A total of 16 consecutive adult patients requiring corrective osteotomy for a symptomatic extra-articular distal radius malunion were prospectively enrolled over a 2-year period from July 2021 to July 2023. The median age of the case series (n=14, 87.5% women and n=2, 12.5% men) was 61 (range 22‐77) years. Exclusion criteria were bilateral distal radius deformity, cognitive impairment, psychiatric illness, active substance abuse, inability to read and understand the language, and any condition known to impair bone healing. The indication for surgery was pain in most patients (n=11, 68.8%). A total of 18.8% (n=3) of the patients experienced loss of range of motion (ROM), but only one of them had this as the only indication for surgery. In addition to corrective osteotomy of the distal radius, 18.8% (n=3) of the patients underwent concomitant ulnar shortening to avoid excessive distraction of the distal radius. The ulnar shortening procedures were also performed using 3D surgical planning and guides. A total of 6.3% (n=1) of the patients withdrew before surgery, and an additional 12.5% (n=2) of the patients were excluded from the final analysis after developing nonunion following a concurrent ulnar shortening procedure. These patients were transferred to standard clinical care and underwent revision fixation of the ulna using longer plates. Because they did not complete the study-specific follow-up, they were excluded from the final analysis.

This study was approved by the Swedish Ethical Review Authority (2021-01974) and performed in accordance with the Declaration of Helsinki. All patients received written and oral information about the study and provided written informed consent prior to enrollment (all participants were adults; therefore, a legal guardian was not required).

Each patient underwent bilateral high-resolution computed tomography of the entire forearm using a standardized scanning protocol (100 kVp, 0.625-mm slice thickness, and convolution kernel: bone). Bilateral 3D virtual models of the forearm bones were created by segmenting the patients’ Digital Imaging and Communications in Medicine images using the Mimics Medical software (Materialise NV). The uninjured forearm was mirrored, aligned at the proximal part of the radius, and used as a template. The maximum planned distraction was limited to approximately 12 to 14 mm. This threshold was used as a technical planning constraint based on clinical experience and published recommendations suggesting that concomitant ulnar shortening should be considered in patients with radial shortening exceeding approximately 12 mm [19]. It should not be interpreted as a formally validated cutoff. The amount of ulnar shortening was then planned to achieve bone contact in the radius osteotomy after fixation. Surgical planning and guide design were conducted in-house by a hand surgeon (KL) and a clinical engineer (CSS) using the 3-matic Medical software (Materialise NV) according to previously described methods [18,20] (Figure 1). Digital files of the in-house–designed guides were sent to a subcontractor (Prototal UK) to be 3D printed in accordance with the medical device manufacturing standard ISO 13485. The printing technology used was selective laser sintering on a 3D printer (EOS P 396; EOS GmbH) with a biocompatible polyamide raw material (PA 2200; EOS GmbH). The guides were quality controlled by the hospital’s clinical engineers upon delivery from the subcontractor, repacked, labeled, and sterilized using autoclave according to the hospital’s clinical routines (134°; 4 minutes). The complete manufacturing process was conducted in accordance with the European Union Medical Device Regulation 2017/745.

All surgeries were performed by 2 senior hand surgery consultants (PA and KL) with experience level 4 according to Tang and Giddins [21]. Both surgeons had prior experience using externally purchased surgical guides. Patients were operated on under general anesthesia, regional anesthesia (brachial plexus block), or a combination of both. After exposure of the volar surface of the radius, the first drilling guide was placed on the bone and fixed with four 1.5-mm K-wires. Proper placement of the first guide was verified using a flat elevator to confirm continuous contact between the guide and the bone surface. The screw holes were predrilled using the guide’s cylindrical sleeves (Figure 2).

When all holes were drilled, the 2 parallel K-wires were left in the bone, and the other 2 K-wires were removed to allow for removal of the drill guide by sliding it off the remaining K-wires. The cutting guide was then positioned by sliding it down to the bone over the 2 parallel K-wires, and the osteotomy was carried out using a power saw. To fixate the bone, either a Variable Angle LCP Two-Column Plate (DePuy Synthes) or a VariAx 2 Distal Radius Plate (Stryker) was used depending on best anatomical fit during the 3D planning phase. The plate was fixated to the distal fragment first with locking screws and then reduced to the radius shaft using the previously drilled holes. If the osteotomy gap lacked bone contact, it was filled with iliac crest graft or residual bone from a concomitant ulnar shortening. If volar bone contact was present, no bone graft was used. Following wound closure, the wrist was immobilized for 2 weeks in a plaster cast. The duration of fluoroscopy use (seconds) was recorded by the radiographic fluoroscopy unit in the surgical room.

Two weeks postoperatively, the cast was replaced with a removable wrist brace, and patients initiated gentle wrist ROM exercises 4 times daily. Light loading was permitted beginning at 4 weeks. After 6 weeks, brace use was typically discontinued, and full wrist loading was allowed provided that radiographic evidence confirmed initial bone healing.

The primary outcome we report is the discrepancy between the 1-year postoperative computed tomography scan and the 3D virtual plan. Volar tilt, ulnar variance, and radial inclination were selected as clinically relevant outcome parameters. On the basis of prior literature, acceptable error margins were predefined as 5° or less for volar tilt and 2 mm or less for ulnar variance [3-5,22]. No specific error margins have been defined for radial inclination as previous studies did not find a significant correlation between radiographic measurements and clinical outcomes [23,24]. Using the 3-matic Medical software, an anatomical reference system was established to ensure clinical interpretability (Figures 3A and 3B) and described in our previous study [18]. Briefly, the origin (0, 0, 0) was positioned at the center of the rim between the lunate fossa and the sigmoid notch, similar to the central reference point used in standard radiographic assessments [18]. The central axis of the radius was defined as the intersection of 2 planes created using a datum-plane function (“two points, perpendicular to view”). The 2 points were manually placed on the radial shaft, one at 30 mm and one at 50 mm proximal to the central reference point using the anteroposterior and lateral views, respectively. The z-axis was aligned parallel to the radial shaft axis at the origin, the x-axis was directed toward the projected tip of the radial styloid and perpendicular to the z-axis, and the y-axis was defined as perpendicular to the x-z plane. Within this reference framework, rotation around the x-axis represented volar tilt, rotation around the y-axis corresponded to radial inclination, and translation along the z-axis represented ulnar variance.

The postoperative and planned radius models were registered in 2 stages. First, the proximal radius was used as the stable reference. Corresponding regions covering approximately the proximal half of the radius were manually selected on the planned and postoperative models using a brush selection tool. The postoperative model was then registered to the planned model using an iterative closest-point algorithm restricted to the selected proximal surfaces.

Following this proximal registration, the coordinate system was reset to the anatomical reference frame of the virtual plan. Corresponding articular regions of the distal fragment were then manually selected on both models, and a second iterative closest-point registration was performed. The transformation between the planned and postoperative positions of the distal fragment was exported as a homogeneous transformation matrix from which the translational and rotational errors were calculated. A transformation matrix shows translation and rotation differences between the planned and postoperative 3D models in all 3 spatial dimensions (x, y, and z).

PROM data were collected using the Disabilities of the Arm, Shoulder, and Hand (DASH) questionnaire [25]; the Patient-Rated Wrist Evaluation (PRWE); and the 100-mm visual analog scale for pain at rest and pain during activity. The total resulting scores from both the DASH and PRWE measures range from 0 to 100, where 0 indicates no disability or pain and 100 indicates the worst possible pain or disability [26].

Clinical outcome measurements were assessed preoperatively and 1 year postoperatively. Wrist ROM in extension, flexion, supination, pronation, radial deviation, and ulnar deviation was assessed using a goniometer. Grip strength was measured using a dynamometer (Jamar STD) [27]. Torque (ie, strength during forearm rotation) was measured in kilograms and converted into newton-meters using a dynamometer with a handle attached to a wall using a connection plate [28]. Lifting strength with one arm was measured in kilograms in a neutral wrist position using a hanging scale dynamometer (HCB 50K20; Kern & Sohn GmbH) [28]. Before strength testing, patients were allowed to familiarize themselves with the dynamometer and the testing procedure. One measurement of each ROM and strength outcome was recorded at each assessment. The same measurement equipment was used throughout the study. Measurements were not repeated or averaged, and intraobserver reliability was therefore not assessed. The measurements were performed by a highly experienced hand surgeon (PA; level 4) [21] following the Swedish national manual for measurements in hand surgery [27].

The equivalence bounds were defined as mean volar tilt error within –5° to +5° and mean ulnar variance error within –2 mm to +2 mm, representing the clinically acceptable margins based on previous literature. To assess whether postoperative radiological outcomes were equivalent to the preoperative virtual plans (ideal=0° or 0 mm), an equivalence test was performed using the two one-sided tests (TOST) procedure using JASP (University of Amsterdam). Equivalence testing uses TOST at an α value of .05 to assess whether the effect is neither too low nor too high. Combining these 2 tests yields a 90% CI, representing values that satisfy both 1-sided tests of 5%. Equivalence could be claimed if both TOST were significant, meaning that the entire 90% CI of the mean postoperative deviation lay within the defined intervals for volar tilt (−5°, +5°) and ulnar variance (−2 mm, +2 mm). Equivalence testing was performed using the signed differences between the planned and achieved outcomes to assess whether the mean directional error was contained within the predefined lower and upper equivalence bounds. In addition, mean absolute errors were calculated and reported descriptively to quantify the magnitude of the deviations irrespective of their direction (positive or negative).

Due to the small sample size, the secondary outcomes were reported using medians and IQRs, and a nonparametric Wilcoxon signed-rank test was applied to assess the differences between preoperative and postoperative measurements. A P value below .05 was considered statistically significant. No corrections for multiple comparisons were applied to secondary outcomes as they were considered exploratory.

For the radiological analysis, of the initial 16 patients, 13 (81.3%) were included (n=1, 6.3% withdrew prior to surgery, and n=2, 12.5% were moved to standard clinical care due to complications). A supplementary table summarizing the available sample size for each outcome and the reasons for missing data is provided in Multimedia Appendix 1, and a study flowchart is shown in Figure 4. An equivalence test was calculated based on the signed values and showed that the error in ulnar variance was considered equivalent to the preoperative plan. Both the lower (–2 mm) and upper (+2 mm) equivalence bounds were statistically significant (P<.001), and the 90% CI for the mean error was included in the equivalence boundaries. The signed mean error in volar tilt was not considered equivalent to the preoperative plan. The test against the upper bound (+5°) was significant (P<.001), but the test against the lower equivalence bound (−5°) was not (P=.33). Thus, the 90% CI for the mean deviation extended beyond the equivalence boundaries. Patient-level errors in ulnar variance and volar tilt, together with the cohort-level mean errors and their 90% CIs used for the equivalence analyses, are shown in Figures 5A and 5B.

Mean absolute errors, representing the magnitude of deviation irrespective of direction, were analyzed for all translational and rotational components. The mean absolute translational errors were below 2 mm in all directions. Among the rotational components, the largest mean absolute error was observed for rotation corresponding to volar tilt. Detailed results for all 6 components are shown in Table 1. Box plots of the signed mean errors in all transformation directions are shown in Figures 6A and 6B.

Preoperative and postoperative PROMs are shown in Table 2. PRWE and DASH scores were analyzed for the same patients included in the radiological analysis (n=13). Pain at rest and pain during activity were analyzed in 76.9% (10/13) of the patients because preoperative data were missing for an additional 23.1% (3/13). Numerical improvements from baseline were observed in PRWE, pain at rest, and pain at activity in all patients except 1 (9/10, 90%). DASH scores improved in all patients except 2 (8/10, 80%). These counts represent any reduction in score and should not necessarily be interpreted as clinically meaningful improvement. Statistically significant improvements were noted in all measurements (P<.05 in all cases).

Preoperative and postoperative clinical outcome measurements are shown in Table 3. Lifting and grip strength were analyzed for the same patients included in the radiological analysis (n=13). Torque measurements were analyzed in 92.3% (12/13) of these patients because preoperative values were missing for 7.7% (1/13). A statistically significant improvement in lifting strength was observed postoperatively compared to preoperatively in 66.7% (8/12) of the patients. The remaining strength measurements were not statistically significant.

The ROM data for the affected arm both pre- and postoperatively are shown in Table 4.

Significant improvements in flexion and ulnar deviation were observed postoperatively relative to preoperatively. All patients except 1 (11/12, 91.7%) showed numerical improvement in these measurements. No significant improvement was observed in extension, supination, pronation, and radial deviation.

The median intraoperative fluoroscopy duration was 56 (IQR 26‐72) seconds (mean 54.9, SD 9.3 seconds) to allow for comparison with other studies. Fluoroscopy data were available for 12 procedures. Two patients who underwent a concomitant ulnar shortening procedure were excluded from the fluoroscopy analysis to avoid including fluoroscopy attributable to the additional procedure. The final fluoroscopy analysis therefore included 10 patients.

In our cohort including 15 patients (1/16, 6.3% withdrew prior to surgery), 4 (26.7%) required additional or revision surgery. Two of these patients developed nonunion following the concomitant ulnar shortening procedure and underwent revision fixation of the ulna using longer plates. Because they were transferred to standard clinical care and did not complete the study-specific follow-up, they were excluded from the final analysis of the primary radiological outcome. Nonunion occurred in the ulna for 2 out of 3 patients treated with concomitant ulnar shortening. In one of these patients, screw loosening in the distal radius also occurred, leading to the partial recurrence of radial deformity.

Implant removal and median nerve decompression were performed in one patient, with relief of symptoms. Another patient had persistent ulnar-sided wrist pain and mild symptoms of carpal tunnel syndrome. Resection of the ulnar styloid, arthroscopic shaving of a central triangular fibrocartilage complex tear, and median nerve decompression yielded good results. Additionally, one patient had transient increased chronic pain and returned to the same pain level as prior to surgery.