- Email: [email protected]
Magnetic Resonance Imaging of Suspected Scaphoid Fractures Using a Low Field Dedicated Extremity MR System
Clinical Radiology (2001) 56: 316±320 doi:10.1053/crad.2000.0657, available online at http://www.idealibrary.com on
Magnetic Resonance Imaging of Suspected Scaphoid Fractures Using a Low Field Dedicated Extremity MR System NIGEL RABY Department of Radiology, Western In®rmary, North Glasgow Hospitals NHS Trust, Glasgow, U.K. Received: 31 May 2000
Revised: 11 September 2000 Accepted: 2 October 2000
AIM: To determine whether magnetic resonance imaging (MRI) using a dedicated extremity system could be utilized for the routine assessment of patients with a clinically suspected scaphoid fracture, and to determine the most cost-eective timing of MRI for these patients. MATERIALS AND METHODS: MRI of the wrist using a 0.2T extremity MR system was performed in two groups of patients. Fifty-six patients were examined within 4 days of injury and 53 with persistent symptoms at 10 days to 6 weeks following injury. A scaphoid series of radiographs was negative in both groups. RESULTS: The early MR group had seven scaphoid, six radial and four other fractures. Management was altered in 89%. The late MR group had 14 scaphoid, nine radial and three other fractures. Management was altered in 69%. A cost model showed that overall costs are less with early rather than late scanning. CONCLUSION: MRI of the wrist when scaphoid fracture is suspected can be undertaken in all patients with negative radiographs and could be performed in most departments with an MRI machine. There are signi®cant patient bene®ts and overall costs would change little from conventional practice. Raby, N. # 2001 The Royal College of Radiologists (2001) Clinical Radiology 56, 316±320. Key words: MR imaging, scaphoid, wrist.
The diagnosis and management of scaphoid fractures remains controversial. There are several hundred papers in the literature on this subject over the past two decades demonstrating the level of continuing debate over this seemingly well-understood clinical problem. Radiological diagnosis of scaphoid fractures remains problematic. Many techniques have been employed to demonstrate scaphoid fractures including the use of comparative views [1], macroradiography [2], isotope bone scintigrams [3], computed tomography (CT) [4], ultrasound [5] and magnetic resonance imaging (MRI). MRI in particular has been shown to be an eective method of diagnosing fractures of the scaphoid [6] but studies published to date have involved relatively small numbers of patients and have been essentially pilot studies. Most have emphasized the usefulness of the technique but have also underlined that clinical application is dicult due to limited access to MRI. In this study two groups of patients were studied with a dedicated extremity MR system located close to the Accident and Emergency Department using MRI as a routine investigation. Author for correspondence and guarantor of study: Dr Nigel Raby, Department of Radiology, Western In®rmary, Dumbarton Road, Glasgow G11 6NT, U.K. Fax: 0141 337 3441; E-mail: N.Raby@ clinmed.gla.ac.uk 0009-9260/01/040316+05 $35.00/0
MATERIALS AND METHODS
Two groups of patients with a history of recent wrist injury were identi®ed. All patients studied were consecutive referrals. Group 1 consisted of 53 patients seen at review or fracture clinic at least 10 days post-injury (range 10 days±6 weeks), some of whom had attended the Accident and Emergency Department at the time of injury. Radiographic examination had not demonstrated a fracture but the patient had been placed in a scaphoid cast due to clinical suspicion of a scaphoid injury. The remainder had not attended hospital initially but had been referred for assessment due to continuing wrist pain. All patients in this group had persistent wrist pain with clinical ®ndings consistent with a scaphoid injury and normal radiographs at the time of clinic attendance. Group 2 consisted of 56 patients with an acute wrist injury attending the Accident and Emergency Department for the ®rst time usually on the day of injury (range 0±4 days). A scaphoid fracture was suspected clinically due to the presence of tenderness in the anatomical snubox but a scaphoid series of radiographs failed to conclusively demonstrate a scaphoid fracture. Patients with a fracture of the scaphoid or any other bone identi®ed on plain radiogaph, were excluded from the study. # 2001 The Royal College of Radiologists
317
MRI OF SUSPECTED SCAPHOID FRACTURES
To avoid any diculties in subsequent data analysis patients were ®rst referred according to the criteria for group 1. When the cohort of 53 patients had been acquired then patients from group 2 were studied. MRI examinations were performed on a 0.2T dedicated extremity system (Artsocan, Esaote, Genoa, Italy). Coronal images only were obtained using the following parameters: Spin echo T1 (TR 450, TE 24, three acquisitions, 3.0 mm slice thickness, 16 cm ®eld of view, matrix 192 192 acquisition time 4.39 min) and STIR (TR 1320, TE 30, TI 70, three acquisitions 3.0 mm slice thickness 16 cm ®eld of view matrix 192 160 acquisition time 6.43 min). A fracture was diagnosed when there was a linear area of low signal traversing bone from cortex to cortex on a T1 weighted sequence with a corresponding area of high signal on the STIR sequence. Areas of diuse signal change with no de®ned linear component on either sequence were designated bone bruises. This is in accordance with previous studies [7,8]. The distal radius and ulna, all the carpal bones and proximal metacarpals were scrutinized for these changes. Patients referred for MR had a form completed by the referring clinician. This asked for details of the clinical ®ndings and interpretation of the radiographs. They were also asked how they would manage the patient on the basis of the clinical and radiological ®ndings. Following MRI the patient returned immediately to the referring clinician with a written report of the ®ndings. This was hand-delivered by the radiologist and the ®ndings discussed if necessary. The clinician was then asked to write on the second part of the form how the patient would now be managed in light of the MRI ®ndings. Patients with a fracture of the scaphoid or any other bone (most commonly the distal radius) were managed and reviewed as per the standard practice for the department. No attempt was made to chart the progress of these patients. A questionnaire was sent to patients from group 2, if they had been discharged following normal MRI, 2±3 months after the initial injury asking if they were now pain-free, how long symptoms had taken to resolve and if they had attended any other practitioner regarding their wrist injury. They were also invited to re-attend for clinical review if they had any persisting symptoms.
RESULTS
MRI Findings In group 1 (Table 1) there were 26 (50%) patients with a fracture demonstrated by MRI. These were all treated in the conventional manner with cast appropriate to the fracture type. Seven patients with bone bruising were treated with a soft splint for symptomatic relief. One was discharged with an invitation to return if symptoms persisted. This patient has not reattended. The other six patients had a soft splint applied for symptomatic relief. These patients were reviewed at a follow-up clinic and all were subsequently discharged pain-free. Of the 20 patients with normal scans
Table 1 ± Patients in Group 1. MRI at 10 days±6 weeks Scaphoid fracture Distal radius fracture Other fracture Bone bruise
14 9 3 7
Normal
20
1 1 1 4 2 1
hamate trapezoid triquetral scaphoid radius hamate
Table 2 ± Patients in Group 2. MRI at 0±4 days post-injury Scaphoid fracture Distal radius fracture Other fracture
7 6 4
Bone bruise
4
Normal
35
1 1 1 1 2 1 1
triquetral trapezoid base of 5th metacarpal ulna styloid scaphoid radius ulna
all were given a soft splint for symptomatic relief. Ten were discharged but invited to return if symptoms did not resolve. None of these patients has reattended. The other 10 patients were kept under review due to clinical concern despite the normal MRI. All of these patients were subsequently discharged pain-free. In group 2 (Table 2), 17 patients (28%) with fractures demonstrated by MRI were placed in the appropriate cast and were reviewed clinically until symptom-free. Four patients with a bone bruise were asked to return for review at 10 days. Three patients did not attend; the fourth patient was subsequently discharged symptom-free. Thirty-one patients with normal MRI were placed in a supportive splint and discharged from further review but with an invitation to return if symptoms did not resolve. None of these patients have re-attended the department. Four patients with normal MRI were asked to return for review at 10 days due to clinical concern about their symptoms. All were pain-free at review and were discharged.
Eect on Patient Management In group 1 without the MRI the management decision was to place 49 of the patients in a scaphoid cast and to review with further X-rays in 10±14 days. Four patients were to have a soft splint and further clinical review. Following the MRI result 26 patients had a de®nitive diagnosis of a fracture. Fourteen had the wrist placed in a scaphoid cast; the rest were placed in a cast appropriate to the fracture type, mostly distal radius. The remaining 27 patients wrists were placed in a soft splint. Seventeen were kept under clinical review and 10 were discharged. Included in these ®gures are the four patients who were to have a soft splint and review. All had a normal MRI. Two were
318
CLINICAL RADIOLOGY
discharged and two retained for further review as previously planned. Thus, there was a change in management in 37 of the 53 patients (69.8%) In group 2 the management plan without MRI was to place all 56 patients in a scaphoid cast and review with radiographs at 10 days. Following MRI 31 patients were discharged with a soft splint, eight were placed in a soft splint and reviewed at a follow up clinic, 10 were immobilized in a cast other than a scaphoid type. A scaphoid cast was applied to the remaining seven patients with a scaphoid fracture demonstrated on MRI. Thus, management was altered in 49 of the 56 patients in this group (87.5%) Questionnaires were sent to 25 patients from group 2 who were discharged. Only ®ve of these were returned. None of these patients reported persistent wrist pain and none had sought advice from either their general practitioner or another Accident and Emergency Department. Due to the poor patient response no further questionnaires were distributed. DISCUSSION
The clinical diagnosis of scaphoid fractures remains unreliable [9,10]. Plain radiographs will fail to detect a number of fractures at initial presentation ÿ16% in a recent analysis of several papers [11]. The long-term complication of avascular necrosis of the proximal pole of the scaphoid following a fracture of the scaphoid waist is well recognized, as is non-union of the fracture. It is for this reason that so much eort is made to identify these injuries even although it is appreciated that these complications occur in only a minority of patients. Estimates of both vary considerably from as low as 5% to as high as 50%. As Barton [12] states `we overtreat a lot of patients to avoid undertreating a few'. MRI has been shown to be a reliable method for detection of fractures at several sites [13]. In particular, several studies have shown that MRI will accurately detect scaphoid fractures [6,11,14]. However, MRI is not yet routinely used in the investigation of these patients largely due to diculties of MR availability and cost. With the advent of relatively cheap dedicated extremity MR systems these problems can be overcome. Such an MR system has been installed in the orthopaedic clinic close to the Accident and Emergency Department. This MR system has time allocated for an MR system examining of patients with clinically suspected scaphoid injuries. How soon can MRI detect a fracture? Previous studies suggest that a fracture will be detected within hours of injury. Meyers and Wiener [13] detected six fractures at several sites less than 48 h after injury; Breitenseher et al. [6] detected 14 scaphoid fractures examined within 7 days of injury (mean 3.6 days, one patient examined at 25 h). A patient has been examined in this department (not within this study group) at 4 h after injury with a positive result. However because data is limited, follow-up of the MRnegative patients who presented within 4 days of injury and who were discharged was attempted by means of a questionnaire. Response to this was very poor; it is hoped
this means that the patients are symptom-free but this cannot be stated with any certainty. However, no patient reattended following discharge despite an open invitation to do so should symptoms not resolve. Since fracture detection beyond 7 days is well proven no attempt was made to follow up those patients in the ®rst group who were discharged following a normal MR at 10 days or more post-injury. The use of coronal T1 and STIR sequences has been advocated and tested by previous authors. Breitenseher et al. [6] evaluated T1, T2*GE and STIR sequences in the coronal plane and concluded that the combination of T1 and STIR sequences provided the highest sensitivity for fracture detection. Van Gelderen et al. [15] also used coronal T1 and STIR sequences and drew similar conclusions. Although many authors have studied MRI of scaphoid fractures using high ®eld strength magnets [16,11] the use of low ®eld magnets has also been tested. Lepisto et al. [17] studied 18 patients demonstrating fractures in 11 on a 0.1T unit. Van Gelderen et al. [15], using a 0.2T system, examined 16 patients. Bretlau et al. [8], using the same scanner as in the current study, examined 52 patients between 2 and 10 days following trauma using follow-up radiographs as the gold standard. Nine scaphoid fractures and six distal radial fractures were detected by MRI. All were subsequently con®rmed on the plain radiographs. There is therefore no apparent dierence in the ability of low and high ®eld MR systems to detect scaphoid fractures. The accuracy of MRI has been reported to have sensitivity and speci®city as high as 100% [6,11,14]. Breitenseher et al. [6] evaluated 42 patients with suspected clinical scaphoid fracture using MRI. The gold standard was plain radiographs obtained at 6 weeks postinjury. Fourteen scaphoid fractures and seven other fractures were detected by MRI and subsequently con®rmed on plain radiograph. The sensitivity and speci®city was 100%; there were thus no false-positives and, more importantly, no false negatives. Hunter et al. [11] examined 36 patients within 7 days of injury, ®nding 13 scaphoid and nine distal radial fractures. Clinical follow-up to pain-free discharge of those with normal MR ®ndings con®rmed that a negative MRI examination excludes a fracture. There were no false-negative examinations in this study. Several other techniques have been employed to detect scaphoid fractures. The most commonly used is bone scintigraphy, which is advocated by some as the investigation of choice [3]. There is little doubt that this is a sensitive technique. A negative isotope study reliably excludes a fracture but a positive scintigram is non-speci®c [18]. Waizenegger et al. [19] found increased isotope uptake in the scaphoid in 19 patients but could only ®nd con®rmatory evidence on late plain radiographs or CT in seven. Fowler et al. [7] compared MRI and scintigraphy in the same patients and found that MRI was more sensitive and speci®c than isotope scanning. Thorpe et al. [20] found MRI and scintigraphy able to detect scaphoid fractures with equal accuracy.
MRI OF SUSPECTED SCAPHOID FRACTURES
In the current study, patients examined by MR early after initial presentation were most commonly found to have normal ®ndings (35/56) with only 17 fractures and four cases of bruising. Those examined at more than 10 days after injury produced more positive MR ®ndings (26/53), with bruising in seven and only 20 normal. This is logical since it is known that the majority of patients do not have a signi®cant injury. After some time has elapsed clinical examination will be normal in many of these patients and they would be discharged. As only those patients who remained symptomatic were sent for MRI these are a selected group in whom a higher proportion of signi®cant injuries are likely to be found. This is after all the rationale for continuing clinical evaluation as a means of identifying those patients with the highest likelihood of an occult fracture. It is thus tempting to conclude that the most costeective time to examine patients is in the group with symptoms persisting 10 days or more after injury. However the following analysis shows that this is not necessarily the case. The cost of MR with this machine is approximately £50.00 per examination. This includes sta and consumable costs. Patients with clinical scaphoid injuries but normal radiographs often require to be reviewed several times before ®nal discharge. Kukla et al. [21] found that these patients re-attended for an average of 2.5 times. Speci®c data on this is not available from this department but if it is assumed that this is a fair estimate then an attempt at relative costing can be made. This can be applied to both of the patient groups to see at which time it is most ®nancially advantageous to scan patients. This hospital is unable to give a cost for attendance at a fracture review clinic. Although some patients with a normal scan or bone bruising were asked to return for clinical review all were subsequently discharged pain-free. With this knowledge in future it would be possible to discharge all of these patients after their MR examination. With these assumptions the following calculations can be applied to the two groups. In group 1, 53 patients had MRI at £50.00 each, incurring an additional total cost of £2650. Fractures were demonstrated in 26 patients. Assuming these would eventually become apparent on conventional radiographs then management is unlikely to be signi®cantly changed and there would be no dierence in the cost of managing these patients. Twenty-seven patients with normal MR or bone bruising only could be discharged from the clinic. They would previously have attended on average a further 2.5 times. Thus, a total of 65 out-patient attendances could be avoided. To remain neutral in terms of overall cost, each clinic attendance would have to have cost £40.00. If it is more than this savings occur, if less then overall costs are increased. In group 2, 56 patients have MRI costing £2800. Seventeen patients with fractures are cost-neutral. Thirty-nine patients with normal MR or bone bruising could be discharged, avoiding 97.5 attendances. The price of a clinic attendance to remain cost-neutral would be less than £30.00. Using this model it is cheaper to examine all
319
patients with MR early rather than selected patients after 10 days. It is also in this group that MRI most often results in a change of management. While there may be little dierence in the cost to the health service there is a perceived bene®t by the patient. The need to wear an immobilizing cast for 10 days can be of economic signi®cance to the patient. Several self-employed manual workers feared inability to work and loss of earnings if encumbered by a cast. One general practitioner who had just started a new appointment was very relieved not to have to wear a cast since this could have precluded being able to perform all of his on-call duties. Previous authors have commented on the ®nding of soft tissue injuries, particularly ligamentous disruption [20]. Such injuries have not been evaluated in this study as they were identi®ed in only a very few cases. This may be a consequence of using a low ®eld MR system. Indeed, Bretlau et al. [8], using the same machine, did not comment on the presence or absence of ligament injuries although Lepisto et al. [17], using a low ®eld system, did ®nd several cases of ligamentous and triangular ®brocartilage disruption. Thus, clinical evaluation will still have a role to play in the evaluation and follow-up of patients with signi®cant clinical ®ndings despite a negative MRI. Nevertheless, as active intervention is only rarely indicated then failure to detect ligamentous injuries may have important clinical consequences in only a very small number of patients. It is likely that it would remain cost-eective to discharge those patients with negative MR ®ndings with an open invitation to review those with persistent symptoms. These results are in accordance with previous authors who have shown that a signi®cant number of patients have a fracture other than a fracture of the scaphoid [8,11]. This is most often a fracture of the distal radius although rarely other carpal bones are involved. It is postulated that in the past this latter group of fractures in particular were never detected radiologically and that they resolved after a period of time. Since almost by de®nition these fractures are undisplaced and do not have a propensity to develop either non-union or avascular necrosis then it is suspected that they may account for some cases of persistent pain which ®nally resolves with normal radiographs throughout. This scenario may also account for some of those patients with increased isotope uptake on bone scintigrams but no evidence of scaphoid fracture on subsequent radiographs. It was initially thought that the number of patients that would require an MRI would be relatively large and hence it was only once the dedicated MR system was in situ that the study commenced. It is now apparent however that the number of patients involved is far fewer than imagined. On average 25±30 patients per month (or less than one per day) are examined. The departmental radiology computer system records requests for a scaphoid series of radiographs separately from requests for radiographs of the wrist. A review of these records has shown that this is the level of requesting for scaphoid injuries that has always existed. There has not yet been any increase in referral rates following the availability of MRI. This level of activity could be managed in any department with a static on-site
320
CLINICAL RADIOLOGY
MR system. There is one slot allocated in the morning to the Accident and Emergency Department for a scaphoid examination. Any patient seen in the preceding 24 h is allocated to this slot and asked to return the following day. The result is given to the Accident and Emergency medical sta and the patient managed accordingly. It can thus be seen that MRI is a practical proposition for the routine investigation of these patients. The dedicated MR system allows even elderly or in®rm patients to be scanned since they merely have to lie supine with their arm abducted. Using a conventional MR machine these patients may not be able to adopt the `superman' position often used to examine the wrist. MR refusal due to claustrophobia is not encountered with the extremity scanner. This study has shown that examination of all patients with a suspected scaphoid fracture using MRI is a realistic proposition. The optimum time to examine the patient by MR is early after injury. This is now the standard practice in my department, where almost all cases of clinically suspected scaphoid fracture but normal radiographs are examined with MRI within the ®rst few days of injury. Acknowledgements. I would like to thank Mr P. Grant, Mr W. Tullett and the sta of the Accident and Emergency Department for their assistance in referring and assessing patients during this study. REFERENCES 1 Abdel-Salam A, Eyres KS, Cleary J. Detecting fractures of the scaphoid: the value of comparative X-rays of the uninjured wrist. J Hand Surg (Br) 1992;17B:28±32. 2 Gaebler C, Kukla C, Breitenseher MJ, Mrkonjic L, Kainberger F, Vecsei V. Limited diagnostic value of macroradiography in suspected scaphoid fractures. Acta Orthop Scand 1998;69:401±403. 3 Tiel-Van Buul M, Van Beek E, Borm J, et al. The value of radiographs and bone scintigraphy in suspected scaphoid fracture. J Hand Surg (Br) 1993;18B:403±406. 4 Jonsson K, Jonsson A, Sloth M, Kopylov P, Wingstrand H. CT of the wrist in suspected scaphoid fracture. Acta Radiol 1992;33: 500±501.
5 Hodgkinson DW, Nicholson DA, Stewart G, Sheridan M, Highs P. Scaphoid fracture: a new method of assessment. Clin Radiol 1993;48:398±401. 6 Breitenseher MJ, Metz VM, Gilula LA, et al. Radiographically occult scaphoid fractures: value of MR imaging in detection. Radiology 1997;203:245±250. 7 Fowler C, Sullivan B, Williams LA, McCarthy G, Savage R, Palmer A. A comparison of bone scintigraphy and MRI in the early diagnosis of the occult scaphoid waist fracture. Skeletal Radiol 1998;27:683±687. 8 Bretlau T, Christensen OM, Edstrom P, Thomsen HS, Lausten GS. Diagnosis of scaphoid fracture and dedicated extremity MRI. Acta Orthop Scand 1999;70:504±508. 9 Waizenegger M, Barton NJ, Davis TR, Wastie ML. Clinical signs in scaphoid fractures. J Hand Surg (Br) 1994;19B:743±747. 10 Esberger DA. What value the scaphoid compression test. J Hand Surg (Br) 1994;19B:748±749. 11 Hunter J, Escobedo E, Wilson A, et al. MR imaging of clinically suspected scaphoid fractures. Am J Roentgenol 1997;168:1287± 1293. 12 Barton NJ. Twenty questions about scaphoid fractures. J Hand Surg (Br) 1992;17:289±310. 13 Meyers S, Wiener S. Magnetic resonance imaging features of fractures using the STIR sequence : correlation with radiographic ®ndings. Skeletal Radiol 1991;20:499±507. 14 Gaebler C, Kukla C, Breitenseher M, et al. Magnetic resonance imaging of occult scaphoid fractures. J Trauma 1996;41:73±76. 15 Van Gelderen W, Gale RS, Steward AH. Short tau inversion recovery magnetic resonance imaging in occult scaphoid injuries: eect on management. Australas Radiol 1998;42:20±24. 16 Imaeda T, Nakamura R, Miura T, Makino N. Magnetic resonance imaging in scaphoid fractures. J Hand Surg (Br) 1992;17:20±27. 17 Lepisto J, Mattila K, Nieminen S, et al. Low ®eld MRI and scaphoid fracture. J Hand Surg (Br) 1995;20B:539±542. 18 Murphy DG, Eisenhauer MA, Powe J, Pavlofsky W. Can a day 4 bone scan accurately determine the presence or absence of scaphoid fracture?. Ann Emerg Med 1995;26:434±438. 19 Waizenegger M, Waistie ML, Barton NJ, Davis TR. Scintigraphy in the evaluation of the `clinical' scaphoid. J Hand Surg (Br) 1994;19:750±753. 20 Thorpe A, Murray A, Smith F, Ferguson J. Clinically suspected scaphoid fracture: a comparison of magnetic resonance imaging and bone scintigraphy. Br J Radiol 1966;69:109±113. 21 Kukla C, Gaebler C, Breitenseher M, Trattnig S, Vicsei V. Occult fractures of the scaphoid: the diagnostic usefulness and indirect economic repercussions of radiography versus magnetic resonance scanning. J Hand Surg (Br) 1997;22:810±813.
Magnetic Resonance Imaging of Suspected Scaphoid Fractures Using a Low Field Dedicated Extremity MR System NIGEL RABY Department of Radiology, Western In®rmary, North Glasgow Hospitals NHS Trust, Glasgow, U.K. Received: 31 May 2000
Revised: 11 September 2000 Accepted: 2 October 2000
AIM: To determine whether magnetic resonance imaging (MRI) using a dedicated extremity system could be utilized for the routine assessment of patients with a clinically suspected scaphoid fracture, and to determine the most cost-eective timing of MRI for these patients. MATERIALS AND METHODS: MRI of the wrist using a 0.2T extremity MR system was performed in two groups of patients. Fifty-six patients were examined within 4 days of injury and 53 with persistent symptoms at 10 days to 6 weeks following injury. A scaphoid series of radiographs was negative in both groups. RESULTS: The early MR group had seven scaphoid, six radial and four other fractures. Management was altered in 89%. The late MR group had 14 scaphoid, nine radial and three other fractures. Management was altered in 69%. A cost model showed that overall costs are less with early rather than late scanning. CONCLUSION: MRI of the wrist when scaphoid fracture is suspected can be undertaken in all patients with negative radiographs and could be performed in most departments with an MRI machine. There are signi®cant patient bene®ts and overall costs would change little from conventional practice. Raby, N. # 2001 The Royal College of Radiologists (2001) Clinical Radiology 56, 316±320. Key words: MR imaging, scaphoid, wrist.
The diagnosis and management of scaphoid fractures remains controversial. There are several hundred papers in the literature on this subject over the past two decades demonstrating the level of continuing debate over this seemingly well-understood clinical problem. Radiological diagnosis of scaphoid fractures remains problematic. Many techniques have been employed to demonstrate scaphoid fractures including the use of comparative views [1], macroradiography [2], isotope bone scintigrams [3], computed tomography (CT) [4], ultrasound [5] and magnetic resonance imaging (MRI). MRI in particular has been shown to be an eective method of diagnosing fractures of the scaphoid [6] but studies published to date have involved relatively small numbers of patients and have been essentially pilot studies. Most have emphasized the usefulness of the technique but have also underlined that clinical application is dicult due to limited access to MRI. In this study two groups of patients were studied with a dedicated extremity MR system located close to the Accident and Emergency Department using MRI as a routine investigation. Author for correspondence and guarantor of study: Dr Nigel Raby, Department of Radiology, Western In®rmary, Dumbarton Road, Glasgow G11 6NT, U.K. Fax: 0141 337 3441; E-mail: N.Raby@ clinmed.gla.ac.uk 0009-9260/01/040316+05 $35.00/0
MATERIALS AND METHODS
Two groups of patients with a history of recent wrist injury were identi®ed. All patients studied were consecutive referrals. Group 1 consisted of 53 patients seen at review or fracture clinic at least 10 days post-injury (range 10 days±6 weeks), some of whom had attended the Accident and Emergency Department at the time of injury. Radiographic examination had not demonstrated a fracture but the patient had been placed in a scaphoid cast due to clinical suspicion of a scaphoid injury. The remainder had not attended hospital initially but had been referred for assessment due to continuing wrist pain. All patients in this group had persistent wrist pain with clinical ®ndings consistent with a scaphoid injury and normal radiographs at the time of clinic attendance. Group 2 consisted of 56 patients with an acute wrist injury attending the Accident and Emergency Department for the ®rst time usually on the day of injury (range 0±4 days). A scaphoid fracture was suspected clinically due to the presence of tenderness in the anatomical snubox but a scaphoid series of radiographs failed to conclusively demonstrate a scaphoid fracture. Patients with a fracture of the scaphoid or any other bone identi®ed on plain radiogaph, were excluded from the study. # 2001 The Royal College of Radiologists
317
MRI OF SUSPECTED SCAPHOID FRACTURES
To avoid any diculties in subsequent data analysis patients were ®rst referred according to the criteria for group 1. When the cohort of 53 patients had been acquired then patients from group 2 were studied. MRI examinations were performed on a 0.2T dedicated extremity system (Artsocan, Esaote, Genoa, Italy). Coronal images only were obtained using the following parameters: Spin echo T1 (TR 450, TE 24, three acquisitions, 3.0 mm slice thickness, 16 cm ®eld of view, matrix 192 192 acquisition time 4.39 min) and STIR (TR 1320, TE 30, TI 70, three acquisitions 3.0 mm slice thickness 16 cm ®eld of view matrix 192 160 acquisition time 6.43 min). A fracture was diagnosed when there was a linear area of low signal traversing bone from cortex to cortex on a T1 weighted sequence with a corresponding area of high signal on the STIR sequence. Areas of diuse signal change with no de®ned linear component on either sequence were designated bone bruises. This is in accordance with previous studies [7,8]. The distal radius and ulna, all the carpal bones and proximal metacarpals were scrutinized for these changes. Patients referred for MR had a form completed by the referring clinician. This asked for details of the clinical ®ndings and interpretation of the radiographs. They were also asked how they would manage the patient on the basis of the clinical and radiological ®ndings. Following MRI the patient returned immediately to the referring clinician with a written report of the ®ndings. This was hand-delivered by the radiologist and the ®ndings discussed if necessary. The clinician was then asked to write on the second part of the form how the patient would now be managed in light of the MRI ®ndings. Patients with a fracture of the scaphoid or any other bone (most commonly the distal radius) were managed and reviewed as per the standard practice for the department. No attempt was made to chart the progress of these patients. A questionnaire was sent to patients from group 2, if they had been discharged following normal MRI, 2±3 months after the initial injury asking if they were now pain-free, how long symptoms had taken to resolve and if they had attended any other practitioner regarding their wrist injury. They were also invited to re-attend for clinical review if they had any persisting symptoms.
RESULTS
MRI Findings In group 1 (Table 1) there were 26 (50%) patients with a fracture demonstrated by MRI. These were all treated in the conventional manner with cast appropriate to the fracture type. Seven patients with bone bruising were treated with a soft splint for symptomatic relief. One was discharged with an invitation to return if symptoms persisted. This patient has not reattended. The other six patients had a soft splint applied for symptomatic relief. These patients were reviewed at a follow-up clinic and all were subsequently discharged pain-free. Of the 20 patients with normal scans
Table 1 ± Patients in Group 1. MRI at 10 days±6 weeks Scaphoid fracture Distal radius fracture Other fracture Bone bruise
14 9 3 7
Normal
20
1 1 1 4 2 1
hamate trapezoid triquetral scaphoid radius hamate
Table 2 ± Patients in Group 2. MRI at 0±4 days post-injury Scaphoid fracture Distal radius fracture Other fracture
7 6 4
Bone bruise
4
Normal
35
1 1 1 1 2 1 1
triquetral trapezoid base of 5th metacarpal ulna styloid scaphoid radius ulna
all were given a soft splint for symptomatic relief. Ten were discharged but invited to return if symptoms did not resolve. None of these patients has reattended. The other 10 patients were kept under review due to clinical concern despite the normal MRI. All of these patients were subsequently discharged pain-free. In group 2 (Table 2), 17 patients (28%) with fractures demonstrated by MRI were placed in the appropriate cast and were reviewed clinically until symptom-free. Four patients with a bone bruise were asked to return for review at 10 days. Three patients did not attend; the fourth patient was subsequently discharged symptom-free. Thirty-one patients with normal MRI were placed in a supportive splint and discharged from further review but with an invitation to return if symptoms did not resolve. None of these patients have re-attended the department. Four patients with normal MRI were asked to return for review at 10 days due to clinical concern about their symptoms. All were pain-free at review and were discharged.
Eect on Patient Management In group 1 without the MRI the management decision was to place 49 of the patients in a scaphoid cast and to review with further X-rays in 10±14 days. Four patients were to have a soft splint and further clinical review. Following the MRI result 26 patients had a de®nitive diagnosis of a fracture. Fourteen had the wrist placed in a scaphoid cast; the rest were placed in a cast appropriate to the fracture type, mostly distal radius. The remaining 27 patients wrists were placed in a soft splint. Seventeen were kept under clinical review and 10 were discharged. Included in these ®gures are the four patients who were to have a soft splint and review. All had a normal MRI. Two were
318
CLINICAL RADIOLOGY
discharged and two retained for further review as previously planned. Thus, there was a change in management in 37 of the 53 patients (69.8%) In group 2 the management plan without MRI was to place all 56 patients in a scaphoid cast and review with radiographs at 10 days. Following MRI 31 patients were discharged with a soft splint, eight were placed in a soft splint and reviewed at a follow up clinic, 10 were immobilized in a cast other than a scaphoid type. A scaphoid cast was applied to the remaining seven patients with a scaphoid fracture demonstrated on MRI. Thus, management was altered in 49 of the 56 patients in this group (87.5%) Questionnaires were sent to 25 patients from group 2 who were discharged. Only ®ve of these were returned. None of these patients reported persistent wrist pain and none had sought advice from either their general practitioner or another Accident and Emergency Department. Due to the poor patient response no further questionnaires were distributed. DISCUSSION
The clinical diagnosis of scaphoid fractures remains unreliable [9,10]. Plain radiographs will fail to detect a number of fractures at initial presentation ÿ16% in a recent analysis of several papers [11]. The long-term complication of avascular necrosis of the proximal pole of the scaphoid following a fracture of the scaphoid waist is well recognized, as is non-union of the fracture. It is for this reason that so much eort is made to identify these injuries even although it is appreciated that these complications occur in only a minority of patients. Estimates of both vary considerably from as low as 5% to as high as 50%. As Barton [12] states `we overtreat a lot of patients to avoid undertreating a few'. MRI has been shown to be a reliable method for detection of fractures at several sites [13]. In particular, several studies have shown that MRI will accurately detect scaphoid fractures [6,11,14]. However, MRI is not yet routinely used in the investigation of these patients largely due to diculties of MR availability and cost. With the advent of relatively cheap dedicated extremity MR systems these problems can be overcome. Such an MR system has been installed in the orthopaedic clinic close to the Accident and Emergency Department. This MR system has time allocated for an MR system examining of patients with clinically suspected scaphoid injuries. How soon can MRI detect a fracture? Previous studies suggest that a fracture will be detected within hours of injury. Meyers and Wiener [13] detected six fractures at several sites less than 48 h after injury; Breitenseher et al. [6] detected 14 scaphoid fractures examined within 7 days of injury (mean 3.6 days, one patient examined at 25 h). A patient has been examined in this department (not within this study group) at 4 h after injury with a positive result. However because data is limited, follow-up of the MRnegative patients who presented within 4 days of injury and who were discharged was attempted by means of a questionnaire. Response to this was very poor; it is hoped
this means that the patients are symptom-free but this cannot be stated with any certainty. However, no patient reattended following discharge despite an open invitation to do so should symptoms not resolve. Since fracture detection beyond 7 days is well proven no attempt was made to follow up those patients in the ®rst group who were discharged following a normal MR at 10 days or more post-injury. The use of coronal T1 and STIR sequences has been advocated and tested by previous authors. Breitenseher et al. [6] evaluated T1, T2*GE and STIR sequences in the coronal plane and concluded that the combination of T1 and STIR sequences provided the highest sensitivity for fracture detection. Van Gelderen et al. [15] also used coronal T1 and STIR sequences and drew similar conclusions. Although many authors have studied MRI of scaphoid fractures using high ®eld strength magnets [16,11] the use of low ®eld magnets has also been tested. Lepisto et al. [17] studied 18 patients demonstrating fractures in 11 on a 0.1T unit. Van Gelderen et al. [15], using a 0.2T system, examined 16 patients. Bretlau et al. [8], using the same scanner as in the current study, examined 52 patients between 2 and 10 days following trauma using follow-up radiographs as the gold standard. Nine scaphoid fractures and six distal radial fractures were detected by MRI. All were subsequently con®rmed on the plain radiographs. There is therefore no apparent dierence in the ability of low and high ®eld MR systems to detect scaphoid fractures. The accuracy of MRI has been reported to have sensitivity and speci®city as high as 100% [6,11,14]. Breitenseher et al. [6] evaluated 42 patients with suspected clinical scaphoid fracture using MRI. The gold standard was plain radiographs obtained at 6 weeks postinjury. Fourteen scaphoid fractures and seven other fractures were detected by MRI and subsequently con®rmed on plain radiograph. The sensitivity and speci®city was 100%; there were thus no false-positives and, more importantly, no false negatives. Hunter et al. [11] examined 36 patients within 7 days of injury, ®nding 13 scaphoid and nine distal radial fractures. Clinical follow-up to pain-free discharge of those with normal MR ®ndings con®rmed that a negative MRI examination excludes a fracture. There were no false-negative examinations in this study. Several other techniques have been employed to detect scaphoid fractures. The most commonly used is bone scintigraphy, which is advocated by some as the investigation of choice [3]. There is little doubt that this is a sensitive technique. A negative isotope study reliably excludes a fracture but a positive scintigram is non-speci®c [18]. Waizenegger et al. [19] found increased isotope uptake in the scaphoid in 19 patients but could only ®nd con®rmatory evidence on late plain radiographs or CT in seven. Fowler et al. [7] compared MRI and scintigraphy in the same patients and found that MRI was more sensitive and speci®c than isotope scanning. Thorpe et al. [20] found MRI and scintigraphy able to detect scaphoid fractures with equal accuracy.
MRI OF SUSPECTED SCAPHOID FRACTURES
In the current study, patients examined by MR early after initial presentation were most commonly found to have normal ®ndings (35/56) with only 17 fractures and four cases of bruising. Those examined at more than 10 days after injury produced more positive MR ®ndings (26/53), with bruising in seven and only 20 normal. This is logical since it is known that the majority of patients do not have a signi®cant injury. After some time has elapsed clinical examination will be normal in many of these patients and they would be discharged. As only those patients who remained symptomatic were sent for MRI these are a selected group in whom a higher proportion of signi®cant injuries are likely to be found. This is after all the rationale for continuing clinical evaluation as a means of identifying those patients with the highest likelihood of an occult fracture. It is thus tempting to conclude that the most costeective time to examine patients is in the group with symptoms persisting 10 days or more after injury. However the following analysis shows that this is not necessarily the case. The cost of MR with this machine is approximately £50.00 per examination. This includes sta and consumable costs. Patients with clinical scaphoid injuries but normal radiographs often require to be reviewed several times before ®nal discharge. Kukla et al. [21] found that these patients re-attended for an average of 2.5 times. Speci®c data on this is not available from this department but if it is assumed that this is a fair estimate then an attempt at relative costing can be made. This can be applied to both of the patient groups to see at which time it is most ®nancially advantageous to scan patients. This hospital is unable to give a cost for attendance at a fracture review clinic. Although some patients with a normal scan or bone bruising were asked to return for clinical review all were subsequently discharged pain-free. With this knowledge in future it would be possible to discharge all of these patients after their MR examination. With these assumptions the following calculations can be applied to the two groups. In group 1, 53 patients had MRI at £50.00 each, incurring an additional total cost of £2650. Fractures were demonstrated in 26 patients. Assuming these would eventually become apparent on conventional radiographs then management is unlikely to be signi®cantly changed and there would be no dierence in the cost of managing these patients. Twenty-seven patients with normal MR or bone bruising only could be discharged from the clinic. They would previously have attended on average a further 2.5 times. Thus, a total of 65 out-patient attendances could be avoided. To remain neutral in terms of overall cost, each clinic attendance would have to have cost £40.00. If it is more than this savings occur, if less then overall costs are increased. In group 2, 56 patients have MRI costing £2800. Seventeen patients with fractures are cost-neutral. Thirty-nine patients with normal MR or bone bruising could be discharged, avoiding 97.5 attendances. The price of a clinic attendance to remain cost-neutral would be less than £30.00. Using this model it is cheaper to examine all
319
patients with MR early rather than selected patients after 10 days. It is also in this group that MRI most often results in a change of management. While there may be little dierence in the cost to the health service there is a perceived bene®t by the patient. The need to wear an immobilizing cast for 10 days can be of economic signi®cance to the patient. Several self-employed manual workers feared inability to work and loss of earnings if encumbered by a cast. One general practitioner who had just started a new appointment was very relieved not to have to wear a cast since this could have precluded being able to perform all of his on-call duties. Previous authors have commented on the ®nding of soft tissue injuries, particularly ligamentous disruption [20]. Such injuries have not been evaluated in this study as they were identi®ed in only a very few cases. This may be a consequence of using a low ®eld MR system. Indeed, Bretlau et al. [8], using the same machine, did not comment on the presence or absence of ligament injuries although Lepisto et al. [17], using a low ®eld system, did ®nd several cases of ligamentous and triangular ®brocartilage disruption. Thus, clinical evaluation will still have a role to play in the evaluation and follow-up of patients with signi®cant clinical ®ndings despite a negative MRI. Nevertheless, as active intervention is only rarely indicated then failure to detect ligamentous injuries may have important clinical consequences in only a very small number of patients. It is likely that it would remain cost-eective to discharge those patients with negative MR ®ndings with an open invitation to review those with persistent symptoms. These results are in accordance with previous authors who have shown that a signi®cant number of patients have a fracture other than a fracture of the scaphoid [8,11]. This is most often a fracture of the distal radius although rarely other carpal bones are involved. It is postulated that in the past this latter group of fractures in particular were never detected radiologically and that they resolved after a period of time. Since almost by de®nition these fractures are undisplaced and do not have a propensity to develop either non-union or avascular necrosis then it is suspected that they may account for some cases of persistent pain which ®nally resolves with normal radiographs throughout. This scenario may also account for some of those patients with increased isotope uptake on bone scintigrams but no evidence of scaphoid fracture on subsequent radiographs. It was initially thought that the number of patients that would require an MRI would be relatively large and hence it was only once the dedicated MR system was in situ that the study commenced. It is now apparent however that the number of patients involved is far fewer than imagined. On average 25±30 patients per month (or less than one per day) are examined. The departmental radiology computer system records requests for a scaphoid series of radiographs separately from requests for radiographs of the wrist. A review of these records has shown that this is the level of requesting for scaphoid injuries that has always existed. There has not yet been any increase in referral rates following the availability of MRI. This level of activity could be managed in any department with a static on-site
320
CLINICAL RADIOLOGY
MR system. There is one slot allocated in the morning to the Accident and Emergency Department for a scaphoid examination. Any patient seen in the preceding 24 h is allocated to this slot and asked to return the following day. The result is given to the Accident and Emergency medical sta and the patient managed accordingly. It can thus be seen that MRI is a practical proposition for the routine investigation of these patients. The dedicated MR system allows even elderly or in®rm patients to be scanned since they merely have to lie supine with their arm abducted. Using a conventional MR machine these patients may not be able to adopt the `superman' position often used to examine the wrist. MR refusal due to claustrophobia is not encountered with the extremity scanner. This study has shown that examination of all patients with a suspected scaphoid fracture using MRI is a realistic proposition. The optimum time to examine the patient by MR is early after injury. This is now the standard practice in my department, where almost all cases of clinically suspected scaphoid fracture but normal radiographs are examined with MRI within the ®rst few days of injury. Acknowledgements. I would like to thank Mr P. Grant, Mr W. Tullett and the sta of the Accident and Emergency Department for their assistance in referring and assessing patients during this study. REFERENCES 1 Abdel-Salam A, Eyres KS, Cleary J. Detecting fractures of the scaphoid: the value of comparative X-rays of the uninjured wrist. J Hand Surg (Br) 1992;17B:28±32. 2 Gaebler C, Kukla C, Breitenseher MJ, Mrkonjic L, Kainberger F, Vecsei V. Limited diagnostic value of macroradiography in suspected scaphoid fractures. Acta Orthop Scand 1998;69:401±403. 3 Tiel-Van Buul M, Van Beek E, Borm J, et al. The value of radiographs and bone scintigraphy in suspected scaphoid fracture. J Hand Surg (Br) 1993;18B:403±406. 4 Jonsson K, Jonsson A, Sloth M, Kopylov P, Wingstrand H. CT of the wrist in suspected scaphoid fracture. Acta Radiol 1992;33: 500±501.
5 Hodgkinson DW, Nicholson DA, Stewart G, Sheridan M, Highs P. Scaphoid fracture: a new method of assessment. Clin Radiol 1993;48:398±401. 6 Breitenseher MJ, Metz VM, Gilula LA, et al. Radiographically occult scaphoid fractures: value of MR imaging in detection. Radiology 1997;203:245±250. 7 Fowler C, Sullivan B, Williams LA, McCarthy G, Savage R, Palmer A. A comparison of bone scintigraphy and MRI in the early diagnosis of the occult scaphoid waist fracture. Skeletal Radiol 1998;27:683±687. 8 Bretlau T, Christensen OM, Edstrom P, Thomsen HS, Lausten GS. Diagnosis of scaphoid fracture and dedicated extremity MRI. Acta Orthop Scand 1999;70:504±508. 9 Waizenegger M, Barton NJ, Davis TR, Wastie ML. Clinical signs in scaphoid fractures. J Hand Surg (Br) 1994;19B:743±747. 10 Esberger DA. What value the scaphoid compression test. J Hand Surg (Br) 1994;19B:748±749. 11 Hunter J, Escobedo E, Wilson A, et al. MR imaging of clinically suspected scaphoid fractures. Am J Roentgenol 1997;168:1287± 1293. 12 Barton NJ. Twenty questions about scaphoid fractures. J Hand Surg (Br) 1992;17:289±310. 13 Meyers S, Wiener S. Magnetic resonance imaging features of fractures using the STIR sequence : correlation with radiographic ®ndings. Skeletal Radiol 1991;20:499±507. 14 Gaebler C, Kukla C, Breitenseher M, et al. Magnetic resonance imaging of occult scaphoid fractures. J Trauma 1996;41:73±76. 15 Van Gelderen W, Gale RS, Steward AH. Short tau inversion recovery magnetic resonance imaging in occult scaphoid injuries: eect on management. Australas Radiol 1998;42:20±24. 16 Imaeda T, Nakamura R, Miura T, Makino N. Magnetic resonance imaging in scaphoid fractures. J Hand Surg (Br) 1992;17:20±27. 17 Lepisto J, Mattila K, Nieminen S, et al. Low ®eld MRI and scaphoid fracture. J Hand Surg (Br) 1995;20B:539±542. 18 Murphy DG, Eisenhauer MA, Powe J, Pavlofsky W. Can a day 4 bone scan accurately determine the presence or absence of scaphoid fracture?. Ann Emerg Med 1995;26:434±438. 19 Waizenegger M, Waistie ML, Barton NJ, Davis TR. Scintigraphy in the evaluation of the `clinical' scaphoid. J Hand Surg (Br) 1994;19:750±753. 20 Thorpe A, Murray A, Smith F, Ferguson J. Clinically suspected scaphoid fracture: a comparison of magnetic resonance imaging and bone scintigraphy. Br J Radiol 1966;69:109±113. 21 Kukla C, Gaebler C, Breitenseher M, Trattnig S, Vicsei V. Occult fractures of the scaphoid: the diagnostic usefulness and indirect economic repercussions of radiography versus magnetic resonance scanning. J Hand Surg (Br) 1997;22:810±813.