Introduction

Anterior cruciate ligament (ACL) injuries are among the most common sports-related injuries in the United States, with reported estimates ranging up to 200,000 annually [31, 36, 51]. Arthroscopic-assisted anterior cruciate ligament reconstruction (ACLR) is the most common surgical technique to manage this condition and has led to favorable long-term outcomes [6, 9,10,11, 21, 34].

Reoperation rates following ACLR range between 6.5 and 34% [16, 26, 36, 54, 59]. However, small numbers or short-term follow-up, inclusion of contralateral procedures, or lack clinical detail for procedures performed during reoperation were reported by many of these studies. Additionally, revision ACLR rates without comment on all ipsilateral reoperations were reported by most large database studies [40, 57, 62]. The purpose of this study was to report mid- to long-term rates of all reoperations on the ipsilateral knee following arthroscopic ACLR. The authors aim to use a large, heterogeneous sample to determine whether a variety of captured demographic and surgeon factors are independent risks for requiring further ipsilateral surgical procedures after initial ACLR. Based on findings from previous large registry studies conducted on reoperation rates following ACLR, the hypothesis was that reoperation rates would be low, high-volume surgeons would have lower reoperation rates, and risk factors for reoperation would include female gender and young age.

Materials and methods

The SPARCS database was established by the New York State Department of Health in 1979, and reporting discharges from all nonfederal, licensed hospitals in New York State was mandated by the New York State Public Health Law in 1986. Emergency rooms and freestanding, licensed ambulatory surgery facilities were included in the subsequent expansion of these requirements. Patient level data on patient characteristics, diagnoses and treatments, and charges are now included for each hospital inpatient and outpatient visit. All New York State Hospitals are required to be 100% compliant with submission of SPARCS data. The SPARCS database has been used to examine trends in numerous peer-reviewed orthopaedic publications [5, 20, 29, 44]. Of note, real numbers of all included meniscus and cartilage reoperations are included in the SPARCS database.

Patients were identified utilizing Current Procedural Terminology (CPT) and International Classification of Diseases (ICD-9) codes. Patients between the ages of 10 and 60 years with a primary diagnosis of ACL tear (717.83 or 844.2) and CPT code 29,888 (arthroscopic ACL reconstruction) were identified with methodology similar to previous ACLR database studies [12, 41]. Patients with prior arthroscopic ipsilateral knee surgery were excluded.

Patients who underwent ACLR were tracked using an encrypted unique patient identifier and followed for a minimum of 2 years for the incidence of additional knee surgery. All ACLR identified between 2003 and 2012 were followed through 2014, thus ensuring minimal 2-year follow-up (maximal 11-year follow-up). For patients who had multiple subsequent surgeries, only the first surgery immediately after the initial ACLR was used for the purposes of data analysis. Similar to previous literature, subsequent open and arthroscopic procedures were identified by common CPT codes [12, 24, 27, 36, 45]. All initial procedures were performed arthroscopically.

The study was exempted by the New York University School of Medicine Institutional Review Board as non-human subject research.

Statistical analysis

Statistical analysis using SAS® 9.3 (SAS Institute, Cary NC) was performed to compare those undergoing additional surgery to those that did not. Differences in continuous variables were evaluated with a Student’s t test, while Fisher Exact test and Chi-squared analysis were used for categorical variables. Age was examined as both a continuous and binary variable. Similar to previous methodology, the number of ACLRs performed in New York State 12 months prior to the index ACLR was used to stratify surgeon volume into high, medium, or low [16, 36]. 52 or more ACLR per year were performed by high-volume surgeons, between 6 and 51 ACLR per year were performed by mid-volume surgeons, and those who performed less than 6 ACLR per year were low-volume surgeons. To explore the impact of concomitant procedures on ACLR reoperation rates, patients were collapsed into four groups based on the nature of their index surgery: isolated ACLR, ACLR with concomitant meniscal procedure, ACLR with concomitant meniscal procedure and another procedure, ACLR with concomitant procedure not involving meniscus [16]. Patients with tobacco-use disorder were identified using ICD-9 code 305.1 [12, 56].

Subgroup analysis was performed on patients who had subsequent surgery to determine if subsequent procedures were correlated with mean age at time of initial ACLR. For this analysis, a one-way ANOVA was performed, and subsequent pairwise comparison was performed for each specific subsequent procedure using “overall mean age at reoperation” for comparison. The same method was used to determine whether significant differences occurred in mean time until reoperation for specific procedures. Kaplan–Meier analysis was performed to determine reoperation-free survival, and survival probabilities were provided for 2-, 5-, and 10-year intervals. Reoperation or final follow-up through the year 2014 for all patients was censored. Multivariate logistic regression that controlled for age, gender, race, insurance, surgeon experience, and nature of initial ACLR was performed to determine independent risk factors for subsequent surgery and for revision ACLR. p values < 0.05 were considered statistically significant for all analyses.

Results

45,231 patients met the inclusion criteria and underwent arthroscopic ACLR between 2003 and 2012. Baseline demographic characteristics for the sample and complete distributions for each covariate can be found in Table 1.

Table 1 Patient demographics for outpatient ACLR in NYS 2003–2012
Full size table

Patients requiring ipsilateral surgery and those who did not are compared in Table 2. After a minimum follow-up of 2 years, subsequent outpatient ipsilateral knee surgery at a mean of 25.9 ± 24.5 months after index ACLR was performed in 10.7% of patients (4,830/45,231). Overall rates of subsequent ipsilateral surgery were found to be higher in females than males (11.2% vs 10.3%, p < 0.002). Higher rates of reoperation were associated with concomitant procedures at the time of index surgery compared to isolated ACLR (11.2% vs 10%, p < 0.001).

Table 2 Comparison of those requiring additional ipsilateral outpatient knee surgery to those who did not
Full size table

The distribution of subsequent procedures and most common diagnoses at first reoperation are detailed in Table 3. In patients undergoing a subsequent procedure, 28.3% had a revision arthroscopic ACLR and 0.8% underwent open ACLR. An overall 3.1% rate of ACLR revisions was represented by these patients (1406/45,231). Patients who had a meniscal repair with initial ACLR were 2.5 times more likely to have a subsequent meniscectomy than those who did not (odds ratio 2.50, confidence interval 2.25–2.77).

Table 3 Distribution of procedures and diagnoses for patients undergoing additional subsequent surgery
Full size table

A subgroup analysis was performed on those 4830 patients who underwent reoperation. Average time in months until reoperation was widely varied based upon nature of subsequent procedure (Fig. 1). As was outlined in the “Methods” section, the reference value for statistical comparisons in this analysis was the mean time to reoperation for the sample (25.9 months; SD 24.9). Stiffness-related procedures (8.9 months; SD 14.9) were performed significantly earlier than reference value (p < 0.05). By comparison, subsequent revision ACLR (27.4 months; SD 23.7), meniscectomy (32.1 months; SD 26.5), and meniscal repair (27.6 months, SD 26.5) were performed on patients after the reference value of 25.9 months (p < 0.01 for all values).

Fig. 1
figure 1

Mean time until reoperation by procedure

Full size image

Mean age at time of initial ACLR was associated with the type of reoperation that patients underwent, as depicted in Fig. 2. Similar to above, the mean age at initial ACLR for all patients undergoing reoperation (28.3 SD 11.4) was used as the reference value for statistical significance. Those undergoing revision ACLR (25.4, SD 10.4) and meniscal repair (22.5, SD, 9.4) were younger at time of initial ACLR when compared to the reference value (p < 0.01). Patients who had only one reoperation were 28.5 years old (11.4) at initial ACLR. By comparison, those having two subsequent reoperations were 27.6 (10.9) and those having three or more reoperations were 26.9 (11.3) at initial ACLR, respectively (p < 0.01 for both).

Fig. 2
figure 2

Mean age at initial ACLR for specific reoperation procedures

Full size image

Female gender was associated with a higher overall reoperation rate (11.2% vs 10.3%, p < 0.002). The differences with regard to specific reoperation procedure existing across genders were revealed by subgroup analysis (Fig. 3). Most notably, revision ACLR (3.41% vs 2.65%, p < 0.001) and meniscectomy (5.77% vs 5.34%, p < 0.05) were found to be significantly more likely in males than females. By comparison, a greater incidence of subsequent cartilage procedures (2.33% vs 1.95%, p < 0.05) and nearly twice the rate of stiffness-related interventions (1.42% vs 0.75%, p < 0.001) were identified in females.

Fig. 3
figure 3

Rates of specific reoperation procedures by gender

Full size image

Results from the Kaplan–Meier survival analyses can be found in Fig. 4. Similar survival probabilities, with 2-, 5-, and 10-year probabilities of 98.4%, 96.7%, and 95.6%, respectively, were demonstrated by contralateral ACLR.

Fig. 4
figure 4

Event-free survival analysis after ACLR

Full size image

Age younger than 19 years (OR 1.41, 1.32–1.51, p < 0.001), female gender (OR 1.09, 1.03–1.16, p = 0.004), worker’s compensation insurance (OR 1.70, 1.54–1.88, p < 0.001), and Caucasian race (OR 1.32, 1.23–1.42, p < 0.001) were suggested to be independent risk factors for any ipsilateral knee reoperation after ACLR by multivariate analysis (Fig. 5). An initial isolated ACLR (OR 0.85, 0.80–0.90, p < 0.001) and having initial ACLR performed by a high-volume surgeon (OR 0.90, 0.83–0.97, p = 0.021) were independently associated with lower rates of reoperation, while tobacco use was not significant (OR 0.89, 0.74–1.10, n.s.).

Fig. 5
figure 5

Risk Factors for reoperation after ACLR

Full size image

A separate multivariate analysis (Fig. 6) was performed to identify risk factors specific for revision ACLR. Age younger than 19 years (OR 2.11, 1.89–2.36, p < 0.001), male gender (OR 1.42, 1.27–1.60, p < 0.001), worker’s compensation insurance (OR 1.36, 1.11–1.66, p = 0.002), and Caucasian race (OR 1.29, 1.14–1.46, p < 0.001) were determined to be independent risk factors for revision surgery. High surgeon volume (OR 1.08, 0.95–1.25, n.s.), initial isolated ACLR (OR 1.03, 0.92–1.15, n.s.), and tobacco use (OR 0.93, 0.64–1.34, n.s.) were not independently associated with revision ACLR.

Fig. 6
figure 6

Risk Factors for revision ACLR

Full size image

Discussion

The most important finding in this study was a 10.7% rate of subsequent surgical intervention on the ipsilateral knee at an average of 25.9 (24.5) months. Overall ACLR revision rate in the sample was found to be 3.1%. Independent risk factors for any reoperation were: age under 19 years, female gender, Caucasian race, index ACLR with concomitant procedure, and worker’s compensation claim. Having the index ACLR performed by a high-volume surgeon was associated with reduced incidence of reoperation.

Age under 19 years as an independent risk factor for reoperation was found in the study (OR 1.41, 1.32–1.51) and has been consistently reported in previous studies [19, 23, 40, 62]. Reasons for this finding were likely due to a combination of younger patients demonstrating the highest participation in athletic activities that had predisposed them to initial ACL injuries, a greater willingness to undergo subsequent surgery to be able to return to play, and predisposition for injury or reinjury due to factors such as alignment [35, 37, 38].

Cartilage procedures and stiffness-related interventions were performed in females in the cohort at a greater rate than males (Fig. 3). Interestingly, similar gender-related differences regarding cartilage damage after ACLR were found in a recent animal study by Kiapour et al. In their animal model, statistically significant greater cartilage damage after ACLR was observed in females, potentially due to decreased vascularity in females as compared to males [30]. Future studies addressing gender-related differences are warranted, particularly when considering that over 75% of young individuals had developed post-traumatic osteoarthritis after ACL injury [30, 61].

Nearly twice the reoperation rate for stiffness-related interventions was found in females, a finding which has been previously reported. A study of 933 knees by Nwachukwu et al. found that females were 2.8 times more likely to develop arthrofibrosis than males [43]. Reasons for this gender difference remain unclear, with some authors attributing differences to variations in hormone expression [18, 32, 53], while others had suggested that females may be more likely than males to seek intervention to restore range of motion (ROM) after ACLR [48]. Stiffness-related interventions were performed at a mean of 8.9 months in this study (Fig. 1), and is in agreement with a mean of 9 months described by Nwachukwu et al. [43].

Caucasian race as an independent risk factor for reoperation was identified (OR 1.32, 1.23–1.42), a finding that has been previously reported [38, 58]. Although insurance status was accounted for in the multivariate analysis, socioeconomic status and access to care were unable to be controlled for and are factors that previous authors have suggested may be responsible for observed race-related differences [13]. Worker’s compensation claims as an independent risk factor for reoperation (OR 1.7, 1.54–1.88) is relatively novel in such a large cohort registry study and is in line with previous authors’ reports that workers’ compensation is associated with lower subjective outcomes after ACLR [8], mirroring reports across other orthopaedic subspecialties [3, 4, 60].

Initial ACLR performed by a high-volume surgeon (> 52 ACLR/year) was associated with lower risk of any reoperation (OR 0.90, 0.83–0.97, p = 0.021), but was not associated with reduced risk for revision ACLR (OR 1.08, 0.95–1.25, n.s.). The second part of the results are similar to a study by Wasserstein et al. that found that surgeon volume was not associated with revision rates after ACLR [62]. Future studies detailing how practice patterns and intraoperative decision-making differ based on surgeon experience could shed greater light on the current findings, particularly when considering recent studies that have found high-volume surgeons provided a greater economic benefit than all other providers while having fewer adverse events [49].

In the present cohort, a decreased risk for future reoperation was found in patients who had undergone an initial isolated ACLR (0.85, 0.80–0.90). These results add to the existing literature that had reported worse long-term outcome scores and higher failure rates in patients undergoing ACLR with meniscal or cartilage injuries as compared to those without concomitant pathology [7, 28, 33, 47, 50]. Additionally, the fact that nearly 58% reoperations in the cohort occurred due to meniscal pathology could represent that surgeons are more agreeable to surgical re-intervention due to the recognition of the role that untreated meniscal damage has in development of knee osteoarthritis [22, 55]. The finding that initial isolated ACLR was not associated with reduced need for revision ACLR (OR 1.03, 0.92–1.15, p = 0.605) could be influenced by nature and high degree of athletic involvement in patients typically requiring revision ACL reconstruction.

The overall ACLR revision rate in this study population was 3.1%, similar to other large registry studies that have reported revision rates ranging from 1.8 to 3.0% [2, 37, 46]. Results from the survival analyses indicate 5-year ACLR revision free rates of 96.5%, or failure rate of 3.5%. (Fig. 4). This is comparable to results from Swedish and Danish ACLR registries reporting 5-year revision rates of 4.1% [1, 35]. Due to difference in cohorts and significantly limited sample sizes in previous studies, it is more difficult to make direct comparisons to the 10-year survival rate of 95.4% found in this study. A systematic review of 14 studies by Crawford et al. had reported 10-year ACL re-rupture rates of 6.2% (93.8% survivorship) [15].

Risk of revision between patellar tendon and hamstring autografts was examined by a registry study performed by Gifstad et al., which included 45,998 primary ACL reconstructions in Scandinavia. At 2 years following primary ACLR, 1.0% in the patellar group and 2.3% in the hamstring group were expected to undergo revision. Additionally, younger patients were found to have worse outcomes, and female gender was associated with increased time without revision [25]. These reports were consistent with this study; however, direct comparison was limited because graft source and activity levels were unknown in this study, and any reoperation was included.

To the best of the authors’ knowledge, there are only two similar studies that use a large registry to determine rates of all subsequent reoperations, not just revision ACLR. In 2009, Lyman et al. performed a study utilizing the SPARCS database identifying 70,547 patients in a 9-year period from 1997 to 2006 [36]. They reported subsequent surgery was performed on either knee within 1 year in 6.5% of cases with a subsequent ACLR on either knee in 1.9% of cases. The most common subsequent procedures were ACLR (28.7%), lysis of adhesions (28%), and meniscectomy (24.5%). They reported higher rates of subsequent surgery and revision ACLR in females. Patients who had index ACLR that was performed by a low-volume surgeon (< 52 ACL per year) had significantly more subsequent knee procedures. Younger age (less than 20) was associated with a significantly increased rate of revision ACLR. Despite their findings, the most significant limitation of their results was an inability to differentiate laterality of subsequent surgery. Their data reported on procedures performed on “either knee”, making direct correlations difficult to ascertain, particularly when considering that a significant percentage of patients may eventually injure the contralateral ACL [40].

The second study by Csintalan et al. had identified 14,522 patients with mean age of 29 years followed over the short to mid-term (mean follow-up of 1.9 years) [16]. Their overall reoperation rate was 3.9%, with the majority of reoperations addressing meniscal pathology (41%). Although the majority of reoperations in both studies involved the meniscus (58%), the reoperation rate presented in this study was significantly higher at 10.7%. The reason for the significantly higher reoperation observed in this study was likely threefold. First, they did not include revision ACLR in their results—thus reducing reoperation rates, as nearly 1/3rd of reoperations in this study involved revision procedures. Second, their mean follow-up was 1.9 years ± 1.5 (range 0–6.7) as compared to the mean follow-up of 5.6 ± 2.9 (range 0–12) in this study. Finally, their patient population consisted of only 48.3% Caucasians, as compared to a proportion of 68.6% Caucasian, which is a potentially confounding factor as Caucasian race has been implicated in higher reoperation rates [39, 40].

This study was subject to several limitations. As with any large-scale registry analysis, there was inherent reliance on the accuracy of the database being analysed [42]. However, the current analysis utilized a source with previously accepted standards of accuracy for evaluation [14, 17, 44, 52]. Additionally, the type of graft (allograft vs autograft) used during ACLR, radiographic findings, or the nature and degree of athletic involvement were unable to be accounted for. While ICD-9 diagnosis codes were used to identify tobacco use, longevity and intensity (packs per day) of tobacco use were not available. Additionally, patients who may have had reoperations outside of New York State could not be identified. Finally, it is important to keep in mind that subjective outcome scores were independent of the results, and thus patient satisfaction, quality of life improvements, or return-to-sport could not be commented upon.

Overall, the results of this study can help providers identify patients at highest risk for reoperation, predict the nature of subsequent interventions, and counsel patients accordingly.

Conclusion

A 10.7% rate of reoperation at an average of 25.9 (SD 24.5) months and an overall ACLR revision rate of 3.1% were observed in this sample of 45,231 patients, with the majority of subsequent interventions addressing meniscal pathology. Independent risk factors for both reoperation and revision ACLR were age 19 years or younger, female gender, Caucasian race, and WC claim, findings which can be used to counsel patients receiving arthroscopic ACL reconstruction.