Optimizing anterior cruciate ligament reconstruction: Individualizing the decision-making process using data from the Kaiser Permanente ACLR Registry: 2018 OREF award paper
Abstract
Despite years of study, controversy remains regarding the optimal graft for anterior cruciate ligament reconstruction (ACLR), suggesting that a single graft type is not ideal for all patients. A large community based ACLR Registry that collects prospective data is a powerful tool that captures information and can be analyzed to optimize surgery for individual patients. The studies highlighted in this paper were designed to optimize and individualize ACLR surgery and have led to changes in surgeon behavior and improvements in patient outcomes. Kaiser Permanente (KP) is an integrated health care system with 10.6 million members and more than 50 hospitals. Every KP member who undergoes an ACLR is entered into the Registry, and prospectively monitored. The Registry uses a variety of feedback mechanisms to disseminate Registry findings to the ACLRR surgeons and appropriately influence clinical practices and enhance quality of care. Allografts were found to have a 3.0 times higher risk of revision than bone-patellar tendon-bone (BPTB) autografts. Allograft irradiation >1.8 Mrad, chemical graft processing, younger patients, BPTB allograft, and male patients were all associated with a higher risk of revision surgery. By providing feedback to surgeons, overall allograft use has decreased by 27% and allograft use in high-risk patients ≤21 years of age decreased 68%. We have identified factors that influence the outcomes of ACLR. Statement of Clinical Significance: We found that information derived from an ACLR Registry and shared with the participating surgeons directly decreased the use of specific procedures and implants associated with poor outcomes.
1 INTRODUCTION
Anterior cruciate ligament (ACL) injuries are often associated with sports participation with a high annual incidence rate estimated at 68.6 injuries/100,000 person-years.1 While ACL deficiency can lead to instability of the knee, some people are able to cope with an ACL deficient knee.2 However, those wishing to return to pivoting and twisting sports often require anterior cruciate ligament reconstruction (ACLR). The incidence rate of ACLR has steadily increased from 33.0/100,000 person-years in 1994 to 45.1/100,000 person-years in 20063 with the most significant increases found in females and adolescents.4
Although ACLR does not return the knee to its preinjury state, it can improve knee stability when compared to knees managed nonoperatively.5 In a large meta-analysis, over 80% of those undergoing ACLR were able to return to sports, although only 65% were able to return to the same level as before their injury.6 In a study of American high school and collegiate athletes, <70% returned to sports and only 43% described themselves as participating at the same level.7 These studies highlight the need for continued study to improve patient outcomes after ACLR. While successful surgery can be difficult to define, revision ACLR surgery is a good endpoint for a failed ACLR, because it is easily definable, and both the patient and the surgeon have agreed that the surgery was not successful.
Many factors may contribute to the outcome of ACLR surgery. Graft choice is one factor that has been extensively evaluated, but many questions remain regarding the best graft for ACLR. In 1985 Clancy described using the middle third of the patellar tendon as a free graft to reconstruct the ACL.8 This bone-patellar tendon-bone (BPTB) construct has been considered the gold standard for over 30 years.9 BPTB autografts are strong and have the advantage of bone to bone healing of the graft within the bone tunnel, but patients may be more prone to anterior knee pain due to the graft harvest site.10 Hamstring autografts gained popularity in the United States and abroad, and are used in over 84% of surgeries performed in Scandinavian countries.11 Hamstring autografts are also strong and may be associated with less kneeling pain, but require tendon to bone healing and have been shown to be associated with some residual hamstring weakness.12 Many studies, including a Cochrane review, have suggested no difference in outcome when comparing BPTB with hamstring autograft,13 but large registry studies have recently shown a higher risk of revision with hamstrings autografts compared to BPTB autografts.11, 14 Allografts, a third alternative, offer several advantages over autografts including no harvest site morbidity, shorter operative time, improved cosmesis, and less postoperative pain. Slower graft incorporation and cost are drawbacks, but usage in the United States has been high.15, 16 While some studies have shown no difference in allografts compared to autografts17-21 others have identified higher rates of revision with allograft use.22-27 Because many of the studies evaluating these questions have been underpowered to detect differences in graft performance due to their small sample sizes, systematic reviews and meta-analyses have been conducted to compile and summarize these different studies. Some reviews have suggested that autografts have better stability or lower revision rates,28-32 yet an almost equal number have found no difference between allografts and autografts.33-37 The reasons for the discrepancy in observed failure rates was identified by the lack of accounting for the age of the patient.38 The strongest evidence regarding allograft performance comes from several large cohort studies, which have found a two to four times higher rate of revision with allografts compared to autografts.14, 38, 39 As a result of the >10-fold number of ACLR patients compared to the entire Multicenter Orthopaedic Outcomes Network (MOON) cohort, our registry can identify potential differences between allograft types and allograft processing.
Despite years of study, controversy remains regarding the optimal graft for ACLR, suggesting that a single graft type is not ideal for all patients. Because of the complexity of this issue (i.e., numerous factors associated with graft performance), a well-designed study with a large, comprehensive patient cohort undergoing ACLR is necessary. A large community based ACLR registry that collects prospective data is a powerful tool that captures this information and can be analyzed to optimize surgery for individual patients.
Orthopedic registries have a long and successful history beginning with the Swedish Hip Register in 1979, which reduced revision rates after hip replacement surgery.40 The value of registries lies in prospectively capturing large cohorts of patients and monitoring their outcomes. Because of the large numbers, prospective data collection, and predefined outcomes and risk factors, registries can compare the effect of multiple factors on the outcome with a high degree of internal validity.41
To demonstrate the effectiveness of registries within the United States and to provide data representative of the US population, we established the Kaiser Permanente ACL Reconstruction Registry (KPACLRR). From its inception, the goal of the KPACLRR was to (1) provide a real world look at clinical practice, patient outcomes, and safety, (2) improve treatment outcomes through feedback to surgeons and hospitals, (3) identify procedures and devices that result in premature failure, and (4) identify prognostic factors that lead to positive and negative outcomes.
This paper represents a compilation of a decade and a half of coordinated contributions of over 340 Kaiser Permanente (KP) sports medicine surgeons across the United States to improve the outcome of ACLR surgery. Since 2010, the KPACLRR has produced 31 peer-reviewed publications and numerous presentations at national and international conferences and led the establishment of the International Society of ACLR Registries in 2010. These findings are integrated into our patient care experience in multiple ways to optimize and individualize patient care including surgeon outcomes reports and a patient risk calculator for shared decision making at the point of care. The culmination of this study including (1) scientific findings, (2) surgeon outcomes reports, and (3) the Risk Calculator that is described in this paper were designed to optimize and individualize ACLR surgery, and have led to changes in surgeon behavior and improvements in patient outcomes.
2 METHODS
2.1 Coverage and study population
KP is an integrated health care system with 10.6 million members and more than 50 hospitals throughout eight geographical regions in the United States (Southern California, Northern California, Pacific Northwest, Mid-Atlantic, Colorado, Georgia, Hawaii, and Ohio). Every KP member in all KP regions who undergoes an ACLR is entered into the Registry, and prospectively monitored.
2.2 Data collection
The KPACLRR was developed through consensus of physicians on key data elements for standardized operative documentation and Registry data collection. This standardized documentation has been integrated in our electronic health record (EHR). In addition to standardized Registry documentation, the ACLR Registry uses EHR, claims data, and other existing administrative databases, to capture patient demographics, comorbidities, anthropometric measures, surgical procedure detail, implant information, and outcomes (Figure 1).
2.3 Quality control and validation
One important feature of the KPACLRR registry data collection process is rigorous quality control and validation of ACLR outcomes. Extensive electronic and manual data quality control checks are applied throughout the data collection, analyses, and reporting process. Data capture is validated through our EHR, claims data, and other administrative data sources. Patients who leave our system are followed up with an annual survey to determine if they have had any additional surgeries.
The registry also has a comprehensive monitoring system of ACLR outcomes consisting of electronic screening algorithms to capture adverse events confirmed with EHR chart review validation by trained clinical experts and application of Agency for Healthcare Research and Quality's Inpatient Quality Indicators.
2.4 Statistical analyses
Numerous statistical methods are used to analyze the registry data including descriptive statistics for reporting and multivariable regression, survival and propensity scores to control for potential confounding variables. Machine learning methods have also been implemented for outlier identification and risk calculator development.
2.5 Registry feedback and clinical decision aid tools (risk calculators)
The registry uses a variety of feedback mechanisms to disseminate Registry findings to the ACLRR surgeons and appropriately influence clinical practices and enhance quality of care. Examples include:
-
Hospital specific quality reports with regional and national benchmarking to monitor infection, deep venous thromboembolism, pulmonary embolism, and revision outcomes
-
Annual registry reports with risk-adjusted hospital ACLR revision rates
-
Surgeon profile reports with patient characteristics, implant utilization, and clinical outcomes benchmarked at the medical center, regional, and national levels
-
Outlier implant reports identifying implants with higher than expected revision rates
-
Implant recall and advisory patient lists for immediate identification and patient management
-
Comparative effectiveness studies identifying clinical best practices. Findings are distributed using electronic newsletters, meetings, webinars, conferences, site visits, and annual Registry reports
-
Risk calculators
To promote personalized preoperative counseling, a risk calculator was developed utilizing the body of evidence built over 12 years. The risk calculator is a prognostic tool that can predict the probability of graft survival within a time period based on specific patient characteristics that we have identified as risk factors of revision. The data set derived from the registry and previous studies was used to develop a risk calculator model.
3 RESULTS
The following studies highlight the systematic investigations using the KPACLR Registry to understand the risk factors associated with positive and negative outcomes after ACLR.
3.1 Description of the KPACLR registry cohort
As of 2016, the KPACLRR registered 39,379 cases. Of these, 36,186 were primary ACLRs and 3193 were revision ACLRs. Over 4000 new primary and 400 new revision cases are performed annually. The registered cohort is racially diverse42 and made up of 63% males and 37% females with a mean age of 29 years. Three hundred and forty-six surgeons contribute cases to the registry, of which 43% are fellowship trained, 77% perform 6–51 ACLRs per year and 67% of hospitals do 24–124 ACLRs per year.43 Tables 1 and 2 present the postoperative outcome and graft choice for this cohort (Tables 1 and 2).
| Primary ACLR | Revision ACLR | |||||
|---|---|---|---|---|---|---|
| Male | Female | Total | Male | Female | Total | |
| Outcomes | N (%) | N (%) | N (%) | N (%) | N (%) | N (%) |
| Total cases registered | 20,093 (62.6) | 12,023 (37.4) | 32,116 | 1745 (62.7) | 1038 (37.3) | 2783 |
| Revised | 663 (3.3) | 420 (3.5) | 1083 (3.4) | – | – | – |
| All reoperations | 1584 (7.9) | 1266 (10.5) | 2850 (8.9) | 163 (9.3) | 129 (12.4) | 292 (10.5) |
| First reoperation site | ||||||
| Same knee | 1008 (5.0) | 820 (6.8) | 1828 (5.7) | 118 (6.8) | 98 (9.4) | 216 (7.8) |
| Other knee | 576 (2.9) | 446 (3.7) | 1022 (3.2) | 45 (2.6) | 31 (3.0) | 76 (2.7) |
| Contralateral ACLR | 420 (2.1) | 368 (3.1) | 788 (2.5) | 19 (1.1) | 20 (1.9) | 39 (1.4) |
| Deep surgical site infection | 49 (0.2) | 28 (0.2) | 77 (0.2) | 10 (0.6) | 3 (0.3) | 13 (0.5) |
| DVT | 66 (0.3) | 22 (0.2) | 88 (0.3) | 5 (0.3) | 4 (0.4) | 9 (0.3) |
| PE | 26 (0.1) | 8 (<0.1) | 34 (0.1) | 1 (<0.1) | 1 (<0.1) | 2 (<0.1) |
| Terminated KP membership | 5246 (26.1) | 2354 (19.6) | 7600 (23.7) | 428 (24.5) | 227 (21.9) | 655 (23.5) |
- Abbreviations: ACLR, anterior cruciate ligament reconstruction; DVT, deep vein thrombosis; KP, Kaiser Permanente; PE, pulmonary embolism.
| Primary ACLR | Revision ACLR | |||||
|---|---|---|---|---|---|---|
| Male | Female | Total | Male | Female | Total | |
| Graft | N (%) | N (%) | N (%) | N (%) | N (%) | N (%) |
| Total graft cases | 19,902 (62.6) | 11,902 (37.4) | 31,804 | 1701 (62.7) | 1012 (37.3) | 2713 |
| Total autograft cases | 12,377 (62.2) | 7071 (59.4) | 19,448 (61.1) | 434 (25.5) | 250 (24.7) | 684 (25.2) |
| BPTB | 5526 (44.6) | 2623 (37.1) | 8149 (41.9) | 243 (56.0) | 129 (51.6) | 372 (54.4) |
| Hamstring | 6438 (52.0) | 4041 (57.1) | 10,479 (53.9) | 161 (37.1) | 96 (38.4) | 257 (37.6) |
| Other | 413 (3.3) | 407 (5.8) | 820 (4.2) | 30 (6.9) | 25 (10.0) | 55 (8.0) |
| Total allograft cases | 7505 (37.7) | 4818 (40.5) | 12,323 (38.7) | 1259 (74.0) | 760 (75.1) | 2019 (74.4) |
| Achilles | 1413 (18.8) | 775 (16.1) | 2188 (17.8) | 263 (20.9) | 158 (20.8) | 421 (20.9) |
| Hamstring | 142 (1.9) | 112 (2.3) | 254 (2.1) | 22 (1.7) | 20 (2.6) | 42 (2.1) |
| Peroneus longus | 496 (6.6) | 345 (7.2) | 841 (6.8) | 64 (5.1) | 47 (6.2) | 111 (5.5) |
| BPTB | 1323 (17.6) | 813 (16.9) | 2136 (17.3) | 340 (27.0) | 183 (24.1) | 523 (25.9) |
| Quad tendon | 36 (0.5) | 25 (0.5) | 61 (0.5) | 2 (0.2) | 1 (0.1) | 3 (0.1) |
| Tibialis anterior | 1842 (24.5) | 1253 (26.0) | 3095 (25.1) | 263 (20.9) | 167 (22.0) | 430 (21.3) |
| Tibialis posterior | 2008 (26.8) | 1349 (28.0) | 3357 (27.2) | 271 (21.5) | 171 (22.5) | 442 (21.9) |
| Other | 245 (3.3) | 146 (3.0) | 391 (3.2) | 34 (2.7) | 13 (1.7) | 47 (2.3) |
| Missing | 20 (0.3) | 13 (0.3) | 33 (0.3) | 8 (0.6) | 2 (0.3) | 10 (0.5) |
- Abbreviation: BPTB, bone-patellar tendon-bone.
We have seen a dramatic increase in the number of ACLRs performed over the past 9 years, especially in adolescent females (Figure 2).
3.2 Surgical timing and impact on ACLR outcomes
Optimal timing for ACL surgery is not clear. Delaying surgery until motion has been restored is important, but the longer one waits the greater the risk of further instability episodes. We have found that compared to being operated within 6 months, finding a medial meniscus tear at the time of surgery is 1.8 times more likely in those operated between 6 and 12 months and greater than two times more likely after 12 months.44 The chance of having a reparable meniscus tear decreased after 1 year, and an increased likelihood of cartilage injury was also noted after 1 year.
3.3 Overall risks of ACLR surgery
Although ACLR surgery has been effective at allowing patients to return to a more active lifestyle, it is not without risk. Using the KPACLRR, in an analysis that included 16,192 ACLRs (15,101 primary and 1091 revisions) cases with a median follow-up of 1.6 years, we determined that 3.7% of primaries had a nonrevision reoperation on the same knee and 1.7% on the contralateral knee.15 The reasons for reoperation included meniscal procedures, cartilage procedures, hardware removal, and arthrofibrosis. Deep surgical site infection developed in 0.3% of primaries and 0.8% of revisions. Symptomatic deep venous thromboses were seen in 0.2% of both primaries and revisions. The overall revision rate for the primary ACLRs was 1.7%, with an annualized rate of revision of 0.9% per year.
3.4 Risk of infection after ACLR by graft type
Although postoperative infections after ACLR are relatively uncommon, the value of a registry is having a large enough cohort with sufficient power to study rare problems and identify possible risk factors for such outcomes. In a study of over 10,000 cases, the overall incidence of surgical site infection was 0.48%.45 Superficial infections were identified in 0.16% and deep infections in 0.32% of the cases. Hamstring tendon autografts were associated with the highest incidence of infection (0.61%), followed by allografts (0.27%) and BPTB autografts (0.07%). After adjusting for age, sex, and BMI, the likelihood of a patient with a hamstring autograft developing a deep surgical site infection was 8.2 times higher than if a BPTB autograft was used.
3.5 ACLR revision risk factors
3.5.1 Impact of age on the risk of revision
Age at the time of ACLR has been shown to be a risk factor for revision in various studies and patient cohorts. We investigated the risk of revision surgery in patients of specific age groups undergoing primary ACLR as well as examined whether the associated risk factors for revision previously identified, including graft type, sex, race, and BMI varied by patient age.46 Of the 21,304 patients evaluated, 33% (N = 7026) were aged <21 years, 27% (N = 5762) were 21–30 years, 22% (N = 4656) were 31–40 years, and 18% (N = 3860) were >40 years. The 5-year revision probability was highest in patients <21 years old (9.0%) and lowest in those >40 years old (1.9%) (Figure 3).
We also determined that patients <21 years of age had a 7.76 times higher risk of revision compared to those >40 years of age, and that graft type, gender, BMI and race varied based on the patients' age. For example, in patients <21 years of age, hamstring autografts were a risk factor for revision compared to BPTB autografts. Also in this cohort, females, black patients and those who were obese had a lower risk of ACLR revision (Table 3). Graft type plays a significant role in the risk of revision surgery, with allografts having a higher risk of revision in all age groups 40 years of age and younger (Figure 4).
| Overall HR (95% CI) | Age <21 HR (95% CI) | Age 21–30 HR (95% CI) | Age 31–40 HR (95% CI) | Age 40+ HR (95% CI) | |
|---|---|---|---|---|---|
| Gender | |||||
| Female vs. male | 0.82 (0.69–0.96)* | 0.76 (0.61–0.93)* | 0.73 (0.50–1.06) | 0.69 (0.39–1.20) | 1.96 (1.05–3.64)* |
| Body mass index (kg/m2) | |||||
| 30–35 vs. <30 | 0.77 (0.65–0.92)* | 0.75 (0.59–0.95)* | 0.86 (0.58–1.27) | 0.98 (0.53–1.80) | 1.01 (0.47–2.17) |
| 35+ vs. <30 | 0.68 (0.54–0.87)* | 0.49 (0.34–0.70)* | 0.86 (0.55–1.36) | 1.20 (0.60–2.40) | 0.89 (0.37–2.14) |
| Race (reference: White) | |||||
| Black | 0.57 (0.41–0.81)* | 0.55 (0.36–0.85)* | 0.70 (0.31–1.60) | 0.69 (0.22–2.16) | Insufficient data |
| Hispanic | 0.76 (0.61–0.96)* | 0.79 (0.61–1.02) | 0.75 (0.46–1.23) | 0.49 (0.26–0.94)* | 1.52 (0.73–3.17) |
| Asian | 0.79 (0.60–1.05) | 0.77 (0.54–1.10) | 1.01 (0.56–1.81) | 0.63 (0.28–1.39) | 0.82 (0.25–2.69) |
| Other/multi | 1.07 (0.73–1.55) | 1.31 (0.86–1.98) | 0.85 (0.35–2.11) | 0.39 (0.05–2.84) | Insufficient data |
| Graft (reference: BPTB autograft) | |||||
| Allograft (any) | 2.63 (2.08–3.33)* | 2.69 (1.94–3.74)* | 2.35 (1.60–3.47)* | 3.04 (1.49–6.18)* | 1.58 (0.53–4.74) |
| Hamstring autograft | 1.43 (1.13–1.80)* | 1.61 (1.20–2.17)* | 1.04 (0.64–1.67) | 1.27 (0.58–2.80) | 0.75 (0.24–2.38) |
- Note: Data are presented as hazard ratios and 95% confidence intervals.
- Abbreviations: ACLR, anterior cruciate ligament reconstruction; BPTB, bone-patellar tendon-bone; CI, confidence interval; HR, hazard ratio.
- * p-value < .05.
3.5.2 Risk of revision by graft type
The most commonly used grafts for ACLR in our cohort are BPTB autograft, hamstring autograft, and allografts. Multiple factors may influence the choice of graft for a particular patient; these could include patient preference, surgeon experience, graft availability, ease of rehabilitation, speed of graft incorporation, harvest site morbidity, and graft survival. In this study, 9817 isolated primary ACLR patients were included. The distribution of grafts utilized in the cohort was 28.4% (N = 2791) BPTB autograft, 30.7% (N = 3012) hamstring autograft, and 40.9% (N = 4014) allograft.14 The revision rate per 100 years of observation was highest in the allograft group (1.23) followed by hamstring autograft (1.10) and BPTB autograft (0.66). The survival for the three grafts are shown in Figure 5.
After adjusting for age, gender, race, and BMI, allografts were found to have a 3.0 times higher risk of revision than BPTB autografts. Hamstring autografts were found to have a 1.8 times higher risk of revision compared to BPTB autografts. Gender specific analysis found that there was a 2.3 times higher risk of hamstring autograft revision in females when compared to BPTB autograft. No higher risk was noted for males. Age was also found to be a significant risk factor for revision, with each year increase in age, the risk of revision decreased by 7%.
3.5.3 Risk of revision based on allograft type and tissue processing
Allografts are a commonly used source of tissue for ACLR and are an attractive option due to ease of use and lack of donor site morbidity. Some allografts are sterilely harvested without any supplemental sterilization methods, but many are put through chemical washes and/or varying doses of irradiation to decrease potential bacterial, fungal, or viral contamination. The lack of clarity in the literature regarding allograft versus autograft performance is due to three primary problems: (1) often age of the patients is not accounted for, (2) many of the studies are underpowered, and (3) those that may have appropriate power often group allografts of different tissue types and processing methods. We performed three studies to help clarify the risk of revision associated with allograft usage.
The first study evaluated the association of allograft processing technique and tissue type with the risk of revision.47 There were 62.9% (N = 3751) allograft ACLRs using soft tissue, 19.9% (N = 1188) with Achilles tendon, and 17.2% (N = 1029) with BPTB. Graft groups included chemical processing (BioCleanse (N = 367), AlloTrue or AlloWash processing (N = 2278)), irradiation >1.8 Mrad (N = 1146), irradiation up to 1.8 Mrad (N = 3637), and no irradiation (N = 1185). After adjustment for patient age, sex, and BMI, the use of BioCleanse (HR = 2.45) and irradiation >1.8 Mrad (HR = 1.64) were associated with a higher risk of revision when compared with other processing methods. BPTB allografts were at higher risk of revision (HR = 1.79) when compared with soft tissue allografts. The use of AlloWash or AlloTrue processing, BMI, and graft donor age did not affect revision rate significantly (Figure 6).
The type of allograft processing and the type of tissue have an impact on the risk of revision after ACLR. Graft irradiation >1.8 Mrad, BioCleanse graft processing, younger patients, BPTB allograft, and male patients were all associated with a higher risk of revision surgery. Patient BMI, graft donor age, and the use of AlloWash or AlloTrue processing did not affect revision risk significantly.
We subsequently compared BPTB autograft with BPTB allograft in the second study, evaluating the risk of revision with the varying types of tissue processing.48 In 18.4% (N = 1029) of cases BPTB allograft (nonprocessed = 160, <1.8 MRads = 543, ≥1.8 MRads = 300) was used and in 81.6% (N = 4557) cases BPTB autograft was used. BPTB allografts had a significantly higher adjusted risk of revision than BPTB autografts (HR: 4.54). This higher risk of revision was consistent with all allograft processing methods when compared to autografts and was also consistently higher in patients with allografts regardless of age (Figure 7).
In the third study, we compared the risk of revision of BPTB and hamstring autografts with soft tissue allografts (the best performing of all of the allografts) while accounting for the various allograft processing methods.49 The cohort included 14,015 cases, of which 32.5% (N = 4557) ACLRs used BPTB autograft, 26.8% (N = 3751) soft tissue allograft, and 40.7% (N = 5707) hamstring autograft. Compared to hamstring autografts, a higher risk of revision was found in allografts with ≥1.8 Mrads without chemical processing after 2.5 years (HR = 3.88), and ≥1.8 Mrads with chemical processing after 1 year (HR = 3.43) and with BioCleanse processed grafts at any time period (HR = 3.02). Compared to BPTB autografts, a higher risk of revision was seen with hamstring autografts (HR = 1.51) and BioCleanse processed allografts (HR = 4.67). Allografts irradiated with <1.8 Mrads with chemical processing (HR = 2.19) and without chemical processing (HR = 2.31) had a higher risk of revision as did allografts with ≥1.8 Mrads without chemical processing after 2.0 years (HR = 6.30) and ≥1.8 Mrads with chemical processing after 1 year (HR = 5.03). Nonprocessed allografts did not have a higher risk of revision compared to autografts (Figure 8).
Irradiated soft tissue allografts were found to have a higher risk of revision compared to BPTB autografts. In comparison with hamstring autografts, grafts irradiated with ≥1.8Mrads had a higher risk of revision. A time dependent interaction was noted and grafts processed with higher doses of irradiation and chemically processed grafts were found to have a higher risk of revision earlier than those without chemical processing. High pressure chemically only processed grafts (BioCleanse) were noted to have a higher risk of revision than both hamstring and BPTB autografts irrespective of time.
3.6 Risk of ACL injury to the contralateral knee
ACLR may help return patients to a more active lifestyle, however, return to activities may expose patients to a re-tear of the reconstructed ACL or a tear of the contralateral knee ACL (CACL), both of which are devastating to the patient. A comparison of the survival of the reconstructed ACL with the contralateral ACLR (CACLR) was undertaken to determine if certain patient factors and graft type at initial reconstruction are risk factors of these events, and to determine whether risk factors for revision ACLR and CACLR are different.50 The 5 year ACL graft survival rate in this study was 95.1% and the CACL survival was 95.8% (Figure 9).
Allografts and hamstring autografts were associated with a higher risk of revision ACLR surgery, whereas BPTB autografts were associated with a higher risk of CACLR (Table 4).
| Total | BPTB autograft | HS autograft | Allograft | |
|---|---|---|---|---|
| Total cases | 17,436 | 4436 | 5568 | 7432 |
| Total person years | 40,678 | 11,615 | 12,393 | 16,577 |
| Revision | ||||
| Revised cases and crude revision rate | 427 (2.45) | 86 (1.94) | 132 (2.37) | 209 (2.81) |
| Incidence density (revision/100 years of observation) | 1.05 | 0.74 | 1.07 | 1.26 |
| CACLR | ||||
| CACLRs and crude CACLR rate | 335 (1.92) | 123 (1.92) | 101 (1.81) | 111 (1.49) |
| Incidence density (CACLR/100 years of observation) | 0.83 | 1.06 | 0.81 | 0.67 |
- Abbreviations: BPTB, bone-patellar tendon-bone; CACLR, contralateral anterior cruciate ligament reconstruction; HS, hamstring.
Males were found to have a higher risk of revision ACLR and females a higher risk of CACLR. Increasing age and increasing BMI decreased the risk of both revision and CACLR. Blacks were found to have a decreased risk of revision ACLR compared to Whites (Figure 10).
The risk of contralateral ACL injury may be overlooked by patients and surgeons during ACL management. Awareness of the risk factors associated with a contralateral knee injury may help guide rehabilitation, with specific consideration to the contralateral limb in those patients at higher risk.
3.7 Minimizing the risk of revision: The KP ACLR risk calculator: A predictive tool for patients and surgeons
Using random survival forest, a machine learning method based on tree ensembles, 22 prospective variables, including patient and operative factors, were screened. Ultimately, the four most predictive variables of revision were chosen for use in the risk calculator. The four variables found to be most predictive of revision ACLR included: age, activity (soccer, basketball, football, skiing, other sport, nonsport), BMI and graft type. These variables can be entered into the Risk Calculator to predict the probability of graft survival through the first 6 years after surgery, and provide comparative risks depending upon selectable risk factors. Figure 11 shows examples of the predicted graft survival for a 16-year-old soccer player and a 40-year-old nonathlete.
Graft survival after ACLR can be predicted using patient specific variables and the graft chosen for reconstruction. The Kaiser Permanente ACLR Risk Calculator (KPARC) provides individualized statistics and can be used by the patient and surgeon to compare modifiable factors that may help decrease the risk of revision.
3.8 Demonstrating the value of registry findings
One purpose of a registry is to identify procedures or devices that have either good or poor outcomes and improve treatment outcomes through feedback to surgeons. Feedback on graft performance was presented through a variety of mechanisms including (1) peer-reviewed publications, (2) internal and external meetings and conferences (3) newsletters of study findings, (4) risk calculators, and (5) confidential individualized reports of surgeon's outcomes. In addition, surgeon champions set a quality improvement goal to reduce allograft usage overall and specifically to decrease the use of high-risk grafts (>1.8 Mrad or BioCleanse processed grafts) and usage in high-risk patient groups (≤21 years of age). Allograft usage was monitored on a quarterly basis to determine if the target was achieved.
Beginning in 2008, the annual proportion of ACLR cases using an allograft increased with a peak of 45% in 2010. By providing feedback to our surgeons, allograft use has decreased in the ensuing years and was 33% in 2015, a decrease of 27%. High-risk allograft usage decreased from 8% in 2011 to 5% in 2015 which is a 38% decrease. Allograft use in patients ≤21 years of age decreased 68% from a high of 28% in 2009 to 9% in 2015. (Figure 12).
Translating registry findings into evidence-based clinical practice is the goal of a registry. We found that information derived from an ACL Registry and shared with the participating surgeons directly decreased the use of specific procedures and implants associated with poor outcomes.
4 CONCLUSION
This body of work represents the combined contributions of over 340 sports medicine surgeons who have collected over 39,000 ACLR's within KP across the United States, who have systematically sought to improve the quality of ACLR for every patient. Collaboration and organization to achieve data harmonization, collection, validation, analysis and feedback for approximately one and a half decades have provided invaluable insights into the current status of ACLR, along with the key factors that influence both positive and negative outcomes. Through these efforts, we have been able to identify factors that influence the outcomes of ACLR, including timing of the surgery, infection and venous thrombosis rate, graft selection, and graft processing, which culminated in a risk calculator for use by patients and surgeons. Our aim has been to continually improve the treatment of ACL injuries that affect so many of our young male and female athletes at an alarmingly increasing rate, and we have demonstrated direct impact on improving the quality of outcomes. Our future work will include efforts at prevention and patient reported outcomes to better understand the impact of ACL injury and treatment on individual patients.
ACKNOWLEDGEMENTS
The authors would like to thank all Kaiser Permanente orthopedic surgeons and the staff of the Department of Surgical Outcomes and Analysis who have contributed to the success of the Kaiser Permanente Anterior Cruciate Ligament Reconstruction Registry.
CONFLICTS OF INTERESTS
This submission and the studies that it contains were funded internally through the Kaiser Permanente Department of Surgical Outcomes and Analysis. None of the authors have any conflicts of interest regarding the material presented in this submission.
ETHICS STATEMENT
All of the studies discussed in this submission were approved by the Kaiser Permanente Institutional Review Board IRB # 5691.
