The Accuracy of the Clinical Examination in the Setting of Posterior Cruciate Ligament Injuries

Transcription

1 The Accuracy of the Clinical Examination in the Setting of Posterior Cruciate Ligament Injuries Richard A. Rubinstein, Jr., MD, K. Donald Shelbourne,* MD, John R. McCarroll, MD, Charles D. VanMeter, MD, and Arthur C. Rettig, MD From the Methodist Sports Medicine Center, Indianapolis, Indiana ABSTRACT Thirty-nine subjects volunteered for this blinded, randomized, and controlled study to assess the clinical examination skills of orthopaedic surgeons with fellowship training in sports medicine. Eighteen of the patients had 19 chronic isolated posterior cruciate ligament tears. The controls were 9 patients with 9 anterior cruciate ligament-deficient knees, 12 subjects with normal knees, and the contralateral normal knees of the ligament-deficient patients. To eliminate preexamination bias, all examiners were blinded from the examinee s history, identity, and diagnosis. The overall clinical examination accuracy for all orthopaedic surgeons was 96%. The accuracy for detecting a posterior cruciate ligament tear was 96%, with a 90% sensitivity and a 99% specificity. The examination accuracy was higher for grade II and III posterior laxity than for grade I laxity. Eighty-one percent of the time, the examiners agreed on the grade of the posterior cruciate ligament tear for any given patient. The posterior drawer test, which included palpation of the tibia-femur step-off, was the most sensitive and specific clinical test. A thorough and precise physical examination, coupled with a patient history, can be considered diagnostic in the majority of isolated posterior cruciate ligament injuries. With this accuracy level known, the natural history of isolated posterior cruciate ligament tears can be reliably documented and studied. The natural history of an isolated PCL tear of the knee is uncertain. Researchers have attempted to determine the * Address correspondence and repnnt requests to K Donald Shelbourne, MD, Methodist Sports Medicine Center, Suite 530, 1815 North Capitol Avenue, Indianapolis, IN No author or related institution has received any financial benefit from this study. long-term course of the PCL-deficient knee 2,7,13,16,18 ; however, there are concerns within each of these reports regarding patient selection bias, inclusion of both isolated and combined injuries in the same study population, inadequate length of followup, deficiencies inherent in a retrospective review, and objective difficulty in confirming the diagnosis. In addition, the seemingly lower incidence of this injury as compared with other knee ligament injuries adds to the difficulty in defining and studying this patient population. The literature contains the descriptions of many clinical tests that are used for detecting posterior and posterolateral instability. 1,4,9-11,17 However, the accuracy of these tests, when used individually or together, has not been substantiated. Daniel et a1.4 described the quadriceps active drawer sign and reported an accuracy of 98% in 42 cases of acute and chronic PCL tears with no false-positives in the normal and ACL-deficient knee controls; however, this study was not blinded or randomized and therefore any conclusions on test accuracy become questionable. When there is uncertainty about the diagnosis of a PCL tear, further evaluation and documentation can be performed. Magnetic resonance imaging (MRI) has been reported to be extremely accurate in the diagnosis of an injury to the PCL. 6,8 The accuracy in diagnosing and grading an isolated PCL knee injury is paramount to the validity of any PCL investigation. The accuracy of the clinical examination for posterior instability is not known and remains a source of controversy among orthopaedic surgeons specializing in knee ligament injuries. Posterior cruciate ligament injuries are frequently missed.2,7,20 Therefore, the true population with isolated PCL laxity is not well defined. To better understand these shortcomings in diagnosis, the confidence level and limitations of the clinical examination need to be established. The purpose of this study was to determine the clinical accuracy of diagnosing and grading isolated PCL tears. It is our hypothesis that the clinical examination is accurate in detecting PCL injuries. To substantiate this, the clinical 550

2 551 examination must be tested in a standardized patient setting using a blinded, randomized, and controlled design. If the reliability of the clinical examination can be firmly established, whether accurate or inaccurate, then future studies will have the necessary groundwork on which to structure their retrospective and prospective evaluations. In addition, this information will help further define the role for adjunctive tests in diagnosing PCL injuries. MATERIALS AND METHODS The study was designed to include a sufficient number of patients and controls to test our hypothesis. A medical statistician was consulted to help determine the appropriate number of patients to study. It was determined that approximately 20 PCL-deficient patient and 20 &dquo;control&dquo; patients were needed. The control group would consist of an approximately equal number of people with &dquo;normal&dquo; knees as well as patients with a proven ACL-deficient knee. Inclusion of the ACL-deficient knees as controls helps eliminate any examiner bias when only a single type of abnormality is tested. The patients contralateral knee, if it was normal, would also act as a control. To avoid uncertainty and controversy in diagnosis, the patients selected for the study currently had chronic unidirectional, isolated ligament laxity. All abnormal knees had to have documented proof of their injuries. This documentation was provided by MRI for the PCL tears and by arthroscopy for the ACL tears. In addition, the patients had previous clinical examinations consistent with the diagnosis. Patients who had surgical reconstruction of their cruciate ligament tears, whether acutely or chronically, were excluded from the study. One hundred thirty-one patients with the clinical diagnosis of PCL injury have been evaluated at the Methodist Sports Medicine Center from 1983 to The majority of these patients had nonoperative treatment of their PCL tears and most of these patients had their diagnosis determined on clinical grounds only. In recent years, MRI has been proven to be a highly accurate, noninvasive method for confirming the diagnosis of PCL tear. The MRI was used as an adjunctive test in 31 of the 131 patients, and the test confirmed their PCL injuries. Eighteen of these 31 patients volunteered to participate in the study. These 18 patients represented 19 PCL injuries, all of which currently had isolated chronic posterior laxity. The average current age of the 14 male and 4 female patients was 26 years (range, 12 to 47). The average time from injury was 3.5 years (range, 5 months to 21 years), with the mechanism of injury being low velocity in 17 (15 occurring in sports and 2 from falls) and high velocity in 2 (1 motorcycle accident and 1 four-wheeler accident). Sixteen of 18 patients had normal contralateral knees, with the exceptions being 1 patient with bilateral PCL tears and 1 patient who had undergone a contralateral ACL reconstruction. This latter patient s contralateral knee was excluded from the data analysis. Four of the patients had a prior arthroscopy (average of 2.3 years previously), 2 for partial meniscectomy and 2 for diagnosis. Nine patients had clinically and arthroscopically proven ACL-deficient knees. Their average current age was 32 years (range, 22 to 41.5), and there were five men and four women. The average time from injury was 12.5 years (range, 4 to 24.5) with the mechanism of injury being low velocity in all (eight related to sports and one secondary to a fall). Each patient had undergone an arthroscopic procedure for meniscal or chondral symptoms at an average time of 1.7 years before. There were six subsequent partial meniscectomies (one combined with a repair) and three debridements. These patients had a normal contralateral knee with the exception of one patient who had a clinically suspected ACL injury but no absolute documentation. Therefore, this latter patient s contralateral knee was excluded from the data analysis. Twelve persons volunteered to participate in the study as normal controls. Their average age was 24.5 years (range, 22 to 28); there were nine women and three men. Eight of these volunteers had no history of significant injury to their knees. One subject noted a hyperextension injury 6 years earlier with no residual problems. This latter subject participated in the study, but her previously injured knee was excluded from the data analysis. Therefore, a total of 39 persons were scheduled for the study: 18 patients with 19 PCL-deficient knees, 9 patients with 9 ACL-deficient knees, and 12 subjects with normal knees. Overall, there were 75 knees for data analysis including 47 uninjured knees (3 knees excluded), although all knees were examined. The average age of all patients was 27 years (range, 12 to 47); there were 22 men and 17 women. All patients were randomly assigned and intermixed into one of two groups assigned to different examination days. All study patients were informed about the research project in writing as well as by telephone conversation, and a consent form was signed by each volunteer. Each person understood that their participation was voluntary, and that they could withdraw at any time. They were informed that since this was an objective clinical examination test, they would not be allowed to see or talk to the orthopaedic surgeons during the examination. Five orthopaedic surgeons, all with additional fellowship training in sports medicine, were selected as examiners for this study. Years of orthopaedic experience, which included their fellowship training in the knee, were as follows: two examiners had 5 years each, one had 10 years, one had 15 years, and one had 20 years. All five orthopaedic physicians would examine all 39 patients once in a random and blinded fashion and on the same day for any one particular patient. The orthopaedic surgeons were not told the number of normal and abnormal knee patients, what types of ligament injuries comprised the abnormal knee group, any patient history, or which side had been injured. The examinations were set up so that the physicians had an equal number of times as first examiner, second examiner, and so on. One physical therapist, certified for using the KT-1000 knee arthrometer, tested each subject in a blinded fashion for anterior and posterior laxity at their quadriceps neutral knee angle, approximately 70 (Fig. 1). The KT-1000 arthrometer was used as an objective measurement of laxity

3 552 Figure 1. The KT-1000 arthrometer is used to objectively measure anterior and posterior laxity at the quadriceps neutral angle (approximately 70 ). at this neutral angle, and the results were recorded in millimeters by the room assistant (Fig. 2). The therapist s assessment of the type of knee instability was recorded as either PCL for abnormal posterior laxity or &dquo;other&dquo; for normal laxity or abnormal anterior laxity. The more standard arthrometer measurement for anterior laxity, performed at 20, was not done. All examiners were instructed as to which clinical tests would be performed (Fig. 3). The standard method for doing each part of the examination was reviewed with each orthopaedic surgeon. The PCL tears were graded by the position and excursion of the tibia in relationship to the femoral condyles with the knee flexed to 90 (posterior drawer test combined with palpation of the tibia-femur step-off) Figure 3. A knee examination worksheet was independently completed for each patient by each examining physician. and the leg and foot in a neutral rotation (Fig. 4). A grade I PCL injury was defined as increased posterior tibial displacement but with the tibia not being flush with the femoral condyles with the knee flexed 90. A grade II PCL injury was posterior tibial displacement in which the anterior tibia was flush with the femoral condyles. A grade III PCL Figure used to record objective knee laxity. 2. The KT-1000 arthrometer data worksheet was Figure 4. An example of the posterior drawer (with palpitation of the tibia-femur step-off) being performed during the patient examination.

4 553 tear denoted posterior displacement such that the anterior tibia was subluxated posterior to the anterior surface of the femoral condyles. Anterior cruciate ligament tears were considered present or absent; the grade was an assessment of excursion on the Lachman examination and not based on ACL integrity. A 0 to 3 mm side-to-side difference was considered normal (grade 0), while knee laxity between 3 and 5 mm, 6 and 10 mm, and 11 and 15 mm was defined as grade I, II, or III, respectively. The orthopaedic surgeons were not allowed to talk to the patients before or during the knee examination. A nonphysician assistant was in each room to make sure that no rules were violated as well as to record and secure the clinical examination data. To assure that no doctors recognized any patient before or during the examination, all patients were brought to the examination rooms privately while the physicians remained in another area. In addition, an opaque sheet was draped in each room at the center of the examination table so that the patient s head was on one side and their legs were on the other side (Fig. 5). Some subjects had knee scars which, if recognized, were often more confusing than helpful. In most cases, the scars were not visible because of their age, small size, or surrounding hair. The examiners also did not know which patients, if any, were required to have a previous arthroscopy or surgery for inclusion in the study. Each orthopaedic surgeon entered the room without introduction to the study patient, examined both knees according to the order of the knee examination worksheet, and, when finished, was allowed to speak to the examinee. The orthopaedic surgeons were only allowed 8 minutes per examination and then were required to go to the next room. There was no communication between physicians regarding the examination findings. The results of the examination were not disclosed to the physicians until the study was completed. The known, documented diagnosis for the 75 knees in the 39 subjects was used for determining clinical accuracy. Three knees were excluded from the data analysis: 1 be- cause of a previous ACL reconstruction and 2 for undocumented diagnosis. One examiner disqualified himself from one of the examinations because he inadvertently saw and recognized a patient s family member. The KT-1000 arthrometer test was not performed on 1 of the examinees because of an oversight during the room changes. One patient became too sore and refused further examination after 3 had been performed; therefore, 2 orthopaedic surgeons and the physical therapist were unable to examine this patient. The data from the first 3 examinations were used in the analysis. The data were analyzed with the assistance of a medical statistician. The previously documented prestudy diagnoses (as confirmed by MRI for the PCL injuries and arthroscopy for the ACL tears) were used as the baseline reference. The sensitivity, specificity, and accuracy of the clinical tests and diagnoses were determined. Sensitivity was defined as the percentage of time the diagnosis or examination test was correct for a given knee ligament injury (PCL or ACL), i.e., the number of positive results over the predicted number of positive results. Specificity was defined as the percentage of time the diagnosis or test was correct in those knees that were predicted not to have the ligament injury in question, i.e., the number of negative results over the predicted number of negative results. Accuracy was defined as the percentage of knees that was correctly identified as either positive or negative for their previously determined diagnosis. The data of the examiners were analyzed individually and collectively. The examiner s orthopaedic experience, computed in years, was also compared with the outcome. The clinical grade of the diagnosis was evaluated by comparing the results of all examiners and determining the number of times that similar grade of laxity was recorded for each patient. The results of the arthrometer readings for each knee were compared with the diagnosis and clinical grade. The clinical measurement of range of motion was one method used to objectively assess interexaminer variability. RESULTS The overall clinical examination accuracy for all examinees was 96% (Table 1). There was no significant difference in outcome when the results of the individual examiners were compared with each other or with the group as a whole. The examination accuracy in the setting of a normal knee was 96%. The accuracy for detecting PCL tears was 96%. When analyzed based on the grade of PCL tear, with grade I being considered low grade and grade II and III being considered TABLE 1 Evaluation of the clinical examination a Figure 5. The appearance of the examination room setup. The patient is supine with an opaque screen acting as a barrier between the patient and physician. Values are percentages; ranges are in parentheses.)

5 554 TABLE 2 KT-1000 arthrometer results for PCL laxity high, the examination sensitivity for the lower grade was only 70%, with a specificity of 99%. The examination sensitivity for the higher grade PCL tears was 97%, with a specificity of 100%. Eighty-one percent of the time, the examiners agreed on the grade of the PCL tear for any given patient (range, 60% to 100%). The overall accuracy for detecting ACL tears was 99%. Seventy percent of the time the examiners agreed on the grade (Lachman excursion) of the ACL laxity for any given patient (range, 50% to 100%). Data analysis based on the ACL grade alone was not possible because of the small number of knees (nine total) with this injury. The overall accuracy of the KT-1000 arthrometer in detecting PCL tears versus &dquo;other&dquo; was 89% (Table 2). The sensitivity of the KT-1000 arthrometer for low-grade PCL tears was 33% (1 of 3) with a specificity of 94% (64 of 68), while the sensitivity for the high-grade tears was 86% (12 of 14) with a specificity of 100%. The millimeters of excursion or side-to-side difference was not determined for anterior laxity. The accuracy of each individual clinical test is reported in Tables 3 and 4. An effusion was reported in 11% of the ligament-deficient knees and 1% of the normal knees. The presence of an effusion occasionally helped in identifying the involved knee, but its absence had little predictive value. No knees had varus or valgus laxity that was greater than 1 +. These findings further confirm that the examined knees had isolated laxities. The examiner variation in estimating total arc of motion averaged 10% (range, 1% to 25%) for any given knee. DISCUSSION TABLE 3 Detection of an isolated chronic PCL injury TABLE 4 Detection of an isolated chronic ACL tear The combined patient history and clinical examination are the most important tools of the physician. Understanding the limitations of the clinical diagnosis is also critical. The recent exponential growth in medical technology has provided the physician with some useful adjunctive tools to aid in both diagnosis and treatment. As expected, it becomes equally important that the clinician understand the judicious usefulness and limitations of these new technologies. Together, they can be applied as an effective means of caring for patients. This study was undertaken to assess the accuracy of the clinical examination in the setting of PCL injuries. There are strengths and weaknesses inherent in this design. It was a blinded, randomized, and controlled study with a simple hypothesis: the accuracy of the clinical examination was being assessed. It is understood that the combination of the detailed history and physical examination will increase the accuracy of the physical examination alone. Without question, valuable diagnostic information is provided from the history of the knee injury. Nevertheless, the logistics of this study were such that each patient already knew and understood his or her diagnosis and this information would invariably be disclosed to and bias the examiner. The restriction of communication and the use of an examination screen kept the physician and patient blinded from one another. Determining the accuracy of clinical examination in acute knee injuries was also not possible with this study format, and therefore the results can be applied to chronic ligament injuries only. Investigating acute ligament tears may be equally beneficial and worthy of future research. Overall, we found that the clinical examination for unidirectional, chronic knee laxity is highly accurate. The lack of a patient history makes this accuracy rate even more remarkable. The examiners were most accurate in diagnosing grade II and III PCL laxities. The subtle grade I laxities were more difficult to detect. While these grade I injuries may be of uncertain clinical significance, their detection is extremely important for studying and determining the natural history of PCL tears. The addition of the patient s history may help in clinically identifying these grade I PCL laxities. Adjunctive tests such as MRI scanning may also be used to confirm a grade I ligament injury. The clinical diagnosis is usually formulated after collaborating the results of several clinical tests. The posterior drawer and posterior sag signs were the most sensitive and specific clinical tests for chronic PCL injuries. The reverse Lachman, absence of a reverse Lachman end point, quadriceps active drawer test, and dynamic posterior shift test were significantly less sensitive, although when they were present they were highly specific. This finding is in contrast to a report by Daniel et al.4 where they found the quadriceps active drawer sign to have a sensitivity of 97% in acute and chronic PCL tears. In that

6 555 study, the examiners were not blinded and the patients were not randomized. Our test results for chronic anterior laxity confirmed what has been previously reported in the acute setting. The Lachman test and absence of a Lachman end point are the most sensitive and specific tests for an ACL tear.5,19 The pivot shift test was also sensitive and specific for anterior laxity. This high accuracy level with the pivot shift test may reflect the chronic nature of these injuries with improved patient relaxation as compared with the poorer relaxation seen in an acute clinical setting.5 The anterior drawer test was the least specific and sensitive for ACL laxity, although it was still present in three fourths of the cases. It also has been found to be less accurate in acute ACL injuries.12,21 Experience in performing these clinical tests will surely improve one s accuracy level and further emphasizes the importance of correctly learning and maintaining clinical skills. Magnetic resonance imaging may be helpful in confirming one s clinical suspicions.8 14 The MRI scan has been shown to be extremely accurate in diagnosing injury to the PCL, with accuracy reported as high as 99%.6 However, these reports did not address whether there is an accuracy difference between acute and chronic tests. In our experience, MRI is extremely accurate in acute PCL injuries but less accurate in detecting chronic injuries. The drawback with the MRI is its high cost as well as it not being able to quantitate the degree of laxity. Arthroscopic documentation of acute PCL injuries was performed by Fowler and Messieh in their report on the nonoperative treatment of isolated PCL tears. Unfortunately, the accuracy of arthroscopy in diagnosing acute and chronic PCL tears is debated. The accuracy probably depends on both the timing of arthroscopy after injury and the arthroscopic technique. A PCL tear may be difficult to detect by standard arthroscopy because the ACL partially obscures the PCL, and even when visualized, it may appear intact despite having clinical laxity. The design of our study requires one to accept the premise that an MRI-proven PCL injury coupled with a physical examination consistent with a PCL injury is adequate documentation of a true, undeniable PCL tear. To further assess the accuracy of the clinical examination, a reliable and reproducible objective test for grading PCL laxity is needed. At present, this objective test has not been found. The reliability of a knee arthrometer for posterior laxity has not been previously established. In our study, the KT arthrometer was relatively accurate in detecting and grading the grade II and III posterior laxities but was not sensitive enough to detect the grade I laxities. There was also a trend toward the higher grade PCL tears having a greater degree of measured laxity, although this was not statistically significant. Another study, which used an arthrometer, also demonstrated this trend, but it was not statistically significant when comparing either isolated and combined posterior instabilities or the spectrum of laxity within each group.&dquo; We believe that the KT-1000 arthrometer can be used to objectively assess the grade II and III posterior laxities. It may be particularly applicable when evaluating and following those PCL tears that are surgically reconstructed. The concern that the results of this study would reflect a single orthopaedic surgeon s clinical abilities was minimized by testing multiple examiners. This design allowed an assessment on both an individual and group basis. This study did not test the reliability of the examiners. It would have been interesting to compare our results, which reflect the skills of fellowship-trained (sports medicine) orthopaedic surgeons, with those of physicians from varying disciplines (e.g., general orthopaedic surgeons, nonorthopaedic sports medicine specialists). However, the logistics of our study design would not allow for multiple additional examiners. Evaluating one or two persons from each discipline would have assessed individual skills only and may not have been representative of a larger group of similarly trained physicians. Another potential bias is that the examiners would have a higher index of suspicion for ligament injuries knowing that this study was attempting to evaluate their accuracy in knee ligament examination. The fact that there was a disproportionately high number of PCL knee injuries within this patient group as compared with a random sample of patients could also raise the examiner s clinical suspicion. However, the examiners did not know the type or number of knee injuries so they would still need to rely on their clinical examination to determine the diagnosis and grade. To understand the clinical examination of the knee and its variability, one needs to first assess it among a population of subjects with &dquo;pure,&dquo; isolated, and proven diagnoses. In the report by Daniel on the limits of knee motion, the test population was a group of 10 subjects with multiple, mixed ligament injuries, 8 having had prior ACL reconstruction. A baseline reference for qualifying and quantifying knee instabilities had not been established. The examiners in Daniel s review were told which knee was the index knee and instructed to quantitate the knee motion in millimeters. The final clinical test assessment was based on the side-to-side difference and grouped according to those greater than 3 mm and those less than or equal to 3 mm. Our study did not disclose which, if any, knee was the index knee, and we sought to determine both the diagnosis and grade. The grade of injury was determined by quantifying the knee joint translation and position. This grading system relies on the qualitative and quantitative feel of the examiner and uses both total excursion and side-to-side comparison. Our study found the accuracy of diagnosing and grading PCL tears was 96% and 81%, respectively. In a laboratory cadaveric setup, Noyes et a1.15 attempted to assess the orthopaedic surgeon s ability to diagnose and quantitate normal and abnormal knee motions and concluded that there were wide interexaminer variations and errors. They recommended that test conditions be standardized and instrumented teaching models and quantitative tools be developed with the latter used as a requirement for reporting clinical results. Their &dquo;clinical&dquo; simulation used two pairs of cadaveric knees with an instrumented linkage system and appears to be testing a specimen with few similarities to a live human model.

7 556 The validity of using these data to make conclusions on the ability to &dquo;clinically&dquo; determine knee motions in noncadaveric knees is questionable and requires further investigation. The orthopaedic examination should be performed and assessed with the combined benefits of a quantitative science as well as a qualitative art, which we believe most closely simulates the practice of clinical medicine. In our hands, the overall accuracy of diagnosis was extremely high, while the objective clinical quantification of laxity was less accurate. In the previously mentioned study by Daniel, the quantitative assessment of knee motion was shown to be poorly reproducible among examiners. This finding was true for the individual clinical test but was not determined in the context of the overall diagnosis or assessment. In our study, the examiners agreed on the overall diagnosis 96% of the time, but they did not agree on the grade of posterior and anterior laxity in 19% and 30% of the cases, respectively. While each knee motion test may be clinician-specific, it is the overall picture that has an impact on the treatment plan or result. Our data support the importance of using multiple and combined subjective and objective evaluation criteria to formulate a final assessment of the diagnosis or result. We identified communication differences among similarly trained orthopaedic surgeons. The grading system for PCL tears included either an assessment of the relationship between the femoral condyles and tibial plateau or an estimation of the amount of total posterior tibial excursion as compared with the contralateral knee, assuming the latter was normal. While some of the examiners considered the posterior drawer test as less when the tibia was internally rotated, other examiners considered it as more for this same position because they were determining the position of the tibia in relation to the femoral condyles (which was now more posterior) and not the amount of or change in excursion. When grading ACL tears, the results could vary depending on whether one assessed excursion or ACL integrity. Some examiners also record the &dquo;normal&dquo; physiologic laxity as grade I while others consider it as grade 0. This physiologic laxity was particularly true when testing the pivot shift or varus laxity at 30. In addition, normal variations were often determined by a side-to-side comparison. For this study, these communication differences were recognized, and therefore the data were recorded and assessed using a common language. A common language is equally as vital when reporting treatment results. The assessment of surgical results depends on both the examiner and the evaluation criteria. Daniel described how one examiner reported seven of eight patients having normal anterior excursion after ACL reconstruction while another examiner stated that none of the same patients had normal anterior motion. We believe that success and failure or normal and abnormal are the sum total of different assessment techniques and tools. One test alone is unlikely to determine the overall result. In addition, the different rating methods will influence the reported outcome. Therefore, the use of common and standardized evaluation methods should provide rntp- rpnrnducihle means nf clinical asnp<<mpnt a more accu- CONCLUSIONS This study was conducted to determine the appropriate level of confidence in the clinical examination when evaluating posterior knee laxity. The results reflect the evaluation skills of orthopaedic surgeons with fellowship training in sports medicine. The physical examination was highly accurate in diagnosing isolated chronic PCL injuries, with a higher confidence level for grade II and III laxity (97%) than for grade I laxity (70%). The posterior drawer test coupled with palpation of the tibia-femur stepoff was the most sensitive and specific clinical test. The examiners agreed on the PCL grade 81% of the time. Currently, an arthrometer may be useful when evaluating and following the higher grade PCL tears. Standardized methods for reproducibly measuring all degrees of PCL laxity are needed. In the setting of a grade I PCL injury, adjunctive tests such as an MRI scan may be helpful in confirming the diagnosis. Nevertheless, there remains no substitute for a thorough and precise physical examination and this, coupled with a patient history, can be considered diagnostic in the majority of isolated PCL injuries. With the clinical accuracy understood, the patient population with these injuries should become better defined and ultimately better studied and treated. ACKNOWLEDGMENTS The authors thank the following people for their input and assistance with this study: Thomas E. Klootwyk, MD, Mark DeCarlo, MS, PT, ATC, Tinker Gray, and Robyn Hahn, RN. REFERENCES 1 Clancy WG, Shelbourne KD, Zoellner GB, et al. Treatment of knee joint instability secondary to rupture of the posterior cruciate ligament J Bone Joint Surg 65A: , Cross MJ, Powell JF: Long-term followup of posterior cruciate ligament rupture: A study of 116 cases Am J Sports Med , Daniel DM Assessing the limits of knee motion Am J Sports Med , Daniel DM, Stone ML, Barnett P, et al Use of the quadriceps active test to diagnose posterior cruciate-ligament disruption and measure posterior laxity of the knee. J Bone Joint Surg 70A , Donaldson WF, Warren RF, Wickiewicz T A comparison of acute anterior cruciate ligament examinations: Initial versus examination under anesthesia Am J Sports Med , Fischer SP, Fox JM, Del Pizzo W, et al Accuracy of diagnoses from magnetic resonance imaging of the knee J Bone Joint Surg 73A 2-10, Fowler PJ, Messieh SS. Isolated posterior cruciate ligament injuries in athletes Am J Sports Med , Grover JS, Bassett LW, Gross ML, et al Posterior cruciate ligament. MR imaging. Radiology , Hughston JC, Bowden JA, Andrews JR, et al: Acute tears of the posterior cruciate ligament. J Bone Joint Surg 62A , Hughston JC, Degenhardt TC. Reconstruction of the posterior cruciate ligament Clin Orthop , Jakob RB, Hassler H, Staeubli H-U Observations on rotatory instability of the lateral compartment of the knee Experimental studies on the functional anatomy and parthomechanism of the true and reversed pivot shift sign. Acta Orthop Scand 52 (Suppl 191) 1-32, Jonsson T, Althoff B, Peterson L, et al: Clinical diagnosis of ruptures of the anterior cruciate ligament: A comparative study of the Lachman test and the anterior drawer sign. Am J Sports Med , Keller PM, Shelbourne KD, McCarroll JR, et al Nonoperatively treated isolated posterior cruciate ligament injuries. Am J Sports Med, 21: , Mink JH, Deutsch AL Magnetic resonance imaging of the knee Clin Orthop 244 : 29-47, Noves FR. Cummings JF. Grood ES. et al. The diagnosis of knee motion

8 557 limits, subluxations, and ligament injury Am J Sports Med , Parolie JM, Bergfeld JA Long-term results of nonoperative treatment of isolated posterior cruciate ligament injuries in the athlete. Am J Sports Med , Satku K, Chew CN, Seow H: Posterior cruciate ligament injuries. Acta Orthop Scand 55: 26-29, Torg JS, Barton TM, Pavlov H, et al: Natural history of the posterior cruciate ligament-deficient knee Clin Orthop , Torg JS, Conrad W, Kalen V Clinical diagnosis of anterior cruciate ligament instability in the athlete Am J Sports Med , Trickey EL Rupture of the posterior cruciate ligament of the knee J Bone Joint Surg 50B , Zelko RR, Abrams S The Lachman sign vs the anterior drawer sign in the diagnosis of acute tears of the anterior cruciate ligament Orthop Trans 6: 196, 1982