Lever sign test for anterior cruciate ligament injuries: a diagnostic meta-analysis

A comprehensive review of the literature was performed utilizing the Embase, PubMed, and Cochrane Library databases from their inception until March 2023. The following different combinations of search terms were utilized: "anterior cruciate ligament," "ACL," "lever sign test," "diagnostic," "sensitivity," "specificity," and "accuracy." Only articles published in English and involving human subjects were included. Additional relevant studies have been determined by screening the reference lists of the detected articles.

The inclusion criteria for the selected studies were as follows: (1) The study aimed to evaluate the Lever Sign test efficacy in diagnosing injuries of the ACL; (2) the investigation provided adequate data to construct 2 × 2 contingency tables of true positives (TP), false positives (FP), true negatives (TN), and false negatives (FN); (3) the study used arthroscopic or MRI examination as the reference standard; (4) the study was conducted in humans; (5) the investigation was published in English.

Investigations were excluded if they dropped in the following specifications: (1) the study was a review article, conference abstract, or case report; (2) the investigation did not report primary data; (3) the study did not report the lever sign test accuracy in diagnosis; (4) the study did not use arthroscopic or MRI examination as the reference standard; (5) the study was conducted in animals or cadavers.

The eligibility of the studies selected was screened by both authors (SH, XW) through an independent assessment of the titles and abstracts. The same two authors retrieved and reviewed full-text articles of potentially relevant studies independently. Discussions with the corresponding author helped to overcome conflicts.

The data extraction process was performed by two authors (XW, SH) in a manner that ensured independence, utilizing a standardized form for the purpose. The following information was extracted: publication years, study population features, design of the study, criteria of exclusion and inclusion, diagnostic criteria for ACL injury, lever sign test methodology, reference standard, and diagnostic performance data.

The authors QW and XW conducted an independent evaluation of the studies' quality employing the revised Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool [10]. The tool is comprised of four distinct domains, namely patient selection, index test, reference standard, and flow and timing. The bias risk for each domain is evaluated, and the first three domains are additionally evaluated for concerns related to applicability. Any disagreements in the quality evaluation were fixed by discussing the corresponding author.

To evaluate the efficacy of the lever sign test in ACL injury detection, MRI or arthroscopy was used as a reference standard. Data on the TP, FP, FN, and TN were obtained independently by two reviewers (XW, QW) from the involved manuscripts. If only specificity and sensitivity were obtainable, the calculation of these numbers was conducted utilizing Web-based tools. Articles that did not provide such data were excluded from the study.

The presence of heterogeneity in meta-analyses of diagnostic accuracy is a frequent occurrence; therefore, the utilization of random effects models is standard. The aforementioned models possess the capability to approximate the mean accuracy of the examination and clarify the variations in this outcome. Due to its inability to account for threshold effects, the traditional I² statistics is not adopted for quantifying heterogeneity in sensitivity and specificity. Exploratory analyses were conducted by visually examining forest and SROC plots to assess whether factors were associated with test accuracy.

A bivariate random-effects model was utilized to conduct a meta-analysis aimed at estimating the combined sensitivity and specificity of the lever sign test in the diagnosis of ACL tears. The utilization of the bivariate model enables the incorporation of the correlation existing across sensitivity and specificity, thereby facilitating the computation of summary receiver operating characteristic (SROC) curves. With values ranging from 0.5 (no discrimination) to 1.0 (perfect discrimination), the area under the curve (AUC) was utilized as a marker of overall diagnostic efficacy.

Analyses were conducted utilizing Onlinemeta v1.0: 2022.3.15 (https://​smuonco.​Shinyapps.​io/​Onlinemeta/​) [11] and Meta-DiSc (www.​metadisc.​es.) [12].

Table 1 indicates that the studies conducted had average to high quality. All studies involved patients who were typically diagnosed with anterior cruciate ligament injury through MRI or arthroscopy. Deveci et al. [13] fail to present the original data, making it unable to calculate the TP, FP, FN, and TN. Therefore, this study was not included in the quantitative analysis though demonstrated in the table. Each study was a cohort study, which reduced the possibility of patient selection bias. However, seven of the investigations did not clearly specify whether blinding was implemented in the patient selection process and the intervention process [4, 5, 14‐18]. The examiners who performed the lever sign test varied in countries and regions. Two studies subdivided the lever sign test performed pre-anesthesia and post-anesthesia, as described by Deveci et al. and Chong et al. [13, 14]. Four investigations included patients with acute ACL injuries (less than 1 month) [4, 5, 17, 19] while another two studies recruited individuals experiencing chronic ACL tears (more than 1 month) [13, 20]. Lelli et al. enlisted both acute and chronic ACL injury participants [3]. Six articles did not specify whether the subjects had an acute or chronic ACL injury [14‐16, 18, 21, 22]. The studies incorporated subjects primarily comprising young adults and adolescents, in accordance with the epidemiology of ACL injuries [23].

Ninety-eight articles were initially identified through a literature search, out of which twenty were deemed possibly eligible according to the screening of their titles and abstracts. Following a comprehensive examination of the texts, a total of 13 articles were found to satisfy the established criteria for inclusion. The meta-analysis consisted of 12 studies, with one article being excluded from the meta-analysis owing to insufficient data on TP, FP, TN, and FN. The overall number of participants included in the meta-analysis was 1365, with 811 being male and 554 being female. Figure 2 illustrates the flowchart detailing the procedure of study selection. Table 2 demonstrates the features of the investigations that have been incorporated.

The efficacy of the lever sign examination for diagnosing ACL tears was assessed through the utilization of forest plots illustrating sensitivity, specificity, positive likelihood ratio, negative likelihood ratio, diagnostic odds ratio, and SROC. The pooled sensitivity and specificity of the lever sign test for diagnosing ACL injuries were 0.810 (95% confidence interval [CI] 0.686–0.893) and 0.784 (95% CI 0.583–0.904), respectively. The positive and negative likelihood ratios were 3.148 (95% CI 1.784–5.553) and 0.210 (95% CI 0.084–0.528), respectively. The diagnostic odds ratio was 17.656 (95% CI 4.800–64.951). The AUC of the SROC was 0.912 (95% CI 0.857–0.967). The summary of the meta-analysis results is presented in Figs. 3, 4, 5, 6, 7 and 8.