Table 5. Summary of results, adverse events, and limitations of the included studies.
APM: arthroscopic partial meniscectomy; MD: mean difference; TKR: total knee replacement; MCID: minimal clinically important difference; PT: physical therapy; ET: exercise therapy; MRI: magnetic resonance imaging; GRC: global rating of change scale; KOOS4: knee osteoarthritis outcome score (pain, symptoms, function in sports and recreation, and quality of life); IKDC: International Knee Documentation Committee; PSFS: patient-specific functional scale; OA: osteoarthritis; LSI: Limb Symmetry Index; NSAIDs: non-steroidal anti-inflammatory drugs
| Author, year | APM/TKR crossover | Adverse events | Study limitations |
| Graaf et al., 2018 [13] | As-treated analysis: delayed APM (N = 47). IKDC score: delayed APM, from 40.8 points at baseline to 63.0 points at 24 months (MD, 21.5 points [95% CI, 15.8–27.3]). Knee pain during weight-bearing: delayed APM, from 66.4 mm at baseline to 36.0 mm at 24 months (MD, 27.5 mm [95% CI, 16.0–39.1]) | Adverse events (e.g., cardiovascular, neurological, or internal medicine conditions, venous thromboembolism, or repeat knee surgery) N = 8. Non-serious adverse events (e.g., reactive arthritis, joint paint resulting in extra consultation or surgical site infection) N = 4 | Screening logs for patients who were not randomized were not maintained. Non-inferiority margin based on 8.8 points, the smallest detectable change, is a conservative estimate of potentially relevant differences. Grouping was not blinded. Non-inferiority testing was intended for the secondary analyses, but no non-inferiority margins were specified in the protocol. MCIDs for the secondary outcomes were not defined until after data analyses. Radiographs were interpreted by a single radiologist. The combination of APM and PT may be more effective than APM alone |
| Berg et al., 2020 [16] | Not reported | Not reported | The radiographic clinics were instructed to follow a standardized protocol; however, we identified some deviations. No radiographic evaluations of the patellofemoral joint were performed. The study did not have sufficient power to detect differences in individual radiographic features. In non-surgical and surgical treatments, one-way crossover is a potential challenge; patients can cross over from ET to APM, but not from APM to ET once they have had surgery |
| Kise et al., 2019 [23] | Not reported | Not reported | This study did not include radiographs appropriate for evaluating varus and valgus alignment. MRI evaluation included degeneration grades 0–3b (lower is better) and measurement of meniscus extrusion. In the subgroup analyses of GRC scale pain and function, small sample sizes, especially for the APM group, might have led to spurious results; this is reflected in the wide 95% CIs. Possible low external validity |
| Kise et al., 2016 [24] | Crossover (1 patients with multiple surgeries) was 12 patients A comparison of KOOS4 at 12 months to 2 years between the crossover and ET groups showed no between-group difference (25.5 vs 25.5, p = 1.0) | 23% of subjects experienced severe pain, swelling, instability, stiffness, and reduced range of motion | The lack of a sham surgery group; we cannot exclude the possibility that the greater placebo effect from surgery on patient-reported outcomes masks a “real” difference in treatment between groups |
| Ahn et al., 2015 [25] | Not reported | Not reported | The follow-up period was short and the cohort size was small. There were significant differences in preoperative demographics and clinical characteristics such as age. Mental health is a component of patient satisfaction, but was not assessed |
| Lim et al., 2010 [26] | Not reported | Not reported | Non-operative treatment was not compared with surgical treatments such as arthroscopic meniscectomy, repair, or osteotomy. Even with high MRI sensitivity, some patients with medial meniscus posterior root tears may be under- and over-diagnosed |
| Yim et al., 2013 [27] | Not reported | Not reported | Other factors affecting the outcomes of non-operative treatment, such as the patient’s occupation and lifestyle, were not assessed. Most participants were women |
| Noorduyn et al., 2020 [28] | Not reported | Cardiovascular events, neurological problems, internal medicine conditions, re-surgery on the affected knee, total knee replacement, and knee pain | Not blinded to treatment; determined based on IKDC and not based on PSFS; patients experiencing knee pain related to MRI-confirmed meniscus tears were recruited. The PSFS has not been validated in this population or in similar populations. PT protocol consisted of general progressive exercises for cardiovascular coordination, coordination, balance, and closed kinetic chain strength of the lower limb, rather than exercises focused on patient-selected relevant activities |
| Ikuta et al., 2020 [29] | Not reported | Not reported | Short-term results were studied over 6 months, and the medium- and long-term outcomes are yet unknown. Small sample size |
| Kudo et al., 2013 [30] | Not reported | Not reported | Participants applied to participate in the ET for knee OA and may have had a strong motivation to exercise. Exercise may be less effective in providing symptomatic relief in cases where flexion contracture is observed |
| Stensrud et al., 2015 [31] | Not reported | Not reported | There were no restrictions due to participation in leisure-time physical activities during the study period, and no differences were reported in terms of type, frequency, or intensity of leisure-time physical activities between the groups. Leisure-time physical activities were self-reported at follow-up, which is limited by recall bias and overestimating fitness level. There is a large difference in time between baseline and intervention initiation between the two groups. The LSI was not reported as an outcome measure in the current study despite common use to express both isokinetic muscle strength 48 and single-leg hop performance |
| Neogi et al., 2013 [32] | Not reported | Not reported | The average follow-up was not long enough. The effect of NSAIDs was not measured. Small sample size |
| Prati et al., 2017 [33] | Not reported | Not reported | The study lacked randomization; a small number of patients were treated. Only a 3-month follow-up was evaluated |
| Sonesson et al., 2020 [34] | Not reported | Not reported | The patients were not blinded to the treatment. Only 70% of patients completed the 5-year follow-up questionnaire |
| Katz et al., 2013 [35] | 30% of patients assigned to the physical therapy group crossed over to the surgery group in the first 6 months | Mild or moderate severity adverse effects occurred in 13 participants in the physical-therapy group, including, death, pain from fall or other trauma, knee bursitis, knee pain, and pain in the back, hip, or foot) | Only 26% of eligible patients were enrolled. The study was not blinded |
| Başar et al., 2021 [36] | Not reported | Not reported | Did not perform long-term follow-up; small sample numbers; the relationship between treatment method and knee OA progression was not investigated |