Abstract
Objectives:
The development of knee pain after anterior cruciate ligament (ACL) reconstruction (ACLR) worsens clinical outcomes and degrades patient satisfaction. Subsequent ipsilateral knee surgery, which has been observed to occur in 4 to 28% of ACLR, is perhaps the most significant risk factor for the development of knee pain in this setting. Thus, the identification of risk factors that predict subsequent surgery is worthwhile to improve the clinical outcomes of ACLR. Although useful in clarifying associations between various clinical variables and an outcome of interest, as well as demonstrating the relative importance of risk factors for a predicted outcome, previously employed multivariable prediction models are less useful in determining the magnitude of effect or, more broadly, establish causal relationships between individual exposures and an outcome. Alternatively, causal mediation analysis (CMA) addresses these two limitations and provides an analytical approach capable of distinguishing: 1. The direct effect of an exposure, from 2. The indirect effect of that same exposure through a mediator (or mediating event) on the ultimate outcome of interest. In doing so, CMA provides a more accurate description of the mechanism(s) of how a singular intervention on a risk factor affects an outcome. Therefore, the purpose of the current study was to determine if lateral meniscus (LM) repair versus excision performed at the time of primary ACL reconstruction (ACLR) decreases the likelihood of significant knee pain 10 years follow up. It was hypothesized that subsequent surgery performed after LM repair or excision at primary ACLR would increase the likelihood of significant KOOS knee pain 10 years follow up. It was further hypothesized that LM repair performed at the time of primary ACLR decreases the likelihood of developing significant knee pain 10 years follow up.
Methods:
The study cohort included all subjects undergoing unilateral, primary ACLR from 2002-2008 enrolled in the Multicenter Orthopaedic Outcomes Network (MOON) without a history of prior medial or LM surgery and contralateral ACLR. Causal mediation analysis (CMA) using R software (Version 4.2.3; Vienna, Austria) was employed to compare the effects of LM repair and excision at the time of primary ACLR on the development significant knee pain at 10 years follow up. Significant knee pain was defined as Knee Osteoarthritis Outcome Score (KOOS) pain < 80. Our inclusion/ exclusion criteria at with 10-year clinical follow up were primary ACLR without a history of prior medial and LM surgery, no revision ACLR, and no history of ACLR in the contralateral knee, no subsequent revision ACLR, and no subsequent surgery in which the stated indication was for hardware removal alone (Figure 1.) A directed acyclic graph (DAG) was constructed to provide a qualitative representation of the influence of known confounders which have previously been shown to affect the outcome of interest (Figure 2.) A stepwise approach to the CMA was employed such that Step 1 determined whether LM procedure (i.e., repair or excision) significantly affected the likelihood of subsequent surgery. Provided that this criterion was satisfied, Step 2 would determine whether subsequent surgery affected the likelihood of KOOS pain < 80 at 10 years follow up, and Step 3 would determine if both the effect of LM repair followed subsequent surgery and direct effect of LM repair were significant. Missing data were multiply imputed using multivariate imputation by chained equations (MICE). All tests were two-sided, assuming a Type I error rate of 0.05.
Results:
We had two thousand three hundred and eighty-seven subjects (1074 females (45%), 1313 males (55%)) ACLRs (Figure 2, Table 1.) At 10 years our follow-up was 76.5% (1825 of 2387). In 1227 (51.4%) cases, there was no lateral meniscus tear reported at the time of primary ACLR. The most common type of LM tear was a complex tear (214 cases, 18.4%.) Of the cases with LM tears, no treatment was performed in 254 (10.6%) cases, excision was performed in 718 (30.1%) cases, and repair was performed in 178 (7.5%) cases. Among those reporting KOOS Pain (N=1825) at 10 years follow up, 252 (13.8%) had KOOS Pain < 80, 1573 had ≥ 80. In the KOOS Pain < 80 group, 75 (29.8%) out of 252 had subsequent surgery. In the KOOS Pain ≥ 80 group, 223 (14.2%) had subsequent surgery. The first step in executing the CMA demonstrated through logistic regression that neither LM repair nor excision significantly affected the likelihood of subsequent surgery after primary ACLR (Chi-Square: 0.11, P = 0.74) (Table 2.)
Conclusions:
Neither LM repair nor excision at the time of primary ACLR significantly affected the likelihood of subsequent surgery and thus had no mediating effect on the primary outcome of interest. This fact precludes the attribution of causality for LM tear treatment (repair or excision) to the likelihood of developing significant knee pain at 10 years follow up. To our knowledge, this is the first study of its kind to employ CMA to assess for a causal relationship between LM treatment at the time of primary ACLR and the likelihood of developing significant knee pain.
