Physical tests for diagnosing anterior cruciate ligament rupture

This is a protocol for a Cochrane Review (Diagnostic test accuracy). The objectives are as follows:

To determine the diagnostic accuracy of physical tests, applied singly or in combination, for detecting anterior cruciate ligament (ACL) tears in people whose symptoms and/or history suggest ACL rupture.

Background

Target condition being diagnosed

The anterior cruciate ligament (ACL) is an internal ligament that lies diagonally within the middle of the knee joint. It originates at the medial side of the lateral femoral condyle and inserts at the medial tibial eminence (Petersen 2007) and consists of two bundles: the anteromedial and the posterolateral bundle (Fu 2015; Petersen 2007). The ACL acts primarily to stabilise anterior translation of the tibia in relation to the femur; additionally, it restricts tibial rotation (LaPrade 2015).

ACL injury is one of the most common knee injuries and usually comprises a complete rupture or tear of the ligament (Zeng 2013). Rupture of the ACL is a severe knee injury (Frobell 2010), and is often treated surgically with ACL reconstruction. In a population‐based study in New Zealand, the incidence rate of ACL surgery was 36.9 per 100,000 person‐years, with 80% of knee ligament surgery involving the ACL (Gianotti 2009). ACL injury is often sustained during sporting activities (Alentorn‐Geli 2009; Gianotti 2009). It frequently affects young and active people (Kiapour 2014), especially if participating in high‐risk sports like alpine skiing, soccer, handball or basketball (Prodromos 2007). ACL rupture might be combined with medial or lateral collateral ligament rupture, meniscal and cartilage injuries or less frequently posterior cruciate ligament rupture (Spindler 2008; Teske 2010). Such concomitant injuries tend to hamper the correct diagnosis of ACL injury, especially during physical examination.

ACL rupture very frequently results in instability (Frobell 2010) and subsequently leads to meniscal and cartilage lesions and/or functional impairments (Frobell 2007; Lohmander 2004; Michalitsis 2013; Sri‐Ram 2013).

Index test(s)

Physical examination tests for the assessment of ACL deficiency are frequently applied in routine care in the assessment of the knee and are part of the diagnostic process and a primary tool for clinicians assessing ACL injuries (Jensen 1990; Van Eck 2013a; Van Eck 2013b). The anterior Lachman test, anterior drawer test and the pivot shift test, which are summarised in Table 1 , are the most commonly known physical tests used to assess the integrity of the ACL (Benjaminse 2006; Leblanc 2015; Malanga 2003; Scholten 2003; Solomon 2001). Other lesser known tests used to assess ACL rupture include the prone Lachman test, loss of extension test, fibular head sign, and lever sign. However, this list is not exhaustive and Swain 2014 refers to "over 25 specific physical tests and numerous features from the clinical history that have been proposed for detection of ACL injury".

Table 1

Most common index tests for ACL rupture

The patient lies supine, the knee being tested flexed to 90 degrees (so that the foot is flat and the hip flexed 45 degrees).

The examiner sits on the patient's foot and grasps the proximal tibia from behind while palpating the tibial plateau. With the index fingers, the examiner palpates the tendons of the hamstring muscle group (both medial and lateral) to ensure relaxation.

The examiner stabilises the distal femur with one hand, while the other hand grasps behind the proximal tibia.

The examiner wraps one arm around the patient's leg to fix it firmly. The examiner flexes the knee 90 degrees, while applying a medial rotation to the tibia with the palm of the other hand (effectively subluxing the lateral tibial plateau).

The examiner slowly extends the knee, while still applying a medial rotation to the tibia with the palm of the other hand.

* These generic descriptions are taken from a secondary source (Stern 2011) and are provided for illustrative purposes only. As indicated in the text, techniques and interpretation of positive response are likely to differ in the literature (Kuroda 2012; Lange 2015; Mulligan 2015). In the review, we will give full descriptions of the actual tests used in the included studies.

Physical examination tests for the assessment of ACL rupture are often conducted and interpreted differently by different examiners (Kuroda 2012; Lange 2015; Malanga 2003; Mulligan 2015). For instance, in their in‐depth study of the pivot shift test, Lane 2008 tabulate 13 examination techniques published in the literature for the pivot shift starting with the 'lateral pivot shift' test described originally in Galway 1980. Another example comes from a review on the reliability of ACL physical tests, which found four different endpoints had been used for the interpretation of Lachman test results (Lange 2015). Furthermore, factors such as the force applied during the conduct of the physical tests, the examiner's hand size and the patient's thigh circumference can influence the results of physical tests. Even if the physical tests, such as the Lachman test, which was originally described in Torg 1976, are being performed using comparable techniques, the actual knee kinematics during test conduct have been found to vary between examiners (Kuroda 2012). Physical examination in general (and thus also the index tests) is to a certain extent subjective (Bossuyt 2013) and an implicit threshold effect is likely because the interpretation of positive or negative test results partly depends on the interpretation by the examiner (Buntinx 2009). Prior training, experience and potentially response to prior test results will also influence the examiner's decision on test positivity (Willis 2012).

Physical examination tests are usually not conducted and interpreted in isolation. Thus, different index tests might be combined to assist in the diagnosis of ACL injury (Rossi 2011; Swain 2014). However, the best combination of physical examination tests for the assessment of ACL rupture has not been established.

The physical examination of the knee and thus the proper conduct of the index test might not be possible in acute situations because of swelling and pain (Frobell 2007; Malanga 2003). However, physical examination immediately after injury, and thus before significant hemarthrosis and painful spasm occur, is considered to be unbiased in terms of diagnosis (Leblanc 2015). Patients may refuse to be examined or index tests might be inconclusive in painful situations, thus the performance and interpretation of physical tests is likely to differ between the acute phase and following phases. Therefore, the timing of the index test will be critically assessed within this review.

Patients with knee pain, knee dysfunction or both pain and dysfunction may present themselves in all kind of clinical settings to different practising clinicians. Similarily, physical examination tests are conducted in all types of care settings. Hence, these tests might be conducted by a general practitioner in a primary care setting, and specialist healthcare professionals, such as orthopaedic surgeons or physical therapists specialised in orthopaedic and sports physical therapy, in secondary or tertiary settings.

Clinical pathway

People with knee pain and/or dysfunction, especially after trauma, will most likely present themselves to a primary care physician (depending on the healthcare system of each individual country). Conversely, acutely injured patients might present themselves to emergency units (tertiary care). In both settings, history taking and careful physical examination is usually the first step in the diagnostic process, and a physician might be able to diagnose an ACL tear without further diagnostic evaluation (Spindler 2008). Usually, however, people diagnosed with a suspected ACL injury, based on their history and results from this initial physical examination, will be referred to an orthopaedic surgeon or other specialised healthcare professional for additional testing and further evaluation of the ACL via further diagnostic procedures. (This will also apply to testing for concomitant knee injuries of the articular cartilage, meniscii, and other ligaments.) Often, magnetic resonance imaging (MRI) is used to confirm the diagnosis of ACL rupture and to screen for concomitant injuries. MRI is especially helpful for diagnosing ACL injury in acute injury presentations, where physical examination might be inconclusive but the patient's history indicates an ACL tear (Frobell 2007). Direct visualisation of the ACL or associated injuries during surgery (arthroscopy or arthrotomy) is generally not performed at this stage, given the risks associated with surgery (e.g. infection, venous thrombosis) (Martin 2013; van Adrichem 2015). Both MRI and diagnostic surgery typically take place in secondary or tertiary care settings.

In the case of verified diagnosis of ACL rupture, depending on various factors (e.g. preferences and activity level of the patient, instability of the knee, concomitant injuries), surgical reconstruction of the ACL might be performed by specialists in tertiary care settings (Cimino 2010; Meuffels 2012; Spindler 2008). However, structured rehabilitation without surgery is also used for treating ACL rupture.

Alternative test(s)

Several other tests are used to assess the integrity of the ACL. Those tests include various instrumented knee laxity testing procedures (such tests are performed manually but with the assistance of instrumented devices in addition to the hands of the examiner, e.g. KT 1000 Arthrometer), ultrasound, (stress) radiography, open and arthroscopic surgery, and MRI. Most of those tests are usually conducted by specialists in secondary or tertiary care settings after initial physical examination conducted in primary care settings. The generally accepted reference standard for assessing the integrity of the ACL is direct visualisation during surgery (arthroscopy or arthrotomy). Furthermore, MRI is considered a valid diagnostic tool for assessing the integrity of the ACL and concomitant knee injury (AAOS 2014; Meuffels 2012). Based on evidence demonstrating that MRI has a high sensitivity and specificity in diagnosing ACL tears, the American Academy of Orthopaedic Surgeons guideline for management of ACL injuries gives a strong recommendation that MRI can provide confirmation of ACL injury (AAOS 2014). Finally, we note that arthroscopy, arthrotomy, MRI or a combination of these have been used as reference standards to confirm ACL tears by previous reviews assessing the diagnostic accuracy of physical tests for detecting ACL rupture (Benjaminse 2006; Leblanc 2015; Scholten 2003; Solomon 2001; Swain 2014; Van Eck 2013b).

Rationale

Several reviews have already assessed the validity of physical examination tests for ACL rupture (Benjaminse 2006; Leblanc 2015; Malanga 2003; Scholten 2003; Solomon 2001; Swain 2014; Van Eck 2013b) but are either outdated, focus on selected physical tests only (e.g. Lachman test, anterior drawer test, pivot shift test), lack appropriate assessment of methodological quality (e.g. did not use QUADAS or QUADAS 2) or apply inappropriate methods for statistical analysis. Notably, the diagnostic accuracy of physical tests for diagnosing ACL rupture varied considerably across studies included in these reviews. Identifying the most accurate physical examination test for assessing the ACL will aid clinicians in using the most appropriate tests according to the results of this review. The use of valid physical examination tests to assess the integrity of the ACL might help in early diagnosis to inform treatment, including whether referral to a specialist for surgery is needed, and could help to reduce the high risk of subsequent meniscal and cartilage lesions and functional impairment from undiagnosed ACL rupture. Using appropriate physical tests right from the first contact with the patient for diagnosing ACL ruptures is highly desirable (Benjaminse 2006) as it can reduce the common delay from injury to adequate diagnosis (Arastu 2015; Bollen 1996; Hartnett 2001). Based on the results of physical examination, patients with false‐negative index tests results might have delays in receiving adequate therapy (surgical or non‐surgical) or even receive no therapy at all until being correctly diagnosed. Patients with false‐positive index test results may undergo unnecessary further diagnostic procedures (e.g. MRI or even arthroscopy) or therapy (e.g. a structured rehabilitation programme, physical therapy), while a correct diagnosis and more appropriate therapy are being hindered at the same time. Both, false‐negative and false‐positive index test results are thus unlikely to improve the patient's health status, but are likely to result in increased costs for the healthcare system and patient. In addition, unnecessary diagnostic procedures (e.g. arthroscopy) are associated with additional risks for patients.

Objectives

To determine the diagnostic accuracy of physical tests, applied singly or in combination, for detecting anterior cruciate ligament (ACL) tears in people whose symptoms and/or history suggest ACL rupture.

Secondary objectives

To assess the diagnostic accuracy of physical tests, applied singly or in combination, for detecting ACL tears in different study populations, in different clinical care settings, and different types of ACL rupture (partial, complete; immediate, acute, chronic; isolated ACL injury, ACL injury together with other knee injuries).

Methods

Criteria for considering studies for this review

Types of studies

Retrospective or prospective cross‐sectional or cohort design studies and randomised comparisons of tests are considered eligible if they compare the results of any physical test performed in the context of a physical examination of the ACL with those of an adequate reference standard. Case‐control studies will be excluded because these have been shown to be at risk of overestimating diagnostic accuracy (Lijmer 1999; Whiting 2004; Whiting 2013). We will only include studies in which the data of 2 x 2 table(s) are available, can be reconstructed from reported summary estimates or can be provided by the authors of primary diagnostic studies.

Studies presenting insufficient data for construction of 2 x 2 tables will be excluded from this review, as will studies reported only in abstract form.

Participants

We will include people whose symptoms and/or history suggest ACL rupture. Studies on patients of any age and in any clinical setting with knee pain, dysfunction or both will be considered eligible.

We will exclude studies evaluating physical (index) tests under anaesthesia, intraoperatively or postoperatively, as well as studies on animals and cadavers.

Index tests

Physical tests used singly or in combination to assess ACL rupture are considered index tests. Studies evaluating index tests will be excluded from this systematic review if they did not name or describe a physical test or did not reference a source that did. However, if the name of the physical test is provided by the authors but not described or referenced in their study, we will include the study in the review. Studies will be excluded if they report the overall accuracy of a group of tests, but individual index tests are not specified/named and therefore individual index tests cannot be clearly identified, or if the authors made use of generic terms, such as physical examination, to denote an unspecified combination of physical tests.

Target conditions

Complete as well as partial rupture of the ACL is defined as the target condition. Studies evaluating acute as well as chronic ACL ruptures will be included, and in addition, studies evaluating ACL ruptures and concomitant knee injuries. However, these aspects will be critically assessed within this review, since it might be that physical test performances will vary according to: timing of the injury (e.g. swelling and pain might affect the conduct of physical tests); whether there is a partial or complete tear (e.g. laxity and pain are likely to differ between these two conditions); or whether there are other injuries (e.g. meniscal lesions).

Reference standards

Arthroscopy, arthrotomy, MRI or a combination of these will be used as reference standards to confirm the target condition.

Search methods for identification of studies

Electronic searches

We will search the Cochrane Register of Diagnostic Test Accuracy, MEDLINE, EMBASE, CINAHL, AMED, the Database of Abstracts of Reviews of Effects (DARE), MEDION (Meta‐analyses van Diagnostisch Onderzoek) and the Aggressive Research Intelligence Facility (ARIF) reviews database. All databases will be searched from inception to present.

We will use a sensitive search strategy as recommended in Chapter 7 of the Cochrane Handbook for Systematic Reviews of Diagnostic Test Accuracy (De Vet 2008) for each database. The MEDLINE strategy is shown in Appendix 1 and will be modified for use in the other databases.

There will be no restrictions based on language or publication status to avoid an overestimation of the modality's diagnostic accuracy.

Searching other resources

We will check the reference lists of all relevant retrieved articles of primary diagnostic studies and systematic reviews. We will use the Science Citation Index (Web of Science) to conduct forward citation tracking of any eligible studies found, in order to identify additional articles relevant for the review.

Data collection and analysis

Selection of studies

Two review authors (CK and TL) will independently screen titles and abstracts of study reports identified by the search strategy. If a decision about the inclusion or exclusion of a study is not possible based on the title and abstract, we will retrieve the full text of the study for assessment. The same two review authors will independently screen the full texts of study reports identified on the basis of the title and abstract screening for eligibility according to the pre‐defined criteria. All disagreements during the screening of titles and abstracts as well as full texts will be solved with the aid of a third review author (JL).

Data extraction and management

Two review authors (CK and TL) will independently extract data using a pre‐defined data collection form. All disagreements will be solved with the aid of a third review author (JL). We will design a review‐specific data collection form, which will be tested on studies of diagnostic accuracy that are not eligible for this review. The following information will be extracted from the included studies:

study characteristics (e.g. author, year of publication, geographical region of study conduct, study design, sample size calculation, a priori study registration, publication of study protocol, examiner/rater characteristics, index test(s), reference standard(s), inclusion/exclusion criteria, clinical setting, adverse events, prevalence of ACL rupture, concomitant injuries, interval between index test(s) and reference standard);

patient characteristics (e.g. number of included patients, age, gender, duration of symptoms, sport participation/injury);

values of diagnostic 2 x 2 table(s) (number of true positive (a), false positive (b), false negative (c), and true negative (d)) that cross‐classified the disease status (as determined by the reference test) and the index test’s outcome for each index test evaluated in the included studies) and number of inconclusive results.

In the case of missing data we will contact the relevant authors of primary diagnostic studies.

Assessment of methodological quality

Two review authors (CK and TL) will independently assess methodological study quality using a scheme based on the revised tool for the Quality Assessment of Diagnostic Accuracy Studies (QUADAS‐2) (Whiting 2011). Any disagreements will be resolved with the aid of a third review author (JL).

QUADAS‐2 is the redesigned and improved version of the Quality Assessment of Diagnostic Accuracy Studies list (QUADAS) (Whiting 2003). It comprises four domains: patient selection, index test, reference standard, and flow and timing. Risk of bias is assessed for each domain, and for the first three domains, a statement on concerns regarding applicability is given. Each key domain has a set of signalling questions to help judge the risk of bias and concerns regarding applicability. Signaling questions are answered as 'yes', 'no', or 'unclear'. Risk of bias is rated as 'low risk of bias', 'high risk of bias', or 'unclear risk of bias'. Concerns regarding applicability are rated as 'low', 'high' or 'unclear'.

One signalling question from the patient selection domain ("Was a case–control design avoided?") and one signalling question from the reference standard domain ("Is the reference standard likely to correctly classify the target condition?") of the QUADAS‐2 tool will be omitted because the inclusion/exclusion criteria of this review render them redundant.

Two questions from the original QUADAS tool (“Was the execution of the index test described in sufficient detail to permit replication of the test?” and "Was the execution of the reference standard described in sufficient detail to permit its replication?") are included to assist appraisal of applicability. As well as noting the adequacy of the descriptions, we will critically appraise the execution of all index tests and reference standards. In particular, since execution and interpretation of physical examination tests is crucial and might differ (Kuroda 2012; Lange 2015; Malanga 2003), this may effect estimates of the diagnostic accuracy of index tests (Whiting 2011) and should be carefully assessed.

In order to improve the reliability of the methodological assessment using the QUADAS‐2 tool, review‐specific guidance is available to the authors on how to assess each signalling question and how to use this information to judge the risk of bias, and in addition judge concern regarding applicability (seeAppendix 2).

Statistical analysis and data synthesis

For each index test, we will plot the observed sensitivities and specificities (with 95% confidence intervals) on forest plots for visual examination of variation in test accuracy across studies using RevMan software. Therefore we will use the reported values of diagnostic 2 x 2 table, which will be extracted from included studies. If complete data of 2 x 2 tables are not available, we will attempt to reconstruct these values from reported summary measures. Because of possible discrepancies in 2 x 2 tables due to rounding errors, data will be considered for inclusion only if the discrepancies in the back‐calculated 2 x 2 table do not exceed 10% in any cell, as specified in a previous Cochrane review (Hanchard 2013). If inconclusive or incomplete index test results are reported in the included studies, such results will be excluded from statistical analysis. We will, nonetheless, summarise the inconclusive results where these occur. Studies will be excluded if there are multiple discrepant analyses.

For the purpose of statistical analysis, complete and partial ruptures of the ACL will be summarised to discriminate from participants without rupture of the ACL. For the analysis of combinations of index tests, any one index test can be positive.

Meta‐analysis using a bivariate logistic regression model with random effects to account for potential between‐study correlation and heterogeneity will be performed, if sufficient data are available (Chu 2006; Reitsma 2005). This will estimate weighted sensitivity and specificity (with 95% confidence intervals) for each diagnostic test of interest. Though it is unlikely that there will be sufficient studies for the direct comparison of tests (Macaskill 2010), the comparison of the results of indirect and direct comparisons is intended and will be performed if sufficient data are available.

Investigations of heterogeneity

Heterogeneity will first be explored through visual examination of forest plots of sensitivities and specificities, and second by including possible sources of heterogeneity as covariates into meta‐regression models if sufficient data are available. We will investigate the following potential sources of heterogeneity:

study population (general population versus well‐defined groups (e.g. young athletic populations, military recruits); prevalence of ACL rupture);

stage of clinical care (primary (generally in the community setting), secondary (referral following preliminary screening) or tertiary (referral to a specialist centre));

type of ACL rupture (partial versus complete; immediate versus acute versus chronic; isolated versus combined injuries);

Sensitivity analyses

If sufficient data are available, we will conduct sensitivity analyses to explore the sensitivity of summary estimates to aspects of methodological study quality, drawing primarily on the results of our assessment of risk of bias using the QUADAS‐2 tool. The effect of excluding studies where the index test or reference standard domain was judged as having a high risk of bias or unclear risk of bias will be explored first, because this is considered the most relevant source of bias. Secondly, we will also explore the effect of excluding studies where two or more domains of the QUADAS‐2 tool were judged as having a high or unclear risk of bias.

Assessment of reporting bias

Testing for publication bias using the Egger test (or other available test for funnel plot asymmetry) is not recommended for diagnostic studies because conducting such tests in systematic reviews of diagnostic test accuracy is likely to result in publication bias being incorrectly indicated by the test far too often (Macaskill 2010). As recently recommended, we will use the Deeks' test to assess publication bias (Van Enst 2014).

Acknowledgements

We thank Nigel Hanchard, Helen Handoll, the editors of the Cochrane Diagnostic Test Accuracy Working Group (particularly Rob Scholten), Lindsey Elstub and Laura MacDonald for their constructive feedback in drafting the protocol. We also thank Joanne Elliott and Anne Eisinga for their help with developing the search strategies.

This project was supported by the National Institute for Health Research (NIHR) via Cochrane Infrastructure funding to the Cochrane Bone, Joint and Muscle Trauma Group, and the Cochrane Diagnostic Test Accuracy Working Group. The views and opinions expressed herein are those of the review authors and do not necessarily reflect those of the Systematic Reviews Programme, NIHR or the United Kingdom National Health Service or Department of Health.

Appendices

Appendix 1. MEDLINE (OVID Online) search strategy

Appendix 2. Review‐specific QUADAS‐2 coding manual

Notes

Withdrawn from publication for reasons stated in the review