Treatment of non-union of the scaphoid is difficult. Several screws have been developed to provide stabilization of these non-unions for long periods of time and to decrease the duration of postoperative immobilization. The double-threaded Herbert screw11 (Zimmer, Warsaw, Indiana) provides fixation of fractures of the scaphoid with both ends of the screw completely buried within the bone. However, accurate placement of the screw is difficult, despite the use of a Huene device, and compression of the screw often produces malrotation of the fragments. Screws inserted from a suboptimum position are difficult to reorient even with the aid of fluoroscopy and a second Kirschner wire placed parallel to the planned location of the screw before insertion to prevent rotation of the scaphoid fragments1. The cannulated screw (Synthes, Paoli, Pennsylvania) can be inserted in the optimum position by placement of a guide-wire before insertion, thereby avoiding the need for the Huene device. The screw also facilitates the placement of a second parallel Kirschner wire, as advocated by Fernandez.
We performed this retrospective study to determine whether the placement of the screw, alignment of the scaphoid, and type of screw were associated with the time to union and the functional result.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Department of Orthopaedics, University of Washington, Box 356500, 1959 N.E. Pacific, Seattle, Washington 98195.
We reviewed the records of forty-eight consecutive patients who had been managed with internal fixation and bone-grafting for non-union of the scaphoid that had been present for more than four months. Fourteen patients were excluded because of evidence of preoperative osteoarthrosis; injury of the contralateral hand that prevented the reporting of functional measurements on the injured side as a percentage of the values for the contralateral side; fracture of the proximal pole of the scaphoid, which necessitates a dorsal approach and prevents the use of the Huene device for insertion of the Herbert screw; concomitant ipsilateral carpal dislocation such as perilunate dislocation; carpal collapse secondary to avascular necrosis; or previous operative treatment of the non-union. Thirty-four patients were included in the study. In order to avoid the effect of a learning curve, we began our study three years after we started using screws to stabilize fractures of the scaphoid.
The patients were divided into two groups depending on the method of fixation. Group 1 consisted of sixteen patients who had been managed from 1986 through 1989 with a Herbert screw, and Group 2 consisted of eighteen patients who had been managed from 1990 through 1992 with a 3.5-millimeter cannulated AO/ASIF screw.
The average age (and standard deviation) of the twelve male and four female patients in Group 1 was 27.4 ± 6.5 years (range, seventeen to forty-eight years). Twelve patients had involvement of the right hand, and ten had involvement of the dominant hand. The average delay from the time of the injury to the operation was 13.9 ± 8.5 months (range, four to twenty-six months).
The average age of the fifteen male and three female patients in Group 2 was 28.1 ± 6.8 years (range, fifteen to forty-five years). Thirteen patients had involvement of the right hand, and twelve had involvement of the dominant hand. The average delay from the time of the injury to the operation was 13.2 ± 8.2 months (range, four to twenty-six months).
The radiographic measurements were made in a blinded fashion, and the final value was arrived at by consensus. The measurements were made on posteroanterior radiographs with the forearm in neutral rotation, lateral radiographs, and oblique radiographs with the forearm in pronation and supination and the wrist in ulnar deviation17,20,22. The radiographic appearance of the two screws was similar.
All patients had a fracture at the waist of the scaphoid that extended from the volar surface distally to the dorsal surface proximally. The fractures were classified as collapsed and unstable on the basis of an abnormal scapholunate, posteroanterior intrascaphoid, or lateral intrascaphoid angle.
The fracture gap was measured at the point of greatest separation of the fracture fragments as seen on the radiograph (Fig. 1). The magnification of the radiographs averaged 20 per cent and varied by 8 per cent, as determined from twenty randomly selected radiographs of wrists that had internal fixation that could be used as a reference marker. However, the reporting of the direct measurement from the radiographs was consistent and is in agreement with measurements in other reports14,24.
Scaphoid and intercarpal alignment was assessed by measurement of the lateral and posteroanterior intrascaphoid angles and the scapholunate angle3 (Figs. 2-A, 2-B, and 2-C). The lateral intrascaphoid angle was measured on a lateral tomogram or a direct sagittal computerized tomography image made while the patients held their hand over their head in the scanner. Reconstructed computerized tomography images were not used as they had too much distortion. The intrascaphoid angles were reproducible to within 7 degrees.
Postoperative radiographs were made at the monthly follow-up examinations. Once trabecular bridging across the fracture was suspected on the basis of the radiographs, standard tomograms or computerized tomography scans were made with use of 0.5-millimeter cuts in the sagittal and coronal planes; primary, not reconstructed, images were made. We tended to make tomograms as soon as union was suspected. Sometimes this was premature and additional tomograms were needed to confirm union. The plain radiographs suggested osseous union an average of six weeks before it was seen on the tomograms. Union was considered to be present if trabecular bridging was seen on at least two coronal or sagittal computerized tomography images.
Postoperative osteoarthrosis was graded as stage 0 (none), stage 1 (mild beaking of the radius without involvement of the radioscaphoid joint), stage 2 (narrowing of the radioscaphoid joint space), or stage 3 (loss of the radioscaphoid joint space). The number of patients who had each stage was too small for us to make meaningful comparisons among stages; therefore, we combined the patients who had stage-1, 2, or 3 osteoarthrosis into one group.
Avascular necrosis was evaluated postoperatively on the plain radiographs. Magnetic resonance images were not made. Avascular necrosis was graded as stage 0 (none), stage 1 (patchy areas of radiodensity of the proximal pole), stage 2 (involvement of the entire proximal pole), or stage 3 (avascular necrosis with carpal collapse). The tourniquet was not released intraoperatively to assess blood flow in the scaphoid fragments because this resulted in a pool of blood in the operative field. Punctate bleeding from the surfaces of the bones while the tourniquet was inflated was measured with the method of Green. Compared with magnetic resonance imaging and visualization of osseous bleeding, plain radiographs lead to an underestimation of the prevalence of avascular necrosis or partial ischemia of the bone8. However, our criteria for the evaluation of radiographs were consistent with those in other reports2,6,15,16.
The placement of the screw was assessed on the posteroanterior, lateral, and oblique radiographs. The screw was considered to be centrally placed if it was located in the central one-third of the proximal pole of the scaphoid (Figs. 3-A and 3-B). Otherwise, it was considered to be peripherally placed, even if the screw extended out of the central one-third on only one radiograph (Fig. 4).
The range of motion, grip strength, and pain were measured. Twenty-eight patients were interviewed by an examiner, who was blinded to the type of internal fixation that had been used. Six patients were examined by a local therapist and were interviewed by one of us by telephone.
The range of motion was reported as both an absolute measurement and a percentage of that on the contralateral side. Flexion and extension of the wrist as well as radio-ulnar deviation were measured postoperatively for both the injured and the contralateral side8. There was no difference between the preoperative and postoperative ranges of flexion and extension of the digits or rotation of the forearm. Maximum grip strength on the injured side was measured with a Jamar dynamometer (J. P. Marsh, Skokie, Illinois) and was reported as a percentage of the maximum strength on the contralateral side. With use of a questionnaire, pain was assessed preoperatively and postoperatively as stage 0 (none), stage 1 (mild discomfort or pain that does not restrict work or sports activities), or stage 2 (pain that restricts work or sports activities). However, there were too few patients in each group for a meaningful analysis; therefore, we divided the patients into those who had pain and those who did not.
The mechanism of injury in Group 1 was a fall for seven patients, a sports-related injury for six, and a motor-vehicle accident for three. The mechanism of injury in Group 2 was a fall for five patients, a sports-related injury for eight, and a motor-vehicle accident for five. There were no work-related injuries. The twenty-six patients who were employed preoperatively returned to their previous occupation after the index operation.
Ten patients in Group 1 and twelve patients in Group 2 had been managed previously with a cast for an average of twelve and ten weeks, respectively. No patient had had previous operative treatment of the non-union of the scaphoid. All of the patients had a fibrous non-union. Six patients in Group 1 and eight patients in Group 2 had cysts in the proximal pole of the scaphoid that measured 0.5 to 2.0 millimeters in diameter.
Operative Technique
Herbert screw: A volar approach20 was used in order to place the Herbert screw. Curets or a high-speed burr with constant irrigation with saline solution were used to remove the fibrous tissue and necrotic bone from the area of the non-union without damage to the cortical shell. Operative dental picks, which were found to cause less comminution of the fracture fragments than Kirschner wires inserted perpendicularly, were used to manipulate and derotate the fragments. Corticocancellous bone graft obtained from the distal aspect of the radius, the olecranon, or the iliac crest was contoured into a trapezoidal shape4,9. The dorsal cortex of the scaphoid was carefully preserved, and the inner surface was notched to accommodate the graft. The fibrous bands attached to the dorsal cortex were left intact because they served as a hinge around which the distal fragment of the scaphoid was rotated dorsally to correct the alignment before insertion of the bone graft. Fluoroscopy was used to confirm reduction of the fracture and placement of the internal fixation. The fracture was then stabilized with a 0.035-inch (0.089-centimeter) Kirschner wire placed along the radial border of the scaphoid perpendicular to the plane of the fracture9. This wire was cut so that it did not interfere with placement of the Huene device. Although the position of the Huene device was confirmed by insertion of a second wire through the device into the scaphoid with the use of fluoroscopy, the final position of the screw was still somewhat unpredictable because of collapse of the scaphoid during placement of the screw. The fracture fragments were then compressed with the Huene device, and the Herbert screw was inserted11. The Kirschner wire used to stabilize the fragments was removed with use of local anesthesia eight weeks postoperatively.
Cannulated screw: The initial exposure, reduction, and placement of the bone graft for the cannulated AO/ASIF screw was the same as those for the Herbert screw. The guide-wire for the cannulated screw was inserted along the axis of the screw with the use of fluoroscopy. The length of the screw to be used was determined once the guide-wire was positioned adjacent to the subchondral bone of the proximal fracture fragment. The guide-wire was driven across the scaphoid into the radius. A second Kirschner wire was placed along the radial border of the scaphoid to control rotation, with use of a technique similar to that for the Herbert screw, avoiding the eventual path of the screw (Fig. 5). The screw was then inserted in the standard manner. The Kirschner wire was left in place for eight weeks.
Four of the bone grafts in Group 1 were from the distal aspect of the radius, two were from the olecranon, and ten were from the iliac crest. Four of the bone grafts in Group 2 were from the distal aspect of the radius, three were from the olecranon, and eleven were from the iliac crest. The grafts measured 2.0 to 2.5 millimeters on the volar surface and approximately one millimeter on the dorsal surface. Cancellous bone was used to fill the cysts.
Postoperatively, the extremity was placed in a below-the-elbow thumb-spica cast, which was worn for six to eight weeks. Patients who were non-compliant with, or unable to follow, instructions or whose lifestyle necessitated strenuous use of both hands wore an above-the-elbow thumb-spica cast for the first four weeks. After removal of the cast, the patients wore a removable thumb-spica splint for four weeks and were advised to avoid lifting more than ten pounds (4.5 kilograms) until there were radiographic signs of union.
Analysis of Data
The potential confounding variables included age, gender, delay until treatment, involvement of the dominant hand, severity of the preoperative symptoms, and severity of the collapse and displacement of the fracture as measured on the radiographs. For continuous variables, multiple linear-regression analyses were used, adjusting for the potential confounding variables. The Cox regression model7 was used to evaluate the time to union and the placement of the screw along with age, gender, delay until treatment, involvement of the dominant hand, and severity of the initial radiographic displacement.
One patient in Group 1 had a persistent non-union and stage-2 osteoarthrosis fifty-four months postoperatively without loosening of the Herbert screw and was not scheduled for additional treatment at the time of the latest follow-up examination. Stage-3 osteoarthrosis developed in another patient in Group 1 and was treated sixteen months postoperatively with arthrodesis of the lunate, triquetrum, hamate, and capitate after excision of the scaphoid. Two patients who were managed with the cannulated screw had removal of the Kirschner wires after pain and irritation developed at the site of insertion. One patient in Group 1 had an intercarpal arthrodesis because of persistent pain at the site of the non-union. Five patients in Group 2 requested removal of the screw: two, because of clinical and radiographic evidence of impingement on the trapezium and three, for personal reasons. Two patients in Group 1 also had the screw removed for personal reasons.
In Group 1, the average duration of follow-up was sixty-nine months (range, forty to eighty-four months) and the average time to union for the fourteen patients who had a union was 7.6 ± 3.6 months (range, 3.0 to 14.0 months). In Group 2, the average duration of follow-up was twenty-nine months (range, twenty-six to forty-four months) and the average time to union was 3.6 ± 1.2 months (range, 2.5 to 7.0 months).
With the numbers available, we were unable to detect any differences between the two groups with regard to the distribution of age and gender, delay until the operation, collapse of the scaphoid, and avascular necrosis. The non-union was treated successfully in all but two patients, both in Group 1.
In thirty-four age and gender-matched normal subjects, the scapholunate angle was 45 ± 4.2 degrees. There was no difference between Group 1 and Group 2 with regard to the severity of the deformity on the radiographs. Preoperatively, the average lateral intrascaphoid angle was 48.8 ± 8.9 degrees in Group 1 and 51.7 ± 5.4 degrees in Group 2; the average posteroanterior intrascaphoid angles were 36.9 ± 4.0 and 35.0 ± 4.9 degrees, respectively; the average scapholunate angles were 51.9 ± 6.8 and 52.0 ± 5.5 degrees, respectively; and the fracture was displaced an average of 1.1 ± 0.5 and 1.3 ± 0.5 millimeters, respectively. The average lateral intrascaphoid angle in both Group 1 and Group 2 was significantly decreased (p < 0.01) postoperatively (to 38.1 ± 5.4 and 35.3 ± 4.0 degrees, respectively). The average scapholunate angle was also significantly decreased (p < 0.05) in both Group 1 and Group 2 (to 50.0 ± 5.0 and 47.8 ± 3.5 degrees, respectively).
Seven of the sixteen patients in Group 1 and nine of the eighteen patients in Group 2 had an abnormal scapholunate angle preoperatively. The angle decreased in five of the patients in Group 1 and in six of the patients in Group 2 and corrected to normal in two of the patients in Group 1 and in three of the patients in Group 2. The average displacement of the fracture was significantly decreased (p < 0.01) to 0.2 ± 0.3 millimeter in Group 1 and to 0.0 millimeters in Group 2, indicating correction of the malrotation of the scaphoid. The lateral intrascaphoid angle was measured on lateral tomograms or on sagittal computerized tomography images made at the time of union; therefore, we could not determine if the measurements changed with time. The scapholunate angle did not change from the time of union to the most recent follow-up examination. Group 2 had significantly greater correction of the lateral intrascaphoid angle and the displacement of the fracture than Group 1 (p < 0.05).
The time to union was evaluated in terms of central placement of the screw compared with peripheral placement. With use of a Cox proportional hazard with backward stepwise adjustment of the baseline clinical and radiographic variables7, central placement of the screw was found to be associated with more rapid healing in both groups at the 95 per cent confidence level (p < 0.05). However, consideration of the series as a whole was biased by the fact that significantly more cannulated screws (seventeen of eighteen) than Herbert screws (seven of sixteen) had been centrally placed (p < 0.01) as well as by the difference between the biomechanical properties of the two screws18,21. Thus, central placement was compared with peripheral placement in Group 1 alone. Again, it was found to be associated with a shorter time to union (p < 0.01; Kaplan-Meier estimate). A Cox regression analysis7 indicated that, in both groups, the probability of a persistent non-union was greater when the screw had been placed peripherally than when it had been placed centrally (p < 0.01).
A Kaplan-Meier estimate12 was used to evaluate the time to union because the duration of follow-up was different for each of the groups. The patients who had been managed with the cannulated screw had a significantly shorter average time to union (3.6 ± 1.2 months; range, 2.5 to 7.0 months) than those who had been managed with a Herbert screw (7.6 ± 3.6 months; range, 3.0 to 14.0 months) (p < 0.05). A shorter time to union was associated with a greater postoperative grip strength and range of motion of the wrist (p < 0.01). The time to union after the index operation was significantly longer for patients who had a longer delay until treatment (p < 0.05). There was no difference between Groups 1 and 2 with regard to the percentage of union at the time of follow-up, after controlling for age, gender, preoperative degree of pain, delay until treatment, displacement of the fracture, and carpal collapse (p = 0.07).
We were not able to identify any predictors of postoperative osteoarthrosis, as it developed in only two patients. The accuracy of plain radiographs in the diagnosis of avascular necrosis has been questioned9,24. Preoperatively, two patients from Group 1 and one from Group 2 had radiographic evidence of avascular necrosis; intraoperatively, there was no punctate bleeding from the proximal pole of the scaphoid in these three patients and in three patients whose radiographs had been interpreted as showing no evidence of avascular necrosis. Two months after the operation, nine patients were diagnosed as having avascular necrosis: it was stage 1 in six and stage 2 in three. Three of these patients had increased radiodensity of the proximal pole of the scaphoid but had had punctate bleeding at the time of the operation. At the most recent follow-up examination, only three patients had avascular necrosis, which was stage 1 in all of them. Preoperative avascular necrosis was not a predictor of a decreased rate of union or time to union. The patients who had avascular necrosis in the postoperative period, however, had a longer time to union (p < 0.05).
Postoperatively, the over-all average total range of motion on the involved side was 80.4 ± 11.4 per cent of that on the contralateral side. This corresponded to 107 ± 19 degrees of flexion and extension of the wrist and 31 ± 7 degrees of radio-ulnar deviation. With the numbers available, there was no significant difference between the range of motion for Group 1 (77.4 ± 14.1 per cent) and that for Group 2 (83.1 ± 7.7 per cent). Logistic regression analysis with adjustment for clinical and radiographic variables showed that the postoperative range of motion increased significantly when the lateral intrascaphoid angle or the degree of displacement of the fracture decreased (p < 0.05). An increased range of motion was positively associated with a decreased time to union (p < 0.05).
Postoperatively, patients regained an average of 73.5 ± 16.8 per cent of the grip strength on the contralateral side. There was no substantial difference between the average grip strength in Group 1 (68.7 ± 21.4 per cent) and that in Group 2 (77.8 ± 10.0 per cent). Increased grip strength was not associated with correction of the intrascaphoid and scapholunate angles, but it was associated with a shorter time to union (p < 0.01).
Over-all, there was a significant difference between the preoperative and postoperative degree of pain (p < 0.01); however, there was no difference between the two groups. Preoperatively all of the patients had stage-1 or 2 pain, and postoperatively four patients (two in each group) had stage-1 or 2 pain. The two patients in Group 1 who had pain postoperatively had a persistent non-union.
The duration of immobilization did not affect any of the parameters used to evaluate the result of treatment. There was also no relationship between the duration of follow-up or the age of the patient and the functional and radiographic results. The type of bone that had been used for the graft had no effect on the time to union or the final alignment of the scaphoid.
Although excellent results have been reported after bone-grafting without internal fixation for treatment of non-union of the scaphoid, little specific information has been provided regarding the time to union and the duration of immobilization needed. Russe reported union in twenty (91 per cent) of twenty-two patients who had been managed with bone-grafting of a non-union of the scaphoid without the use of internal fixation; however, the duration of immobilization was not clearly defined. Green reported a 76 per cent rate of union in forty-five patients with use of the same treatment. We were not able to find any reports regarding whether bone-grafting alone could correct scaphoid and carpal malalignment. Amadio et al.3 evaluated forty-five patients and found that malunited fractures were associated with a twofold increase in the prevalence of post-traumatic osteoarthrosis.
The rate of union in the present study was 94 per cent (thirty-two of thirty-four patients) and is most likely related to the strict inclusion criteria that enabled us to select an ideal group of patients. We excluded patients who had severe avascular necrosis, fracture of the proximal pole, or osteoarthrosis, all of which have been associated with a decreased rate of union5,23. Our results are comparable with reported rates of union of 75 to 91 per cent after bone-grafting alone5,10,20 and of 97 per cent (147 of 151 fractures) after bone-grafting and fixation with Kirschner wires23. In a study of eighty-six patients, Cooney et al.5 demonstrated that there was a trend toward a greater rate of union when Kirschner wires were incorporated with bone-grafting than when bone-grafting was used alone, but those authors used several techniques of bone-grafting and no statistical comparison was possible. Adams et al. reported a higher rate of union if the internal fixation was optimum; however, they did not define optimum fixation. We considered internal fixation to be optimum when alignment of the scaphoid was improved.
Because it is difficult to design a comparative study in which the number of patients is large enough for the rate of union to be used as the end point, we studied the time to union. We also used tomography to confirm the union. The importance of using tomography was demonstrated by Cooney et al.6, who found that patients managed with screw fixation who were thought to have a healed non-union on the basis of plain radiographs often needed another bone-grafting procedure because of persistent instability. In another study, Cooney et al.5 used a variety of techniques, including Kirschner wires, in forty-four patients in whom volar inlay grafts were used and demonstrated a time to union of eighteen weeks. Stark et al.23 reported an average time to union of seventeen weeks (maximum, thirty-three weeks) in 147 of 151 patients who had had fixation with Kirschner wires in addition to bone-grafting. As far as we know, the shortest average time to union was ten weeks, as reported by Fernandez. The average time to union in our study (5.3 months) was longer, which may be related to the use of tomography to determine union. On the average, plain radiographs predicted union six weeks before solid osseous bridging was confirmed on at least two coronal or sagittal computerized tomography scans.
Several reports have suggested that the time to union is associated with the duration of immobilization for patients who are managed with bone-grafting alone or with bone-grafting and Kirschner wires6,10,20,23. In reports of bone-grafting alone, however, the duration of immobilization was not clearly defined, although it was indicated that immobilization was continued for as much as one year10,20. Cooney et al.5 reported an average duration of immobilization of seventeen weeks after treatment with bone-grafting with or without Kirschner wires, and Stark et al.23 reported an average of seventeen weeks (maximum, thirty-three weeks) of immobilization after bone-grafting and insertion of Kirschner wires. Fernandez reported immobilization in a cast for two weeks after fixation with a standard non-cannulated AO/ASIF screw. Herbert and Fisher used no immobilization after treating non-unions and acute fractures with a double-threaded screw. We chose six to eight weeks of immobilization on the basis of our preliminary findings that immobilization of less than six weeks was associated with a longer time to union. This period of immobilization is well tolerated, and patients can subsequently use a removable brace to avoid excessive loading of the scaphoid until healing has been confirmed on tomograms.
It is difficult to compare functional results among studies because many authors did not report these results as a percentage of the values on the contralateral side.
The range of motion after bone-grafting alone was reported as normal, although specific data were not provided10,20. In their report of the results of non-unions treated with bone-grafting either alone or with internal fixation, Cooney et al.5 noted that the motion of the wrist changed little from preoperatively to postoperatively; however, specific measurements were not given. Stark et al.23 noted that the range of motion returned to normal after bone-grafting and internal fixation, but they also did not record specific measurements. Fernandez noted a flexion-extension arc of 136 degrees and radio-ulnar deviation of 37.5 degrees but did not compare these measurements with the preoperative values or with the values for the contralateral side. These results appear to be better than those found in the present study, in which the flexion-extension arc was 107 degrees and radio-ulnar deviation was 31 degrees. However, a direct comparison is not possible because the data regarding motion of the wrist in the study by Fernandez cannot be normalized.
We could not compare the grip strength of our patients with that of patients in other studies either because specific measurements were not recorded9,23 or because comparisons were not made with the contralateral sides8.
The relief of pain in the present study was comparable with that reported in other studies. Cooney et al.5 noted that fourteen (21 per cent) of sixty-eight patients in whom the fracture had healed continued to have occasional mild pain. This prevalence was similar to that reported by Stark et al.23, who reported mild postoperative pain in thirty-four (23 per cent) of 147 patients who had a union. Fernandez reported that two of twenty patients who had been managed with screw fixation had residual mild pain, which was not relieved by removal of the screw.
Green stated that avascular necrosis was more accurately determined by an intraoperative finding of no punctate bleeding than by findings on plain radiographs. Our results were similar to those reported in other studies, regardless of the treatment used. Fernandez noted that three of twenty patients had avascular necrosis after treatment with screw fixation but that it had resolved by the time of follow-up. Stark et al.23 noted that avascular necrosis developed postoperatively in sixteen of 151 patients but that it resolved after union. The presence of avascular necrosis preoperatively or postoperatively was not associated with the functional outcome in our study.
The prevalence of postoperative osteoarthrosis in our study (two of thirty-four) was similar to that reported in other studies regardless of the type of fixation used5,6,10,19,23. Because of the low rate of non-union, we were unable to associate the development of osteoarthrosis with either the type of screw fixation or with union. Cooney et al.5 noted an increased prevalence of osteoarthrosis when the fracture had failed to unite. Stark et al.23 noted little or no progression of radiocarpal osteoarthrosis after the fracture had healed.
None of the studies that we reviewed indicated whether alignment of the scaphoid was improved after bone-grafting alone or after bone-grafting with internal fixation6,9,10,20,22. One of the major factors associated with the result in our study was improvement in the alignment of the scaphoid. Amadio et al.3 noted that patients who had a malunion of the scaphoid had a twofold increase in the prevalence of osteoarthrosis; however, a multivariate regression analysis did not show an association between an increased scapholunate angle and an increased prevalence of osteoarthrosis. All of our patients had collapse of the scaphoid with alteration of the posteroanterior intrascaphoid angle or the lateral intrascaphoid angle, and the alignment of the scaphoid improved in twenty-eight patients postoperatively. We found that a decreased lateral intrascaphoid angle corresponded with an improvement in the over-all range of motion, indicating that even the mechanics of the wrist were improved in the immediate setting. Many studies of non-union of the scaphoid have not indicated changes in the scapholunate angle15,9-13,16,19,20. Stark et al.23 reported that twenty-five (17 per cent) of 151 patients had an abnormal scapholunate angle preoperatively. After bone-grafting and fixation with Kirschner wires, the angle decreased in fifteen patients, but it increased in nine; one patient was not included because the fracture did not unite. Cooney et al.6 noted that the scapholunate angle improved from 65 to 54 degrees. None of the patients in our study had an increase in the scapholunate angle postoperatively, suggesting that fixation with a screw may maintain correction of the scaphoid more effectively than fixation with Kirschner wires.
Our study showed that the time to union after fixation with the AO/ASIF cannulated screw was shorter than that after fixation with the Herbert screw. The rate of central placement in the proximal pole of the scaphoid was higher for the cannulated screw than for the Herbert screw. Scaphoid and carpal alignment were improved postoperatively, and this improvement was maintained at the latest follow-up examination. Improved alignment of the lateral intrascaphoid angle was associated with an over-all improvement in the range of motion.