Limb Symmetry Index Cutoffs After Injury: Why a 90% Pass Can Still Mean the Athlete Isn't Ready

A high school soccer player is eight months out from ACL reconstruction. Her clinic runs the standard battery — isokinetic quad strength, a triple hop for distance, a crossover hop — and every number on the involved leg comes back at or above 90% of the uninvolved leg. She's cleared for full training. Eleven weeks later, cutting off that same leg in a scrimmage, the graft fails again. Nobody misread the test. The test itself was measuring the wrong thing.

That's the trap built into the limb symmetry index (LSI) as most clinics run it. LSI is a ratio — involved-limb value divided by uninvolved-limb value, times 100 — and a ratio can climb for two completely different reasons. It can rise because the injured leg genuinely got stronger. Or it can rise because the other, supposedly uninjured leg got weaker, deconditioned across eight months where the whole program was built around protecting and rebuilding one side while the other quietly lost the loading it used to get from running, cutting, and heavy bilateral lifting. Both cases produce the same 90-something percent on a printout. Only one of them means the athlete is actually ready.

The mistake that keeps repeating across rehab programs isn't sloppy testing — it's treating a passed LSI threshold as a finish line rather than a question with half the information missing. Ninety percent of what, measured against what value, is the question nobody asks once the ratio clears the number a return-to-sport checklist wants to see. This piece works through why that gap exists, what the research shows about how often it produces a false pass, and the absolute-value check that closes it.

Wellsandt, Failla, and Snyder-Mackler (2017), publishing in the Journal of Orthopaedic & Sports Physical Therapy, tested this exact question directly. Working from patients enrolled in a larger ACL reconstruction trial, they compared quadriceps strength and single-leg hop performance measured the standard way — involved limb against the uninvolved limb at the same follow-up visit — against involved-limb performance measured against each patient's own uninvolved-limb strength recorded before surgery, back when both legs were presumably healthy. The two comparisons told different stories. A meaningful share of patients who cleared 90% LSI against their current uninvolved leg fell well short of 90% against their own pre-injury baseline, because the uninvolved leg had lost strength over the rehab timeline too — general detraining, reduced bilateral loading, months of a program organized around one knee. The ratio looked fine. The athlete, measured against who she used to be, was not.

The limitation is worth stating plainly: this was a secondary analysis of an existing ACLR cohort rather than a purpose-built trial, and the pre-injury value came from a single early testing session rather than a longitudinal pre-injury record — a proxy, not a perfect measurement of true baseline capacity. It's still the clearest direct evidence that LSI's denominator can move in ways that flatter the numerator.

The second piece of evidence comes from Grindem et al. (2016), the Delaware-Oslo ACL cohort study of 106 athletes published in the British Journal of Sports Medicine. Their return-to-sport decision rule combined quadriceps strength LSI, hop-test LSI, and a self-reported knee function score, requiring at least 90% on each before clearance. Athletes who met all criteria before returning to pivoting sport had an 84% lower rate of second ACL injury than athletes who returned without meeting them, and delaying return-to-sport further, up to nine months post-surgery, cut reinjury rate more. That result is real and it's a strong argument for using LSI thresholds at all — but the study bundled three criteria together, so it can't isolate how much of the protection came from the strength ratio specifically versus the hop tests or the self-report score, and it didn't check the ratio against a pre-injury baseline the way Wellsandt's group did. Put the two studies side by side and the honest read is that LSI thresholds work as part of a bundled decision rule, while the ratio alone can still be fooled by a weakened denominator.

The uninvolved leg doesn't weaken because anyone neglects it on purpose. It weakens because almost every incentive in a rehab program points the other way. Early post-op weeks are built around protecting the graft — partial weight-bearing, brace-limited range of motion, exercises chosen for the injured knee's tissue timeline rather than the uninvolved leg's training needs. Bilateral lifts that would normally maintain the healthy leg's strength — back squats, deadlifts, loaded carries — get scaled down or removed for months because the involved leg can't yet tolerate the load, and the uninvolved leg gets scaled down right along with it even though it could handle far more.

General conditioning follows the same path. An athlete who was running, cutting, and jumping five or six days a week before surgery drops to two or three sessions of rehab-focused work, most of it single-leg and submaximal. The uninvolved leg loses the high-velocity, high-force stimulus it used to get from sport itself — not because a clinician removed it deliberately, but because the whole training context shrank around the injury. Add a layer of fear-avoidance — athletes who unconsciously load the uninvolved leg more cautiously too, worried about overloading anything near the surgical knee during a single-leg hop test — and the supposedly uninjured limb can arrive at the six-month mark measurably below where it stood on the day of injury, even though nothing was technically ever wrong with it.

This is why an LSI reading needs a second data point to interpret honestly: is the uninvolved limb's absolute value holding steady, improving, or drifting down against its own pre-injury number? A ratio moving toward 100% while the denominator drifts down is not the same finding as a ratio moving toward 100% while the denominator holds — and most clinic reporting only records the ratio, not the trend underneath it, a blind spot similar to what shows up in jump asymmetry and injury prediction research on the general lower-limb side.

Closing the gap starts with capturing an absolute value, not just a ratio, and doing it before the injury whenever the program allows for it. Preseason testing is the cheapest fix here: an isokinetic dynamometer session — concentric quad and hamstring torque at 60°/s and 180°/s — or, where a full isokinetic unit isn't available, a handheld dynamometer with a fixed, belt-stabilized setup, run on every roster athlete before the season starts. That single number, banked per athlete, is what a post-injury LSI should eventually be checked against — not just the uninvolved leg on test day.

When no pre-injury baseline exists, which is the more common situation in club and school settings, the fallback is a normative reference value rather than a raw guess: peak knee extensor torque normalized to body mass (Nm/kg), compared against published age- and sport-matched norms, or peak vertical force per kilogram from a force-plate countermovement jump, normalized the same way. Neither substitutes for a true individual baseline, but both catch the specific failure mode this article is about — a ratio that looks fine while both absolute numbers sit below where a healthy athlete of that profile should test.

Field protocol for the combined check: (1) isokinetic or handheld dynamometer quad strength test, three trials per leg at a fixed testing angle, best trial recorded per side; (2) single-leg hop battery — single hop for distance, triple hop for distance, crossover hop for distance, 6-meter timed hop — averaged across two trials per leg per test; (3) calculate LSI as involved ÷ uninvolved × 100 for each measure; (4) separately calculate the uninvolved leg's current value as a percentage of either the pre-injury baseline or the matched normative reference; (5) cross-reference both numbers before making any return-to-sport call.

Retest cadence should track rehab phase, not a fixed calendar. Early return-to-run phase: dynamometer and hop testing every 4-6 weeks, since strength changes fast during that window and a stale number leads to bad phase-progression calls. Late-stage, pre-return-to-sport phase: retest at every clearance milestone rather than waiting for a scheduled date, because this is exactly the phase where the false-pass pattern shows up — strength gains slow down for the involved leg while general conditioning gains continue for the uninvolved one, which is precisely when the ratio can drift upward for the wrong reason.

For the actual clearance decision, treat the paired reading from the table above as the real gate, not the LSI number in isolation. An athlete sitting at 91% LSI with the uninvolved leg tracking within 5% of a documented preseason baseline is a legitimate pass. An athlete at that same 91% LSI whose uninvolved leg has drifted 12% below its own preseason number needs a different plan entirely — not more single-leg accessory work on the involved side, but a bilateral strength block that brings the uninvolved leg back up first, because clearing that athlete to cut and pivot now sends two under-recovered legs back into contact instead of one.

None of this replaces the qualitative side of a return-to-sport decision — a pain-free clinical exam, movement-quality screening under fatigue, and a psychological-readiness measure like the ACL-RSI scale still belong in the same file. What the absolute-value check adds is a way to catch the specific failure mode a ratio-only protocol misses: the pass that looks clean on paper because the wrong leg quietly got weaker too.