Abstract
Background:
Evidence for the success of a meniscal repair performed alone versus combined with anterior cruciate ligament reconstruction (ACLR) is equivocal. No large-scale comparative studies exist regarding this issue.
Hypothesis:
In the general population, meniscal repair in a presumed stable knee has the same rate of reoperation as meniscal repair performed with ACLR.
Study Design:
Cohort study; Level of evidence, 3.
Methods:
All meniscal repairs performed with ACLR in Ontario, Canada, between July 2003 and March 2008 in patients aged 15 to 60 years were identified using administrative billing, diagnostics, and procedural coding. This cohort was matched 1:1 for sex, age, and calendar year of surgery with a cohort of patients who underwent meniscal repair alone. The McNemar test of matched pairs was used to compare reoperation rates (debridement or repair) within 2 years of the index procedure. Conditional logistic regression analysis was used to identify potential risk factors for reoperation among unmatched patient (socioeconomic status surrogate, comorbidity) and provider (surgeon volume, academic hospital status) factors.
Results:
Of 1332 patients who underwent meniscal repair and ACLR, 1239 (93%) were matched with patients who underwent meniscal repair alone. The rate of meniscal reoperation was 9.7% in the combined cohort compared with 16.7% in the repair alone cohort (P < .0001). In the regression analysis, only ACLR was protective against meniscal reoperation (odds ratio, 0.57; P < .0001). Surgeon volume of meniscal repair did not influence outcome.
Conclusion:
A meniscal repair performed in conjunction with ACLR carries a 7% absolute and 42% relative risk reduction of reoperation after 2 years compared with isolated meniscal repair.
Although not always indicated, meniscal preservation through the repair of traumatic tears remains an important goal in knee injury management. Many factors are thought to influence the healing of meniscal tears, including type, chronicity, and vascular supply. Knee stability and concurrent reconstruction of the anterior cruciate ligament (ACL), when injured, may confer additional healing benefits, but the published evidence is limited. 35 Biomechanical data 2 and early reports of the natural history of meniscal tears in the ACL-deficient knee 21 suggest meniscal repair (MR) in isolation should not be performed in the ACL-deficient knee. Gallacher et al 12 published a series of 24 patients who underwent staged repair of a torn meniscus and ACL, beginning with an MR, and the success was 50%. They compared these patients with 148 of their own patients who had undergone simultaneous MR with ACL reconstruction (ACLR) and had a 72% success rate.
At present, there is no high-level comparative evidence for a difference in meniscal healing rates when repair is performed with or without ACLR. However, within some case series, authors have incidentally found MR to be more successful among those undergoing concomitant ACLR compared with MR in stable knees.4,6,13,14,18,19,38 Other series have failed to identify ACLR as a factor in MR success.22,32,35
Recently, published case series of MR performed with ACLR quote very high success rates by either repeat arthroscopy or clinical outcome, ranging between 83% and 96%.1,10,13,37,39 However, the type and location of tears and the methods of fixation employed in these studies were variable. No comparisons with MR in isolation were made.
The influence of patient age on MR is also controversial. Published case series suggest that MR of tears in the vascular zone with or without ACLR has a similar healing prognosis irrespective of age up to 58 years.32,35 In contrast, a histological study 28 has demonstrated poorer cellularity and healing potential of menisci in patients over age 40 years, suggesting there is an effect of age on healing potential. One prospective study of 1485 meniscal tear patterns in patients with stable knees demonstrated a higher likelihood of encountering a meniscal tear not suitable for repair with increasing age. 29
Another factor in the outcomes of elective orthopaedic surgery is the role of provider volume. Population data from total joint arthroplasty, including shoulder, 18 knee, 17 and hip 20 replacement, suggest that lower surgeon volume is associated with higher revision rates and morbidity/complications. The effect of surgeon volume on MR has never been evaluated.
Based on the current state of knowledge, we hypothesized that the success of MR alone in a stable knee would be the same as the success of MR with ACLR as measured by the meniscal reoperation rate in 2 population groups of equal age and sex. To test this hypothesis, we designed a population-based matched-cohort study examining reoperation after MR in patients who underwent a concomitant ACLR compared with patients who underwent MR alone. Our secondary objective was to examine the influence of additional patient and provider factors on reoperation after MR. We hypothesized that lower surgeon volume of the procedure would be associated with a higher risk of reoperation after MR.
Materials and Methods
The data for this study were obtained via health records reported in Ontario and federally (government of Canada). Data were administered by the Institute for Clinical Evaluative Sciences (www.ices.on.ca). All persons between the ages of 15 and 59 years who underwent ACLR in Ontario between July 1, 2003, and March 31, 2008, were identified from the Ontario Health Insurance Plan (OHIP) Physician’s Claims database with the use of fee codes (see Appendix 1, available in the online version of this article at http://ajs.sagepub.com/supplemental/). The OHIP provides universal health care coverage for all residents of Ontario, and it is estimated that 95% of physicians in Ontario are paid on a fee-for-service basis by the OHIP. All OHIP data include demographic information for each patient via the Registered Persons Database and diagnosis codes associated with each fee code. Although the procedural and demographic data are considered highly accurate, the diagnostic codes of the OHIP are considered vague and nonspecific and therefore were not used. 40
All patients identified via OHIP fee codes have been assigned an anonymous number and were linked to hospital admission databases (Same Day Surgery [SDS], Discharge Abstracts Database [DAD]) administered by the Canadian Institute for Health Information (CIHI). The SDS/DAD contained International Classification of Diseases, Tenth Revision (ICD-10) diagnostic codes and Canadian Classification of Health Interventions (CCI) procedural codes. These data originate from each hospital’s medical records department.
The cohort was gathered using OHIP fee codes with a service date corresponding to an SDS/DAD record. Relevant ICD-10 and CCI codes (see Appendix 1, available online) were used to further define study inclusion criteria, exclusion criteria, and outcomes. Basic entry required the OHIP fee code for knee ligament reconstruction and CCI procedural code for “cruciate ligament.” Exclusion criteria were applied based on factors deemed to be clinically relevant. Patients 14 years and younger, for fear of the influence of open physes on ACLR, and 60 years and older were excluded. Patients admitted through the emergency room were excluded because of extremely low numbers and the high risk that these patients represented polytrauma patients or knee dislocations that were inappropriately categorized. Those with 2 or more ligament reconstruction OHIP fee codes associated with the same hospital admission were excluded on the basis of either a bilateral procedure or billing anomaly. Patients who were non-Ontario residents at the service date (eg, visiting from another province) were excluded on the basis of poor expected follow-up care. Patients whose index event included an OHIP fee code for revision ligament reconstruction, collateral ligament surgery, high tibial osteotomy, or microfracture were excluded. Patients in whom a nonstandard ACLR (thermal shrinkage, suture repair) was identified by CCI coding were excluded. Finally, patients were excluded if the hospital admission of their index event contained an ICD-10 diagnosis code for posterior cruciate ligament injury.
The OHIP fee codes were then used to identify a subgroup within this cohort of ACLR patients who underwent a concurrent MR. Those who underwent meniscal debridement were excluded.
Matching
The unexposed group was developed using only OHIP fee codes for MR and not hospital admission data. These patients were matched 1:1 by sex, age within 5 years, and surgery in the same calendar year. Patients were excluded if they appeared in the ACLR cohort or had another procedure billed with the same service date.
Main Outcome
The main outcome was a repeat meniscal operation identified via an OHIP fee code as either MR or meniscal debridement within 2 years from the index surgery. Information on side of surgery was incomplete and therefore excluded from the main analysis. However, to obtain a better understanding of the number of reoperations that were in fact contralateral surgery, we obtained CCI data on side of surgery for every patient identified as having a positive main outcome. These patients were then classified as having an ipsilateral, contralateral, or side-unknown reoperation.
Data Analysis
The McNemar test for paired data was used to compare the reoperation rate between groups. Conditional logistic regression analysis was performed on all patients (combined as 1 group) to identify unmatched confounding variables. These included comorbidity, income quintile (a surrogate for socioeconomic status), and surgeon volume of MR.
Surgeon volume was calculated anonymously using OHIP fee codes submitted by the index surgeon in the prior year from July 1 to June 30. Cases where the index surgeon billed as an assistant were excluded from the calculation. Although generated as a continuous variable, surgeon volume was converted to a categorical variable approximately equally distributed by the number of patients into quartiles, a commonly employed technique in published population data.20,25 However, volumes changed each year; therefore, average hard cutoff points were chosen for each quartile: group 1 = 1-4 per year; group 2 = 5-10 per year; group 3 = 11-20 per year; and group 4 = 21+ per year, with a maximum of 89 per year for the highest-volume single surgeon in any single year. A separate category for surgeons who had billed zero MR in the year prior was created, after it was discovered that this occurred in approximately 10% of index cases. This group was thought to reflect very low-volume surgeons and/or those new to practice. A power analysis for the secondary hypothesis was not performed because the variance for success of MR by hospital status and surgeon volume is unknown.
An established method to estimate patient income, as a reasonable surrogate for socioeconomic status, was employed. Using 2001 census data, Statistics Canada has calculated the average income per single-person equivalent in geographic enumeration areas. From this distribution, corresponding mean income quintiles have been established across Canada. 5 The national income quintile associated with each patient’s home postal code associated with the index hospital admission was determined and used as a demographic parameter in the analysis.
The hospital where the index operation was performed was identified via CIHI hospital discharge data and categorized as either “teaching” or “nonteaching” based on the membership of that hospital in the Council of Academic Hospitals of Ontario.
The modified Charlson-Deyo index 8 was used to assign a comorbidity score between 0 and 6. This index was calculated according to established methods using a 3-year look-back window. In joint replacement literature, this index has been shown to predict complication rates. 23
A further stratified analysis was performed to adjust for the effect of surgeon volume by applying the additional matching criterion of surgeon volume, and McNemar paired analysis was repeated. All reported P values are 2-tailed with an α of .05. Odds ratios (ORs) and 95% confidence intervals (CIs) are presented for the results of logistic regression. Analyses were performed using SAS version 9.1 for UNIX (SAS Institute, Cary, North Carolina).
Results
Based on inclusion/exclusion criteria, we identified 1332 patients who underwent ACLR with concurrent MR (Table 1) and matched 1239 of these cases 1:1 to control patients who underwent MR without ACLR in the same study period. The 93 patients (7.0%) who could not be matched were excluded from the study. These patients were not found to be different than those included by sex distribution, comorbidity, hospital type, surgeon volume, or rate of meniscal reoperation (see Appendix 2, available online); however, the unmatched MR patients were significantly older (P < .001 by post hoc independent-sample t test). Table 2 outlines the baseline characteristics of both groups, including items not matched (surgeon volume, hospital status, income quintile, and comorbidity).
Inclusion/Exclusion Criteria for the Development of the Case Cohort a
CCI, Canadian Classification of Health Interventions; OHIP, Ontario Health Insurance Plan; ACLR, anterior cruciate ligament reconstruction; ICD-10, International Classification of Diseases, Tenth Revision.
Baseline Characteristics of MR + ACLR Cohort and MR Cohort a
MR, meniscal repair; ACLR, anterior cruciate ligament reconstruction.
Five patients in the MR group did not have income quintile data and were excluded from the regression analysis.
The rate of meniscal reoperation was significantly lower in the cohort that underwent the MR + ACLR (9.7%) than in the cohort that underwent MR alone (16.7%) (P < .0001). The absolute risk reduction of meniscal reoperation was 7.0%, and the relative risk reduction was 42%.
The time to meniscal reoperation was significantly shorter in the MR cohort (mean ± standard deviation, 321 ± 200 days) compared with the MR + ACLR cohort (389 ± 170 days) by the Student t test (P = .002). In both cohorts, there was a trend toward earlier meniscal reoperation if the second procedure was a meniscal re-repair compared with partial meniscectomy. In the MR + ACLR cohort, partial meniscectomy occurred after a mean of 395 ± 164 days compared with 360 ± 192 days if a re-repair was performed. In the MR cohort, partial meniscectomy occurred after 332 ± 191 days compared with 272 ± 230 days if a re-repair was performed.
Less than half (43.8%) of all MRs occurred in teaching hospitals; however, the distribution between the 2 cohorts was not equal. Surgeons in teaching hospitals performed significantly more MR + ACLR procedures (51.7%) compared with MR without ACLR (36.0%) (P < .001). In addition, a significantly greater number of MR + ACLR cohort surgeries were performed by the highest-volume surgeons (Table 2). However, conditional logistic regression modeling revealed that surgeon volume, comorbidity score, income quintile, and hospital teaching status were not significant risk factors for meniscal reoperation (Table 3). Anterior cruciate ligament reconstruction was a protective factor on reoperation for MR (OR 0.57; 95% CI, 0.42-0.76; P = .0002).
Conditional Regression Analysis of a Matched Cohort (Age, Sex): Risk of Meniscal Reoperation a
Five pairs were deleted from conditional regression analysis because of missing value on income quintile. OR, odds ratio; CI, confidence interval; ACL, anterior cruciate ligament.
The stratified analysis that matched patients who underwent MR + ACLR to patients who underwent MR alone by age, sex, year of surgery, and surgeon volume yielded 988 matched pairs. The MR + ACLR group had 90 repeat meniscal surgeries (9.1%) compared with 172 (17.4%) in the MR group, which was still significantly different by the McNemar test (P < .0001).
Side of surgery was examined in all patients with a reoperation in the main (not stratified) cohorts. Seventy-seven percent (92/120) of patients in the MR + ACLR cohort had a side of surgery listed for both the index operation and reoperation. Among these, 90% (83/92) had their “reoperation” on the ipsilateral side and 10% (9/92) on the contralateral side. Fewer patients from the MR cohort who underwent reoperation had CCI codes specifying side of surgery: only 115 of 207 (56%). Among these, 89% (102/115) had their “reoperation” on the ipsilateral side and 11% (13/115) on the contralateral side.
Among patients in the MR + ACLR group, we calculated the number of ACL revisions (using OHIP fee codes) within the 2-year follow-up period and found that 25 (2.0%) had a revision ACLR. Among those, 17 (68%) had a concurrent repeat meniscal operation. In comparison, only 9 (0.7%) patients among 1239 in the MR group had an ACLR in either knee within 2 years of their index MR.
Discussion
After accounting for age, sex, and provider volume, we have demonstrated that MR has a significantly lower rate of reoperation when performed with ACLR than without. Furthermore, surgeon or hospital volume and additional patient factors (comorbidity, socioeconomic surrogate) did not influence reoperation. This finding represents the highest current level of evidence supporting this concept.
The universal health care system in Ontario lends itself well to medical coding research. The government is the only payer of orthopaedic operations and the only administrator of hospital visits and admissions. By combining both OHIP fee codes and associated diagnostic and procedural codes from hospital admissions, our goal was to maximize the accuracy of cohort development while still maintaining a high capture rate. Meniscal surgery, both as the index event and outcome, can be reliably confirmed through coding, as it is represented by a very specific OHIP fee code whose remuneration is identical to meniscectomy.
There are numerous potential explanations for the different reoperation rates observed in our matched cohorts. One possibility is a slower or more standardized rehabilitation process in the MR + ACLR cohort because of the ACLR. 2 Despite a lack of published evidence, some authors suggest that menisci heal better when an ACLR is performed because of a biological healing influence brought into the knee by drilling bony tunnels. This theory has spawned proponents of bone marrow stimulation, including via notch microfracture, as a means to augment healing in the treatment of isolated repairs. 11 The literature supporting the use of biological augments such as fibrin clots is equivocal in animals3,34 and limited to uncontrolled older case series in humans.15,16
Another potential explanation for the differential rate of reoperation between these 2 groups may relate to occult instability. Knee instability is believed to decrease the success of MR. 36 We have assumed that patients in our MR cohort have a stable knee, but this may not be true. Some of these knees may have been ACL deficient, with the surgeon and patient electing only to address meniscal injury; thus, occult instability could account for group differences. Nevertheless, we demonstrated that only 9 patients in the MR group, compared with 20 patients in the MR + ACLR group, went on to have a subsequent ACLR. This suggests that the rate of symptomatic instability, and by extension potentially occult instability, was likely very low in the MR group.
Patterns of meniscal injury may also be different between ACL-intact knees, ACL-deficient knees, and in the acute phase of ACL injury. Magnetic resonance imaging and arthroscopic surgery case series have described general patterns including a propensity toward posterior horn medial meniscal tears in chronic ACL-deficient knees7,31,33 and lateral meniscal tears in the acute ACL injury. 7 The potential difference in tear patterns between cohorts is a consideration in this study, as tears in different locations may have different healing potentials.
Another consideration in the differential success of MR with and without ACLR relates to differences in the quality of meniscal tissue. Meister et al 27 histologically examined meniscal specimens from 44 meniscectomies and found a greater proportion of meniscal tissue with a degenerative quality among the samples from ACL-intact knees. They hypothesized that the meniscal tissue quality is worse in isolated MR patients, as these injuries represent a point on the continuum of the natural degenerative process. In contrast, menisci in ACL-deficient knees were found to be healthier and thus hypothesized to fail because of the biomechanical influence of an ACL-deficient knee.
Although reoperation is a practical outcome, the absence of reoperations does not equate to healing. The literature suggests that some patients have menisci that do not heal after MR but remain asymptomatic. Studies utilizing second-look arthroscopy after ACLR and MR identify an incidence of asymptomatic, but incompletely healed, menisci of approximately 10%.1,10,37 The validity of correlating healing and/or clinical success is unknown; however, it seems likely that asymptomatic or minimally symptomatic patients are unlikely to pursue further surgery and can be deemed as a success. The proportion of mildly symptomatic patients who do not pursue further surgery is also unknown.
We chose not to include side of surgery as a requirement in the analysis of these data. Doing so would have decreased the sample size and potentially introduced further bias. Our goal was to develop as complete a population cohort as possible. Among patients with a reoperation and available data, we found approximately 90% of patients had a reoperation on the ipsilateral side to the index procedure, irrespective of which cohort they entered. Even though the side of surgery was not available in all patients, this finding strengthens our main conclusion. In addition, it informs that the differential reoperation rate concluded in our study between the 2 matched cohorts is only minimally inflated (by ~10%).
We noted that a significantly greater number of MR + ACLR procedures were performed by high-volume surgeons, but this did not influence the risk of reoperation. The difference identified in the distribution of surgeon volume between cohorts may be explained by practice patterns. Musahl et al 30 demonstrated that sports fellowship–trained surgeons in the United States were more likely to perform MR with ACLR than surgeons trained in other subspecialty areas or those without fellowship training. Presumably, sports fellowship–trained surgeons compose much of the high-volume surgeon group in our study.
We noted that healthy patients seek medical attention for both MR and ACLR; every patient had a modified Charlson score of zero. This index has been used more successfully in total joint replacement cohorts, which are significantly older with more comorbidity. A better system for classifying comorbidity in younger, healthier population-based data is warranted.
Socioeconomic factors have been correlated with surgical outcomes in mixed private and public health care systems. 24 In theory, equal access in a universal health care system suggests that socioeconomic factors should not affect patients’ access to care, quality of care, and follow-up. Using income quintile (estimated from postal code/address) as a surrogate for socioeconomic status, we showed that this had no influence on reoperation; however, no individual measure of income or similar socioeconomic status indicator was available. Conclusions based on a population-level measure should be cautioned, as they may not apply on the individual level. Of note, almost half of the study patients were from the 2 highest national income quintiles. Although we found no influence of socioeconomic status on reoperation, this suggests that socioeconomic status may still play a role in access to care in our public health system, which is a concept that warrants further research.
Finally, our results refer to within 2 years postoperatively and do not inform the long-term natural history of MR. A clinically relevant length of follow-up after MR remains unknown. Two series have examined up to 10 years’ follow-up and have shown an approximately 25% failure rate by repeat meniscal surgery in ACL-stable knees.9,26
Many other factors that potentially influence the success rate of MR were beyond the scope of this study. These include tear location (medial or lateral), size of tear, chronicity, zone and pattern, repair technique (inside-out vs all-inside), use of biological augmentation, and postoperative rehabilitation protocols. Nevertheless, our findings are representative of the true population and the largest dataset on this topic.
We conclude that concomitant ACLR is statistically and clinically significantly protective against reoperations after MR, irrespective of provider factors, with an absolute risk reduction of 7%. Research should be directed at understanding why this phenomenon occurs so as to emulate the more favorable scenario when repairable torn menisci are seen in the setting of ligament-intact knees.
Footnotes
Presented at the 37th annual meeting of the AOSSM, San Diego, California, July 2011.
One or more of the authors has declared the following potential conflict of interest or source of funding: This study was supported by the Institute for Clinical Evaluative Sciences, a nonprofit research institute sponsored by the Ontario Ministry of Health and Long-Term Care.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
