Predicted probability of meniscus tears : comparing history and physical examination with MRI

BACKGROUND: The indication for surgical treatment of a meniscal lesion should not only rely on magnetic resonance imaging (MRI) findings, but also on a detailed history and a thorough clinical examination. However, various intra-articular lesions may often produce similar symptoms. So, what kinds of symptoms are more associated with a meniscal tear? Is MRI worth doing? OBJECTIVE: The aims of this study were to identify sensitive and specific clinical tests and elements of patients’ history with a high predictive value, and to assess the combined diagnostic accuracy of sensitive and specific clinical tests and elements of patients’ history with MRI. METHODS: Data from 281 consecutive knee arthroscopies to investigate and treat suspected internal knee pathologies were retrospectively collected between March 2009 and April 2010. The study group consisted of 262 knees. Statistically significant factors in the clinical diagnosis of meniscal tears were screened by a chi-square test. Logistic regression analysis was used to determine which factors associated with meniscal tears found during arthroscopy. The diagnostic values of MRI and the sensitive and specific clinical tests and elements of patients’ history with high predictive value for meniscal tears were calculated. RESULTS: The overall diagnostic value of MRI for meniscal tears was: accuracy, 88.8%; sensitivity, 95.7%; specificity, 75.8%; positive predictive value (PPV), 88.2%; and negative predictive value (NPV), 90.4%. Giving way, locking and McMurray’s test were independent diagnostic factors with a predicted correct percentage of 80.0% (p <0.05) for the diagnosis of meniscal tears found during arthroscopy. Locking, McMurray’s test and MRI increased the predicted correct percentage of meniscal tears found during arthroscopy to 91.6% (p <0.05). For the diagnosis of meniscal tears found during arthroscopy, giving way, locking and McMurray’s test had the following values for accuracy (49.2, 60.9, 76), sensitivity (43.5, 55.2, 75.8), specificity (84, 96, 76.9), PPV (94.4, 98.8, 95.1) and NPV (19.4, 25.8, 35.1). Combining MRI, the diagnostic values of giving way, locking, and McMurray’s test were: accuracy, 88.3,89.9,89.4; sensitivity, 95.7,97.4,97.4; specificity, 74.2,75.8,74.2; PPV, 87.5,88.4,87.7; and NPV, 90.2,94,93.9. CONCLUSIONS: Giving way, locking and McMurray’s test are independent clinical diagnostic factors for the diagnosis of meniscal tears. MRI has higher accuracy, sensitivity and NPV for the diagnosis of meniscal tears than giving way, locking and McMurray’s test. The combination of giving way, locking, McMurray’s test and MRI for confirmation is typical for a meniscal lesion diagnosis. Based on these findings, MRI should be used in a standard manner to detect meniscal tears found during arthroscopy.


Introduction
Meniscus tears, seen in young and old patients alike, are an extremely common cause of knee pain [1].Incidental meniscal findings on magnetic resonance imaging (MRI) of the knee are common in the general population and increase with increasing age.Among persons with radiographic evidence of osteoarthritis (Kellgren-Lawrence grade 2 or higher), the prevalence of a meniscal tear was found to be higher than the corresponding prevalence among persons without radiographic evidence of osteoarthritis [2].Arthroscopic surgery in patients with moderate-tosevere osteoarthritis of the knee provides no additional benefit to optimised physical and medical therapy [3].It is important to make an accurate diagnosis of meniscus tears so that the appropriate treatment can be given.A detailed history and physical examination can help differentiate patients who have a meniscus tear from those whose knee pain arises from other conditions [4][5][6].MRI is the test commonly used on patients with meniscus tears of the knee because of its ability to diagnose meniscal tears [7][8][9].MRI is commonly used because various intra-articular lesions historically have had common symptoms, patient history alone is in adequate as a diagnostic tool, and the diagnostic accuracy of clinical tests for meniscal tears has often been questioned.A review of the available literature reveals conflicting results as to their usefulness [10][11][12][13][14][15].If the patient had no clinical symptoms, even though MRI showed meniscal tears of a patient, knee arthroscopy was unnecessary.In a high percentage of cases, relying on MRI alone without using clinical judgment may have led to inappropriate treatment.In any case, MRI did not prevent "unnecessary surgery" [16].The aims of this study were to identify sensitive and specific clinical tests and elements of patients' history with high predictive value, and to assess the combined diagnostic accuracy of sensitive and specific clinical tests and elements of patients' history with MRI by using arthroscopy as a gold standard for control.

Study population
Institutional review board approval was obtained before beginning the study, allowing a retrospective review of patient records and images without informed consent.We retrospectively collected data from 281 consecutive knee arthroscopies performed by 2 experienced knee surgeons between March 2009 and April 2010 to investigate and treat suspected internal knee pathologies [17].A total of 19 knees which had had previous meniscectomies, knee ligament repair, or reconstruction and knee arthroscopy were excluded from the study (fig.1).The study group consisted of 262 knees and their medical records were subsequently reviewed for relevant clinical history, MRI reports and operative findings.There were 115 male and 135 female patients in the age group of 7-78 years, with a median age of 41.There were 138 (53%) left knees and 124 (47%) right knees in the study group.The mechanisms of tear for the 262 knees included beyond recall (88 knees, 33.59%), twisting (101 knees, 38.55%), jumping-landing (39 knees, 14.88%), trauma (29 knees, 11.07%), and valgus-varus stress (5 knees, 1.91%).

Clinical history and index items
Clinical data (including patient demographics, waiting period between MRI and arthroscopy), suggestive symptoms (including pain, swelling, limited motion, giving way locking, and clicking), physical examination, mechanism of tear, clinical diagnosis and operative details were re-

MRI examinations and arthroscopy
All MRI examinations were performed at our institution with a 1.5-T superconducting magnet (Signa, GE Healthcare, USA) and a dedicated knee extremity coil.For most of the examinations, high-resolution FSE proton density-weighted and fat-suppressed, FSE T2-weighted paired sequences were performed in the sagittal and coronal planes.Fat-suppressed, FSE T2-weighted images were also obtained in the axial plane.FSE proton density-weighted sequence parameters were as follows: 2 signals acquired; TR/TE, 2,000/18; echo-train length, 4; receiver bandwidth, 32 kHz; matrix, 256 × 256; field of view, 14-16 cm; and slice thickness, 3 mm.Fat-suppressed, FSE T2-weighted sequence parameters were 2 signals acquired;TR/TE, 4,000/50; echo-train length, 8; receiver bandwidth, 20 kHz; matrix, 256 × 256; field of view, 14-16 cm; and slice thickness, 3 mm.There were 179 knees with suspected meniscal tears that underwent MRI within 6 weeks before arthroscopy.MRI reports of 179 knees with suspected meniscal tears were reviewed and analysed.Arthroscopies were performed under general anaesthesia and the usual technique by 2 experienced knee surgeons [18,19].Arthroscopy findings served as the study's gold standard.

Statistical analysis
The composite data was tabulated on a Microsoft Excel spreadsheet and studied for correlation.The diagnostic accuracy of MRI and the sensitive and specific clinical tests and elements of patients' history with high predictive value for meniscal tears were determined by calculating sensitivity (Se), specificity (Sp), positive predictive value (PPV) and negative predictive value (NPV).Continuous variables were summarised using median values.Categorical variables were summarised using frequency and were compared using the chi-square or McNemar test as appropriate.Logistic regression analysis was used to examine correlations with regard to suggestive symptoms, clinical tests and MRI results associated with meniscal tears.Correlation of the clinical examination and MRI with arthroscopy from the pooled data of 262 knees was expressed as a percentage.All the tests were 2-sided, and a p-value of less than 0.05 was considered to be statistically significant.Analyses were performed using SPSS version 13.0 (SPSS, Chicago, IL, USA).

Results
Of the 262 knees, 189 knees with a meniscal tear were identified at surgery in 180 patients.There were 89 male and 91 female patients in the age group of 7-77 years, with a median age of 41.There were 85 (45%) left knees and 104 (55%) right knees.There were 82 lateral meniscus tears (and 36 discoid lateral meniscus accompanied men-

Original article
Swiss Med Wkly.2011;141:w13314 iscus tears) and 77 medial meniscus tears.There were 179 knees with a suspected meniscal tear that underwent MRI within 6 weeks before arthroscopy (median = 1.3 weeks).The operative diagnoses of 262 knees were meniscal tear (189), osteoarthritis (134), synovial disease (122), loose body (66) and others (30) as listed in table 1.Of 179 knees that underwent MRI, the operative diagnoses were meniscal tear (117), osteoarthritis (87), synovial disease (68), and loose body (26).Positive findings of MRI were meniscal tear (112), osteoarthritis (11), synovial disease (3), and loose body (4).Of the 179 knees with an MRI diagnosis of suspected meniscal tears, 112 were confirmed on MRI and arthroscopy (true positive); 47 knees had no evidence of meniscal tear on MRI and arthroscopic procedure (true negative); 15 knees which didn't have meniscal tears confirmed by arthroscopy were misinterpreted by MRI (false positive); and 5 knees with meniscal tears confirmed by arthroscopy were not diagnosed by MRI (false negative).MRI diagnosis had an accuracy of 88% for meniscal tears (sensitivity 96%, specificity 76%, positive predictive value 88% and negative predictive value 90%) (table 2).By performing the chi-square test it was found that items such as swelling, giving way, locking, quadriceps atrophy, joint line tenderness test, lack of flexion or extension, Apley's test, McMurray's test, Kellogg-Speed test, pain on hyperextension, and pain on hyperflexion were common in meniscus tears (p <0.05).The results of univariate analysis by chi square analysis are given in table 3. Multivariate logistic regression analysis of those significant clinical variables showed that giving way, locking and McMurray's test were independent diagnostic factors with a predicted correct percentage of 80.0% (p <0.05) for the diagnosis of meniscal tears found during arthroscopy (table 4).We knew that MRI was also a diagnostic factor, and multivariate logistic regression analysis of all the significant variables including clinical items and MRI showed that locking, McMurray's test and MRI were independent diagnostic factors with a higher predicted correct percentage (91.6%)(p <0.05) (table 5).By using logistic regression analysis of the significant variables (

Discussion
The frequent association of meniscal tears with other intraarticular lesions has been observed.Meniscal tears and other intra-articular lesions also have common symptoms [20][21][22][23].We used 262 samples from 250 individuals with or without meniscal lesions to gain a better estimation of history and physical examination's specificity, and to thus more accurately identify optimal clinical tests to predict the presence of a meniscal tear found in arthroscopy.
As it has been found that McMurray's, joint line tenderness and Apley's [24] tests are not diagnostically accurate when used alone, for example, joint line tenderness was sensitive

Original article
Swiss Med Wkly.2011;141:w13314 (75%) but not specific (27%), while the McMurray's test was specific (97%) but not sensitive (52%) [11], the diagnostic value of a history plus a physical examination is limited.In unclear situations, the clinician requests an MRI for additional information to aid planning the operation and to predict the prognosis.However, many studies have shown no significant differences in the clinical and MRI diagnosis of meniscal tears [16,[25][26][27][28][29][30].For example, the sensitivity of MRI for detecting tears of the posterior horn of the lateral meniscus is lower than for tears at other meniscal locations [31].One study even showed that, when the MRI was normal, high clinical suspicion and a skilled clinical examination were more reliable [24]; another study showed that 37% of the operations supported by a significant disorder on MRI were unjustified [32].Several investigators combined these tests with other components of the physical examination such as patient history and imaging [24,26,33].Unfortunately, no conclusions can be drawn when examining these studies, due to their vast differences.In this study, univariate analysis by the chi-square test revealed that history and physical findings such as swelling, giving way, locking, quadriceps atrophy, joint line tenderness test, lack of flexion or extension, Apley's test, McMurray's test, Kellogg-Speed test, pain on hyperexten- sion, and pain on hyperflexion were diagnostic factors for a meniscal tear.By logistic regression analysis of the significant clinical variables, giving way, locking and McMurray's test were independent diagnostic factors with a predicted correct percentage of 80.0% (p <0.05) for exhibiting a meniscal tear confirmed by arthroscopy.For the time being, history taking and physical examination, albeit of limited use, should be considered with individual patient demands to provide the basis for further evaluation; synthesis of a group of examination manoeuvres and historical items may be required for adequate diagnosis [34].Abdon et al. [35] reported that clinical accuracy in diagnosing meniscal tears was 61%, but a combination of patient reported symptoms increased the predictive value of identifying a meniscal lesion to 70-80%.Researchers in that study had been able to determine variables of significant importance by discriminate analysis.However, patient history and physical examination alone did not provide clinicians with levels of certainty in the 90th percentile that a patient had a meniscal tear.As clinical diagnosis was difficult, all possible steps should be taken to avoid errors [35].
The MRI accuracy for meniscal tears achieved in our study corresponds to that found by others [28,[36][37][38].In this study, if an MRI was included in the model, the results of logistic regression analysis were that locking, McMurray's test and MRI were statistically higher risk factors in knee arthroscopy individuals with meniscal tears and that the presence of these increased the predicted correct percentage for the diagnosis of meniscal tears found during arthroscopy to 91.6% (p <0.05).
Though MRI was more sensitive for meniscal tears, Ben-Galim et al. [26] demonstrated that MRI had a false positive rate of 65% for identifying medial meniscal tears and 43% for lateral meniscus tears compared with surgical findings.MRI, in addition to clinical evaluation, is recommended because of its high accuracy and negative predictive value [38]. In

Limitations
Our study had several limitations: one major limitation was the retrospective design of the study.It also included a wide range of patient ages.This study only included patients admitted for arthroscopy and therefore the selected patient population is likely to have a high rate of meniscal tears.A total of 5 cases waited 6 weeks between the MRI and the arthroscopy, and this time could have permitted meniscus lesions to heal, thus producing false positive MRI.
Funding / potential competing interests: No financial support and no other potential conflict of interest relevant to this article were reported.

Figures (large format)
Diagram of patients enrolled in the study.

Figure 1
Figure 1Diagram of patients enrolled in the study.

Table 1 :
Demographic data of 189 knees with meniscus tears and operative diagnoses of 262 knees.

Table 2 :
Diagnostic values of MRI for meniscal tears.

Table 3 :
Results of univariate analysis by chi square analysis.

Table 4 :
Results of multivariate logistic regression analysis of the significant clinical variables.

Table 5 :
Results of multivariate logistic regression analysis of the significant clinical variables and MRI.
Swiss Medical Weekly • PDF of the online version • www.smw.ch

Table 6 :
Reliability of giving way, locking, McMurray's test, and combining MRI, for diagnosis of meniscal tears (179 knees).