Abstract
Background:
Revision anterior cruciate ligament (ACL) reconstruction is poorly described because of its rare incidence and mainly small case series presented in the literature. The Danish ACL reconstruction registry has monitored the development in revision ACL reconstruction since 2005.
Hypothesis:
We hypothesized that younger patients had a higher risk of revision ACL reconstruction than older patients and that subjective clinical outcome was worse after revision ACL reconstruction than after primary ACL reconstruction.
Study Design:
Cohort study; Level of evidence, 2.
Methods:
All clinics performing ACL reconstructions in Denmark report to the national ACL reconstruction registry. The revision rate after primary ACL reconstruction (n = 12,193 procedures) and re-revision rate after revision ACL reconstruction (n = 1099 procedures) were calculated for the period of 2005 to 2010. Outcome at 1-year follow-up for the revision cohort was reported using the Knee Injury and Osteoarthritis Outcome Score (KOOS), Tegner function score, and objective knee stability measurement.
Results:
The rate of revision ACL reconstruction was 4.1% after 5 years. Revision occurred most frequently after 1 to 2 years. Patients below 20 years of age at the time of primary ACL reconstruction had a higher risk of revision (8.7%) than did patients older than 20 years of age (2.8%) (adjusted relative risk, 2.58; 95% confidence interval, 2.02-3.30). The KOOS scores 1 year after revision ACL reconstruction (mean ± standard deviation) were 73 ± 18 for symptoms, 78 ± 17 for pain, 84 ± 16 for activities of daily living, 52 ± 28 for sports, and 48 ± 21 for quality of life. All these scores were significantly lower than for primary ACL reconstruction: 77 ± 17 for symptoms, 84 ± 15 for pain, 89 ± 13 for activities of daily living, 62 ± 25 for sports, and 59 ± 21 for quality of life. Side-to-side difference in knee laxity improved from 5.8 mm before revision ACL reconstruction to 1.9 mm 1 year after revision ACL surgery. The use of allograft tissue for the revision procedure resulted in a higher risk of re-revision than did autograft tissue (relative risk, 2.05; 95% confidence interval, 1.5-2.4) (P < .01). The rate of re-revision after 5 years was 5.4%.
Conclusion:
In this observational population-based study, the 5-year revision ACL reconstruction rate was 4.1%. Despite achieving acceptable knee stability after revision ACL reconstruction, subjective outcome is less favorable than after primary ACL reconstruction.
Keywords
Reconstruction of the anterior cruciate ligament (ACL) is increasingly performed for restoration of knee stability in typically young, athletic, active patients. In the age group that is mainly prone to ACL injuries (15-40 years), the incidence of surgery is 85 per 100,000. 7 Numerous factors affect a good outcome after ACL reconstruction such as surgical technique, graft choice, graft fixation, postoperative rehabilitation, and patient education. 5 An ultimate result of failure of ACL reconstruction is the need for revision ACL reconstruction.
However, the indication for revision ACL reconstruction is not universally defined, and not all patients who have poor outcomes after ACL reconstruction end up having a revision ACL reconstruction or will benefit from it. The reasons for this are numerous. As ACL-injured patients get older, their work career and establishing a family often result in lower sport activity levels, so patients with poor functioning ACL reconstructions accept their unstable knee and therefore do not go on to revision ACL reconstruction. Also, if a poor outcome is primarily caused by postoperative pain and not by instability, a revision ACL reconstruction is rarely indicated. Because revision ACL reconstruction is performed rarely, less than 10% of all knee ligament reconstructions, 9 the possibilities to perform high evidence level studies are very limited. Therefore, knowledge about outcomes after revision ACL reconstruction is based on small case series studies. Recently, national registries have been established in Scandinavia, and these registries can generate demographic and outcome data for revision ACL reconstruction for entire nations, thereby providing much more reliable data for the true outcome after revision ACL reconstruction.7,9 Another multicenter cohort of prospectively followed patients with ACL reconstruction and ACL revision procedures has been established in the Multicenter Orthopedic Outcome Network (MOON) and Multicenter ACL Revision Study (MARS) in the United States.14,21
This study aimed to describe the epidemiology and risk factors for revision ACL reconstruction after primary ACL reconstruction based on data from the Danish national ACL reconstruction registry. Further, the rate of re-revision ACL reconstruction after revision ACL reconstruction was studied. Patient characteristics and clinical outcomes after revision ACL reconstruction were compared with clinical outcomes after primary ACL reconstruction. We hypothesized that younger patients had a higher risk of revision ACL reconstruction than older patients and that subjective clinical outcome was worse after revision ACL reconstruction than after primary ACL reconstruction.
Materials and Methods
Study Design
Our study is based on the Danish ACL reconstruction registry, which is a prospective, nationwide, and Web-based clinical database initiated on July 1, 2005. The registry contains data on primary ACL procedures and revisions of primary ACL procedures performed in Denmark. Both public (n = 24) and private (n = 27) hospitals provide data to this registry. The primary aim of the registry is to measure, maintain, and improve the quality of knee ligament surgery. 9
Registration in the Danish ACL reconstruction registry is compulsory for all public and private hospitals, according to the notification from 2006 on the improvement of nationwide and regional clinical quality databases passed by the National Board of Health. Nevertheless, 88% of all ACL operations performed in Denmark were recorded in the registry. This completeness is calculated by identifying missing procedures using the Danish national registry of patients as a gold standard. The use of personal identification numbers in Denmark enables identification of all patients in the country.
Preoperative, operative, and 1-year follow-up data in the registry, described elsewhere, 9 are recorded by the operating surgeon using a standardized form and a secured Internet portal. A 1-year follow-up visit with objective assessments is advocated and performed in 48% of cases. Furthermore, the patients independently report subjective scores regarding knee function using the self-assessment Knee Injury and Osteoarthritis Outcome Score (KOOS) 13 and Tegner functional score. 15 These data are Web recorded by the patient before surgery and 1 year after surgery. According to the annual report in 2011 published from the Danish ACL reconstruction registry, 39% and 28% of patients entered subjective data preoperatively and 1 year after surgery, respectively. Because of the low completeness of patient-based scores, a validation study was performed to demonstrate any difference in subjective outcome between patients who did report their score at 1-year follow-up and patients who did not. A total of 200 patients from each group were sent scores by mail; 60% responded. The KOOS4 score (a modified KOOS score, defined below) in the group that did report at 1-year follow-up was 72 points; in the nonreport group, it was 70 points. We have subsequently concluded that the subjective scores received from patients represent the scores from the entire cohort. The 5-year revision rate is reported for all 1099 revisions and all 12,193 primary cases. The Danish National Board of Health and the Danish Data Protection Agency approved this study.
Patients
From the Danish ACL reconstruction registry, we identified a study population of 12,193 primary ACL reconstruction procedures performed between July 1, 2005, and December 31, 2010. From this cohort, revision rates were determined, and subjective outcome data are used to compare with outcome data from an ACL revision cohort of 1099 revision ACL procedures, defined as a replacement of a primary ACL reconstruction. Not all revision ACL procedures have their primary ACL reconstruction registered in the Danish ACL reconstruction registry because the primary ACL procedure has occurred before July 1, 2005, or because of a lack of registration of the primary ACL reconstruction, although it has occurred after July 1, 2005.
Statistical Analyses
We calculated proportions to describe epidemiological patient characteristics for the primary and revision ACL reconstruction groups. We used the Kaplan-Meier method to calculate the overall risk of revision and risk of re-revision for primary ACL procedures and revision ACL reconstructions, respectively. Thus, all primary ACL reconstruction procedures are followed up from the day of primary ACL reconstruction and ended on the day of revision of primary ACL reconstruction if revision occurred or on the day of the time of death, emigration, or September 1, 2011, whichever came first.
Likewise, all revision ACL procedures are followed up from the day of ACL revision and ended on the day of re-revision of ACL revision if it occurred or on the day of the time of death, emigration, or September 1, 2011, whichever came first. The cause for revision was determined by the operating surgeon’s best judgment based on radiographic and history assessments.
We studied the effect of age, gender, and cause of injury on the risk of revision ACL reconstruction. A Cox regression analysis was used to calculate hazard ratios as a measure for relative risk (RR) with 95% confidence intervals (CIs). We present overall risk of revision ACL reconstruction as well as risk of revision at different time points: 0 to 1 year, 1 to 2 years, 2 to 3 years, 3 to 4 years, and 4 to 5 years after surgery.
Finally, both primary ACL procedures and revision ACL reconstructions were linked to KOOS and Tegner score data preoperatively and 1 year after surgery, if reported. The KOOS and Tegner scores have been calculated according to published standards.13,15 The KOOS subscores are presented as mean values with standard deviations. Tegner scores are presented as median values and ranges. A Wilcoxon 2-sample test was used to compare postoperatively the KOOS and Tegner scores between study populations, with statistical significance at P = .05.
To calculate a single measure for the effect of ACL reconstruction based on KOOS data, a new parameter has recently been defined. This measure is designated KOOS4 and is defined as the average of the 4 most responsive KOOS subscores from preoperatively to postoperatively. The included subscores are symptoms, pain, sports, and quality of life. 4 A potential subjective failure criterion by KOOS data is a quality of life score <40. This parameter is used as a subjective failure criterion for ACL revision patients.
Results
Epidemiology
Male patients represented 54% of ACL revision cases compared with 60% of primary ACL reconstruction cases. Revision ACL reconstruction was performed at a younger age than primary ACL reconstruction (Figure 1). The time from primary ACL reconstruction to revision ACL surgery averaged 1.8 ± 1.0 years.
Age distribution between revision anterior cruciate ligament (ACL) reconstructions and primary ACL reconstructions. The ACL revisions are performed in a younger age group than are primary ACL reconstructions. Data are presented as percentages for each age interval.
The primary cause for graft failure after primary ACL reconstruction leading to revision ACL reconstruction was new trauma (38%), followed by unknown cause for graft failure (24%), and poor femoral tunnel placement (20%) (Table 1). Sport was the most frequent type of new trauma, registered in 83% of cases.
Causes for Graft Failure Leading to Revision (n = 1099 Procedures) a
Based on the operating surgeon’s evaluation.
Allografts were used in 21% of revision ACL reconstruction cases. Of the remaining autograft procedures, hamstring tendon grafts were used in 42% of cases, and patellar tendon–bone grafts were in 28% of cases, leaving 9% to other autograft types. Anteromedial femoral tunnel placement was used in 25% of cases and transtibial technique in the remaining cases. No double-bundle revision procedures were performed for ACL revisions.
New meniscal lesions were seen in 26%, and cartilage lesions larger than 2 cm2 were demonstrated in 20% of revision ACL reconstruction patients at the time of surgery. Previous meniscal surgery was performed in 17% of revision ACL reconstruction patients.
Risk of Revision ACL Reconstruction After Primary ACL Reconstruction
At 5 years, 95.9% of primary ACL reconstructions were still intact, corresponding to a revision rate of 4.1% (Figure 2). Nevertheless, the mean revision rate during the entire follow-up period was 4.7%.
Kaplan-Meier survival curve with 95% confidence interval (indicated by the thin adjoining lines) for primary anterior cruciate ligament reconstructions performed from 2005 to 2010.
The time of revision ACL reconstruction after primary ACL reconstruction can be seen in Figure 3. In the time periods of 0 to 1 year, 1 to 2 years, 2 to 3 years, 3 to 4 years, and 4 to 5 years after primary ACL reconstruction, 0.6%, 1.2%, 0.7%, 0.5%, and 0.3% were revised, respectively. Thus, the incidence of revision ACL reconstruction peaks between 1 to 2 years and levels off to a very low level at 5 years.
The incidence of anterior cruciate ligament (ACL) revision at yearly intervals after primary ACL reconstruction.
We investigated the influence of age, sex, and cause of initial ACL injury (sports/nonsports) on the risk of revision ACL reconstruction. Patients younger than 20 years at the time of primary ACL surgery had a significantly higher risk of revision ACL reconstruction than patients older than 20 years (adjusted RR, 2.58; 95% CI, 2.02-3.30) (Figure 4). However, sex and cause of primary injury (sports/nonsports) did not influence the risk for ACL revision.
Kaplan-Meier survival curve divided into patients younger and older than 20 years at the time of surgery for primary anterior cruciate ligament reconstructions performed from 2005 to 2010.
Clinical Outcome After Primary and Revision ACL Reconstructions
When comparing outcomes between primary ACL reconstruction and revision ACL reconstruction, the following findings were made. Of 12,193 primary ACL reconstructions and 1099 revision ACL reconstructions, 1-year KOOS data were available for 4799 primary ACL reconstructions (39%) and 303 ACL revision reconstructions (28%). Subjective outcome scores based on the KOOS score were significantly lower 1 year after revision than after primary ACL surgery (P < .001) (Table 2). Using a KOOS quality of life score <40 as a subjective clinical failure criterion, 20% of primary ACL reconstructions and 38% of ACL revision reconstructions were subjective failures.
KOOS and Tegner Scores Before and 1 Year After Surgery for Primary and Revision ACL Reconstruction a
Reconstructions occurred between July 1, 2005 and December 31, 2009. ACL, anterior cruciate ligament; KOOS, Knee Injury and Osteoarthritis Outcome Score.
KOOS subscores are presented as mean ± standard deviation values.
KOOS4 is a modified KOOS score defined as the average of the subscores of symptoms, pain, sports, and quality of life subscores.
Tegner scores are presented as median (range) values.
Significant difference between ACL primary and ACL revision parameters at 1-year follow-up.
Functional activity based on the Tegner score was also lower 1 year after revision than after primary ACL surgery. Knee laxity measured as side-to-side difference at 1-year follow-up was 1.5 mm for primary ACL reconstructions and 1.9 mm for ACL revision reconstructions. This difference was not significantly different.
Risk of Re-revision
The main failure parameter for revision ACL reconstruction that can be extracted from this national clinical database is the incidence of re-revision reconstruction presented in Figure 5. The 5-year re-revision rate was 5.4%. The use of allograft tissue for the revision procedure resulted in a higher risk of re-revision than autograft tissue (RR, 2.05; 95% CI, 1.5-2.4) (P < .01). Sex did not influence the risk for re-revision.
Kaplan-Meier survival curve of revision anterior cruciate ligament (ACL) reconstructions with the end point of re-revision reconstructions. Data for all registered Danish ACL revisions from 2005 to 2010 are indicated by thin lines adjoining the curve.
Another indication of failure after revision ACL reconstruction is a lack of improvement in knee stability from preoperatively to postoperatively. In the revision ACL cohort, no improvement of sagittal knee stability was found in 15% of cases. In comparison, the proportion of primary ACL reconstruction cases that did not experience improvement in knee stability was 12%.
Discussion
In this observational population-based study, the 5-year revision rate after primary ACL reconstruction was 4.1%. The rate of revision peaked 1 to 2 years after primary ACL reconstruction. Patients younger than 20 years of age at the time of primary ACL reconstruction had a higher risk of revision ACL reconstruction than patients older than 20 years of age.
Despite achieving acceptable knee stability after revision ACL reconstruction, subjective outcome was less favorable than after primary ACL reconstruction at 1-year follow-up. The risk for re-revision was 5.4% after 5 years. Allograft usage was a risk factor for re-revision.
This is the first nationwide and population-based study that is able to present patient characteristics, graft choices, and clinical outcomes for patients with revision ACL reconstruction. In Denmark, approximately 3000 knee ligament reconstructions are performed yearly, and more than 85% of these operations are registered in the Danish ACL reconstruction registry. This provides a good validity of the results presented.
The epidemiological characteristics of a large ACL revision cohort of 460 patients have been described by the MARS group. 21 That cohort differs from our cohort because of differences in surgical strategies between the United States and Europe. Most importantly, the use of allografts is much more prevalent in the United States. The primary findings regarding the risk of revision ACL reconstruction after primary ACL reconstruction are that patients younger than 20 years have a higher risk of revision ACL reconstruction than patients older than 20 years and that revision ACL reconstruction is performed in a younger age group than for primary ACL reconstruction. Age as a risk factor has been found in a cohort of 338 primary ACL reconstruction patients. In that study, patients younger than 20 years had a revision risk of 13% compared with 1% if the patients were older than 20 years. 10
The most likely cause for this finding is that young patients more often wish to return to preinjury sport activities after ACL reconstruction. Contact sports are by far the most frequent cause for ACL injury in teenagers. Especially for female athletes, there is an increased risk for ACL injury compared with male athletes.1,6,9 Typically, there is a strong wish for teenaged athletes to return to their previous sport activities and at the same level. As previous studies have demonstrated a high risk for reinjury and subsequent re-revision 12 of up to 25% for female handball team players, we consider the above issues as the main reasons for why younger patients have a higher risk of revision ACL surgery. Another finding that supports the theory that a return to sports is a significant factor for reinjury of the ACL is the time point at which the patients are revised. As demonstrated in Figure 2, the peak period for revision ACL reconstruction is between 1 and 2 years after primary ACL reconstruction, which is the typical time for a return to pivoting sports after primary ACL reconstruction, taking into account the lack of time between reinjury and scheduling the revision surgery.
Another important finding of the present study is that the clinical outcome after ACL revision based on subjective outcome scores is less favorable than after primary ACL reconstruction. Similar findings have been seen in larger case series of revision ACL reconstructions when compared with reference outcome data for primary ACL reconstructions and also from multicenter cohorts such as the MOON study.9,10,14,18,19 Poorer outcome is found for the patient-based parameters in the KOOS score and for functional activity. However, knee stability as evaluated by follow-up instrumented knee laxity measurements did not differ between primary and revision reconstruction patients. Similar findings were seen in a German study comparing 166 revision ACL reconstruction cases with matched primary ACL reconstructions. 18 We suggest the use of a KOOS quality of life subscore <40 as a subjective failure criterion. With use of this parameter, as many as 20% of primary ACL reconstructions and 38% of ACL revision reconstructions are subjective failures. This indicated that patient perception of life limitations after ACL surgery is much more pronounced than what is expected from reoperation rates.
Revision procedures are able to restore knee stability, but symptoms and pain reduce knee function, resulting in poorer subjective outcome. As suggested by a French study, this can most likely be explained by accumulated meniscal and cartilage injuries in the revision reconstruction patients. 16 In the MARS ACL revision cohort, meniscal and cartilage injuries were seen in 90% of patients, and meniscal injury at the primary reconstruction resulted in an increased risk of cartilage deterioration at the time of revision surgery.3,19,21 The negative effect of intra-articular degenerative changes on clinical outcome is supported by a study that has demonstrated a correlation between radiographic osteoarthritis changes and subjective outcome after ACL revision. 2
To the best of our knowledge, the present study is the first nationwide and population-based study presenting 5-year survival profiles after ACL reconstruction. The incidence of revision ACL reconstruction at 5 years was found to be 4.1%. However, the true failure rate is higher because a clinical failure on primary ACL reconstruction not resulting in revision ACL reconstruction may occur in a number of cases. Some patients have insufficient restoration of knee laxity after primary ACL reconstruction but accept a change in activity level instead of repeat surgery. Another group of patients suffer from postoperative chronic pain or pain in relation to sport activities, which in many cases cannot be relieved by revision ACL reconstruction. Thus, the incidence of revision ACL reconstruction underestimates the true failure rate of primary ACL reconstruction.
A recent French multicenter study investigating descriptive data of ACL revisions in 293 patients demonstrated that the main causes for ACL graft failure were femoral tunnel position (36%), new trauma (30%), and unknown cause (15%), 16 which agree with our findings. The incidence of new trauma with subsequent new knee instability is probably the only valid estimate of the listed causes. Also, the incidence of new trauma has been found to be comparable in the French, the MARS, and the present ACL revision cohort. Recent intraoperative evaluation studies have demonstrated that the most frequent graft failure characteristic was graft elongation and not graft rupture. 17 Such graft elongation can be a result of impingement due to poor placement, multiple minor trauma, and biochemical host factors that cause structural deterioration of the graft.
However, the cause of ACL failure is difficult to assess. In a national registry, the determination of cause is based on the operating surgeon’s best judgment. Although definitions of poor tunnel placements and tunnel widening can be defined and evaluated objectively on radiographs, many patients with poor tunnel placement and significant tunnel widening have well-functioning ACL reconstructions, so a specific link between these phenomena and ACL graft failure cannot be established. There are, however, studies that demonstrate a correlation between nonanatomic graft placement and increased risk of ACL revision. 8
The risk of re-revision in the present study was found to be 5.4% at 5-year follow-up. This is a risk comparable with the revision risk after primary ACL reconstruction. Preliminary data from the MOON cohort at 2-year follow-up demonstrated a reoperation risk of 15%, of which 2.5% were re-revisions. 20 However, the latter prospective cohort was limited to 39 patients, making the statistical background for the re-revision rate potentially weak. Acceptable objective knee stability after ACL revision was achieved, with an average side-to-side difference in sagittal knee laxity of 1.9 mm. However, 15% of patients did not have any improvement in knee stability after the revision procedure. This measure of failure could contribute to poorer subjective outcome findings. We found that revision reconstructions performed with allografts had 2 times the risk of re-revision surgery. One study has compared allograft and autograft usage for ACL revision reconstruction. That study had only 14 autograft and 15 allograft patients and 2 years’ follow-up, so re-revision rates could not be compared. Yet, more extension problems were seen in the autograft group, and more lateral osteoarthritis and tunnel widening were seen in the allograft group. 11
The clinical importance of this study is that the data can serve as reference material for ACL revision incidence and outcome after revision ACL reconstruction. Therefore, outcomes after future improvements of both primary and revision ACL reconstruction treatment strategies and surgical techniques can be compared to results from the present study.
The data from this national clinical registry have several limitations. One of the problems is data completeness and compliance of subjective patient registrations. The Danish national registry of patients was established in 1977 and keeps records of all contacts and treatments at public somatic hospitals in Denmark. This enables a determination of patient registration completeness in the national clinical registries on the individual level. We have registration completeness higher than 80%, which is considered acceptable for a national prospective cohort. The task of having patients report subjective scores over the Internet has been very challenging. An average of 39% and 28% of patients reported data preoperatively and at 1-year follow-up, respectively. This is not optimal for the evaluation of subjective data. For that reason, a validation study has been performed to test whether there was a difference in subjective outcome between responders and nonresponders at a follow-up time of 2 to 3 years. The validation study did not demonstrate any such difference, and we, therefore, consider our subjective data to be valid and representative for the entire cohort.
Clinical registries are population based with large sample sizes, providing a high precision of estimates, and ensure generalization. They enable multiple subgroup comparisons that normally would not be feasible in randomized controlled trials. Collection of data is done prospectively and independently of the future research aim, reducing both selection and information bias. However, difficulties with validation of the data, misclassification, and lack of potential confounding data are the issues in registry-based studies. Registry-based studies have the possibility to use epidemiological and statistical methods to prevent confounding, including matching, restriction, stratification, standardization, and regression analyses.
Conclusion
In this observational population-based study, the 5-year revision ACL reconstruction rate was 4.1%. The rate of revision ACL reconstruction peaked 1 to 2 years after primary ACL reconstruction. Patients younger than 20 years of age at the time of primary ACL reconstruction had a higher risk of revision ACL reconstruction than did patients older than 20 years of age. Despite achieving acceptable knee stability after revision ACL reconstruction, subjective outcome was less favorable than after primary ACL reconstruction. Because revision ACL recontruction is performed in young patients, there is a strong need to monitor the incidence and outcome to be able to improve the future outcome of the procedure. This is the first study to present revision rates and outcomes after ACL revision based on a national cohort with the diversity in surgeons and techniques that a national cohort represents.
Footnotes
One or more of the authors has declared the following potential conflict of interest or source of funding: This study was supported by the Danish Rheumatism Association.