinosus and gracilis tendon autografts for anterior cruciate ligament reconstruction. Am J Sports Med
39. Spindler KP, Kuhn JE, Freedman KB, Matthews CE, Dittus RS, Harrell FE Jr. Anterior cruciate ligament reconstruction autograft choice: bone-tendon-bone versus hamstring: does it really matter? A systematic review. Am J Sports Med
40. Webster KE, Feller JA, Hameister KA. Bone tunnel enlargement following anterior cruciate ligament reconstruction: a randomised comparison of hamstring and patellar tendon grafts with 2-year follow-up. Knee Surg Sports Traumatol Arthrosc
41. Yunes M, Richmond JC, Engels EA, Pinczewski LA. Patellar versus hamstring tendons in anterior cruciate ligament reconstruction: a meta-analysis. Arthroscopy