Study objectives

• To determine the diagnostic accuracy of the Thessaly Test for meniscal tears in the knee and whether or not this test can obviate the need for further investigation by arthroscopy or MRI.
• To determine how the Thessaly Test compares with clinical history and to other commonly used physical examinations (McMurray’s Test, Apley’s Test and the joint line tenderness Test) in diagnosing meniscal tears.
• To determine if the presence of arthritis or other knee pathologies influences the accuracy of the Thessaly Test.
• To determine if the use of combinations of physical tests (such as the Thessaly Test, McMurray’s Test, Apley’s Test and the joint line tenderness Test) provides better specificity and sensitivity than a single test alone in the diagnosis of meniscal tears.
• To determine the ability of non-specialist GPs to use the Thessaly Test in comparison with specialist knee clinicians.

Patient selection

Two cohorts of patients were recruited to the study between October 2012 and March 2014 (Figure 7). The first group had unspecified knee pathology and were typical of the target patient group that a primary care clinician may wish to assess using the Thessaly Test. These patients were primarily drawn from patients referred to a specialist knee clinic at Glasgow Royal Infirmary by GPs and other primary care clinicians. A proportion of this group were recruited directly from a single GP practice (n = 11). This group were used as an internal control to verify that subjects recruited at Glasgow Royal Infirmary were representative of the target population of patients who attend general practice with knee pain. The second control group consisted of patients with no current or recent knee pathology. These subjects were included in order to test the specificity of the Thessaly Test. Control subjects were recruited from two sources – from the hand injury clinic at Glasgow Royal Infirmary and from advertisement posters placed within Glasgow Royal Infirmary and local universities.

FIGURE 7

Study flow diagram. Note: patients who did not attend for a MRI scan were classed as lost to follow-up.

Only one knee from any one individual was included in the study.

Written informed consent was obtained from each patient.

As all patients attending the knee clinic were essentially eligible for the study, there were no screening failures from the knee clinic. Similarly, patients in the control group responding to our advert were effectively self-screened and therefore there were no screen failures from this source either.

Sample size

Sample size calculation was based on the following assumptions. Assuming the sensitivity of the Thessaly Test is around 75%, the study would need around 300 subjects to estimate the sensitivity to within ± 5%. A similar calculation for the width of the CI for a binomial proportion is appropriate for the specificity (e.g. if the specificity was around 90%, the required sample size to estimate the specificity to within ± 8% would be n = 50 participants). The power for the pairwise comparison of tests, or combinations of tests, will depend on the degree of disagreement between the tests [e.g. with around 220 pairs of measurements the study would have 90% power to detect a difference in proportions of 0.10 when the proportion of discordant pairs is expected to be 0.15 (using McNemar’s Test)].

Patient assessment

Patients were examined by two types of independent practitioners: a trained orthopaedic musculoskeletal specialist (n = 11) and a primary care clinician (n = 12). The orthopaedic musculoskeletal specialist was either an orthopaedic consultant knee surgeon (n = 3) or an orthopaedic extended scope physiotherapist with a specialist interest in the knee (n = 8). Primary care clinicians were either a GP (n = 4) or a community physiotherapist (n = 8). There was a broad range of experience within the primary care clinician group ranging from newly qualified practitioners to a practitioner with more than 30 years of experience. We believe that the diverse mix of experience reflects reality in the primary care setting.

Each clinician performed the following physical tests: the Thessaly Test, McMurray’s Test, Apley’s Test and the joint line tenderness Test. In addition, each clinician took a clinical history from the patient. The order in which physical tests were carried out was randomised to avoid bias in knowledge gained from performing any previous test. This was achieved using a random sequence generator to determine the order that each test should be performed in. The sequences were generated in advance and printed on study data collection forms. The forms were handed out in the order they were printed by the study co-ordinator (KB) to sequential examiners.

All patients in both groups were asked to attend for a MRI scan of their knee, any patient not attending for MRI was excluded from the final analysis. MRI diagnosis of meniscal tears was used as the gold standard reference for each of the physical tests examined in this study. All MRI scans were reported by radiologists who were not linked to the study and who were blind to the patients’ status with regard to clinical history and examination findings. MRI scans for the knee pathology patients were carried out within 6 weeks of assessment.

Magnetic resonance imaging scan sequences are detailed below:

• T1 spin-echo Sagittal
• T2 fl2d Sagittal (flash-gradient ECHO)
• proton-density turbo-spin echo fat saturation transverse.
• proton-density turbo-spin echo fat saturation coronal.

In order to determine if the presence of osteoarthritis of the knee influenced the outcome of physical tests for meniscal tears, knee radiography was performed on all patients with knee pathology. No radiography was performed on control patients as it was deemed unethical to expose control patients to unnecessary X-rays. Degree of osteoarthritis of the knee was graded from knee radiography using the Kellgren and Lawrence system.³² Kellgren and Lawrence grading was undertaken by a single-blinded examiner (AP).

A proportion of patients with a knee pathology subsequently had a routine arthroscopy as part of their treatment/diagnosis of their knee condition. In these patients we recorded the presence of any meniscal tears at the time of arthroscopy. Arthroscopy was normally carried out within 6 weeks of the MRI scan. These data were then correlated with the diagnosis of meniscal tears obtained from MRI images of the patient, in order to verify the accuracy of MRI imaging for meniscal tears by computing specificity, sensitivity, negative and positive predictive values (NPVs and PPVs) for MRI, with arthroscopy as the gold standard.

Physical tests

The Thessaly Test was carried out as described by Karachalios et al.²⁵ The Thessaly Test, McMurray’s Test, Apley’s Test and the joint line tenderness Test were all carried out as described in Chapter 1.

Standardised clinical history

A simple standardised clinical history was developed for this study (see Standardised minimal clinical history data set). Our concept was that any test that proved successful in this study required to be simple enough for national roll-out without requiring an extensive training programme. All clinicians were therefore given basic instruction in the use of the standardised clinical history in Standardised minimal clinical history data set and asked to use this as a minimum in their diagnosis.

Primary outcome measures

• Sensitivity and specificity of the Thessaly Test in determining the presence of meniscal tears when employed by primary care clinicians.

Secondary outcomes measures

• Sensitivity and specificity of the Thessaly Test in determining the presence of meniscal tears when employed by specialist musculoskeletal clinicians.
• Sensitivity and specificity of McMurray’s Test, Apley’s Test, the joint line tenderness Test and clinical history in determining the presence of meniscal tears.
• Determination of the influence of osteoarthritis on the sensitivity and specificity of the Thessaly Test, McMurray’s Test, Apley’s Test and the joint line tenderness Test.
• Determination of the influence of other knee pathologies such as ACL damage or patellofemoral instability on the sensitivity and specificity of the Thessaly Test, McMurray’s Test, Apley’s Test and the joint line tenderness Test.
• Determination of the optimal combination of tests for most accurate diagnosis of meniscal tears.

Statistical analysis

The full statistical analysis plan is described in detail in Appendix 1. In brief, we calculated the specificity, sensitivity, PPV and NPV for each of the following individual tests: Thessaly Test, McMurray’s Test, Apley’s Test, the joint line tenderness Test and clinical history. We have used evidence of meniscal tear on MRI as our ‘gold standard’ reference for diagnosis. We have compared the relative rates between each individual test. These results are reported separately for specialist musculoskeletal clinicians and for primary care clinicians.

The following subanalysis was also performed:

• Patients with and without osteoarthritis of the knee were compared to determine if osteoarthritis influences the outcome of the Thessaly Test or any of the other physical tests. Other patient factors were also considered [body mass index (BMI), sex, age and previous surgery].
• A comparison of results of all four physical tests between specialist musculoskeletal clinicians and primary care clinicians was performed to determine if specialised training influences the outcome of test results.
• We have examined whether or not a combination of physical tests can produce greater diagnostic accuracy than a single test alone.
• In a subset of patients, we have compared the accuracy of MRI scan diagnosis of meniscal tears to that achieved using arthroscopy.

Statistical tests applied to data

A chi-squared test was used for comparisons between musculoskeletal clinicians and primary care clinicians for each physical test and clinical history, as well as to compare the sensitivity or specificity of the four physical tests along the different bins of the categorical (or categorised) covariates.

McNemar’s Test was used to for comparisons between musculoskeletal clinicians and primary care clinicians for each physical test and clinical history. McNemar’s Test assesses the significance of the difference between two correlated proportions, where the two proportions are based on the same sample of subjects. In this study we wished to determine the difference between the proportion of successful diagnoses obtained by the two groups of observers; hence, the choice of McNemar’s Test over intraclass correlation.

Logistic regression was used to assess the influence of different covariates on the sensitivity or specificity of the four physical tests for musculoskeletal clinicians and primary care clinicians (all patients or knee patients).

Logistic regression using a stepwise procedure from the covariates and the four physical tests to build a model to predict MRI.

Guide to likelihood ratio graphs

The diagnostic abilities of two or more diagnostic tests are traditionally compared by their respective sensitivities and specificities. Comparison is relatively simple if both tests have similar sensitivity, but one has a superior specificity. However, comparison is more complicated when the specificities and sensitivities are both different and a trade-off of one over the other is required.

Likelihood ratio graphs are a simple graphic that readily facilitates comparison between two or more diagnostic tests, allowing an assessment of whether or not a decrease in specificity may be offset by a sufficient gain in sensitivity to yield a test that, nevertheless, has superior diagnostic ability.

For all likelihood ratio graphs in this report, the Thessaly Test is plotted as the reference test (solid black lines). Any comparator test plotted in zone I is superior in all aspects to the Thessaly Test. Comparators plotted in zone II are superior at detecting the absence of a meniscal tear, but inferior at detecting the presence of a tear. Comparators plotted in zone III are superior at detecting the presence of a tear, but inferior at detecting the absence of a tear. Any comparator plotted in zone IV is inferior to the Thessaly Test (Figure 8).

FIGURE 8

Likelihood ratio graph.