Abstract
Objective:
To evaluate changes in knee MRI findings over the course of 18 months in subjects with osteoarthritic change (OA) and meniscal tear (MT) treated with arthroscopic partial meniscectomy (APM) or non-operatively with physical therapy (PT).
Methods:
We used 18-month follow up data from the Meniscal Tear in Osteoarthritis Research (MeTeOR) Trial. MRIs were read using the MRI OA Knee Score (MOAKS). We focused on 18-month change in bone marrow lesions (BMLs), cartilage thickness, cartilage surface area, osteophytes size, effusion-synovitis, and Hoffa-synovitis. We used multinomial logistic regression to assess associations between MRI-based changes in each feature and treatment type.
Results:
351 subjects were randomized and 225 had both baseline and 18-month MRI. In both treatment groups, patients experienced substantial changes in several MRI-based markers. In 60% of the APM group, vs. 33% of the PT group, cartilage surface area damage advanced in ≥2 subregions (adjusted odds ratio (OR) 4.2 (95% CI 2.0, 9.0). Patients who underwent APM also had greater advancement in scores for osteophytes and effusion-synovitis. We did not find significant associations between treatment type and change in cartilage thickness, BMLs, or Hoffa-synovitis.
Conclusion:
This cohort with meniscal tear and OA had marked advancement in MRI-based features over 18 months. Those treated with APM had more advancement in some features compared to those treated non-operatively. The clinical relevance of these early findings is unknown and requires further study.
Introduction
Recent estimates suggest that 14 million adults in the United States have knee osteoarthritis (OA), including 8 million persons under age 65.(1) Of these, approximately 80% have a concomitant meniscal tear.(2) Non-operative treatment of symptomatic meniscal tear typically includes a physical therapy (PT) regimen (muscle strengthening, endurance, flexibility and balance training); surgical treatment typically consists of arthroscopic partial meniscectomy (APM). Data from several large clinical trials have suggested that patients with meniscal tear and osteoarthritic changes treated with APM plus PT have similar pain relief as compared to patients treated with PT alone, though crossover from PT to APM makes interpretation challenging in several of these trials.(3–7)
In patients with OA, meniscal tear has been shown to be an independent risk factor for progression of cartilage damage.(8) Observational studies have suggested that a history of APM may be associated with a higher risk of incident osteoarthritis(9,10). It is unclear in these studies whether the risk of progression is attributable to the initial meniscal damage or to the surgical treatment. This question can best be addressed in a clinical trial setting, in which all subjects have knee pain, meniscal tear and are deemed surgical candidates by their orthopedic surgeons.
The aim of our study was to evaluate early MRI-based changes in persons with meniscal tear and OA treated with APM and those treated non-operatively.
Patients and Methods
Study Sample
We used data from the Meniscal Tear in Osteoarthritis Research (MeTeOR) Trial, a multi-center randomized controlled trial (RCT) of APM plus PT vs. PT alone to treat symptomatic meniscal tear in OA(4). Subjects were recruited from orthopedic surgery clinics in seven US referral centers. Eligible subjects were ≥45 years old, had evidence of meniscal tear on magnetic resonance imaging (MRI), evidence of OA changes on MRI or x-ray, and knee symptoms. The full inclusion and exclusion criteria are published elsewhere(11).
Subjects were randomized to APM with PT or PT alone. For subjects undergoing APM, the surgeon used standard arthroscopic portals and trimmed the damaged meniscus back to a stable rim(11). Surgeons also trimmed loose fragments of cartilage and bone but did not penetrate the subchondral bone. Subjects randomized to the PT arm followed a standardized, strengthening-based PT protocol including both weekly sessions with a therapist and home-based exercises; generally the program lasted 6 weeks (4,11). Subjects were permitted to see their orthopaedic surgeons throughout the study and could discuss with the surgeon the option of crossing over to receive APM, if symptoms persisted despite PT.
Outcome
Subjects underwent MRI at baseline as part of routine clinical care. Each of the centers performed cartilage sensitive sequences, permitting us to assess the MRIs with semi-quantitative methods. At 18 months subjects underwent MRI using the same sequences as performed at baseline. Baseline and 18 month MRIs were read using the MRI OA Knee Score (MOAKS) in pairs, unblinded to time by an experienced musculoskeletal radiologist who is an expert in semi-quantitative MRI analysis of knee OA(12). In a sample of 10 subjects the MOAKS total OA scores for this reader were closely associated to the total OA scores of another highly experienced reader, with an interclass correlation of 0.98.(13) The reader was blinded to the subject treatment and all other demographic information. We focused on 18-month change in several joint features: bone marrow lesions (BMLs), cartilage surface area, cartilage thickness, osteophyte size, effusion-synovitis, and Hoffa-synovitis. Given the biomechanical models demonstrating greater contact pressures associated with APM, we envisioned the most striking effects would be observed in cartilage damage (with contact pressures transmitted directly to cartilage) and osteophytes (bony enlargement in response to greater contact pressure)(14,15). In the MOAKS system, each joint feature is divided into subregions, and each subregion is scored on an ordinal scale from 0–3. We assessed change in each feature as follows:
Bone Marrow Lesions
BML size is assessed in 14 subregions and thus the change in number of subregions affected has a theoretical range of −14 to 14, as BMLs can both develop and resolve. We assessed the change in the number of subregions affected by any BML (i.e., with a score > 0). We assessed the maximum advancement in BML size score across all subregions, which was grouped into no change, advancement by 1 grade, and advancement by 2+ grades. We also assessed whether there were any subregions with improvement in score, and whether there were any subregions with advancement in score.
Cartilage
Cartilage surface area and thickness were analyzed separately. We assessed the number of subregions with advancement, the number of subregions with new cartilage damage (i.e., a score of 0 at baseline and >0 at follow-up), and the maximum advancement across all subregions. The number of subregions with advancement and the number of subregions with new damage have a possible range of 0 to 14; based on distribution these were grouped into 0, 1, 2+ subregions. Maximum advancement was grouped into no change, advancement by 1 grade, and advancement by 2+ grades.
Osteophytes
We assessed the number of locations with advancement, the number of locations with new osteophytes, and the maximum advancement across all locations. Osteophytes are scored in 12 locations and thus the number of locations with advancement and the number of locations with new osteophyte has a possible range of 0 to 12. Based on distribution, the number of locations with advancement was grouped into 0, 1, 2+ subregions, and the number of locations with new osteophyte and the maximum advancement across all subregions were grouped into no advancement vs. any advancement.
Synovitis
Effusion-synovitis represents a combination of effusion and synovial thickening and Hoffa-synovitis is seen as hyperintensity on fat-suppressed water sensitive sequences and is a sensitive but not specific surrogate marker for the true synovitis. They are each rated on an ordinal scale from 0–3. Changes were classified as improvement, no change, and advancement.
Statistical Analyses
We first evaluated the association between baseline characteristics and treatment group to ensure that the groups were balanced after excluding crossovers. Baseline KLG grade was imbalanced between the treatment groups and was thus adjusted for in multivariable models. For each joint feature considered, we used multinomial logistic regression with structural advancement of that feature as the dependent variable and treatment group as the independent variable. We calculated odds ratios with associated 95% confidence intervals, where the odds ratio represents the increased odds of experiencing structural advancement for subjects receiving APM vs. PT. To adjust for multiple testing we used the Holm step-down procedure(16,17).
The primary analysis compared subjects randomized to and receiving APM with subjects randomized to and receiving PT, without crossover to APM. The analysis is mechanistic in focus; consequently, it examines how the treatment actually received (surgical vs. non-operative) affects progression. Subjects who crossed over from PT to APM had higher baseline pain and slower initial clinical improvement; since these factors may be associated with more rapid structural progression, we did not include the cross-overs in the primary analysis(18). We performed a secondary as-treated analysis, also mechanistic in focus, in which subjects crossing over from PT to APM within 6 months of randomization were analyzed in the surgical group. Finally, we performed an intention to treat (ITT) analysis in which subjects were analyzed according randomization group, irrespective of treatment received. This analysis was considered important because baseline factors, known and unknown could be more important and have a bigger impact than the intervention. Subjects crossing over from PT to APM between 6 and 18 months were excluded from all analyses, to ensure that subjects analyzed in the surgical arm in as-treated analyses were exposed to surgery for at least twelve months. Subjects randomized to the APM arm who did not receive surgery were also excluded. This group was very small, and unlike the PT to APM crossovers where we have information on treatment received (number of PT visits, date of surgery), we do not know we do not know what other treatment courses, if any, this group pursued. When these subjects were included in the analyses the results did not differ meaningfully.
Sensitivity analyses to examine the impact of missing data:
For each outcome, we used multiple imputation (MI) to impute outcome for those subjects missing baseline and/or 18 month MRI data(19,20). We did this under two assumptions: (1) Missing at Random (MAR): we assumed that the missing data was associated with observed covariates (treatment group, KLG, sex, and race, baseline MOAKS if available) and (2) Missing Not at Random (MNAR): we assumed that the missing data was associated both with observed covariates (as in (1)) and with unobserved outcomes; that is, that structural progression may be better or worse than expected based on observed covariates alone. We took a so-called “tipping point” approach – how severe must the missing data mechanism be in order to change the study’s conclusions?(21) To do this, we imputed data under various not at random mechanisms ranging from more (MNAR1) to less (MNAR5) plausible. Details of each mechanism are described in the appendix.
All analyses were conducted using SAS 9.4 (SAS Institute, Cary NC).
Results
Cohort Characteristics
351 subjects were randomized and 225 subjects had both baseline and 18 month MRI. Of the 225 with paired MRI, 9 subjects crossed over from PT to APM between 6 and 18 months and 4 subjects were randomized to APM but did not undergo surgery (Figure 1). These 13 (5.8%) subjects were excluded from all analyses. One hundred and seventy-five subjects were included in the primary analysis: 103 were randomized to and underwent APM and 72 were randomized to PT and did not cross over. An additional 37 patients were randomized to PT and crossed over to APM in the first six months. Thus the secondary as-treated analysis consisted of 212 subjects (140 subjects in the APM group and 72 in the PT group) and the intention to treat analysis consisted of 212 subjects (103 in the APM group and 109 in the PT group). The included subjects did not differ on baseline characteristics compared to the subjects excluded (Appendix Table 1, Appendix Table 2).
Figure 1. Sample Details and analytic cohorts.
351 subjects were enrolled and randomized in the MeTeOR trial, and 225 had both baseline and 18 month MRI. 13 were excluded from all analyses, leaving 103 in the APM group, 37 in the APM to PT cross-over group, and 72 in the PT group. The primary analysis is APM (box I) vs. PT (box III). The first secondary analysis is as-treated: APM + APM to PT cross-over (box I + II) vs. PT (box III). The second secondary analysis is ITT: APM (box I) vs. APM to PT cross-over + PT (box II + III)
The primary analytic sample was 56% female and 89% white. The mean age was 59 (standard deviation (SD) 7) years and mean baseline KOOS pain (0–100 scale, 100 worst) was 45 (SD 16). The treatment groups were balanced on baseline demographics and clinical characteristics with the exception of KLG; the APM group had a higher percentage of patients with KLG3 and a lower percentage of patients with KLG2 compared to the PT group (Table 1). The treatment groups were balanced on baseline MOAKS (Appendix Table 3).
Table 1.
Cohort Characteristics. Frequency (%) presented for categorical variables, mean (SD) presented for continuous variables.
| Treatment | ||
|---|---|---|
| APM + PT (n=103) | PT alone (n=72) | |
| Sex | ||
| Male | 44 (43%) | 33 (46%) |
| Female | 59 (57%) | 39 (54%) |
| Race | ||
| Non-white | 12 (12%) | 8 (11%) |
| White | 91 (88%) | 64 (89%) |
| Age | 58.9 (7.9) | 58.4 (6.1) |
| BMI | 29.8 (6.1) | 30.0 (5.3) |
| Baseline KOOS Pain | 44.7 (15.4) | 46.1 (17.2) |
| Baseline WOMAC Pain | 37.8 (17.2) | 40.7 (17.8) |
| Baseline WOMAC Function | 35.7 (17.5) | 38.0 (19.5) |
| Baseline KLG | ||
| 0 | 23 (22%) | 14 (19%) |
| 1 | 26 (25%) | 19 (26%) |
| 2 | 23 (22%) | 22 (31%) |
| 3 | 31 (30%) | 17 (24%) |
| Meniscal Tear Category* | ||
| None or signal abnormality on meniscus | 0 (0%) | 2 (3%) |
| Non-degenerative simple tear | 13 (13%) | 9 (13%) |
| Short degenerative complex tear | 42 (41%) | 24 (33%) |
| Long degenerative complex tear | 33 (32%) | 19 (26%) |
| Meniscal root tear | 15 (15%) | 18 (25%) |
Based on central readings; patients were enrolled on the basis of readings at local centers.
APM= Arthroscopic partial meniscectomy.
PT=Physical therapy.
BMI=body mass index.
KOOS=Knee Injury and Osteoarthritis Outcome Score. Scored 0–100, 100 worst.
WOMAC= Western Ontario and McMaster Universities Osteoarthritis Index. Scored 0–100, 100 worst.
KLG= Kellgren and Lawrence Grade.
The median number of days between randomization and intervention start (surgery or first PT visit) was 21 for the APM group and 9 for the PT group. The median time between baseline and 18 month MRI was 579 days (19.1 months); between randomization and 18 month MRI was 542 days (17.9 months).
Change in joint features
Cartilage
The number of subregions with advancement in cartilage surface area score ranged from 0 to 7 with a mean (SD) of 1.9 (1.9). The mean (SD) number of subregions with advancement in cartilage surface area score was 2.3 (1.9) in the APM group compared to 1.3 (1.6) in the PT group (Figure 2A). Among subjects undergoing APM, 19% had 0 subregions with advancement in cartilage surface area score, 21% had 1 subregion with advancement, and 60% had 2+ subregions with advancement. Among subjects in the PT arm, 43% had 0 subregions with advancement in cartilage surface area score, 24% had 1 subregion with advancement, and 33% had 2+ subregions with advancement. This translates to a 2-fold increased odds of 1 subregion with advancement (OR 2.0, 95% CI 0.9, 4.8) and a 4.2-fold increased odds of 2+ subregions with advancement for APM vs. PT (OR 4.2, 95% CI (2.0, 9.0); Table 2). We also found significant increased odds of advancement for APM vs. PT when evaluating the number of subregions affected by cartilage surface area damage and maximum advancement in damage score (Table 2).
Figure 2. Early MRI-Based Advancement in Cartilage and Osteophyte by Treatment Group (Primary Analysis).
Each panel shows the distribution of MRI-based advancement by treatment group, for A) Cartilage Surface Area, B) Cartilage Thickness, C) Osteophytes. Number of subregions with advancement is along the Y-axis and treatment group (APM vs. PT) is along the X-axis. Each small circle represents one participant. The diamond indicates the mean and the square indicates the median.
Table 2.
Association between Treatment Group and Advancement, Primary Analysis (adjusted for baseline KL)
| APM n=103 | PT n=72 | p-value* | OR (95% CI) APM vs. PT | |
|---|---|---|---|---|
| Cartilage Surface Area | ||||
| Number of SRs with advancement in cartilage surface area score | 0.0008* | |||
| 0 SRs with advancement | 19 (19%) | 31 (43%) | ||
| 1 SR with advancement | 21 (21%) | 17 (24%) | 2.03 (0.85, 4.82) | |
| 2+ SRs with advancement | 61 (60%) | 24 (33%) | 4.22 (1.99, 8.96) | |
| Number of additional SRs affected by any cartilage surface area damage | 0.0075* | |||
| 0 additional SRs affected | 35 (35%) | 42 (58%) | ||
| 1 additional SR affected | 35 (35%) | 16 (22%) | 2.82 (1.32, 6.01) | |
| 2+ additional SRs affected | 31 (31%) | 14 (19%) | 2.67 (1.21, 5.87) | |
| Maximum advancement in cartilage surface area score across all SRs | 0.0015* | |||
| No Change | 19 (19%) | 31 (43%) | ||
| Advance by 1 grade | 33 (33%) | 22 (31%) | 2.50 (1.13, 5.53) | |
| Advance by 2+ grades | 49 (49%) | 19 (26%) | 4.19 (1.91, 9.20) | |
| Cartilage Thickness | ||||
| Number of SRs with advancement in cartilage thickness score | 0.1666 | |||
| 0 SRs with advancement | 39 (39%) | 38 (53%) | ||
| 1 SR with advancement | 24 (24%) | 15 (21%) | 1.55 (0.70, 3.45) | |
| 2+ SRs with advancement | 38 (38%) | 19 (26%) | 1.98 (0.96, 4.10) | |
| Number of additional SRs affected by any cartilage thickness damage | 0.1896 | |||
| 0 additional SRs affected | 52 (51%) | 43 (60%) | ||
| 1 additional SR affected | 21 (21%) | 18 (25%) | 1.01 (0.48, 2.16) | |
| 2+ additional SRs affected | 28 (28%) | 11 (15%) | 2.09 (0.92, 4.75) | |
| Maximum advancement in cartilage thickness score across all SRs | 0.1787 | |||
| No Change | 39 (39%) | 38 (53%) | ||
| Advance by 1 grade | 32 (32%) | 19 (26%) | 1.66 (0.80, 3.45) | |
| Advance by 2+ grades | 30 (30%) | 15 (21%) | 1.97 (0.90, 4.32) | |
| Osteophytes | ||||
| Number of locations with advancement in osteophyte score | 0.0097* | |||
| 0 locations with advancement | 19 (18%) | 23 (32%) | ||
| 1 locations with advancement | 10 (10%) | 13 (18%) | 0.89 (0.31, 2.51) | |
| 2+ locations with advancement | 74 (72%) | 36 (50%) | 2.64 (1.25, 5.58) | |
| Any additional locations affected by any osteophyte | 0.0230 | |||
| No | 30 (29%) | 33 (46%) | ||
| Yes | 73 (71%) | 39 (54%) | 2.10 (1.11, 3.99) | |
| Any advancement in osteophytes score across all locations | 0.0388 | |||
| No | 19 (18%) | 23 (32%) | ||
| Yes | 84 (82%) | 49 (68%) | 2.13 (1.04, 4.35) | |
| Bone Marrow Lesions | ||||
| Change in number of SRs affected by any BML | 0.3595 | |||
| Improvement | 19 (19%) | 14 (20%) | ||
| No change | 38 (37%) | 33 (46%) | 0.90 (0.38, 2.09) | |
| 1 additional SR affected | 23 (23%) | 16 (23%) | 1.10 (0.42, 2.87) | |
| 2+ additional SR affected | 22 (22%) | 8 (11%) | 2.10 (0.71, 6.23) | |
| Maximum advancement in BML size score across all SRs | 0.2543 | |||
| No Change | 39 (38%) | 36 (51%) | ||
| Advance by 1 grade | 31 (30%) | 19 (27%) | 1.53 (0.73, 3.22) | |
| Advance by 2+ grades | 32 (31%) | 16 (23%) | 1.85 (0.85, 4.02) | |
| Any SRs with improvement in BML size score | 0.9042 | |||
| No | 53 (52%) | 38 (54%) | ||
| Yes | 49 (48%) | 33 (46%) | 1.04 (0.56, 1.92) | |
| Any of SRs with advancement in BML size score | 0.1102 | |||
| No | 39 (38%) | 36 (51%) | ||
| Yes | 63 (62%) | 35 (49%) | 1.68 (0.89, 3.16) | |
| Hoffa’s Synovitis and Synovitis-Effusion | ||||
| Change in Hoffa-synovitis | 0.6610 | |||
| Improvement | 27 (26%) | 17 (24%) | ||
| No Change | 62 (60%) | 47 (66%) | 0.80 (0.39, 1.64) | |
| Advance | 14 (14%) | 7 (10%) | 1.19 (0.39, 3.62) | |
| Change in effusion-synovitis | 0.0063* | |||
| Improvement | 32 (31%) | 37 (51%) | ||
| No Change | 46 (45%) | 29 (40%) | 1.84 (0.94, 3.59) | |
| Advance | 25 (24%) | 6 (8%) | 4.99 (1.80, 13.85) | |
indicates statistically significant p-value after Holm correction.
APM= Arthroscopic partial meniscectomy.
PT=Physical therapy.
SR=Subregion.
The number of subregions with advancement in cartilage thickness score ranged from 0 to 6 with a mean (SD) of 1.2 (1.4). The mean (SD) number of subregions with advancement in cartilage thickness score was 1.5 (1.6) in the APM group and 0.8 (1.0) in the PT group (Figure 2B). Thirty-eight percent of subjects undergoing APM had 2+ subregions with advancement compared to 26% of those receiving PT. Compared to subjects receiving PT, subjects receiving APM had approximately 2-fold elevated odds of having 2+ subregions with advancement, but these associations did not reach statistical significance (OR 1.98 (95% CI 0.96, 4.10; Table 2). Similarly, patients who had APM had two-fold greater odds of advancement in the number of subregions affected by reduced cartilage thickness scores and maximum advancement in cartilage thickness score; these associations did not reach statistical significance (Table 2).
Osteophytes
The number of locations with advancement in osteophyte score ranged from 0 to 11 with a mean (SD) of 3.2 (3.0). The mean number of locations with advancement in osteophyte score was 3.8 (3.2) in the APM group compared to 2.4 (2.7) in the PT group (Figure 2C). Osteophyte advancement was frequent: 82% of subjects in the APM group and 68% of subjects in the PT group experienced advancement in MRI-based osteophyte score in at least one subregion. Subjects undergoing APM had a 2.6-fold increased odds of having 2+ subregions with advancement in osteophyte score compared to 0 subregions (OR 2.6, 95% CI 1.3, 5.6; Table 2).
Bone marrow lesions
The change in the number of subregions affected by BML ranged from −4 to 7 with a mean (SD) of 0.4 (1.4). The mean (SD) change in the number of subregions with any BML was 0.6 (1.6) subregions in the APM group and 0.2 (1.1) in the PT group (Figure 3A). Treatment was not significantly associated with change in BML (Table 2).
Figure 3. Early MRI-Based Advancement in BML, Hoffa-Synovitis, and Effusion-Synovitis by Treatment Group (Primary Analysis).
Each panel shows the distribution of MRI-based advancement by treatment group, for A) Bone Marrow Lesion (BML), B) Hoffa-synovitis, C) Effusion-Synovitis. Advancement is along the Y-axis and treatment group (APM vs. PT) is along the X-axis. Advancement is measured in number of subregions for BML and in change in score for Hoffa-synovitis and Effusion-synovitis. Each small circle represents one participant. The diamond indicates the mean and the square indicates the median.
Synovitis
Fourteen percent of subjects undergoing APM experienced advancement in Hoffa-Synovitis compared to 10% of subjects receiving PT (Figure 3B). This difference was not statistically significant. Twenty-four percent of subjects undergoing APM experienced advancement in effusion-synovitis, 45% had no change, and 31% improved (Figure 3C). Eight percent of subjects receiving PT experienced advancement, 40% no change, and 51% improvement. The adjusted odds of advancement vs. improvement associated with APM was 5.0 (95% CI 1.8, 13.8).
Secondary Analysis
The subjects included in secondary analysis did not differ on baseline characteristics compared to the subjects excluded (Appendix Table 4, Appendix Table 5).
As treated
Secondary analysis was performed for the as-treated sample, which included the 37 subjects who crossed over from PT to APM in the APM group. Results were similar to the main analysis, with statistically significant differences between treatment arms in cartilage surface area, osteophytes, and effusion-synovitis (Appendix Table 6). As in the primary analysis, we did not find significant associations between treatment group and changes in BML, cartilage thickness, or Hoffa-synovitis.
Intention to treat
The ITT analysis included the 37 subjects who crossed over from PT to APM in the PT group. Results were similar to the main analysis, with increased odds of advancement of cartilage surface area and advancement in effusion-synovitis in the APM vs. PT groups. The odds of advancement in osteophytes were in the same direction but attenuated in this analysis compared to the primary and as-treated analyses and did not reach statistical significance (Appendix Table 7). As in the primary and secondary analyses, we did not find significant associations between treatment group and changes in BML, cartilage thickness or Hoffa-synovitis.
Sensitivity Analysis for Missing Data
Sensitivity analysis with multiple imputation for missing data demonstrated similar associations as the main analysis under a missing at random (MAR) mechanism. That is, if we assume that patients missing 18-month cartilage change data are similar to those with data, or that we can reasonably impute change from subject characteristics (age, sex, KL, baseline MOAKS if available) then our conclusions do not change (Table 3). As we change the missing data mechanism and assume that subjects in the APM group with missing data are progressing less than observed APM subjects and/or PT subjects with missing data are progressing more than observed PT subjects, the associations are attenuated. Generally, in order to change the conclusion about the association between treatment group and progression we would have to assume an extreme missing data mechanism: that PT subjects with missing data progress in the same fashion as observed APM subjects and APM subjects with missing data progression in the same fashion as observed PT subjects.
Table 3.
Association between Treatment Group and Progression, Sensitivity Analysis with Multiple Imputation for Missing Data. Odds ratios (95% CI) for progression for APM vs. PT presented in cells.
| Primary analysis | Secondary As-Treated Analysis | MAR | MNAR (1) | MNAR (2) | MNAR (3) | MNAR (4) | MNAR (5) | |
|---|---|---|---|---|---|---|---|---|
Less extreme  More extreme | ||||||||
| Cartilage Surface Area | ||||||||
| Number of SRs with advancement in cartilage surface area score | ||||||||
| 0 SRs with advancement | ref | ref | ref | ref | ref | ref | ref | ref |
| 1 SR with advancement | 2.0 (0.9, 4.8) | 1.7 (0.8, 3.8) | 1.7 (0.7, 4.0) | 1.2 (0.5, 2.7) | 1.4 (0.6, 3.0) | 1.3 (0.5, 3.2) | 1.4 (0.7, 2.9) | 1.0 (0.4, 2.2) |
| 2+ SRs with advancement | 4.2 (2.0, 9.0) | 3.3 (1.7, 6.5) | 3.3 (1.6, 6.5) | 2.5 (1.2, 5.2) | 2.0 (1.01, 3.9) | 2.0 (0.97, 4.2) | 2.2 (1.2, 4.0) | 1.4 (0.7, 2.8) |
| Number of additional SRs affected by any cartilage surface area damage | ||||||||
| 0 additional SRs affected | ref | ref | ref | ref | ref | ref | ref | ref |
| 1 additional SR affected | 2.8 (1.3, 6.0) | 2.4 (1.2, 4.8) | 2.6 (1.3, 5.2) | 2.0 (0.98, 4.1) | 1.9 (0.9, 4.0) | 1.8 (0.9, 3.5) | 1.7 (0.9, 3.4) | 1.4 (0.7, 2.6) |
| 2+ additional SRs affected | 2.7 (1.2, 5.9) | 2.1 (1.0, 4.5) | 2.3 (1.1, 5.0) | 1.8 (0.8, 3.7) | 1.8 (0.8, 4.0) | 1.5 (0.7, 3.3) | 1.7 (0.8, 3.6) | 1.3 (0.6, 2.6) |
| Maximum advancement in cartilage surface area score across all SRs | ||||||||
| No Change | ref | ref | ref | ref | ref | ref | ref | ref |
| Advance by 1 grade | 2.5 (1.1, 5.5) | 2.0 (1.0, 4.2) | 1.9 (1.0, 3.9) | 1.3 (0.6, 2.6) | 1.6 (0.8, 3.2) | 1.5 (0.7, 3.4) | 1.6 (0.8, 3.1) | 1.1 (0.5, 2.1) |
| Advance by 2+ grades | 4.2 (1.9, 9.2) | 3.3 (1.6, 6.7) | 3.2 (1.6, 6.5) | 2.4 (1.2, 4.9) | 2.1 (1.1, 4.2) | 2.1 (1.0, 4.5) | 2.1 (1.1, 4.0) | 1.4 (0.7, 2.8) |
| Cartilage Thickness | ||||||||
| Number of SRs with advancement in cartilage thickness score | ||||||||
| 0 SRs with advancement | ref | ref | ref | ref | ref | ref | ref | ref |
| 1 SR with advancement | 1.6 (0.7, 3.4) | 1.4 (0.7, 3.1) | 1.4 (0.7, 2.9) | 1.0 (0.4, 2.2) | 1.2 (0.6, 2.6) | 1.4 (0.6, 3.1) | 1.3 (0.6, 2.8) | 1.1 (0.5, 2.4) |
| 2+ SRs with advancement | 2.0 (1.0, 4.1) | 1.9 (1.0, 3.8) | 1.9 (1.0, 3.6) | 1.5 (0.7, 3.0) | 1.4 (0.7, 2.9) | 1.4 (0.7, 3.0) | 1.5 (0.8, 3.0) | 1.1 (0.6, 2.3) |
| Number of additional SRs affected by any cartilage thickness damage | ||||||||
| 0 additional SRs affected | ref | ref | ref | ref | ref | ref | ref | ref |
| 1 additional SR affected | 1.0 (0.5, 2.2) | 0.9 (0.5, 1.9) | 0.8 (0.4, 1.6) | 0.7 (0.3, 1.6) | 0.7 (0.3, 1.6) | 0.9 (0.4, 1.7) | 0.9 (0.4, 2.0) | 0.9 (0.4, 1.7) |
| 2+ additional SRs affected | 2.1 (0.9, 4.8) | 1.9 (0.9, 4.1) | 1.9 (0.9, 4.1) | 1.7 (0.8, 3.5) | 1.7 (0.8, 3.7) | 1.5 (0.7, 3.4) | 1.5 (0.7, 3.1) | 1.2 (0.6, 2.6) |
| Maximum advancement in cartilage thickness score across all SRs | ||||||||
| No Change | ref | ref | ref | ref | ref | ref | ref | ref |
| Advance by 1 grade | 1.7 (0.8, 3.4) | 1.7 (0.8, 3.3) | 1.6 (0.8, 3.3) | 1.2 (0.6, 2.5) | 1.5 (0.7, 3.2) | 1.5 (0.7, 3.0) | 1.3 (0.7, 2.7) | 1.2 (0.6, 2.3) |
| Advance by 2+ grades | 2.0 (0.9, 4.3) | 1.8 (0.8, 3.7) | 1.8 (0.9, 3.7) | 1.4 (0.7, 2.8) | 1.4 (0.7, 2.9) | 1.5 (0.7, 3.2) | 1.4 (0.7, 2.7) | 1.3 (0.6, 2.6) |
| Osteophytes | ||||||||
| Number of locations with advancement in osteophyte score | ||||||||
| 0 locations with advancement | ref | ref | ref | ref | ref | ref | ref | ref |
| 1 locations with advancement | 0.9 (0.3, 2.5) | 0.8 (0.3, 2.0) | 0.7 (0.3, 1.8) | 0.7 (0.3, 1.6) | 0.6 (0.2, 1.7) | 0.7 (0.3, 1.9) | 0.8 (0.3, 2.1) | 0.7 (0.3, 1.9) |
| 2+ locations with advancement | 2.6 (1.3, 5.6) | 2.5 (1.2, 5.0) | 2.4 (1.1, 5.0) | 1.9 (1.0, 3.7) | 1.7 (0.9, 3.4) | 1.7 (0.9, 3.3) | 1.9 (1.0, 3.6) | 1.2 (0.6, 2.3) |
| Any additional locations affected by any osteophyte | ||||||||
| No | ref | ref | ref | ref | ref | ref | ref | ref |
| Yes | 2.1 (1.1, 4.0) | 2.0 (1.1, 3.6) | 1.9 (1.1, 3.4) | 1.6 (0.8, 3.0) | 1.3 (0.7, 2.4) | 1.5 (0.8, 2.8) | 1.5 (0.8, 2.6) | 1.2 (0.7, 2.1) |
| Any advancement in osteophytes score across all locations | ||||||||
| No | ref | ref | ref | ref | ref | ref | ref | ref |
| Yes | 2.1 (1.0, 4.4) | 2.0 (1.0, 3.9) | 2.0 (1.0, 3.9) | 1.8 (0.8, 3.7) | 1.6 (0.8, 3.1) | 1.5 (0.8, 2.9) | 1.5 (0.8, 3.0) | 1.3 (0.7, 2.5) |
| Bone Marrow Lesions | ||||||||
| Change in number of subregions affected by any BML | ||||||||
| Improvement | ref | ref | ref | ref | ref | ref | ref | ref |
| No change | 0.9 (0.4, 2.1) | 1.2 (0.6, 2.8) | 1.2 (0.6, 2.6) | 0.9 (0.4, 1.9) | 0.9 (0.4, 2.1) | 1.0 (0.4, 2.3) | 0.9 (0.4, 1.9) | 0.8 (0.4, 1.8) |
| 1 additional SR affected | 1.1 (0.4, 2.9) | 1.3 (0.5, 3.3) | 1.1 (0.4, 2.9) | 0.9 (0.3, 2.2) | 1.0 (0.4, 2.2) | 1.0 (0.5, 2.4) | 0.9 (0.4, 2.2) | 0.9 (0.4, 2.1) |
| 2+ additional SR affected | 2.1 (0.7, 6.2) | 2.1 (0.7, 6.0) | 2.0 (0.7, 5.6) | 1.4 (0.5, 4.4) | 1.6 (0.5, 5.0) | 1.6 (0.5, 5.2) | 1.5 (0.5, 4.3) | 1.3 (0.5, 3.5) |
| Maximum advancement in BML size score across all SRs | ||||||||
| No Change | ref | ref | ref | ref | ref | ref | ref | ref |
| Advance by 1 grade | 1.5 (0.7, 3.2) | 1.5 (0.7, 2.9) | 1.4 (0.7, 2.9) | 1.1 (0.6, 2.2) | 1.3 (0.6, 2.8) | 1.3 (0.6, 2.7) | 1.3 (0.6, 2.6) | 1.1 (0.5, 2.2) |
| Advance by 2+ grades | 1.9 (0.9, 4.0) | 1.3 (0.6, 2.8) | 1.3 (0.6, 2.9) | 1.0 (0.5, 1.9) | 1.1 (0.6, 2.3) | 1.1 (0.5, 2.4) | 1.1 (0.6, 2.1) | 1.1 (0.5, 2.2) |
| Any SRs with improvement in BML size score | ||||||||
| No | ref | ref | ref | ref | ref | ref | ref | ref |
| Yes | 1.0 (0.6, 1.9) | 0.8 (0.4, 1.4) | 0.8 (0.5, 1.5) | 0.7 (0.4, 1.3) | 1.0 (0.5, 1.8) | 0.9 (0.5, 1.7) | 0.9 (0.5, 1.6) | 1.0 (0.5, 1.9) |
| Any of SRs with advancement in BML size score | ||||||||
| No | ref | ref | ref | ref | ref | ref | ref | ref |
| Yes | 1.7 (0.9, 3.2) | 1.4 (0.8, 2.5) | 1.5 (0.8, 2.6) | 1.1 (0.6, 2.0) | 1.3 (0.7, 2.3) | 1.3 (0.7, 2.3) | 1.2 (0.7, 2.3) | 1.2 (0.6, 2.2) |
| Hoffa’s Synovitis and Synovitis-Effusion | ||||||||
| Change in Hoffa-synovitis | ||||||||
| Improvement | ref | ref | ref | ref | ref | ref | ref | ref |
| No Change | 0.8 (0.4, 1.6) | 0.8 (0.4, 1.6) | 0.8 (0.4, 1.7) | 0.7 (0.3, 1.5) | 0.7 (0.4, 1.4) | 0.8 (0.4, 1.6) | 0.8 (0.4, 1.6) | 0.8 (0.4, 1.6) |
| Advance | 1.2 (0.4, 3.6) | 1.7 (0.6, 4.7) | 1.8 (0.6, 5.5) | 1.3 (0.5, 3.5) | 1.5 (0.5, 4.4) | 1.5 (0.5, 4.2) | 1.2 (0.4, 3.6) | 1.1 (0.4, 3.1) |
| Change in effusion-synovitis | ||||||||
| Improvement | ref | ref | ref | ref | ref | ref | ref | ref |
| No Change | 1.8 (0.9, 3.6) | 1.9 (1.0, 3.6) | 1.9 (1.02, 3.4) | 1.6 (0.9, 3.0) | 1.6 (0.9, 2.9) | 1.4 (0.8, 2.5) | 1.6 (0.8, 3.0) | 1.2 (0.6, 2.1) |
| Advance | 5.0 (1.8, 13.8) | 4.1 (1.5, 10.9) | 4.1 (1.5, 11.4) | 3.7 (1.3, 10.6) | 2.6 (1.0, 6.5) | 2.7 (1.0, 7.4 | 2.5 (1.0, 6.2) | 1.7 (0.7, 4.2) |
MAR: Missing at Random
MNAR: Missing Not at Random
APM= Arthroscopic partial meniscectomy.
PT=Physical therapy.
SR=Subregion.
Discussion
We evaluated data from an RCT of APM with PT vs. PT alone, and found that both treatment groups had substantial early advancement of MRI-based biomarkers of each of the structural joint features examined. Patients undergoing APM had greater early advancement in MRI-based markers over 18 months than those treated non-operatively for cartilage surface area, osteophytes, and effusion-synovitis.
Two RCTs found no differences in radiographic advancement between subjects treated with APM and those treated non-operatively for degenerative meniscal tear.(5,6) However, radiographic OA grade is an insensitive marker of structural change(22,23); both studies found radiographic advancement rates of less than 5%. Roemer et al. found that both meniscal damage and partial meniscectomy were associated with incident OA (Kellgren Lawrence 2) over 4 years in a nested case-control sample from the Osteoarthritis Initiative.(10) In addition, the authors used MOAKS to evaluate MRI-based cartilage progression, defining progression as any increase in either size or thickness of cartilage damage. In the incident OA cases, partial meniscectomy was associated with worsening cartilage damage compared to knees with meniscal damage and without meniscectomy, and to knees without meniscal damage. However, only 26 knees underwent meniscal surgery and had MRI. Our analysis builds on the work of Roemer et al. by taking advantage of the large MeTeOR trial cohort. In MeTeOR, all subjects had documented meniscal tear and all had substantial enough pain and functional limitation that subjects and their enrolling surgeons were prepared to proceed to APM. This balancing of structural and symptom severity between treatment groups is difficult to achieve in observational studies. The surgeries in MeTeOR were done in a uniform manner and follow-ups were at regular intervals post-randomization. Like Roemer et al., we found associations between APM and subsequent cartilage advancement.
To our knowledge, this is the first study evaluating early MRI-based changes in a follow up evaluation of data from of an RCT of APM vs. non-operative therapy. We found that MeTeOR participants who had APM had higher likelihood of MRI-based advancement in cartilage surface area, osteophytes, and effusion-synovitis. We did not find significant associations between treatment type and advancement in BMLs or Hoffa-synovitis. The lack of association between treatment and BMLs and Hoffa synovitis may reflect the transient nature of BMLs and synovitis; these features do not reflect cumulative damage as do cartilage damage and osteophytosis. The results were similar in the main per-protocol analysis and in the secondary as-treated and intention to treat analyses.
The clinical relevance of these early MRI findings remain uncertain. It will be important to determine whether subjects who demonstrate these changes in imaging findings over 18 months are at higher risk of worsening in symptom severity, functional limitation and total joint replacement over subsequent follow-up. The findings underscore the importance of clinical follow up of this cohort and, more generally, of individuals with meniscal tear treated either operatively or non-operatively.
Only 225 out of 351 (64%) randomized had MRI data available at both baseline and 18 months. While our analyses found associations between some aspects of structural disease progression and treatment group, we were concerned about the amount of missing data. Our tipping point sensitivity analysis with multiple imputation suggested that it would take an extreme missing data mechanism to change the conclusions; we would have to assume that PT patients missing data are actually advancing at rates similar to the APM patients, and that APM subjects missing data actually progress at rates similar to PT subject. While this extreme scenario does not seem plausible, we can never rule out a missing data mechanism with 100% certainty.
These results should be interpreted within the context of the study limitations. The primary analytic sample was limited to those subjects undergoing MRI at 18 months. Thirty-one percent of patients randomized to PT crossed over to APM within 6 months of randomization. Subjects crossing over to APM had shorter symptom duration and greater baseline pain; thus the balancing of confounders inherent to randomization may have been disrupted. We evaluated differences in known potential confounders between the groups, and adjusted where necessary. However, we cannot be certain that the groups were balanced on unknown confounders. To further minimize risk of bias, we conducted three sets of analyses, all adjusted for KL grade and factors imbalanced at baseline: the primary analysis that excluded crossovers, a sensitivity as-treated analysis that included cross-overs in the APM group; and an intention to treat analysis that included cross overs in the PT group. All three of these analyses yielded similar conclusions. Due to the multinomial nature of many of the outcomes variables we used logistic regression and present odds ratios. Odds ratios overstate relative risks, especially when the prevalence of the outcome is high, as in this analysis(24). Thus, these odds ratios should not be interpreted as relative risks. Caution should be taken in generalizing these results to a more general knee OA cohort. First, Inclusion criteria for the MeTeOR trial included evidence of meniscal tear on MRI and symptoms consistent with torn meniscus (i.e., clicking, catching, popping). Each patient had to be willing to undergo APM if randomized to the APM group.(11) Thus, the MeTeOR trial may be more generalizable to patients with knee OA and meniscal tear with symptoms. Patients were recruited from academic medical centers and only 26% of eligible subjects agreed to participate in the MeTeOR RCT.(4) Finally, this analysis is a secondary analysis of an RCT and as such we did not conduct a formal power analysis.(25) To address the uncertainty in our parameter estimates, we included 95% confidence intervals.(26)
In discussing treatment options for symptomatic meniscal tear, patients and providers must weigh the potential benefits and risks of treatment options, including this information on structural advancement. Future work will assess the association between early structural advancement and subsequent pain, function, and risk of total knee replacement. Clinicians should be aware that regardless of treatment, there was MRI evidence of progression. At this point the clinical meaning of the MRI-based changes documented in this study is unknown. Assessing the relevance of these MRI-based changes is an important research priority.
Supplementary Material
Significance and Innovations.
This is the first study to use data from a randomized controlled trial (RCT) of surgical vs. non-operative treatment for patients with knee OA and meniscal tear to evaluate the early MRI-based changes over 18-month period for patients undergoing APM or PT.
We found marked MRI-based advancement in both groups. In addition, we found that patients who had APM had higher odds of advancement in cartilage surface area, osteophytes, and effusion-synovitis, while the data did not provide sufficient evidence to establish an association between treatment type and change in cartilage thickness, BMLs, or Hoffa-synovitis.
The clinical relevance of these findings requires further study and should be considered a research priority.
Acknowledgments
Supported by: RRF Investigator Award, NIAMS R01AR05557, K24 AR 057827, P30 AR 072577
Trial Registration: ClinicalTrials.gov NCT00597012
References
- 1.Deshpande BR, Katz JN, Solomon DH, Yelin EH, Hunter DJ, Messier SP, et al. Number of persons with symptomatic knee osteoarthritis in the us: Impact of race and ethnicity, age, sex, and obesity. Arthritis Care Res (Hoboken) 2016;68:1743–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Bhattacharyya T, Gale D, Dewire P, Totterman S, Gale ME, McLaughlin S, et al. The clinical importance of meniscal tears demonstrated by magnetic resonance imaging in osteoarthritis of the knee. J Bone Joint Surg Am 2003;85-A:4–9. [DOI] [PubMed] [Google Scholar]
- 3.Thorlund JB, Juhl CB, Roos EM, Lohmander LS. Arthroscopic surgery for degenerative knee: Systematic review and meta-analysis of benefits and harms. Br J Sports Med 2015;49:1229–35. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Katz JN, Brophy RH, Chaisson CE, De Chaves L, Cole BJ, Dahm DL, et al. Surgery versus physical therapy for a meniscal tear and osteoarthritis. New England Journal of Medicine 2013;368:1675–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Herrlin SV, Wange PO, Lapidus G, Hallander M, Werner S, Weidenhielm L. Is arthroscopic surgery beneficial in treating non-traumatic, degenerative medial meniscal tears? A five year follow-up. Knee Surg Sports Traumatol Arthrosc 2013;21:358–64. [DOI] [PubMed] [Google Scholar]
- 6.Yim JH, Seon JK, Song EK, Choi JI, Kim MC, Lee KB, et al. A comparative study of meniscectomy and nonoperative treatment for degenerative horizontal tears of the medial meniscus. Am J Sports Med 2013;41:1565–70. [DOI] [PubMed] [Google Scholar]
- 7.Kise NJ, Risberg MA, Stensrud S, Ranstam J, Engebretsen L, Roos EM. Exercise therapy versus arthroscopic partial meniscectomy for degenerative meniscal tear in middle aged patients: Randomised controlled trial with two year follow-up. BMJ 2016;354:i3740. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Englund M, Guermazi A, Roemer FW, Aliabadi P, Yang M, Lewis CE, et al. Meniscal tear in knees without surgery and the development of radiographic osteoarthritis among middle-aged and elderly persons: The multicenter osteoarthritis study. Arthritis Rheum 2009;60:831–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Englund M, Roos EM, Lohmander LS. Impact of type of meniscal tear on radiographic and symptomatic knee osteoarthritis: A sixteen-year followup of meniscectomy with matched controls. Arthritis Rheum 2003;48:2178–87. [DOI] [PubMed] [Google Scholar]
- 10.Roemer FW, Kwoh CK, Hannon MJ, Hunter DJ, Eckstein F, Grago J, et al. Partial meniscectomy is associated with increased risk of incident radiographic osteoarthritis and worsening cartilage damage in the following year. Eur Radiol 2017;27:404–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Katz JN, Chaisson CE, Cole B, Guermazi A, Hunter DJ, Jones M, et al. The meteor trial (meniscal tear in osteoarthritis research): Rationale and design features. Contemporary clinical trials 2012;33:1189–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Hunter DJ, Guermazi A, Lo GH, Grainger AJ, Conaghan PG, Boudreau RM, et al. Evolution of semi-quantitative whole joint assessment of knee oa: Moaks (mri osteoarthritis knee score). Osteoarthritis Cartilage 2011;19:990–1002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.MacFarlane LA, Yang H, Collins JE, Guermazi A, Jones MH, Teeple E, et al. Associations among meniscal damage, meniscal symptoms and knee pain severity. Osteoarthritis Cartilage 2017;25:850–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Bedi A, Kelly NH, Baad M, Fox AJ, Brophy RH, Warren RF, et al. Dynamic contact mechanics of the medial meniscus as a function of radial tear, repair, and partial meniscectomy. J Bone Joint Surg Am 2010;92:1398–408. [DOI] [PubMed] [Google Scholar]
- 15.Zhang AL, Miller SL, Coughlin DG, Lotz JC, Feeley BT. Tibiofemoral contact pressures in radial tears of the meniscus treated with all-inside repair, inside-out repair and partial meniscectomy. Knee 2015;22:400–4. [DOI] [PubMed] [Google Scholar]
- 16.Aickin M, Gensler H. Adjusting for multiple testing when reporting research results: The bonferroni vs holm methods. Am J Public Health 1996;86:726–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Holm S A simple sequentially rejective multiple test procedure. Scandinavian Journal of Statistics 1979;6:65–70. [Google Scholar]
- 18.Katz JN, Wright J, Spindler KP, Mandl LA, Safran-Norton CE, Reinke EK, et al. Predictors and outcomes of crossover to surgery from physical therapy for meniscal tear and osteoarthritis: A randomized trial comparing physical therapy and surgery. J Bone Joint Surg Am 2016;98:1890–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Rubin D Multiple imputation for nonresponse in surveys Vol 1371989. [Google Scholar]
- 20.Bell ML, Fairclough DL. Practical and statistical issues in missing data for longitudinal patient-reported outcomes. Stat Methods Med Res 2014;23:440–59. [DOI] [PubMed] [Google Scholar]
- 21.Liublinska V, Rubin DB. Sensitivity analysis for a partially missing binary outcome in a two-arm randomized clinical trial. Stat Med 2014;33:4170–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Hunter DJ, Zhang W, Conaghan PG, Hirko K, Menashe L, Reichmann WM, et al. Responsiveness and reliability of mri in knee osteoarthritis: A meta-analysis of published evidence. Osteoarthritis Cartilage 2011;19:589–605. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Wirth W, Duryea J, Hellio Le Graverand MP, John MR, Nevitt M, Buck RJ, et al. Direct comparison of fixed flexion, radiography and mri in knee osteoarthritis: Responsiveness data from the osteoarthritis initiative. Osteoarthritis Cartilage 2013;21:117–25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Davies HT, Crombie IK, Tavakoli M. When can odds ratios mislead? BMJ 1998;316:989–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hoenig JM, Heisey DM. The abuse of power: The pervasive fallacy of power calculations for data analysis. The American Statistician 2001;55:19–24. [Google Scholar]
- 26.Ranstam J Why the p-value culture is bad and confidence intervals a better alternative: Elsevier; 2012. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.