AdMIRable Review Summer 2024

What is normal for a knee varies with age, as it does for shoulders (Summer 2023 issue of the AdMIRable Review), necks, low backs, hair color, and facial wrinkles. Do doctors have knowledge of what is normal at different ages, or do they assume any finding on a knee MRI must be the cause of the patient’s symptoms?

A 2021 study of “Cognitive Biases in Orthopaedic Surgery” (Janssen 2021) concluded that cognitive biases affect decision-making and reasoning in orthopaedic surgeons, just as they do in all humans. Two common biases that affect diagnosis and surgical recommendations are base rate neglect and confirmation bias. Base rate neglect is deciding any finding noted on a knee MRI is the source of the patient’s symptoms and failing to consider how common that same finding is in the knees of similarly-aged, ASYMPTOMATIC patients. Confirmation bias is exemplified in the medical records of patients whose history and physical exam findings change after the results of a knee MRI are known to the doctor, so the patient after the MRI now has the expected symptoms and physical exam findings for the condition seen on MRI. An example of this is the knee physical exam is recorded as showing no instability, an MRI is then obtained showing no anterior cruciate ligament (previously ruptured and either apparently new or “chronic and pre-existing”), and the office visit note after the MRI states the patient who had a “stable” knee before the MRI now has anterior instability on physical exam, with an anterior cruciate ligament reconstruction now recommended.

In discussing base rate neglect, a preliminary discussion of knee imaging and injury/pathology diagnosis will be helpful.

The most common knee surgery done in the United States is related to the knee menisci. The menisci attach to the knee joint capsule that surrounds the periphery of the joint. There are also anterior and posterior attachments to the tibia (bone) called roots. The menisci are fibrocartilage (thus not visible on x-ray) “C” shaped structures that help distribute force (body weight plus muscle force plus impact force) between the femur and the tibia (Feeley 2018, Lau 2018).

Menisci can develop a “tear” either from an acute traumatic event (typically a twist or pivot while weight bearing), or from degeneration (“fatigue failure” for the engineers reading this) (New Zealand ACC 2013, Englund 2009, Englund 2001). Vertical, radial, and perhaps “Parrot beak” tears are typically considered potentially traumatic, while horizontal and complex tears are typically degenerative and correlate with age, genetics, and BMI (obesity), as they are part of the degenerative or osteoarthritis process.

Meniscal vertical and radial tears that are near the periphery of the meniscus (near the knee capsule or in the “red” zone) where there is a blood supply to the meniscus tear, are in the zone for potential scar tissue formation and tear healing. Meniscal repair can be performed for these tears, but this requires a period of non-weight-bearing. Since tears eligible for repair are usually most common in young athletes with high energy mechanisms of injury (sports), meniscal repairs are mostly undertaken in this demographic.

Meniscal “tears” in the more central portion of the meniscus (“white” zone) are in the portion of the meniscus that does NOT have a blood supply and thus cannot repair itself or form scar and heal, either spontaneously or after suture repair. Thus, these irreparable tears persist, or they can be part of the meniscus removed at partial meniscectomy surgery. The more meniscus that is removed, the more the contact force between the femur and the tibial increases, and the more quickly osteoarthritis typically develops (Reito 2022).

A systematic review (Beals 2016) of 14 publications using MRI on 295 asymptomatic young (average age 31) athletes found meniscal degeneration in 27% and a meniscal tear in 4%.

A study of 1.5T MRI in 23 asymptomatic marathon runners from California (average age = 40, average years running 10, average miles run/week 41 – Shellolck 1991) found 9% had a meniscal tear, which was a lower percentage than the 20% rate in a comparison non-runner athlete group, and also lower than the 16% in the comparison asymptomatic non-athletes. Clearly having a meniscal tear does not automatically exclude high level activity performance.

A study (Stahl 2008) of 3.0 Tesla MRI in 10 runners 2-3 days before and 2-3 days after running a single marathon found no MRI change in runners acclimated to this running distance. Similar studies documenting no significant knee injury by MRI in experienced marathon runners (Schueller,-Weiderkamm 2006) and novice runners (first marathon – Horga 2020) have been published, suggesting that new symptoms after activity to which the individual is not acclimated but without specific injury incident are likely not due to structural knee pathology (“delayed onset muscle soreness” is common).

In older adults, pain with no injury or with low violence (minimal events) is the usual first presentation of osteoarthritis symptoms. In those with minimal events, this is “when,” but not “why,” symptoms begin.

A study from Germany with 1.0 Tesla MRI (Zanetti 2003) found a meniscal tear in 57% of 100 patients in their symptomatic knee, and in 36 of those 57 (63%), the patients had a meniscal tear in their asymptomatic, contralateral knee.

Meniscal tears increase in prevalence with age and with osteoarthritis severity. One of the best epidemiological studies comes from the Framingham, Massachusetts heart disease study with the town population followed for the development of heart disease. When 991 adults from this population were evaluated for both knee symptoms and by knee MRI (Englund 2008):

The prevalence of a meniscal tear or of meniscal destruction in the right knee as detected on MRI ranged from 19% among women 50 to 59 years of age to 56% among men 70 to 90 years of age; prevalences were not materially lower when subjects who had had previous knee surgery were excluded. Among persons with radiographic evidence of osteoarthritis (Kellgren–Lawrence grade 2 or higher, on a scale of 0 to 4, with higher numbers indicating more definite signs of osteoarthritis), the prevalence of a meniscal tear was 63% among those with knee pain, aching, or stiffness on most days and 60% among those without these symptoms. The corresponding prevalences among persons without radiographic evidence of osteoarthritis were 32% and 23%. Sixty-one percent of the subjects who had meniscal tears in their knees had not had any pain, aching, or stiffness during the previous month.

A later publication of 771 adults from this cohort who had no evidence of knee osteoarthritis at a mean age of 62 (Guermazi 2012) found 24% of 710 adults had a meniscal lesion, 42 had knee pain, and 120 had no knee pain.

In a subsequent publication (Englund 2016) on the same population, the presence of hand osteoarthritis correlated significantly with knee meniscal tears, suggesting a “common systemic/genetic predisposition and/or a common environmental risk factor for radiographic hand OA and meniscus damage in the middle-aged and elderly. The association remained robust even after accounting for subject characteristics, radiographic OA of the knee, and history of knee injury.”

A systematic review (Culvenor 2018) combined 63 studies of 5,397 asymptomatic and uninjured knees having an MRI and found in those under age 40, 2%-7% had a meniscal tear, while in those older than 40, 13-26% had a meniscal tear. Articular cartilage defects were common (Less than 40, 6%-17%, while over 40, 29%-57%), as were bone marrow lesions (marrow edema or "microfracture on MRI reports - under 40, 6%-24%, while over 40 14% -31%).

Newer MRI machines have higher strength magnetic fields and yield images with higher resolution (somewhat like having more pixels on your television). A study (Horga 2020) using 3.0 Tesla MRI (high MRI magnet strength) on predominantly white British asymptomatic sedentary adults found 97% of the 230 knees had at least one abnormality, with 30% having meniscal tears. Degenerative tears (horizontal and complex) were present in 26%, and likely traumatic tears (vertical, radial, and bucket handle) were present in 5% of these knees.

Recent studies on unilateral knee injury or pain and bilateral MRI (including MRI of asymptomatic, contralateral knee) are instructive. A Dutch study of 134 patients, mean age 41, seeing their general practitioner for unilateral knee injury who had bilateral knee 1.0 Tesla MRI (Boks 2006) found ligament injury almost exclusively in the injured knee. Two radiologists reviewed each MRI, and the inter-observer kappa was 0.84 to 1.0 for the multiple findings evaluated.

“The prevalence of meniscal tears was high in both symptomatic (medial, 26%; lateral, 19%) and contralateral knees (medial, 17%; lateral, 8%). Of the 45 patients with a meniscal tear on the symptomatic side, 19 (42%) also had a meniscal tear on the contralateral side (13 horizontal and 6 other tears).” “Of the 87 patients without a meniscal tear on the symptomatic knee, 10 (12%) had a meniscal tear on the contralateral side (9 horizontal and 1 other tear). Nearly all (18 of 19) bilateral meniscal tears were seen in patients aged 40 years and older.”

In a Texas study of workers with unilateral knee symptoms attributed to injury having routine bilateral knee MRIs (Liu 2017), 43% of these workers had worse MRI pathology in the knee with symptoms attributed to injury. Thus, 57% had worse MRI pathology in the contralateral, uninjured, asymptomatic knee.

A similar cohort of 30 Texas workers with unilateral knee symptoms attributed to injury had bilateral knee MRI (Crijns 2023). Musculoskeletal Radiologist reports of both knees were reviewed by 76 academic medical center surgeons. On average the surgeons correctly guessed which knee was the symptomatic knee in 61% of the cases (slightly better than “flipping a coin”). Adding the patient’s age or a description of history and exam did not improve the surgeons’ guess accuracy. The kappa for inter-observer agreement was 0.17 or “slight agreement.”

From the above discussion, base rate neglect has a profound effect of surgeon choice and assessment.

Confirmation bias affects the surgeon’s assessment of symptoms, examination, and diagnostic studies.

Surgeons have traditionally felt that the outcomes for partial meniscectomy in patients with degenerative meniscal tears associated with a traumatic onset were better than in those with pain but no history of trauma. But a small European series (Kim 2013) of degenerative (horizontal) meniscal tears in adults under 40 (low prevalence of osteoarthritis) found no difference in outcome scores in those with the trauma history, compared to those without any history of trauma.

Surgeons have been taught that symptomatic meniscal tears produce “mechanical symptoms” such as locking, clicking, popping, pain localized to the joint line, pain with pivoting, and giving way. Thus, surgeons have traditionally recommended partial meniscectomy for patients with “mechanical symptoms,” even if the knee had obvious osteoarthritis.

The evidence points against this recommendation. A Boston study (Deshpande 2016) evaluated 84 patients with knee pain by 1.5 or 3.0 T MRI. A meniscal tear was present in 80% of the patients for whom the surgeons had a high level of confidence of a meniscal tear, 87% of those for whom the surgeons had medium confidence, and 64% of those for whom the surgeons had low confidence.

A larger Boston study (Farina 2021) of 565 patients having knee arthroscopy found pre-operative “meniscal” or “mechanical” symptoms correlated strongly with the burden and severity of osteoarthritic change in the articular cartilage and not with any specific meniscal pathology. Stated bluntly, “mechanical” meniscal symptoms are typically due to osteoarthritis and falsely attributed by surgeons to meniscal pathology.

In addition, physical exam provides little correlation with the presence of a meniscal tear (Smith 2015, Blyth 2015, New Zealand 2018).

Since neither symptom assessment nor physical examination provide definitive information, meniscal treatment decisions are usually based on MRI images.

The largest series looking at outcomes of surgery for those with and those without “mechanical symptoms” comes from Finland (Sihvonen 2016). They reported on 900 consecutive patients having arthroscopic partial meniscectomy with a 12-month follow up. Pre-operatively the groups only varied in that there was more osteoarthritis, lower function, and more pain in the group with the “mechanical” symptoms. Patients with “mechanical symptoms,” compared with patients without “mechanical symptoms,” had a lower percentage of “satisfied” (61% versus 75%) and “improved” (79% versus 88%) outcomes, rather than the better outcomes surgeons expected. They noted the mean improvement in these surgical patients on the WOMET (a validated knee disease specific quality of life – Sihvonen 2012) was identical to the improvement that occurred in patients having a pretend or sham arthroscopy in their prior publication comparing partial meniscectomy to sham surgery (Sihvonen 2013).

If meniscal tears in patients over age 35 to 40 with no history of high-violence injury are degenerative (usually horizontal or complex), are these middle-aged or elderly patients with knee pain after no injury (alleged cumulative trauma) or after minor injury improved by partial meniscectomy? Multiple systematic reviews of published studies, considering mainly randomized trials of patients in middle-aged or older-age groups of arthroscopic partial meniscectomy versus either physical therapy (exercise) or actual sham surgery (skin incision, no actual intra-articular surgery) have concluded that on average there is no outcome difference of significance (no benefit to surgery). (Khan 2014, Thorlund 2015, Brignardello-Petersen 2017, Abram 2019, Lee 2020, Blom 2021, O’Connor 2022, Wijn 2023, Fernandez-Matias 2023, Migliorini 2023, Meng 2024).

In workers’ compensation, middle-aged or elderly patients with degenerative meniscal tears, arthroscopic partial meniscectomy typically leads to the injured worker having a subsequent knee replacement sooner than if they had not had surgery and had chosen to live with an arthritic knee-(Tsourmas 2024). The knee replacement now becomes work compensable, instead of being attributed to age, obesity, genetic risk, etc.

So why then do surgeons and the American Academy of Orthopaedic Surgeons (Leopold 2021) still recommend arthroscopic partial meniscectomy in the middle-aged and elderly worker with knee osteoarthritis?

Here is where biases must be addressed. The surgeons’ experience misleads them. They see patients improve with surgery (confirmation bias) and don’t remember the published studies that say the same outcomes would occur without surgery (base rate neglect).

To bring the subject into perspective, workers’ compensation is frequently called an “adversarial system.” If the authorized treating physician (surgeon) recommends surgical partial meniscectomy, and utilization review denies the surgery, the patient feels inadequately treated and expects not to improve (nocebo effect). If, however, the patient or surgeon is successful on appeal, the patient ultimately gets the surgery that the patient wanted and expected, and this results in “improvement.” Patient expectations of treatment success are correlated with patient-reported surgical outcomes for many conditions (like knees, hips, backs, etc. – Haanstra 2012, Waljee 2014, Mooiweer 2024). Thus, the patient’s objective outcome may not be better, but the patient’s perception and patient-reported outcomes may be better after surgery that is as effective as placebo - reinforcing the surgeon’s confirmation bias. Both nocebo and placebo effects on the injured worker’s part reinforce the surgeon’s choice (confirmation bias).

Will this change in the future? When Max Planck was asked why scientists don’t accept the theory of quantum mechanics as more and more confirmation of its predictions accumulated, he said “a new scientific truth does not triumph by convincing its opponents, but because its opponents eventually die, and a new generation grows up that is familiar with it.”

It is hoped that a realistic assessment of the evidence and an examination of the surgeon’s biases can be effectively communicated to the surgeon and to the injured worker to overcome the injured worker’s perceptions and improve overall scientifically-based decisions, to reduce the unnecessary risks and complications of surgery before another generation.

Jump to references for this article.