Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

DOI: 10.1148/radiol.2341031062
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Oei, E. H. G.
Right arrow Articles by Hunink, M. G. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Oei, E. H. G.
Right arrow Articles by Hunink, M. G. M.
Related Collections
Right arrowRelated Articles
(Radiology 2005;234:125-133.)
© RSNA, 2005

Emergency Radiology

Acute Knee Trauma: Value of a Short Dedicated Extremity MR Imaging Examination for Prediction of Subsequent Treatment1

Edwin H. G. Oei, MD, MSc, Jeroen J. Nikken, MD, PhD, Abida Z. Ginai, MD, PhD, Gabriel P. Krestin, MD, PhD, Jan A. N. Verhaar, MD, PhD, Arie B. van Vugt, MD, PhD and M. G. Myriam Hunink, MD, PhD

1 From the Program for the Assessment of Radiological Technology (ART Program) (E.H.G.O., J.J.N., M.G.M.H.) and Departments of Radiology (E.H.G.O., J.J.N., A.Z.G., G.P.K., M.G.M.H.), Epidemiology and Biostatistics (E.H.G.O., J.J.N., M.G.M.H.), Orthopaedic Surgery (J.A.N.V.), and Traumatology (A.B.v.V.), Erasmus MC, University Medical Center Rotterdam, Room EE21-40a, Dr Molewaterplein 50, 3015 GE Rotterdam, the Netherlands. Received July 7, 2003; revision requested September 24; revision received April 5, 2004; accepted May 19. Supported in part by the Revolving Fund of the University Hospital Rotterdam and by Esaote, Genoa, Italy. Address correspondence to M.G.M.H. (e-mail: m.hunink@erasmusmc.nl).


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
PURPOSE: To assess the predictive value of a short magnetic resonance (MR) imaging examination, in addition to or instead of radiography, performed in patients with acute knee trauma to identify those who require additional treatment versus those who do not and can be discharged without further follow-up.

MATERIALS AND METHODS: The randomized controlled trial and use of collected data for prediction modeling were approved by the institutional review board; informed consent was obtained. Patients with recent knee injury were included in the trial if radiography was ordered. They were randomized into a group undergoing only radiography and a group undergoing radiography plus immediate MR imaging. A 0.2-T dedicated extremity MR imager and four short pulse sequences were used. Univariable and multivariable logistic regression analysis was used to evaluate patient characteristics, trauma mechanism, and findings at radiography and MR imaging for their value in prediction of need for subsequent treatment within the 6-month follow-up.

RESULTS: Data in 189 patients (123 male patients, 66 female patients; mean age, 33.4 years), 109 of whom underwent treatment after their initial visit, were analyzed. Age of 30 years or older, indirect trauma mechanism, radiographic results, and MR imaging results were significant predictors of need for treatment in univariable and multivariable analyses (P < .05). In the multivariable analysis, only abnormal MR imaging results were significantly predictive of need for treatment, and only when MR imaging replaced radiography (odds ratio, 2.61; 95% confidence interval: 1.12, 6.06).

CONCLUSION: Implementation of a dedicated extremity MR imaging examination, in addition to or instead of radiography, performed in patients with traumatic knee injury improves prediction of the need for additional treatment but does not significantly aid in identification of patients who can be discharged without further follow-up. Value of a short MR imaging examination in the initial stage after knee trauma is limited.

© RSNA, 2005


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Traumatic injuries of the knee joint, which are often caused by sports activities, constitute a large proportion of musculoskeletal trauma encountered in the emergency department. The initial evaluation of knee injuries usually consists of taking the clinical history and performing a physical examination, which involves various manipulative tests (1,2). Although physical examination may aid in establishing the diagnosis at a later stage or with the patient anesthetized, its accuracy in the acute stage has been questioned, especially for meniscal tears, and that accuracy is influenced by many factors (3–5). Moreover, it has been well recognized that thorough physical examination of a recently injured knee with acute hemarthrosis often is difficult because of swelling, pain, and guarding (6–8).

If a fracture is suspected, radiography usually is added to the diagnostic work-up, and such an addition often is based on criteria previously published as the Ottawa Knee Rule (9,10). Traumatic abnormalities, other than a fracture, that cannot be visualized by using radiography may be present in the knee joint. Unlike radiography, magnetic resonance (MR) imaging provides information on soft-tissue damage and has been accepted widely for the evaluation of internal knee derangements. A recent systematic review and meta-analysis demonstrated that MR imaging has a very good performance in diagnosis of tears of the menisci and cruciate ligaments (11). Nonetheless, the routine use of MR imaging in the initial work-up of acute knee injuries has been hampered by the high costs, the limited availability, and the long duration of an MR imaging examination. Low-field-strength dedicated extremity MR imagers have been developed to overcome some of these problems. Because of the compact design and low field strength of these types of imagers, the costs can be kept relatively low compared with the costs of high-field-strength whole-body units. Therefore, low-field-strength dedicated extremity MR imaging has created the possibility to apply MR imaging more routinely in the initial evaluation of knee trauma.

Application of MR imaging as an initial examination tool after knee trauma in the emergency department setting can potentially yield benefit to the patient and the health care system. Detailed imaging information in an early stage may result in more timely diagnosis and treatment in patients who would otherwise have been followed up, with delay of definitive therapy. Conversely, patients could possibly be identified who are unlikely to require specific treatment and who will thus not need reassessment or follow-up. In the light of cost minimization and efficiency, this identification would be beneficial, since these patients do not need to return to the outpatient department.

The purpose of this study was to assess the predictive value of a short MR imaging examination, in addition to or instead of radiography, performed in patients who present with acute knee trauma to identify those who require additional treatment versus those who do not and can be discharged without further follow-up.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Study Design and Population
This study was part of a prospective randomized controlled trial aimed at assessment of the value and cost-effectiveness of dedicated extremity MR imaging in the initial evaluation of all patients with acute knee, ankle, or wrist injury. Data about the three joints were analyzed and reported separately, and the patients included in this study were different from those included in the studies about the wrist and the ankle joints. Although it is not necessary to perform a randomized controlled trial to evaluate the predictive value of MR imaging in acute knee trauma, we used the data from such a trial to do so. Where applicable, we followed published guidelines for reporting the results of randomized controlled trials (12), and otherwise, we used those for reporting results of diagnostic studies (13). The randomized controlled trial, as well as the use of the collected data for prediction modeling, was approved by the institutional review board of Erasmus MC, University Medical Center Rotterdam, Rotterdam, the Netherlands.

From August 1999 to May 2001, we recruited patients from the emergency department of our university hospital who were referred for radiography by a traumatologist, orthopedic surgeon, or emergency physician because of traumatic knee injury within the preceding 7 days. We excluded patients who had additional injuries of the head, back, thorax, or abdomen; patients with a compound knee fracture; patients requiring urgent treatment (eg, in case of a threat to the circulation of the leg); and patients with preexisting symptoms in the same knee. Patients were included 7 days per week from 8:00 AM to 11:00 PM. All eligible patients were provided with written and oral information about the aim of the study. After informed consent was obtained, patients were randomized to a group who would undergo the current diagnostic work-up (only radiography) and a group who would undergo radiography and a short dedicated extremity MR imaging examination immediately afterward. Randomization was performed by research staff who were not involved in the further treatment of the patient with use of consecutively numbered sealed envelopes containing computer-generated random assignments. We applied block randomization with a block size of 20 patients to achieve equal numbers of patients in both study arms.

Imaging
Radiography was performed in the anteroposterior and lateral views. In a few patients, additional patellar or tunnel views were obtained because a patellar fracture or a loose body was suspected by the examining physician. For the MR imaging examination, we used a low-field-strength dedicated extremity MR imaging system (Artoscan M; Esaote Biomedica, Genoa, Italy) with a 0.2-T permanent magnet and a dual phased-array knee coil. A larger linear knee coil was used in patients with a very large knee. In regard to the imaging protocol, we aimed at providing a quick examination in the emergency setting that would minimize the costs and burden to the patient. With these goals in mind, we adapted several imaging parameters to reduce total examination time. This adaptation resulted in some loss of image quality, but in a pilot study, the imaging protocol as described in Table 1 was found to be of acceptable quality for the detection of most lesions (14). Average total acquisition time was 6 minutes 7 seconds, with a total examination time, including start-up of the MR imaging unit and patient positioning, of shorter than 15 minutes.


View this table:
[in this window]
[in a new window]

  		TABLE 1. Parameters of Short MR Imaging Protocol

 
According to daily practice in our hospital, the radiographs were initially assessed by the treating physician in the emergency department. On the next day, all radiographs were reassessed by one of two musculoskeletal radiologists (including A.Z.G.) who had at least 4 years of experience in musculoskeletal radiology at a daily meeting with the treating physician. If the reassessment differed from the initial interpretation, the treatment decision was based on the interpretation of the musculoskeletal radiologist, and the patient was called back for treatment if necessary. The MR images were immediately interpreted by an experienced musculoskeletal radiologist (A.Z.G., with 25 years of experience in musculoskeletal radiology) who was unaware of the results of radiography. The findings on MR images were reported to the treating physicians in the emergency department so that they could aid in the decision of whether to treat or follow up the patient. In a minority of patients (eg, during the weekends and in the evenings), the MR images were assessed by one of the radiology residents, who were in their 2nd–5th year of training, and were reassessed by the musculoskeletal radiologist on the next day. The treating physician was notified of any discrepancies in interpretation, and the patient was called back for treatment if indicated. For both radiographs and MR images, the interpretation of the musculoskeletal radiologist was used in the analysis.

For the purpose of our analysis, the results of radiography and MR imaging were interpreted as abnormal or normal on the basis of the presence or absence of recent traumatic abnormalities that required treatment. Results of radiography were considered abnormal if a fracture, an epiphysiolysis, a dislocation, or an osteochondral lesion was visible. Results of MR imaging were interpreted as abnormal if one or more of the following were present: a fracture, an epiphysiolysis, a dislocation, an osteochondral lesion, a meniscal tear, a total rupture of the anterior cruciate ligament, a total rupture of the posterior cruciate ligament, a total or partial rupture of the medial collateral ligament, or a total or partial rupture of the lateral collateral ligament. Although a partial rupture of the anterior or the posterior cruciate ligament is obviously a traumatic lesion, we considered the MR imaging results as normal, for the purpose of our analysis, if this was the only visible abnormality, since an isolated partial tear of a cruciate ligament does not usually need specific treatment. A partial rupture of the collateral ligament, however, usually is treated with a knee brace. Similarly, results of radiography and MR imaging were interpreted as normal (ie, not requiring specific treatment) if only joint effusion, hemarthrosis, soft-tissue edema, bone marrow edema, meniscal degeneration, or osteoarthritis was seen or if the findings suggested an old traumatic lesion. Bone marrow edema does not require specific treatment, but it may be suggestive of an intraarticular lesion, and therefore its presence may increase the likelihood of treatment. Therefore, we reanalyzed the data with isolated bone marrow edema interpreted as an MR imaging finding that required treatment.

Follow-up and Outcome
The outcome measure was additional treatment within 6 months after the initial visit to the emergency department. Treatment was defined as any specific additional therapy, which included surgery, physical therapy with supervised and structured rehabilitation exercises, a plaster cast, and taping.

Patients were followed up as long as they had daily complaints of the affected knee, and this follow-up time was as long as a maximum of 6 months after initial presentation. Data were collected by searching the emergency department records, hospital records, outpatient clinic records, and electronic hospital information system. In addition, we sent questionnaires to all patients, and the questionnaires included questions about whether they had undergone treatment at another institution. These questionnaires were mailed at 1, 6, and 12 weeks and at 6 months after the initial visit to the emergency department. If the questionnaires were not returned, we attempted to interview the patients by telephone. Patient records were reviewed, and patients were interviewed by two of the authors (E.H.G.O. and J.J.N.) and research support staff. The final diagnosis was determined by reviewing all information from diagnostic imaging, follow-up, questionnaires, or arthroscopy or surgery reports, if available.

Since one may argue that the indications for physical therapy often are not well defined, we performed a separate analysis in which physical therapy was not considered a specific treatment.

Statistical Analysis
The data were analyzed according to intention to diagnose and treat by using univariable and multivariable logistic regression analysis. We evaluated the following independent variables for their statistical significance in the prediction of whether treatment was required within 6 months after the initial hospital visit: age (both continuous and dichotomous at various threshold levels), sex, side of trauma, trauma mechanism (indirect or direct trauma), and the results of both radiography and MR imaging (abnormality that required treatment vs no abnormality). The absence of MR imaging information, which was the case in half of our patient population because of the randomized study design, was used as the reference category for the MR imaging results in the analysis (normal or abnormal vs no MR imaging information). This allowed us to determine the additional value of the MR imaging results compared with no MR imaging information.

To account for the fact that the findings at radiography and MR imaging were correlated, we created what we called an overall imaging variable, in which the results of both the radiographic and MR imaging examinations were integrated. In this variable, the following five possible combinations of radiographic and MR imaging results were coded by using four dummy variables: (a) normal radiographic results and no MR imaging information (reference category), (b) normal results of both radiography and MR imaging, (c) abnormal results of radiography and no MR imaging information, (d) discrepant results of radiography and MR imaging, and (e) abnormal results of both radiography and MR imaging. By using the overall imaging variable, one takes into account the diagnostic interaction between radiography and MR imaging results, which may be involved if two imaging modalities are used together to visualize the same lesion. It is, therefore, a more appropriate manner of analyzing the results of radiography and MR imaging combined, rather than using the two variables separately, in the same regression model.

We first assessed each variable separately with univariable logistic regression analysis and calculated the odds ratio and the 95% confidence interval. Subsequently, we performed multivariable logistic regression analyses, which included all variables that had a P value of less than .10 ({chi}2 test) in the univariable analysis. Because we wanted to assess the value of the imaging modalities, we created four models that included the following: radiographic results (model 1), MR imaging results (model 2), radiographic and MR imaging results (model 3), and the overall imaging variable described previously (model 4). We used the Hosmer-Lemeshow goodness-of-fit test to evaluate the calibration of the models. A Hosmer-Lemeshow goodness-of-fit test result that was not significant (P > .05) indicated that the model fit well with the observed data (15). The area under the receiver operating characteristic curve was calculated to assess the discriminatory power of each model in distinguishing patients who required treatment from those who could be discharged. A larger area under the receiver operating characteristic curve indicated a higher discriminatory power (16).

To compare the area under the receiver operating characteristic curve across the models, we used the method described by Hanley and McNeil (17), which provides adjustment for the fact that the models were correlated since they were derived from the same sample of patients. The likelihood ratio test, which can be used for comparing nested models, was used to assess the value of MR imaging in addition to radiography with a comparison of models 1 and 3. To compare models that are not nested, one needs to calculate the Akaike information criterion (AIC) for each model and assess the difference in AIC between the models. The AIC of a model is calculated by subtracting two times the degrees of freedom from the {chi}2 estimate of the model (18). A higher AIC indicates better model performance. We used this technique to assess the value of MR imaging results as a replacement for radiographic results by comparing the AIC of model 1 with that of model 2 and also to assess the value of MR imaging results in addition to radiographic results by comparing the AIC of model 1 with that of model 4. A difference in AIC larger than 2.00 was considered significant. All analyses were performed by using software (SPSS for Windows, release 10.0.0; SPSS, Chicago, Ill).

Sample Size Calculation
The data for this study were obtained from a randomized controlled trial in which the costs and effectiveness of early MR imaging of knee trauma were evaluated. Hence, the number of independent variables that could be analyzed in the multivariable regression analysis was determined by the number of patients included in the randomized controlled trial: 109 included patients received treatment, and 80 patients did not need treatment. Since at least 10 patients with an event and 10 patients without an event are necessary for each independent variable in the multivariable logistic regression analysis (19), a maximum of eight independent variables could be analyzed.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
DISCUSSION
 REFERENCES
 
One hundred ninety-six patients were initially included in the study. Although inclusion was intended to be consecutive, only about half of the eligible patients were included. The main reason that only half were included was that radiology technologists on service who were supposed to identify eligible patients and ask them to participate in the study often forgot to do so. Six patients were incorrectly included, and they did not fulfill the inclusion criteria at reascertainment (ie, the research staff discovered afterward that these patients should not have been included). One patient presented to our hospital more than 7 days after knee injury, three had a history of preexisting complaints of the knee joint, and another two patients had knee pain of nontraumatic origin. These six patients, together with one patient in whom automutilation was strongly suspected, were excluded from the analysis. Of the 189 remaining patients (123 male patients, 66 female patients; mean age, 33.4 years), 96 patients were randomized to the group who underwent radiography as the diagnostic strategy and 93 patients were randomized to the group who underwent radiography followed by MR imaging. Female patients (mean age, 36.1 years; range, 16.6–72.9 years) were older than male patients (mean age, 31.9 years; range, 12.6–74.6 years). This difference was of borderline significance (P = .047, t test). A diagram that shows the flow of subjects passing through the study is presented in Figure 1.



View larger version (29K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 1. Diagram shows flow of patients passing through the study.

 
The lesions that were found at radiography and MR imaging, together with the corresponding frequencies, are listed in Table 2. Figures 2 and 3 present illustrative examples. During the follow-up period, 109 patients underwent treatment, consisting of arthroscopic surgery in 21 patients, which included anterior cruciate ligament reconstruction in one; open surgery in five; drainage in one; and plaster cast immobilization, sometimes supplemented with physical therapy, in 23. Partial meniscectomy was performed in 18 of 21 patients who underwent arthroscopic surgery. Of these 18 patients, 10 had been randomized to the MR imaging diagnostic strategy, and in all these patients the meniscal tear was diagnosed at MR imaging. In the five patients in whom open surgery was performed, the procedure consisted of resection of a fragment of the anterior cruciate ligament combined with reefing of the medial collateral ligament, suturing of the tendon of the rectus femoris muscle, osteosynthesis of a fracture of the tibial plateau and spongiosaplasty, reinsertion of the medial collateral ligament on the tibia, and fixation of a fragment from osteochondritis dissecans with bone pins. In 59 patients, physical therapy was the only treatment.


View this table:
[in this window]
[in a new window]

  		TABLE 2. Frequency of Lesions Diagnosed with Radiography and MR Imaging in 189 Patients with Knee Injury

 


View larger version (127K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2a. (a) Sagittal T1-weighted spin-echo MR image (710/24, 192 x 216 matrix, 5-mm section thickness) shows tear (arrow) of anterior cruciate ligament. (b) Short inversion time inversion-recovery MR image (1160/24, 192 x 128, 6-mm section thickness) shows that associated bone marrow edema (arrows) is visible in the lateral femoral condyle and the posterolateral tibial plateau.

 


View larger version (129K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 2b. (a) Sagittal T1-weighted spin-echo MR image (710/24, 192 x 216 matrix, 5-mm section thickness) shows tear (arrow) of anterior cruciate ligament. (b) Short inversion time inversion-recovery MR image (1160/24, 192 x 128, 6-mm section thickness) shows that associated bone marrow edema (arrows) is visible in the lateral femoral condyle and the posterolateral tibial plateau.

 


View larger version (154K):
[in this window]
[in a new window]
[Download PPT slide]
  		Figure 3. Sagittal T1-weighted spin-echo MR image (680/24, 192 x 216 matrix, 5-mm section thickness) shows tear of the posterior horn of the lateral meniscus. A displaced fragment (arrow) is seen posterior to the anterior horn.

 
The odds ratios and 95% confidence intervals of all demographic, clinical, and imaging variables that were assessed in the univariable logistic regression analysis are presented in Table 3. Age, trauma mechanism, radiographic results, MR imaging results, and the overall imaging variable had a P value of less than .10 ({chi}2 test) and were subsequently used in the multivariable analysis. Age of the patient was a significant predictor of treatment as a continuous variable, but the most significant association was found when age was dichotomous, with 30 years as the threshold value (P = .01). Compared with no MR imaging information, abnormal MR imaging results were predictive of the need for treatment, whereas normal MR imaging results were predictive of no treatment required. Similarly, in regard to the overall imaging variable, the category in which radiographic and MR imaging results were both abnormal was significantly predictive of treatment, and the category in which radiographic and MR imaging results were both normal was significantly predictive of no treatment required. The two intermediate categories had no significant association with treatment. As expected, sex and side of trauma were not significant in the univariable analysis (P = .36 and .61, respectively).


View this table:
[in this window]
[in a new window]

  		TABLE 3. Results of Univariable Logistic Regression Analysis for Prediction of Need for Treatment after Acute Knee Injury

 
We created four multivariable models that included the significant variables from the univariable analysis, with different combinations of results of the two imaging modalities (Table 4). All variables were significantly contributive to these models (P < .05 for all, likelikood ratio test). Model 1 consisted of age, trauma mechanism, and the results of radiography. In model 2, age and trauma mechanism were included, but the results of radiography were replaced with the MR imaging results, and the MR imaging results were significant predictors of the need for treatment only if they were abnormal; however, a normal MR imaging result could not be used to predict that treatment was not required, which was reflected by a 95% confidence interval of the odds ratio containing unity. Model 3 consisted of age, trauma mechanism, and the results of both radiography and MR imaging using the separate variables for radiographic and MR imaging results (additive combination of models 1 and 2). In this model, the odds ratios of the MR imaging results variable were not significant for both a normal and an abnormal MR imaging result, which suggested that a positive or a negative MR imaging result, versus no MR imaging information available, did not significantly contribute to the prediction of the need for treatment when it was controlled for age, trauma mechanism, and radiographic results. Model 4 consisted of age, trauma mechanism, and the overall imaging variable. Although this variable did significantly contribute to the prediction of treatment (P = .002), only the odds ratio of the category in which both radiographic and MR imaging results were abnormal was significant.


View this table:
[in this window]
[in a new window]

  		TABLE 4. Multivariable Prediction Models for Prediction of Need for Treatment after Acute Knee Injury

 
All model performance statistics are tabulated in Table 4. The Hosmer-Lemeshow goodness-of-fit test result was not significant for any of the models, and this finding indicated that all models fit well with the observed data. The area under the receiver operating characteristic curve, which is a measure of the discriminatory power of the model, was not significantly different when model 1 was compared with model 2, model 1 was compared with model 3, and model 1 was compared with model 4 (P = .53, .91, and .93, respectively). Thus, neither the replacement of radiographic results with MR imaging results (comparison of model 1 and model 2) nor the addition of MR imaging results to radiographic results (comparison of model 1 and model 3 or 4) improved the discriminatory power. Addition of MR imaging results to model 1 (and thus creation of model 3) significantly improved the model performance (P = .04, likelihood ratio test), indicating that information from MR imaging improved model performance when added to radiographic results. However, as was noted previously, this did not result in a better discrimination between patients who required treatment and patients who could be discharged without follow-up, since both a normal and an abnormal MR imaging result compared with no MR imaging information had a 95% confidence interval that included unity. The AIC of model 2 was found to be lower than that of model 1, which suggested that MR imaging results as replacements for radiographic results did not improve model performance. Similarly, the performance of model 4 (which included information from MR imaging in the overall imaging variable) was not significantly better than that of model 1.

Repeating the analysis with isolated bone marrow edema defined as an MR imaging finding that required additional treatment did not change the significance of any of the predictors in the univariable and multivariable analyses (data not shown). Similarly, in the separate analysis in which physical therapy was not considered a specific treatment, we found the same significant predictors in the univariable analysis. In the multivariable analysis, age was only a significant predictor in model 1, but the significance of the imaging variables in all models was not affected (data not shown). Also, all odds ratios of the imaging variables and the model performance statistics were similar.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
REFERENCES
 
We evaluated the value of MR imaging as a routine screening tool in the initial stage after knee trauma, a setting in which MR imaging has not been widely implemented thus far. The purpose of performing imaging tests in the acute stage is to distinguish patients who require additional treatment from those who do not. This allows patients to be treated appropriately and others to be discharged from follow-up without the need for additional visits or work-up. MR imaging has been advocated as a screening tool prior to therapeutic arthroscopy, since it has proved to be very reliable for noninvasive detection of internal knee derangements (11,20). Application of MR imaging as an initial examination tool in the emergency department setting, however, has been limited because of the associated high costs and other problems, such as burden to the patient, long imaging time, and limited availability of an MR imaging unit. We used a dedicated extremity MR imaging system, the design of which could overcome most of these problems. Because of the low field strength and the relatively small size of the equipment, such a unit does not require infrastructural changes prior to installation, so that the costs of installation and maintenance of such an MR imaging unit are low compared with those of a high-field-strength whole-body unit. Furthermore, it has been shown that lower magnetic field strength does not substantially reduce the diagnostic performance for most internal knee derangements (11,21–25).

With four multivariable logistic regression models, we assessed the value of a dedicated extremity MR imaging examination, when performed in addition to and instead of radiography, for the prediction of the need for additional treatment in patients with acute knee trauma. Contrary to what we expected, the results of this study suggested that such an MR imaging examination has limited additional value for the prediction of the need for treatment. Although the MR imaging results variable was significant in the prediction of the need for treatment when added to the model with radiographic results, compared with no MR imaging information available, neither abnormal nor normal MR imaging results had significant added predictive value. As an alternative to radiographic results, MR imaging results were only significantly predictive of treatment if they were abnormal, and they did not contribute to the identification of patients who would not require treatment and who could be discharged without follow-up.

Although inclusion was intended to be consecutive, only about half of the eligible patients were randomized. Patients could be included daily from 8:00 AM to 11:00 PM. Many different radiology technologists on service were responsible for asking patients to participate, and they forgot to do so regularly despite multiple efforts by the investigators to improve recruitment. In many instances, patients who were not included could not be traced, so that we could not determine the exact number of missed eligible patients. Since we could not identify a systematic pattern in regard to the characteristics of missed patients, we assume that this suboptimal patient inclusion did not cause substantial selection bias, but it cannot be ruled out.

We used a short MR imaging protocol, which may partly account for the limited additional value of MR imaging that we found in our study. We believe, however, that a more expanded protocol or a protocol that is used in daily practice (routine) for nonemergency patients is not feasible for routine application in the emergency setting because of the excess burden to the patient, with the longer examination time and the higher costs. Furthermore, the aim was to use this protocol as a triage tool to distinguish patients who require treatment from patients who can be discharged without further follow-up. Considering this purpose, we think that a less extensive MR imaging technique than is commonly used for a regular MR imaging examination should suffice.

Our method of interpretation of the MR imaging results also may have influenced the results of this study. We considered an MR imaging study to be abnormal only if one or more lesions that generally require therapy were visible. Visible lesions that are usually not treated, such as a partial rupture of the anterior cruciate ligament, were not counted as abnormal when we used this definition (26,27). Although this is justified in the majority of cases, we recognize that exceptions are possible in specific patient groups, such as professional athletes. Therefore, the results of this study must be judged by taking into consideration the classification method of the MR imaging findings. Although bone marrow edema does not require specific treatment, it can be an indirect manifestation of another lesion; therefore, such a finding may increase the likelihood of treatment. We performed a separate analysis in which isolated bone marrow edema was interpreted as an MR imaging finding that required additional treatment, but results of this analysis did not change the significance of any predictor.

Apart from radiologic techniques, we assessed the predictive value of clinical factors such as age and trauma mechanism. Older age is associated with a higher prevalence of degenerative knee disorders, and this higher prevalence, in turn, is associated with increased likelihood of structural damage after trauma. For example, a meniscal tear occurs more easily in patients who have a meniscus with myxoid degeneration than in those who have an unaffected meniscus. In our analysis, an age of older than 30 years was found to be significantly predictive of treatment required. Older age also is associated with a higher prevalence of preexisting asymptomatic meniscal tears, but symptoms in a patient with such a condition often are not due to that tear itself and will usually resolve within a few months if there are no other abnormalities in the knee joint. Because of the long waiting period for arthroscopy in our institution, it is therefore unlikely that preexisting asymptomatic meniscal tears were unnecessarily treated in our study.

We also recorded whether the mechanism of injury was a direct blow or an indirect trauma, usually due to rotational force or varus or valgus stress, which has been shown to give rise to different patterns of findings at MR imaging (28,29). Whereas a direct hit more often gives rise to a fracture, an indirect trauma more frequently causes meniscal tears, ligament ruptures, and osteochondral lesions and was found to be a significant predictor of the need for treatment in this study.

A potential limitation of this study was that we did not assess the predictive value of findings at physical examination. The accuracy of physical examination of the knee joint, however, has been controversial, especially if performed in the acute stage after injury. It has been widely recognized that extensive joint swelling, pain, or guarding render physical examination without anesthesia inaccurate in the early stage after injury (6–8), which was the period of interest in this study. Moreover, it has been reported that the accuracy of physical examination of the knee is influenced by the experience of the physician (30).

Another limitation of this study was that we defined treatment within the follow-up period as the outcome measure, assuming that this truly reflected the need for treatment in all patients. It is possible, however, that patients were incorrectly discharged from further treatment and follow-up or that patients received unnecessary treatment. We believe that the assumption is justified for patients who were identified as those who did not require therapy. If treatment were incorrectly withheld in the initial stage, these patients would likely have returned and received treatment within the 6 months of follow-up because of persisting symptoms. On the other hand, the appropriateness of a treatment that a patient has already undergone is difficult to assess in retrospect, and we were unable to identify patients who had been treated unnecessarily. Since the treatment of the various lesions was not standardized, the decision to treat and the type of treatment varied across physicians. We would not expect this to influence the results of our study, however, since differences in type of treatment did not affect the outcome measured. Only a difference in the choice of whether or not to treat a patient influenced the outcome, as discussed previously.

Follow-up of patients in this study was limited to a period of 6 months after their initial visit to the emergency department. Although one might argue that we may have incorrectly classified patients in whom treatment was initiated after this 6-month period as having undergone no treatment, we believe that this is rarely the case, since therapy for traumatic lesions of the knee joint is generally initiated soon after the injury, if necessary. Nonetheless, we found that in some cases surgery was performed after the 6-month follow-up period because of the waiting time for arthroscopy. This was the reason why we did not use the need for surgery as an alternative outcome measure in this study. In all of these patients, however, physical therapy had already been initiated within 6 months after first presentation so that these patients were dealt with correctly in our analysis. We acknowledge that the indications for physical therapy are often not well defined, so it can be argued that physical therapy should not count as a specific treatment. Therefore, we repeated the analysis in which physical therapy was not considered a specific treatment, but results of this repeat analysis did not change the conclusions of our study.

With results of this study, one can conclude that a prediction of the need for treatment in patients who present after acute knee trauma can be made on the basis of age, trauma mechanism, and the radiographic results. To apply this prediction rule in clinical practice would require choosing the optimal threshold score, which requires consideration of the likelihood of a treatable lesion and weighing the benefit gained by making a correct prediction against the consequences of making an incorrect prediction, in terms of expected effectiveness and costs. The optimal threshold value of the prediction rule represents the combination of true- and false-positive ratios that yield the greatest expected utility for the patient at acceptable costs to society (16).

We conclude that the value of a dedicated extremity MR imaging examination in the initial stage after knee trauma is limited. A short MR imaging examination, in addition to or instead of radiography, improves the prediction of the need for additional treatment in patients with traumatic knee injury but does not significantly aid in the identification of patients who can be discharged without further follow-up.


   ACKNOWLEDGMENTS
 
The authors thank Caroline van Bavel-van Hamburg and Wibeke van Leeuwen for their support in collecting the data.


   FOOTNOTES
 
Abbreviation: AIC = Akaike information criterion

Author contributions: Guarantor of integrity of entire study, M.G.M.H.; study concepts and design, all authors; literature research, E.H.G.O., J.J.N.; clinical studies, E.H.G.O., J.J.N.; data acquisition and analysis/interpretation, all authors; statistical analysis, E.H.G.O., J.J.N.; manuscript preparation, definition of intellectual content, revision/review, and final version approval, all authors; manuscript editing, E.H.G.O.

See also the articles by Nikken et al in this issue.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 

  1. Solomon DH, Simel DL, Bates DW, Katz JN, Schaffer JL. The rational clinical examination: does this patient have a torn meniscus or ligament of the knee? value of the physical examination. JAMA 2001; 286:1610-1620.[Abstract/Free Full Text]
  2. Smith BW, Green GA. Acute knee injuries. I. History and physical examination. Am Fam Physician 1995; 51:615-621.
  3. Scholten RJ, Deville WL, Opstelten W, Bijl D, van der Plas CG, Bouter LM. The accuracy of physical diagnostic tests for assessing meniscal lesions of the knee: a meta-analysis. J Fam Pract 2001; 50:938-944.[Medline]
  4. Kim SJ, Kim HK. Reliability of the anterior drawer test, the pivot shift test, and the Lachman test. Clin Orthop 1995; 317:237-242.
  5. Corea JR, Moussa M, al Othman A. McMurray’s test tested. Knee Surg Sports Traumatol Arthrosc 1994; 2:70-72.[CrossRef][Medline]
  6. Hardaker WT, Jr, Garrett WE, Jr, Bassett FH, 3rd. Evaluation of acute traumatic hemarthrosis of the knee joint. South Med J 1990; 83:640-644.[CrossRef][Medline]
  7. Mitsou A, Vallianatos P. Clinical diagnosis of ruptures of the anterior cruciate ligament: a comparison between the Lachman test and the anterior drawer sign. Injury 1988; 19:427-428.[CrossRef][Medline]
  8. Simonsen O, Jensen J, Mouritsen P, Lauritzen J. The accuracy of clinical examination of injury of the knee joint. Injury 1984; 16:96-101.[CrossRef][Medline]
  9. Stiell IG, Greenberg GH, Wells GA, et al. Derivation of a decision rule for the use of radiography in acute knee injuries. Ann Emerg Med 1995; 26:405-413.[CrossRef][Medline]
  10. Stiell IG, Wells GA, Hoag RH, et al. Implementation of the Ottawa Knee Rule for the use of radiography in acute knee injuries. JAMA 1997; 278:2075-2079.[Abstract/Free Full Text]
  11. Oei EH, Nikken JJ, Verstijnen AC, Ginai AZ, Hunink MG. MR imaging of the menisci and cruciate ligaments: a systematic review. Radiology 2003; 226:837-848.[Abstract/Free Full Text]
  12. Moher D, Schulz KF, Altman D. The CONSORT statement: revised recommendations for improving the quality of reports of parallel-group randomized trials. JAMA 2001; 285:1987-1991.[Abstract/Free Full Text]
  13. Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. Radiology 2003; 226:24-28.[Abstract/Free Full Text]
  14. Nikken JJ. MRI for the initial evaluation of acute wrist, knee, and ankle trauma Thesis. Rotterdam, the Netherlands: Erasmus MC, 2003.
  15. Hosmer DW, Lemeshow S. Assessing the fit of the model In: Applied logistic regression. New York, NY: Wiley, 1989; 135-175.
  16. Hunink MG, Glasziou PP, Siegel JE, et al. Decision making in health and medicine: integrating evidence and values Cambridge, England: Cambridge University Press, 2001.
  17. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983; 148:839-843.[Abstract/Free Full Text]
  18. Akaike H, ed. Information theory as an extension of the maximum likelihood principle Budapest, Hungary: Akademia Kiado, 1973.
  19. Peduzzi P, Concato J, Kemper E, Holford TR, Feinstein AR. A simulation study of the number of events per variable in logistic regression analysis. J Clin Epidemiol 1996; 49:1373-1379.[CrossRef][Medline]
  20. Vincken PW, ter Braak BP, van Erkell AR, et al. Effectiveness of MR imaging in selection of patients for arthroscopy of the knee. Radiology 2002; 223:739-746.[Abstract/Free Full Text]
  21. Kersting-Sommerhoff B, Gerhardt P, Golder W, et al. MRT des kniegelenks: erste ergebnisse eines vergleichs von 0,2-T-spezialsystem mit 1,5-T-hochfeldmagnet. [MRI of the knee joint: first results of a comparison of 0.2-T specialized system and 1.5-T high field strength magnet.] Rofo Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 1995; 162:390-395.
  22. Vellet AD, Lee DH, Munk PL, et al. Anterior cruciate ligament tear: prospective evaluation of diagnostic accuracy of middle- and high-field-strength MR imaging at 1.5 and 0.5 T. Radiology 1995; 197:826-830.[Abstract/Free Full Text]
  23. Kladny B, Gluckert K, Swoboda B, Beyer W, Weseloh G. Comparison of low-field (0.2 tesla) and high-field (1.5 tesla) magnetic resonance imaging of the knee joint. Arch Orthop Trauma Surg 1995; 114:281-286.
  24. Rutt BK, Lee DH. The impact of field strength on image quality in MRI. J Magn Reson Imaging 1996; 6:57-62.[Medline]
  25. Barnett MJ. MR diagnosis of internal derangements of the knee: effect of field strength on efficacy. AJR Am J Roentgenol 1993; 161:115-118.[Abstract/Free Full Text]
  26. Sommerlath K, Odensten M, Lysholm J. The late course of acute partial anterior cruciate ligament tears: a nine to 15-year follow-up evaluation. Clin Orthop 1992; 281:152-158.
  27. Kannus P, Jarvinen M. Conservatively treated tears of the anterior cruciate ligament: long-term results. J Bone Joint Surg Am 1987; 69:1007-1012.[Abstract/Free Full Text]
  28. Hayes CW, Brigido MK, Jamadar DA, Propeck T. Mechanism-based pattern approach to classification of complex injuries of the knee depicted at MR imaging. RadioGraphics 2000; 20(special issue):S121-S134.[Abstract/Free Full Text]
  29. Sanders TG, Medynski MA, Feller JF, Lawhorn KW. Bone contusion patterns of the knee at MR imaging: footprint of the mechanism of injury. RadioGraphics 2000; 20(special issue):S135-S151.[Abstract/Free Full Text]
  30. Boeree NR, Ackroyd CE. Assessment of the menisci and cruciate ligaments: an audit of clinical practice. Injury 1991; 22:291-294.[CrossRef][Medline]

Related Articles

Acute Wrist Trauma: Value of a Short Dedicated Extremity MR Imaging Examination in Prediction of Need for Treatment
Jeroen J. Nikken, Edwin H. G. Oei, Abida Z. Ginai, Gabriel P. Krestin, Jan A. N. Verhaar, Arie B. van Vugt, and M. G. Myriam Hunink
Radiology 2005 234: 116-124. [Abstract] [Full Text] [PDF]

Acute Ankle Trauma: Value of a Short Dedicated Extremity MR Imaging Examination in Prediction of Need for Treatment
Jeroen J. Nikken, Edwin H. G. Oei, Abida Z. Ginai, Gabriel P. Krestin, Jan A. N. Verhaar, Arie B. van Vugt, and M. G. Myriam Hunink
Radiology 2005 234: 134-142. [Abstract] [Full Text] [PDF]



This article has been cited by other articles:


Home page
 RadiologyHome page
J. J. Nikken, E. H. G. Oei, A. Z. Ginai, G. P. Krestin, J. A. N. Verhaar, A. B. van Vugt, and M. G. M. Hunink
Acute Peripheral Joint Injury: Cost and Effectiveness of Low-Field-Strength MR Imaging--Results of Randomized Controlled Trial
Radiology, September 1, 2005; 236(3): 958 - 967.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Oei, E. H. G.
Right arrow Articles by Hunink, M. G. M.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Oei, E. H. G.
Right arrow Articles by Hunink, M. G. M.
Related Collections
Right arrowRelated Articles

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE