| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 University Department of Radiology, Addenbrooke's Hospital, Box 219, Hills
Rd., Cambridge CB2 2QQ, United Kingdom.
2 Centre for Applied Medical Statistics, Department of Public Health and Primary
Care, Institute of Public Health, Cambridge CB2 2SR, United Kingdom.
3 Department of Orthopedics, Addenbrooke's Hospital, Cambridge CB2 2QQ, United
Kingdom.
Received June 22, 2005;
accepted after revision October 10, 2005.
Address correspondence to P. W. P. Bearcroft.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. A prospective controlled observational study was performed with 91 consecutive referrals (53 males, 38 females; age range, 13-76 years; mean age, 40.6 years) from a single orthopedic foot and ankle surgeon to a regional teaching hospital for ankle MRI. Measurements of diagnostic confidence using a visual analogue scale (VAS) were made both before and after the MRI examination in each case. Proposed management was also recorded before imaging and then with benefit of the MRI result. Finally, the surgeon provided a subjective statement regarding the perceived usefulness of the MRI examination when the patient was subsequently reviewed in the clinic.
RESULTS. There was a significant overall change in diagnostic confidence in 65 (71%) cases. The number of diagnoses reduced from an average per patient of 2.3 (95% CI, 2.09-2.44) to 1.2 (95% CI, 1.05-1.39), and in 157 (69%) of all diagnoses entertained, there was a significant change in the confidence with which they were considered after imaging. Management plans changed in 32 (35%) of the patients, and in 31% of those patients for whom plans were changed, a less invasive plan was adopted. Overall, the surgeon thought that understanding of the patient's disease had either depended on MRI or had been substantially improved by it in 60 (66%) cases.
CONCLUSION. MRI of the ankle influences clinicians' diagnoses and management plans.
Keywords: ankle • diagnostic confidence • MRI • orthopedic surgery
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The referring clinician was required to complete a structured request form for the initial referral for ankle MRI. The data collected included relevant demographic data, clinical history, diagnosis, diagnostic certainty, and proposed management. A visual analogue scale (VAS) was used to record clinician's estimates of preimaging probability for the most common diagnoses: osteochondral defect; osteoarthritis; or disorders of the peroneal, Achilles, posterior tibial, and flexor hallucis and digitorum longus tendons. Other diagnoses were specified on a separate VAS if applicable (Table 1). The VAS was exactly 100 mm in length and was labeled 0% at one end and 100% at the other end but otherwise had neither end anchors nor distance markers along the length.
|
The intended management before imaging was divided into five groups, each represented by a single check box: discharge to primary care; review at outpatient clinic, physiotherapy, treatment with an orthosis or plaster cast; other imaging (including CT and bone scintigraphy); arthroscopic surgery; and open surgery. In the case of open surgery, the clinician was required to fill in a text field to indicate which operation was being proposed. For the assessment of whether MRI changed patient care, the data were analyzed as two groups: nonsurgical and surgical care.
Once the MRI examination had been performed, the findings and the images were presented to the surgeon at a meeting with the radiologist. The surgeon had access to the clinical details and the preimaging assessment of the patient. Taking the radiologic information into consideration, the surgeon then completed an identically structured form to the preimaging request form, which recorded diagnosis, diagnostic certainty on a VAS, and proposed management. The patient was not seen before reviewing the radiologist's report so that changes in diagnosis and management could be attributed to the MRI result.
Diagnostic certainty was considered to have increased only if the diagnosis was considered to be definite (> 90% probability) or very unlikely (< 10% probability) after MRI. Less-pronounced changes in diagnostic certainty were not included.
Changes in management were considered relevant when there was a shift toward or away from surgery, and therefore such shifts were assessed separately. However, in cases where surgery was proposed both before and after MRI, the surgeon was asked to indicate whether the specific surgical approach had changed.
After the patient had been reviewed in the outpatient clinic by the surgeon and after definitive treatment was planned, the clinician's subjective assessment of the usefulness of the MRI in each case was evaluated by asking which one of the following statements applied to MRI in each patient. "Has imaging been: confusing, leading to investigations that I would not have otherwise have done; confusing, but has not led to other investigations; little or no effect on my understanding of this patient's disease; substantially improved my understanding of this patient's disease; or my understanding of this patient's disease depended on diagnostic information provided by MRI (unavailable from any other nonsurgical procedure)." At the end of the study period, the hospital notes were reviewed to confirm the accuracy of the information on the preimaging, postimaging, and follow-up forms.
Imaging Protocols
MRI was performed with a system operating at 1.5 T (LX Horizon EchoSpeed,
GE Healthcare) or at 1.0 T (Gyroscan, Philips Medical Systems). The standard
imaging protocol for both scanners involved a coronal T1 sequence, an axial
dual-echo sequence, and sagittal fat-saturated fast spin-echo T2 and
gradient-echo T2* sequences. All cases were reported by a single
experienced musculoskeletal radiologist.
Data Handling and Analysis
Data analysis was performed using SPSS statistical package, version 11.0
(SPSS). The pre- and postimaging VAS measurements were graded as shown in
Table 1 and treated as ordinal
variables. This grading system has been described previously and used in a
number of studies of the effectiveness of MRI of the wrist
[5], shoulder
[6], and knee
[7], and the reliability of VAS
measurement of diagnostic certainty has previously been shown
[8]. A shift from unlikely,
uncertain, or probable to definite (confirmed) or very unlikely (excluded) was
interpreted as an increase in diagnostic certainty. The McNemar test was used
to examine the hypothesis that diagnostic certainty increased after imaging.
The Wilcoxon's signed rank test was used to examine the hypothesis that fewer
diagnoses were considered after imaging than before. The McNemar test was used
to examine the hypothesis that the number of patients for whom surgery was
planned changed after imaging.
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Before imaging, more than one diagnosis was considered in 75 (82%) of the 91 patients. There was a significant reduction in the overall number of diagnoses being considered after MRI (mean [95% CI] number of diagnoses before and after MRI: 2.26 [2.09-2.44] and 1.22 [1.05-1.39], respectively; p < 0.001; Wilcoxon's signed rank test z = -7.01; Fig. 1). The preimaging diagnosis and the MRI findings are summarized in Table 2.
|
Table 3 shows the changes in diagnostic certainty after MRI. Entries in bold along the diagonal represent patients' diagnoses that were not significantly altered by imaging. Those entries to the left of the diagonal represent diagnoses that were considered less likely after imaging than before imaging, and conversely, those to the right of the diagonal represent diagnoses for which imaging increased the diagnostic likelihood. The data are summarized graphically in Figure 2, where it can be seen that there is a significant shift of diagnostic certainty from the middle of the graph before imaging (clinicians uncertain about the diagnosis) to the sides after imaging (representing increased diagnostic confidence). There was a change in the diagnostic certainty in 65 (71%) patients. Nineteen new diagnoses were made in 16 patients after the MRI; 16 (84%) of these were considered probable or definite. After imaging, 114 diagnoses were excluded in 75 patients; 27 (24%) of these had been considered probable or definite before imaging. Of the total 226 diagnoses, the number of diagnoses considered very unlikely or definite increased from 46 (20%) before MRI to 199 (88%) after MRI (p < 0.0001, McNemar test).
|
|
Table 4 summarizes the management plans before and after MRI. Entries in bold along the diagonal represent management plans that were not significantly altered by imaging, whereas entries on either side of the diagonal document a corresponding alteration in management plans. Management plans changed in 32 (35%) patients: nine (14%) of the 65 patients in whom surgery was planned before imaging were treated nonsurgically. In these nine patients, there were two new diagnoses considered after the MRI, but diagnostic confidence increased in seven (78%). Six (23%) of the 26 patients in whom nonsurgical plans were considered before MRI were treated surgically. In this group of patients, two new diagnoses were considered postimaging, but diagnostic confidence increased in all cases. The shift from or to surgery was not significant (McNemar test p value = 0.61). Ten patients were discharged after the MRI; five (50%) of these had been considered for surgery before imaging. In six of the 32 cases in which open surgery was proposed both before and after imaging, the surgeon indicated that the surgical approach would change in light of the results of the MRI; no change in approach was proposed in the remaining 26.
|
A change in management was associated with a change in diagnosis in 18 (56%) of 32 patients and an increase in diagnostic certainty in 31 (97%). In only one (3%) of the patients in whom the management changed did neither the diagnosis nor diagnostic certainty change.
In the 59 patients in whom management did not change after MRI, eight (14%) had new diagnoses considered, and diagnostic confidence improved in 37 (63%).
Table 5 shows the clinician's assessment of the usefulness of MRI of the ankle. In 60 (66%) cases, the surgeon thought that diagnostic confidence had either been substantially improved by or had depended on the results of the MRI.
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Precautions were taken to minimize the effect of bias inherent in such before-and-after studies [12]. The data were collected prospectively, and all patients were recruited consecutively. A prior power calculation was performed to ensure sufficient patients were recruited into the study. The initial power analysis had shown that to up to 70 patients would be needed to achieve 80% power and 90 patients for 90% power. To reach this more rigorous latter figure, the study was performed over an 18-month period and 91 qualifying patients were recruited.
Guyatt et al. [12] stressed the tendency for observational studies to overestimate the therapeutic benefits of interventions when compared with randomized controlled trials. To counter this tendency, diagnostic certainty was considered to have increased only if the diagnosis was considered to be definite (> 90% probability) or very unlikely (< 10% probability) after MRI. Less-pronounced changes in diagnostic certainty were not included. For the assessment of whether MRI changed patient care, the data were analyzed as two groups: nonsurgical and surgical care.
Before imaging, the majority of patients (82%) had more than one proposed diagnosis, which reflects the clinician's diagnostic uncertainty that prompted the initial referral for MRI. After imaging (Fig. 1), in the majority of patients (56%), a single diagnosis was considered and 14% of the results were considered normal. In this regard, MRI was shown to be a useful tool of exclusion, allowing the clinician to exclude many diagnoses that had been considered before imaging. In addition, there was an increase in the diagnostic certainty with which the remaining diagnoses were held, with the number of diagnoses considered very likely or definite increasing from 20% before imaging to 88% after imaging (Table 3 and Fig. 2).
The value of refining a clinician's diagnosis is realized when there is a resulting change in patient management. In this regard, management plans changed in 35% of patients. Although there was no overall statistical difference in the number of surgical procedures performed in the patient population as a whole (Table 4), a number of individual patients passed out of or into the surgical group: 14% of patients in whom surgery had been planned were treated conservatively, and 23% of patients who were to have been treated conservatively were considered to need surgery after imaging. In 7% of patients in whom the decision to operate was not altered, the surgical approach was changed by imaging. Therefore, it is concluded that imaging had a greater effect on the management plans of individual patients than can be deduced from overall population statistics. Although it is not possible to conclude decisively that the management plans were altered only as a result of the MRI, it is difficult to explain the observed increase of diagnostic certainty associated with the management plans by another influence because the patients were not seen by the clinician between referral for MRI and receiving the results. Therefore, no clinical factors should be contributing to the changes in planned management.
Although MRI was shown to influence treatment plans, 54 of the 65 patients in whom surgery was planned before imaging still went on to undergo surgery, which is a higher percentage than was found in similar studies of the effectiveness of MRI in the shoulder [6], wrist [5], and knee [7]. In many cases, this is because the working diagnosis was not one that can easily be made with conventional MRI (as opposed to MR arthrography), and the MRI examination was requested to exclude other abnormalities. Specifically, disorders of the anterolateral gutter (synovitis associated with impingement and meniscoid lesions) require ankle MR arthrography for their detection and delineation [13]. The fact that no abnormality was seen in the anterolateral gutter on conventional MRI was not accepted by the referring clinician as sufficient evidence that the clinical diagnosis was refuted. Ankle arthroscopic surgery was undertaken in these cases.
It is important to note that it may be inappropriate to assume that the effectiveness of MRI of the ankle calculated in one institution will be identical in another. In this study, one experienced musculoskeletal radiologist has been linked to one foot and ankle surgery specialist in a teaching hospital setting. In other settings, the effectiveness may differ for a number of reasons including the case mix of patients and the experience and practice of the referring clinician [14]. Nevertheless, these data give a useful indication of the effectiveness of MRI of the ankle and help to explain why MRI may have a reduced impact on diagnostic thinking and patient management in certain contexts.
In this study, the authors did not attempt to show a causal link between the diagnostic and therapeutic effects of MRI of the ankle and the eventual patient outcome. In part this is due to the problem that any major effect on health will ultimately be mediated through the complications and effectiveness of the clinician's treatment rather that through the direct effect of the MRI examination itself. Furthermore, there are difficulties in demonstrating changes in patient health with an observational study design in which all patients are examined [11] because without a control group we cannot know how the patients may have fared without MRI. To show an effect of imaging on patient outcome, a comparative study of different imaging studies would be preferred, with some patients undergoing the examination and others receiving nothing or a different imaging study [15]. Such an approach would not be appropriate in this context because there is no suitable alternative imaging technique.
Ultimately, a litmus test of whether an imaging study has been worthwhile can be gauged by asking the clinician to give a value judgment in retrospect once the patient episode is finished. The clinician can see the impact of the MRI examination in the light of all the other sources of information available at the time of treating the patient and also in the light of how the patient subsequently responded to the treatment. The fact that the surgeon thought that diagnostic confidence either had been substantially improved by or had depended on the results of the MRI in 66% of the patients is a testament to the usefulness of ankle MRI in this practice.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
