Fractures of the distal part of the radius in younger adults are usually the result of high-energy trauma and have intra-articular involvement. The goal of treatment should be anatomical reduction and stable fixation in order to minimize the risk of posttraumatic arthritis. At the present time, there are two generally accepted methods of fixation. Before the advent of volar fixed-angle locking screw and plate systems, the more popular method of treatment was external fixation. External fixation is often supplemented with percutaneous fixation with Kirschner wires to maintain the reduction of the articular fragments1. However, the reduction achieved indirectly by ligamentotaxis can be variable and is determined largely by the degree of comminution. Complications include pin-track problems, loss of reduction, complex regional pain syndrome, and finger stiffness1,2.
Open reduction and plate fixation of comminuted intra-articular fractures of the distal part of the radius did not have the same popularity as external fixation in the past. The high rates of complications and the great surgical difficulty of achieving stable fixation have discouraged some orthopaedic surgeons from using this method3,4. Nevertheless, there have been more recent attempts to evaluate the use of plating for the treatment of distal radial fractures5,6. A surgical technique involving the application of conventional T-plates on both the dorsal and volar aspects of the distal part of the radius has also been proposed7.
The ideal method of fracture fixation should produce good results and be able to be mastered by most orthopaedic surgeons. So far, most studies have evaluated limited numbers of fractures that have been treated in a variety of ways. The purpose of the present study was to test the hypothesis that external fixation combined with percutaneous pin fixation is a better method than open reduction and plate fixation for the treatment of intra-articular fractures of the distal part of the radius in young adults. The two treatment groups were compared with use of standardized clinical and radiographic measures.
We conducted a multicenter prospective randomized controlled trial involving three hospitals, including Queen Mary Hospital in Hong Kong, Tan Tock Seng Hospital in Singapore, and Chang Gung Memorial Hospital in Taiwan. The study was approved by the medical ethics review board at each center. Written informed consent was obtained from each patient.
All patients between sixteen and sixty years of age with an acute intra-articular AO group-C1, C2, or C3 distal radial fracture (i.e., a complete articular fracture) were considered for inclusion in the study. Patients with pathological fractures and those with a history of premature osteoporosis, drug abuse, or alcohol abuse were excluded. Patients with open fractures (Gustilo and Anderson8 type-II or III fractures) and patients who presented more than eight hours after the injury were also excluded. Patients who were unlikely to cooperate or to attend all scheduled follow-up visits, such as migrant workers, were not recruited.
One attempt at closed reduction was performed, and patients who had an unacceptable reduction were recruited into the study. The criteria for an acceptable closed reduction included dorsal tilt of =10°, volar tilt of =20°, an articular gap or step of =2 mm, radial inclination of =10°, and radial shortening of =5 mm.
Between March 2002 and March 2005, 137 patients who had a total of 144 complete intra-articular distal radial fractures were recruited for the study. The study group included eighty-five male patients (62%) and fifty-two female patients (38%) with a mean age of forty-two years (range, seventeen to sixty years). Sixty-eight fractures (47%) occurred in the right wrist, and seventy-six (53%) occurred in the left wrist. Seventy-four fractures (51%) involved the nondominant side, and seventy (49%) involved the dominant side, with seven patients having a bilateral fracture.
Among the 144 fractures, fifty-nine (41%) were the result of low-energy trauma and eighty-five (59%) were the result of high-energy trauma. Sixty-six fractures (46%) involved the ulnar styloid. Associated injuries included one ipsilateral elbow dislocation, one scaphoid fracture, one contralateral fracture of the surgical neck of the humerus, and one contralateral fracture of the lateral humeral condyle. Fifty-six fractures resulted from a fall from a height below the level of the head, seventeen resulted from a fall from a height above the level of the head, sixty-seven resulted from a motor-vehicle accident, two resulted from an assault, and two resulted from a crush injury. Eleven fractures were classified as grade-I open fractures according to the classification system of Gustilo and Anderson8.
The AO classification system has been reported to have low reproducibility and interobserver reliability, especially for the categorization of AO groups (C1, C2, and C3) and subgroups (C1.1 through C3.3)9. For the classification of type-C (complete articular) fractures, we adopted the guidelines of site and degree of comminution as follows. C1 represented fractures that were simple in both the articular and metaphyseal regions, C2 represented those that were simple in the articular region and comminuted in the metaphyseal region, and C3 represented those that were comminuted in the articular region.
All fractures were classified by two of us (F.L. and Y.T.). Using this simple guideline, we were able to achieve an interobserver agreement of 0.78 and an intraobserver agreement of 0.84. There were thirty-six C1 fractures, fifty C2 fractures, and fifty-eight C3 fractures (Table I).
Surgical procedures were performed after randomization. A permuted block randomization method was adopted with blocks of four allocations, with each block containing two external fixation and two plate fixation procedures in random order. This allocation was based on a computer-generated list from an independent research assistant. Permuted block randomization ensures that treatment group numbers are evenly balanced at the completion of each block10. Each clinic could only recruit patient numbers in multiples of four. The allocation was conveyed to the clinic by fax, and the principal investigator (F.L.) was responsible for allocation concealment and monitoring of the randomization process.
All procedures were performed within the first week after the fracture, with the exception of two procedures that were performed within ten days after the injury. All patients received prophylactic antibiotics. Seventy-four fractures (51%) were stabilized with external fixators combined with percutaneous pins, and the other seventy fractures (49%) were stabilized with plates. Ulnar styloid fractures were treated nonoperatively. In both groups, the quality of reduction was assessed intraoperatively under image intensification. A successful reduction was defined as dorsal tilt of =10°, volar tilt of =20°, an articular gap or step of =2 mm, radial inclination of =10°, and radial shortening of =5 mm.
Open Reduction and Plate Fixation
Surgical exposure of the fracture was performed through a volar or dorsal incision, depending on the extent of displacement and comminution of the fracture. A combined volar and dorsal approach was used in cases in which a satisfactory reduction was not achieved through a single approach. When metaphyseal support of the articular fragments was compromised by comminution, autogenous cancellous bone graft was used to support the articular fragments. Conventional, nonlocking stainless steel 3.5-mm T-plates (Synthes, Bettlach, Switzerland) were used, and combined volar and dorsal plate fixation was done for more comminuted fractures.
External Fixation and Percutaneous Kirschner Wire Fixation
A small AO/ASIF external fixator (Synthes) was used. Two half-pins were inserted in the second metacarpal through stab incisions, and two pins were placed in the radial aspect of the shaft of the radius through a limited incision while avoiding the extensor tendons and the sensory branches of the radial nerve. Reduction was achieved with ligamentotaxis and percutaneous fracture fragment manipulation with Kirschner wires. With traction applied, the wrist was put in either extension or flexion, depending on the fracture pattern. For impacted die-punch fragments that could not be reduced percutaneously, a limited dorsal incision was made and a small pointed awl was used to achieve a direct reduction11. Cancellous bone from the iliac crest was applied as a graft to support articular fragments and to prevent them from settling in the early postoperative period. Kirschner wires were applied in most cases, usually obliquely from the radial styloid and also transversely to support the die-punch fragments. Finally, the external fixator was applied in a joint-bridging fashion.
In the external fixation group, bone graft was used for sixteen fractures (22%) and Kirschner wires were used for sixty-six fractures (89%). In the plate fixation group, forty fractures (57%) were fixed with a volar plate, twelve (17%) were fixed with a dorsal plate, and eighteen (26%) were fixed with both dorsal and volar plates. Bone graft was used in twelve cases (17%), and supplementary Kirschner wires were applied in thirty cases (43%).
Postoperative Aftercare
After plate fixation, a plaster slab was applied for two to three weeks. All patients received supervised physiotherapy, including early finger and elbow mobilization and wrist range-of-motion exercises at the time of cast or external fixation removal. Removal of dorsal plates was routinely performed at six months to avoid the possibility of tendon attrition. The external fixation was removed at six weeks, or one week earlier if Kirschner wires were used.
Outcome Measurement
The patients were assessed clinically and radiographically in the early postoperative period and at six, twelve, and twenty-four months. One hundred and seventeen patients (123 fractures; 85.4 %) were available for follow-up at twelve months, and ninety-eight patients (103 fractures; 71.5%) were available for follow-up at twenty-four months (Table I). With use of the "intention-to-treat" principle, outcomes were analyzed according to the original randomization although patients may have crossed over to the other group.
Clinical assessment was performed with use of two scoring systems. The Gartland and Werley point system for Colles fractures12 is a demerit system in which the result is ranked according to point range on the basis of four parameters: residual deformity (0 to 3 points), subjective evaluation of pain and disability (0 to 6 points), range of motion (0 to 5 points), and complications (0 to 5 points), including arthritic change, median nerve complications, and finger stiffness. The modified clinical scoring system of Green and O'Brien13 was also used in the present study. This 100-point scale has four categories, with 25 points allocated equally to pain, functional status, range of motion, and grip strength.
A short questionnaire in Chinese was used to measure the patient-assessed effect of the distal radial fracture on general health and function (see Appendix). Arthritis was graded on radiographs according to the modified criteria of Knirk and Jupiter (see Appendix). The original arthritis grading according to the method of Knirk and Jupiter14 was modified to include subchondral sclerosis and scalloping of the lunate or scaphoid facets15.
An assigned physiotherapist or occupational therapist from each clinic received training on the assessment method during the planning phase of the trial and was responsible for the follow-up assessment. The clinical assessment was performed at the individual clinic by this assigned therapist, who was not blinded to the treatment as it would not have been feasible to perform an accurate assessment after masking the treatment method and as the method of fixation (e.g., external fixation) could be determined during follow-up.
All radiographs were evaluated by the first two authors (F.L. and Y.T.). We were able to achieve an interobserver agreement of 0.82 and intraobserver agreement of 0.88 with regard to the modified Knirk and Jupiter criteria. The radiographic evaluation was not performed in a blinded fashion as the method of fixation could be seen on the postoperative radiographs.
Statistical Analysis
To ensure that the sample size would be powerful enough to reject the null hypothesis, the statistical power was set at 90% and the level of significance was set at 0.05. On the basis of the Gartland and Werley point system and the modified Green and O'Brien scoring system, the total sample size needed for chi-square analysis for an effect size of 0.3, representing a 30% difference in the expected proportion of outcomes between the two groups, was approximately 120. The Pearson chi-square test was used to test for difference between scores in the two fixation groups. Differences were considered to be significant when the p value was <0.05.
The two groups were comparable with regard to age, gender, the side of the injury (dominant or nondominant), and fracture type (AO classification).
According to the Gartland and Werley point system, 67% of the wrists in the plate fixation group were rated as excellent, 30% were rated as good, 3% were rated as fair, and none were rated as poor at the time of the twenty-four-month follow-up. In the external fixation group, 39% were rated as excellent, 55% were rated as good, 6% were rated as fair, and none were rated as poor at twenty-four months. This difference between groups was significant (p = 0.04) in favor of plate fixation.
When the individual AO groups were compared, plate fixation had a significantly better result than external fixation for AO group-C2 fractures (p = 0.01) at the time of the twenty-four-month follow-up (Table II). With the numbers available, the differences in groups C1 and C3 were not significant (p = 0.43 and 0.09, respectively).
According to the Green and O'Brien clinical scoring system, 55% of the wrists in the external fixation group were rated as excellent, 26% were rated as good, and 19% were rated as fair at twenty-four months. In the plate fixation group, 69% were rated as excellent, 10% were rated as good, 17% were rated as fair, and 4% were rated as poor at twenty-four months (Table III). There was no significant difference between the two groups. When the individual AO groups were compared again, there was no significant difference between the two groups.
With regard to the arthritis grade, 44% of the wrists in the plate fixation group had grade-0 arthritis, 52% had grade-1 arthritis, and 4% had grade-2 arthritis at twenty-four months. In the external fixation group, 20% of the wrists had grade-0 arthritis, 65% had grade-1 arthritis, and 15% had grade-2 arthritis at twenty-four months. Overall, plate fixation demonstrated significantly better results in terms of the arthritis grade than external fixation did (p = 0.01). When the amount of arthritis in the individual AO groups was analyzed at twenty-four months, plate fixation was significantly better than external fixation for the treatment of AO group-C2 fractures (p = 0.001) (Table IV). With the numbers available, the differences between plate fixation and external fixation were not significant for group-C1 and C3 fractures (p = 0.39 and 0.82, respectively).
With regard to subjective assessment, 92% of the patients in the plate fixation group reported that the wrist injury did not interfere or only slightly interfered with their daily activities at the time of the latest follow-up, whereas 86% of those in the external fixation group reported the same. Eighty-one percent of the patients in the plate fixation group reported that the wrist injury did not interfere or only slightly interfered with their social activities, whereas 76% of the patients in the external fixation group reported the same. Eighty-four percent of the patients in the plate fixation group reported that the wrist did not interfere or slightly interfered with their normal work, whereas only 76% of those in the external fixation group reported the same.
Thirty-three plates were removed at an average of twenty-two weeks. Thirty dorsal plates were removed according to the protocol to avoid tendon attrition. Three volar plates were removed at the request of the patient.
Complications
All fractures healed uneventfully. Loss of reduction occurred in five wrists that had been treated with plate fixation. One of these five wrists had a reoperation to remove a screw that had penetrated into the joint; another had a reoperation for revision of the fixation. Loss of reduction also occurred in five wrists that had been treated with external fixation. Four of the five wrists subsequently underwent open reduction and plate fixation. These wrists were still assessed in the external fixation group in order to follow the "intention-to-treat" principle.
Five pin-track infections and three superficial infections occurred in the external fixation group. All of the patients who had an infection received antibiotics. Two patients had surgical débridement and early removal of the external fixator. In the plate fixation group, three patients had a superficial infection. All three received antibiotics, and two had removal of implant. All fractures healed without additional complications.
In the external fixation group, three patients had development of superficial radial nerve irritation that subsided with time. One patient had posttraumatic carpal tunnel syndrome and underwent surgical release at the time of fixation. Another three patients had development of carpal tunnel syndrome postoperatively. Three patients in the external fixation group had development of a frozen shoulder. One patient had development of type-1 complex regional pain syndrome three months postoperatively. The syndrome subsided eight weeks later after treatment with a combination of an anticonvulsant and a nonsteroidal anti-inflammatory medication.
In the plate fixation group, one patient had a carpal tunnel syndrome prior to fixation. Two patients who were managed with dorsal plate fixation experienced paresthesias in the median nerve distribution. Both were managed effectively with wrist splinting in the neutral position at nighttime. One patient had paresthesias in the ulnar nerve distribution after volar plate fixation. These symptoms had subsided at the time of the six-month assessment.
One month after plate removal, one patient had another fall, which resulted in a radial shaft fracture through the most proximal screw-hole. Another patient who had been managed with dorsal plating had symptoms of extensor pollicis longus tendinitis. During removal of the implant, the tendon was noted to be attenuated.
Nonunion of the ulnar styloid occurred in three patients in the external fixation group and one patient in the plate fixation group. All nonunions were asymptomatic. One patient in the external fixation group also had a concomitant scaphoid fracture that was not identified initially and subsequently required surgical treatment of the nonunion.
Following a distal radial fracture, the attainment and maintenance of anatomical reduction of the articular surface is crucial to the preservation of wrist function. Trumble et al.16 stated that the degree to which articular step-off, gapping between fragments, and radial shortening can be improved with surgery correlates strongly with improved outcome. Hence, a treatment method that is more likely to achieve these goals will result in better function.
We have demonstrated that open reduction and plate fixation is a better way to treat intra-articular distal radial fractures than is external fixation and percutaneous fixation. At twenty-four months of follow-up, the results for the plate fixation group were significantly better than those for the external fixation group when assessed with the Gartland and Werley clinical grading system and the arthritis grade. We could not identify a difference between the two methods with use of the modified Green and O'Brien scoring system because it is purely clinical and is more stringent with regard to range of motion and strength than is the Gartland and Werley system.
There have been few prospective studies in which external fixation with percutaneous pin fixation has been compared with open reduction and plate fixation. In a multicenter randomized control trial, Kreder et al.17 noted that indirect reduction and percutaneous fixation was associated with a more rapid return of function and a better functional outcome in comparison with open reduction and internal fixation. Those authors recommended that open reduction should be preceded by an attempt at minimally invasive percutaneous reduction. In practice, as the fracture becomes more comminuted, the chance of a successful indirect reduction decreases and the need for additional procedures increases. Rogachefsky et al.18 recommended open reduction and combined internal and external fixation for the treatment of AO group-C3 fractures.
Our results differ from those of the comparative study by Kreder et al.17. First, the methods of fixation may have differed. In their plate fixation group, additional external fixation was allowed at the surgeon's discretion. Second, in their study, a dorsal approach was used primarily. In our study, only 17% of the fractures in the plate fixation group had dorsal fixation alone and no additional external fixation was used in the plate fixation group. Third, in our series, all dorsal plates were removed at six months. We believe that the limited use of dorsal plates and their early removal improved the effectiveness of the plate fixation method.
We have also demonstrated that improvement in the clinical assessment following intra-articular distal radial fractures continues for at least twenty-four months. This finding was noted in association with both the Gartland and Werley system and the modified Green and O'Brien system (Tables II and III). On the basis of this finding, we disagree with the recommendation of Kreder et al.17 that functional scores are stable after one year and that additional follow-up beyond one year is not indicated.
The present study also evaluated the different patterns of AO type-C fractures. It is interesting to note that plate fixation demonstrated significantly better clinical and radiographic results than external fixation did when used for the treatment of AO group-C2 fractures. Both types of fixation were similarly effective for the treatment of AO group-C1 fractures, which are characterized by large fracture fragments. The intrinsic difficulty of fixation of the more comminuted group-C3 fractures makes it very difficult for either plating or external fixation to achieve and maintain an accurate reduction and stable fixation.
The present study had several limitations. In both treatment groups, fixation was heterogeneous. The choice of surgical approach for plating was decided by the surgeon and included dorsal, volar, and combined approaches. The need for bone-grafting was also decided by the surgeon. In practice, it would not be possible to standardize such additional procedures that may affect the results because the determining factors such as the fracture pattern, the size of the osseous defect, and the bone quality of the patient will vary from case to case.
Another drawback of the present study is that a more contemporary and validated patient-assessed scoring system such as the Disabilities of the Arm, Shoulder and Hand (DASH)19 questionnaire was not used because a validated Chinese version of the DASH or other similar scoring system did not exist at the time of the study. Although a higher proportion of patients in the plate fixation group reported that the wrist injury had no or little effect on general health and function at the time of the latest follow-up, a statistical analysis was not possible. The use of DASH, which is a more detailed questionnaire, likely would have improved the validity of the study.
In the present study, substantial bias could have occurred both in the evaluation of clinical outcomes and radiographs. The clinical evaluation could not be blinded as the method of fixation (e.g., external fixation) could be discerned during follow-up. Similarly, the radiographic evaluation was not blinded as the method of fixation could be seen on the postoperative radiographs. Catalano et al.15 included the use of computed tomography measurements in their arthritis grading system. However, such measurements were not included in our multicenter trial because of resource limitations.
To our knowledge, the present study represents the first large-scale multicenter trial on fracture treatment in East Asia conducted in three different countries. Loss of follow-up was a major problem in all centers. Our one-year follow-up rate of >85% was satisfactory. However, our goal of a follow-up rate of 80% could not be achieved at two years. It is possible that the results could have been different if these patients were included in the final assessment.
Improvement in the design of both internal and external fixation devices has occurred recently. Locking plate fixation providing angular stability and the ease of volar application has been documented20-22. The concept of fragment-specific fixation with use of mini-plates and smaller implants is also being explored23,24. In the meantime, a nonbridging type of external fixation has also been devised with the aim of achieving early wrist motion after surgical fixation25. Clearly, future studies should compare the results of these newer fixation methods with our results.