Necessity of grafting in scaphoid nonunion fixation: A comparative outcome analysis

Abstract

BACKGROUND

Scaphoid nonunion (SNU) presents substantial functional limitations, frequently necessitating surgical intervention to restore wrist mobility and prevent degenerative changes. While the traditional approach to managing SNUs has relied on open fixation combined with bone grafting, the universal necessity of grafting, especially in stable nonunion (NU), is increasingly questioned. Emerging evidence indicates that graft-less fixation can deliver favorable outcomes, with notable healing observed in such cases.

AIM

To compare the outcomes of graft-less vs graft-augmented fixation in stable SNU, focusing on union rates (URs), healing time (HT), and functional recovery.

METHODS

A retrospective analysis of 44 patients with stable SNUs (Slade grades I-V) managed by either graft-less (n = 21) or graft-augmented (n = 23) fixation. Subgroup analysis included non-vascularized (n = 12) and vascularized grafts (n = 11). Clinical and radiological outcomes were assessed over a minimum 24-month follow-up. HT and UR were compared across groups concerning NU type, bone resorption, duration, and anatomical location.

RESULTS

Overall UR was 81.8%. URs did not differ significantly between groups (P > 0.05), whereas HT was significantly shorter with graft augmentation (P < 0.001). Graft-less fixation yielded superior grip strength (P < 0.001), radial/ulnar tilt (P = 0.004, < 0.001), and Mayo modified wrist scores (P = 0.01). Graft augmentation particularly improved outcomes in cystic SNUs, those with ≥ 5 mm bone loss, and NUs ≥ 1 year. Non-vascularized and vascularized grafts demonstrated comparable outcomes (UR: 91.7% vs 81.8%, P = 0.5). Smoking showed a significant association with delayed healing in graft-augmented fixation (P = 0.04), whereas no such relationship was observed in graft-less fixation.

CONCLUSION

Fixation without grafting is viable in selected stable SNUs, offering comparable union and superior function, albeit with longer HT. Bone grafting remains advantageous in cystic, long-standing, or biologically compromised NUs. Treatment should be tailored based on NU characteristics and patient factors.

Key Words: Scaphoid nonunion; Graft-less fixation; Vascularized graft; Mini-dorsal wrist approach; Scaphoid union; Wrist capsulotomy

Core Tip: This study compares graft-augmented and graft-less fixation in stable scaphoid nonunions. While grafting significantly shortened healing time, graft-less fixation yielded superior functional outcomes, particularly in grip strength and wrist motion. Subgroup analysis highlights the role of bone loss, nonunion duration, and smoking in guiding treatment choice. These findings support individualized fixation strategies and question the routine necessity of bone grafting in select cases.

INTRODUCTION

Scaphoid nonunion (SNU) remains a persistent challenge in orthopedic practice due to its insidious progression and impact on wrist biomechanics. If left untreated, it predisposes to functional deficit and, ultimately, carpal collapse with degenerative arthritis. Surgical intervention is frequently required to promote healing, preserve alignment, and prevent long-term sequelae such as SNU advanced collapse wrist[1]. SNU is broadly categorized by structural stability. Stable nonunion (NU) is often maintained by a cartilaginous shell bridging the fragments, preserving scaphoid contour and carpal congruity. In contrast, unstable variants exhibit deformity, loss of containment, or interfragmentary motion[2]. The Slade and Dodds classification further stratifies these lesions based on radiographic and biological features. Grade I denotes delayed union (4-8 weeks), grade II indicates fibrous union, and grade III involves minimal sclerosis with < 1 mm bone resorption. Grades IV and V exhibit cystic changes without deformity and resorption < 5 mm and 5-10 mm, respectively. Grade VI reflects extensive resorption, instability, and deformity[3].

Traditionally, internal fixation combined with bone grafting (BG) has been the mainstay of SNU management, providing mechanical and biological support. BGs offer osteoconductive and osteo-inductive benefits, especially in cases with voids or cystic changes[4,5]. However, recent reports suggest that in selected stable, well-aligned Nus[6-10], rigid fixation alone may be sufficient to achieve union[11-15]. The current study was designed to evaluate the outcomes of stable SNUs treated either with graft-augmented fixation or with fixation alone. By comparing union rates (URs), healing times (HTs), and complication profiles over a minimum two-year follow-up (FU), we aim to clarify the role of BG in stable SNU and define criteria for when it may be safely omitted.

MATERIALS AND METHODS

Study design and patients

This retrospective study was conducted in accordance with the Declaration of Helsinki and received approval from the institutional ethics committee. It included patients with neglected stable SNUs who underwent surgical fixation and followed from July 2019 to July 2025. Eligible cases were classified as grades I to V according to the Slade and Dodds system, with stability confirmed either preoperatively by magnetic resonance imaging (MRI) or intraoperatively by the presence of an intact cartilaginous shell. All patients had a minimum clinical and radiological FU of 24 months. Exclusion criteria comprised unstable SNUs, grade VI NUs, humpback deformity, dorsal intercalated segment instability, prior wrist surgeries, or inadequate FU. A detailed flowchart outlining the patient selection process is provided in Figure 1.

Figure 1 Flowchart outlining the patient selection process. SNAC: Scaphoid nonunion advanced collapse; DISI: Dorsal intercalated segment instability; SNU: Scaphoid nonunion; MRI: Magnetic resonance imaging; FU: Follow-up.

Patients were allocated to graft-less or graft-augmented fixation based on preoperative factors and intraoperative judgment. To mitigate selection bias inherent to the retrospective design, strict inclusion criteria were applied, all procedures were performed with the same fixation device by four senior hand surgeons (> 10 years’ experience), and subgroup analyses were conducted to account for confounders. BG either non-vascularized (NVBG) or vascularized (VBG) was generally favored in patients with potentially impaired healing (female sex, older age, prolonged NU-duration), whereas graft-less fixation was preferred in cases with minimal cystic change and preserved subchondral bone stock. The decision between NVBG and VBG was guided by intraoperative assessment of bone quality and vascularity, determined by punctate bleeding after curettage/debridement and, when available, preoperative MRI. NVBG was employed when vascularity was adequate and bone loss limited, while VBG was reserved for larger defects or questionable perfusion. Baseline demographic and clinical variables were compared between groups, and subgroup analyses performed, to enhance comparability, although residual selection bias cannot be excluded.

Data collection

All data on patient demographics were retrieved from electronic medical records. Data were collected on age, sex, injured side, hand dominance, NU duration, NU pattern (morphology, bone resorption, anatomical location), and smoking status. All patients were fully counseled regarding surgical options, including risks and alternatives, and provided written informed consent prior to surgery.

Surgical technique

All procedures were performed under regional supraclavicular anesthesia; spinal anesthesia was added if iliac BG was needed. Patients were positioned supine. Prophylactic IV antibiotics were given 10 minutes before limb exsanguination and tourniquet inflation. The surgical approach was selected based on fracture location and surgeon preference. For proximal pole (PP) NUs (n = 21, 47.7%), a dorsal approach was consistently used - either traditional or mini-dorsal - regardless of BG use. In waist NUs (n = 23, 52.3%), surgeons opted for either dorsal or volar approaches.

In the traditional dorsal approach (n = 25), an 8-10 cm incision was made just ulnar to Lister’s tubercle, in line with the third metacarpal. Thick skin flaps were elevated to preserve dorsal sensory branches. The fourth compartment tendons were retracted medially, and the second and third compartments radially. A Z-shaped capsulotomy, based on a radial-based flap, was utilized to access the scaphoid. For the mini-dorsal approach (n = 12), a 3-4 cm incision was placed radial to Lister’s tubercle between the axes of the second and third metacarpals. The distal retinaculum was longitudinally released between the third and fourth compartments. Second and third compartment tendons were gently retracted laterally using a Gelpi retractor, while the fourth compartment remained intact. Dissection proceeded beneath it to reach the dorsal capsule. Care was taken to preserve terminal branches of the superficial radial nerve. An inverted L-shaped capsulotomy was performed under fluoroscopic guidance to identify the scapholunate interval. The capsulotomy consisted of a 1.5 cm horizontal limb at the radio-scaphoid (RS) joint and a 2 cm vertical limb just ulnar to the scapholunate ligament (SLL) (Figure 2A). For the volar approach (n = 7), a longitudinal incision was made over the flexor carpi radialis (FCR), extending distally from 4 cm proximal to the wrist crease toward the scaphoid tuberosity. After identifying and retracting the FCR medially, the deep fascia was opened between the FCR and radial artery. An oblique volar capsulotomy was then performed along the scaphoid axis, dividing the radio-scapho-capitate ligament to expose the scaphoid.

Figure 2 Radiographic evaluation of waist nonunion managed via the mini-dorsal approach. A: Intraoperative clinical image demonstrating the inverted L-shaped capsulotomy used in the mini-dorsal approach. The transverse limb (white dashed line) is located at the radio-scaphoid interval, while the longitudinal limb (blue dashed line) lies ulnar to the scapholunate ligament interval; B: Preoperative anteroposterior radiograph showing a waist nonunion; C and D: Postoperative anteroposterior (C) and scaphoid-view (D) radiographs at seven months, showing complete healing.

SNU stability was confirmed intraoperatively in all instances via direct visualization of an intact bridging cartilaginous shell. Patients with gross fragment mobility were excluded (n = 7). In the graft-less group, the cartilaginous envelope was preserved. In contrast, in the graft-augmented group, it was intentionally opened to allow thorough preparation of the NU site using a low-speed burr and small curettes. Three BG types were utilized. The first was cancellous NVBG from the distal radius (DR) (n = 5). The second was cortico-cancellous NVBG from the iliac crest (IC) (n = 7), employed exclusively via the volar approach. The third was VBG from the DR (n = 11), harvested dorsally and based on the 1,2 intercompartmental supra-retinacular artery, following the technique described by Zaidemberg et al[16].

Screw entry points were determined per surgical approach to ensure optimal positioning. In volar approach, the scapho-trapezial (ST) joint was exposed by dividing the ST ligament and volar capsule. With the wrist extended, a guide wire was inserted retrogradely through the distal scaphoid tuberosity in a dorsal-ulnar trajectory. Wrist extension facilitated trapezial dorsal translation and guide wire alignment. For dorsal and mini-dorsal approaches, maximal wrist flexion improved SLL visualization and guide wire access. The guide wire was introduced antegradely from the PP approximately 2 mm radial to the SLL insertion, and advanced in a volar-ulnar direction along the central scaphoid axis.

Fluoroscopy confirmed guide wire placement along the longitudinal axis and across the NU site. In dorsal approaches, the wire tip was placed 2 mm short of the distal ST joint surface; in volar cases, 2 mm short of the proximal RS joint surface to protect joint integrity. Following reaming, a headless compression Herbert screw was inserted over the guide wire. Screw length was determined using an alternative guide wire of equal length or a measuring device, and the head was countersunk to prevent ST- or RS-joint irritation. In one case, a 5 mm drill fragment was retained within the proximal scaphoid, after it was broken while reaming, to avoid vascular compromise.

BG employment followed scaphoid reaming. DR- or IC-NVBG grafts were impacted into the NU-site before screw insertion. VBG pedicle was positioned and temporarily stabilized with a 0.9 mm K-wire. Final screw positioning and length were confirmed fluoroscopically, and wrist range of motion (ROM) was checked to exclude screw impingement.

Capsulotomies were repaired per approach: Z-shaped dorsally and inverted L-shaped mini-dorsally. In volar cases, the ST capsule, divided radio-scapho-capitate ligament, and volar soft tissues were approximated with interrupted sutures, followed by capsulotomy repair and FCR-sheath closure. After hemostasis, layered closure was performed and a short-arm splint was applied for two weeks. Operative time (from incision to closure) was recorded and compared across groups.

Postoperative protocol

Active finger motion was encouraged from postoperative day one. Sutures were removed after two weeks, and the splint was replaced with a scaphoid cast for an additional four weeks. Thereafter, patients wore a removable orthosis until radiographic union was confirmed. Wrist mobilization was initiated following cast removal, while contact sports and heavy loading were restricted until complete radiological healing. Standard wrist radiographs (anteroposterior, lateral, and scaphoid views) were obtained at two weeks and every three weeks thereafter until union was evident.

Outcomes and definitions

Mean HTs and URs were compared between groups. The influence of bone resorption, NU type, location, and duration on both outcomes was also analyzed. Union was defined by bridging trabeculae and cortical continuity in at least three cortices on plain radiographs (Figures 2B-D and 3). Initial assessment was performed by the operating surgeon, with confirmation by two independent hand surgeons blinded to treatment allocation. Final confirmation was achieved by multi-slice computed tomography. Union failure was identified if no trabecular continuity, persistent cysts, or a visible NU gap were present at six months.

Figure 3 Radiological evaluation of graft-less fixation for proximal pole nonunion. A: Preoperative anteroposterior wrist radiograph indicating a proximal pole nonunion; B-E: Sagittal and coronal computed tomography slices illustrating the nonunion with cyst changes and bone resorption; F and G: Follow-up radiographs at three-months (anteroposterior and lateral views) demonstrating healing after graft-less fixation; H: Anteroposterior wrist radiograph at five-months confirming full consolidation.

At final FU, clinical assessments included wrist ROM (flexion, extension, radial/ulnar deviation) measured via goniometer. Pain was scored using the visual analogue scale[17], and function was assessed with the Mayo modified wrist (MMW) score[18]: 0-64 (poor), 65-79 (fair), 80-89 (good), 90-100 (excellent). Grip strength was expressed as a ratio of the operated to the contralateral hand. Complications such as infection or union failure were recorded.

Statistical analyses

Data was analyzed using IBM SPSS Statistics v29.0.2.0 (IBM Corp., 2023, Armonk, NY, United States). Descriptive statistics [means ± SDs, n (%)] were reported. Normality was assessed with the Kolmogorov-Smirnov test. Between-group comparisons used independent t-tests for normally distributed variables, and Mann-Whitney U tests otherwise. Categorical data were analyzed using the χ² test. A significance threshold of P < 0.05 was applied. Multiple linear regression (enter method) was used to assess whether NU-duration, bone resorption, smoking, or age predicted HT.

RESULTS

Patient’s characteristics at baseline

A total of 44 patients with stable SNU were included in the final analysis. Of whom, 27 were manual laborers (61.4%), five were students (11.4%), four were professional athletes (9.1%), four were housewives (9.1%), three were salesmen (6.8%), and one was a police officer (2.3%). Included patients were allocated to graft-less (n = 21) and graft-augmented (n = 23) fixation groups. The latter was subdivided into NVBG (n = 12) and VBG (n = 11). While fracture type distribution (linear or cystic) was similar between the two main cohorts (P = 0.8), the degree of bone resorption varied significantly (P = 0.03), reflecting surgeons’ preference to employ grafting in cases with greater resorption. Subgroup analysis revealed significant differences in fracture type (P = 0.03) and resorption extent (P = 0.01). Comprehensive demographics and preoperative details are summarized in Table 1.

Table 1 Demographic characteristics of the study participants, n (%)/mean ± SD.

Variables	All participants (n = 44)	Fixation groups			Grafting subgroups
		Graft-less group (n = 21)	Graft-augmented group (n = 23)	P value	NVBG subgroup (n = 12)	VBG subgroup (n = 11)	P value
Age (years)	28.9 ± 6.6	26.3 ± 5.5	31.4 ± 6.7	0.009a	30.2 ± 6.6	32.6 ± 6.9	0.4
Gender
Male	40 (90.9)	20 (95.2)	20 (87)		12 (100)	8 (72.7)
Female	4 (9.1)	1 (4.8)	3 (13)	0.6	0 (0)	3 (27.3)	0.09
Affected side
Right	26 (59.1)	12 (57.1)	14 (60.9)		6 (50)	8 (72.7)
Left	18 (40.9)	9 (42.9)	9 (39.1)	0.3	6 (50)	3 (27.3)	0.06
Dominance
Non-dominant	16 (36.4)	9 (42.9)	7 (30.4)		4 (33.3)	3 (27.3)
Dominant	28 (63.6)	12 (57.1)	16 (69.6)	0.4	8 (66.7)	8 (72.7)	1.00
Previous traumatic event	23 (52.3)	9 (42.9)	14 (60.9)	0.2	8 (66.7)	6 (54.5)	0.6
SNU duration (month)	13.5 ± 6.6	13.1 ± 6.7	13.8 ± 6.6	0.7	14.8 ± 5.4	12.7 ± 7.8	0.4
SNU duration
< 1 year	21 (47.7)	11 (52.4)	10 (43.5)		2 (16.7)	8 (72.7)
≥ 1 year	23 (52.3)	10 (47.6)	13 (56.5)	0.5	10 (83.3)	3 (27.3)	0.01a
Fracture type
Cystic	31 (70.5)	15 (71.4)	16 (69.6)		6 (50)	10 (90.9)
Linear	13 (29.5)	6 (28.6)	7 (30.4)	0.8	6 (50)	1 (9.1)	0.03a
Bone resorption
< 5 mm	22 (50)	14 (66.7)	8 (34.8)		7 (58.3)	1 (9.1)
≥ 5 mm	22 (50)	7 (33.3)	15 (65.2)	0.03a	5 (41.7)	10 (90.9)	0.01a
Nonunion site
Waist	23 (52.3)	14 (66.7)	9 (39.1)		8 (66.7)	1 (9.1)
Proximal pole	21 (47.7)	7 (33.3)	14 (60.9)	0.06	4 (33.3)	10 (90.9)	0.005a
Smoking	23 (52.3)	11 (52.4)	12 (52.17)	0.97	9 (75)	3 (27.3)	0.11

^aP < 0.05.

NVBG: Non-vascularized bone graft; VBG: Vascularized bone graft; SNU: Scaphoid nonunion.

Radiological outcome

Study participants completed a mean FU of 27.52 ± 3.1 months. Overall UR was 81.8%, with no significant difference between groups (P > 0.05). However, HT was significantly shorter in the graft-augmented group (P < 0.001). Detailed radiological findings are listed in Table 2.

Table 2 Clinical and radiological outcomes following scaphoid nonunion management, n (%)/mean ± SD.

Clinical and radiological findings	All participants (n = 44)	Fixation groups			Grafting subgroups
		Graft-less group (n = 21)	Graft-augmented group (n = 23)	P value	NVBG subgroup (n = 12)	VBG subgroup (n = 11)	P value
Operative time (minutes)	83.2 ± 35.9	47.6 ± 8.3	115.8 ± 11.8	< 0.001a	112.08 ± 15.3	120.00 ± 3.8	0.1
Surgical approach	Dorsal (n = 25), volar (n = 7), mini-dorsal (n = 12)	Mini-dorsal (n = 12), dorsal (n = 9)	Dorsal (n = 16), volar (n = 7)		Dorsal (n = 5), volar (n = 7)	Dorsal (n = 11)
Follow-up duration (month)	27.52 ± 3.1	28.0 ± 3.3	27.08 ± 2.9	0.3	27.9 ± 3.1	26.18 ± 2.7	0.3
Radiological findings
Healing
Nonunion	8 (18.2)	5 (23.8)	3 (13)	0.3	1 (8.3)	2 (18.2)	0.5
Union	36 (81.8)	16 (76.2)	20 (87)		11 (91.7)	9 (81.8)
Healing time (weeks)	14.1 ± 2.5 (n = 36)	16.2 ± 2 (n = 16)	12.5 ± 1.4 (n = 20)	< 0.001a	12.2 ± 1.25 (n = 11)	13 ± 1.6 (n = 9)	0.2
Clinical findings
Flexion ROM (°)	72.05 ± 6.4	73.5 ± 7.7	70.6 ± 4.5	0.1	70.8 ± 5.1	70.4 ± 4.1	0.8
Extension ROM (°)	63.75 ± 8.7	64.04 ± 10.2	63.4 ± 7.2	0.8	63.7 ± 6.4	63.2 ± 8.4	0.8
Total ROM (°)	135.8 ± 14	137.6 ± 17	134.1 ± 10.9	0.4	134.5 ± 10.9	133.6 ± 11.4	0.8
Radial deviation (°)	13.86 ± 3.8	15.5 ± 3.8	12.3 ± 3.1	0.004a	13.4 ± 2.4	11.09 ± 3.4	0.07
Ulnar deviation (°)	26.18 ± 6.8	30.9 ± 5.7	21.8 ± 4.6	< 0.001a	22.5 ± 4.8	21.2 ± 4.4	0.5
VAS
Median (min-max)	0 (0-3), 0.48 ± 0.9	0 (0-3), 0.38 ± 0.86	0 (0-3), 0.56 ± 1	0.5	0 (0-2), 0.16 ± 0.57	0 (0-3), 1 ± 1.2	0.5
Grip strength (kg)	44.56 ± 7.7	46.3 ± 7.04	43.02 ± 8.15	0.1	45.4 ± 3.1	40.4 ± 11.0	0.1
Grip strength ratio (%)	88.38 ± 5.9	91.3 ± 4.6	85.6 ± 5.7	< 0.001a	86.2 ± 5.2	85.1 ± 6.4	0.6
MMW score	90.8 ± 8.7	94.3 ± 8.2	87.6 ± 8.1	0.01a	90.4 ± 7.5	84.5 ± 7.8	0.08

^aP < 0.05.

NVBG: Non-vascularized bone graft; VBG: Vascularized bone graft; ROM: Range of motion; VAS: Visual analogue scale; MMW score: Mayo modified wrist score.

NU-bone resorption and healing outcome

In cases with < 5 mm bone resorption (n = 19), HT averaged 14.25 ± 2.7 weeks and UR was 84.2%. For ≥ 5 mm resorption (n = 25), HT was similar (13.95 ± 2.45 weeks) with 80% UR (P = 0.74). In < 5 mm resorption cases, grafting significantly shortened HT (12.28 ± 1.4 weeks vs 15.7 ± 2.4 weeks; P = 0.01) and improved UR (87.5% vs 81.8%). In ≥ 5 mm resorption cases, grafting similarly reduced HT (12.6 ± 1.4 weeks vs 16.28 ± 2.2 weeks; P = 0.003) and enhanced UR (86.6% vs 70%).

NU-anatomical location and healing outcome

For PP SNUs (n = 21), mean HT was 14.16 ± 2.5 weeks with 85.7% UR. Although UR was similar across groups, grafting significantly reduced HT (12.75 ± 1.54 weeks vs 17 ± 1.2 weeks; P = 0.0009). In waist SNUs (n = 23), mean HT was 14 ± 2.6 weeks and UR was 78.2%. Grafting resulted in shorter HT (12.25 ± 1.3 weeks vs 15.4 ± 2.6 weeks; P = 0.01) and higher UR (88.8% vs 71.4%). No significant difference was found between graft-less fixation of PP and waist NUs (P = 0.35). Likewise, graft-augmented fixation showed no significant variation between the two sites (P = 0.5).

NU-morphology and healing outcome

In linear NUs (n = 13), HT averaged 14.41 ± 2.99 weeks with 92.3% UR. Grafting significantly shortened HT (12.16 ± 1.6 weeks vs 16.66 ± 2.25 weeks; P = 0.013), though UR was slightly higher in graft-less cases (100% vs 85.71%). In cystic NUs (n = 31), mean HT was 14 ± 2.3 weeks with 77.41% UR. Grafting improved both HT (12.71 ± 1.43 weeks vs 15.9 ± 2 weeks; P = 0.001) and UR (87.5% vs 66.66%). There was neither significant difference in HT between graft-less fixation of linear and cystic NUs (P = 0.44), nor between graft-augmented fixation of linear and cystic NUs (P = 0.38).

NU-duration and healing outcome

Stratifying by SNU-duration, < 1-year cases (n = 21) had 14.5 ± 2.5 weeks HT and 80.9% UR, while ≥ 1-year cases (n = 23) had 13.6 ± 2.5 weeks HT and 82.6% UR (P = 0.28). In the graft-less group, duration had minimal impact on outcomes (< 1 year: 15.7 ± 2.4 weeks and 81.8%; ≥ 1 year: 16.2 ± 2.2 weeks and 70%; P = 0.87). Grafting significantly improved HT and UR in both categories (< 1 year: 13.1 ± 1.7 weeks and 80%; ≥ 1 year: 12.1 ± 1.19 weeks and 92.3%; P = 0.24). Compared to graft-less fixation, these differences were statistically significant (< 1 year: P = 0.03; ≥ 1 year: P = 0.002).

Avascular necrosis

Although vascularity could not always be confirmed perioperatively, BG still achieved favorable healing rates, with union in 91.7% of NVBG cases and 81.8% of VBG cases. However, postoperative osteonecrosis could not be evaluated, as FU MRI was not routinely performed. Consequently, the potential influence of surgical approach, graft utilization, or graft type on avascular necrosis (AVN) risk remains undetermined in this investigation.

Functional outcome

Graft-less group showed better radial and ulnar deviation (P = 0.004 and P < 0.001), grip strength (P < 0.001), and MMW scores (P = 0.01). Full clinical comparisons are listed in Table 2. No significant flexion-extension differences were noted between L-shaped and Z-shaped capsulotomy techniques. However, L-shaped capsulotomy yielded slightly better ulnar deviation and grip strength but was associated with longer HT (P = 0.01) and lower URs (Table 3).

Table 3 Wrist function outcomes according to capsulotomy design, mean ± SD.

Outcome variables	L-shaped capsulotomy (mini-dorsal approach) (n = 12)	Z-shaped capsulotomy (dorsal approach) (n = 25)	P value
Flexion ROM (°)	73.75 ± 7.7 (60-80)	71.6 ± 6 (55-80)	0.33
Extension ROM (°)	63.75 ± 12 (45-80)	64.2 ± 7.4 (50-80)	0.71
Total ROM (°)	137.5 ± 19.4 (110-160)	135.8 ± 12 (115-160)	0.52
Radial deviation (°)	15 ± 4.2 (8-20)	13.5 ± 3.8 (7-20)	0.29
Ulnar deviation (°)	29.9 ± 6.3 (20-35)	25.2 ± 6.9 (17-36)	0.07
VAS, median (min-max)	0 (0-3), 0.5 ± 1	0 (0-3), 0.48 ± 0.9	0.9
Grip strength (kg)	47.79 ± 3.6 (40-52.5)	42.96 ± 9.5 (20-53)	0.16
Grip strength ratio (%)	91.21 ± 5 (78.7-97.22)	87.92 ± 5.9 (71.9-97)	0.05

ROM: Range of motion; VAS: Visual analogue scale.

Correlation analyses

Univariate regression analysis revealed a weak negative correlation between age and HT across all participants (r = -0.32, P = 0.05), indicating a tendency for faster healing in younger patients. However, this association was not significant within the graft-less (r = 0.16, P = 0.53) or graft-augmented groups (r = -0.33, P = 0.15). SNU duration showed no significant correlation with HT in the overall cohort (r = -0.12, P = 0.47), graft-less (r = 0.20, P = 0.42), or graft-augmented group (r = -0.11, P = 0.62). Similarly, bone resorption ≥ 5 mm was not significantly associated with HT in the total population (r = -0.22, P = 0.19), graft-less (r = 0.09, P = 0.74), or graft-augmented group (r = 0.13, P = 0.56). Notably, smoking showed a significant negative correlation with HT only in the graft-augmented group (r = -0.45, P = 0.04), while no significant association was observed in the graft-less group (r = 0.16, P = 0.53) or overall cohort (r = -0.13, P = 0.44).

Multiple linear regression analysis was also conducted across the entire cohort and within categories. For the overall cohort, the model was not statistically significant [F (4, 31) = 1.08, P = 0.38; R² = 12.3%]. Similarly, group analyses for graft-less and graft-augmented fixation yielded non-significant models, explaining 5.6% and 24.4% of the variance, respectively. The waist NU group demonstrated moderate explanatory capacity, with the model accounting for 21.4% of HT variance; however, this result was also not statistically significant (P = 0.5). In contrast, the model for PP NUs reached statistical significance [F (4, 13) = 3.19, P = 0.05; R² = 49.5%]. Within this group, smoking was identified as a significant negative predictor of HT (B = -2.566, P = 0.031).

Complications

In one waist NU case, the drill bit tip fractured intraoperatively during reaming, leaving a 5 mm fragment embedded in the proximal scaphoid to preserve bone vascularity. Fixation proceeded using impacted cancellous DR BG, resulting in successful union without complications (Figure 4). In the graft-less group, one patient developed a superficial wound infection, resolved with oral antibiotics. In the graft-augmented group, one patient experienced radial wrist pain due to bony overgrowth following healing with iliac NVBG graft. Arthroscopic debridement relieved symptoms and restored function.

Figure 4 Healing outcome of waist nonunion fixed with cancellous bone grafting. A: Intraoperative fluoroscopic image showing fixation of a waist nonunion using cancellous bone grafting harvested from the distal radius; a small fragment of a broken drill bit is embedded in the proximal pole; B-F: Postoperative anteroposterior and scaphoid-view radiographs and corresponding sagittal and coronal computed tomography slices at three months, showing complete healing.

Five patients in the graft-less group (23.8%) experienced union failure. Four were manual workers and one housewife. Bone resorption ≥ 5 mm was observed in three cases, mainly at the scaphoid waist, with one at the PP. Four patients were smokers, and all SNUs were cystic. In the graft-augmented group, union failure occurred in three cases (13%). Two received DR VBGs and exhibited ≥ 5 mm resorption at either the waist or PP. The third, treated with an IC NVBG, failed to heal despite < 5 mm resorption and a linear NU. This patient reported mechanical wrist pain, with imaging confirming ST joint violation from a sunken Herbert screw, resulting in ST joint osteoarthritis (Figure 5). A proximal row carpectomy was performed. The remaining NU cases resumed daily activities without significant symptoms and declined further intervention.

Figure 5 Failure of union after graft-augmented fixation for waist nonunion. A: Postoperative anteroposterior radiograph one month following fixation of a waist nonunion using cortico-cancellous graft from the iliac crest via the dorsal approach; B: Follow-up radiograph at two months revealing a sunken Herbert screw; C-G: Anteroposterior wrist radiograph and computed tomography slices at eight months showing union failure, graft resorption, and screw penetration into the scapho-trapezial joint.

DISCUSSION

This study demonstrated significantly shorter HT with graft utilization (P < 0.001), consistent with Atilgan et al[19] (P < 0.001), but not with Zengin N et al[20] (P = 0.41). Prior graft-less fixation series reported union times of 11.6-22 weeks[6-14], whereas graft-augmented techniques ranged between 10.8-16 weeks[21-27], supporting the biological advantage of BG via osteo-conduction, osteo-induction, and osteogenesis[5]. Table 4 summarizes findings from earlier reports. Although healing was faster with grafting in this investigation, URs did not differ significantly (P = 0.3). Similar outcomes were noted by Atilgan et al[19] (84.2% vs 87.5%) and Zengin N et al[20] (75% vs 88%, P = 0.34). Zhong et al[28] even reported higher URs in graft-less fixations for grade III and IV NUs, albeit non-significant (P = 0.52 and 0.46). These comparable URs in our investigation may be due to internal autografting generated from reaming, alternative to open curettage and debridement[7,12], enhanced stability from compression screw minimizing micromotion at NU site[7,12], and preserved peri-scaphoid ligaments, particularly with the mini-dorsal approach.

Table 4 Summary of case series reporting graft-less and graft-augmented fixation for scaphoid nonunion.

Ref.	n (M:F)	Age (years)	Site	NU-duration	Approach	Follow-up	Clinical results	Healing time	Union rate/judgement
Graft-less fixation of scaphoid nonunion
Tada et al[14], 2021, retrospective study	19 (19:0)	29.2	Waist/PP/distal pole	14.8 weeks	Open (volar/dorsal)	NR	NR	3.7 months	17/19 (89.4%); X-ray, CT
Belloti et al[6], 2020, prospective study	12 (NR)	R: 27-40	Waist/PP	> 6 months	PC (volar/dorsal)	6 months	Flexion: 59.1°, extension: 48.7°; DASH: 6.99, PWRE: 7.97	NR	12/12 (100%); X-ray
Gurger et al[9], 2018, retrospective study	12 (10:2)	27	Waist/PP	7.6 months	PC (volar)	4.5 months	Flexion: 68°, extension: 66°; MMW: Excellent (n = 8) and good (n = 3)	3.8 months	11/12 (91.6%); X-ray, CT
Cha et al[8], 2017, retrospective study	15 (15:0)	28.9 ± 6.2	Waist	10.8 months	PC (volar)	12 months	Flexion: 69°, extension: 68°; DASH: 7 ± 3.9	5.5 months	15/15 (100%); X-ray, CT
Somerson et al[13], 2015, retrospective study	14 (13:1)	21.1 ± 10.6	Waist/PP	> 6 months	PC (volar/dorsal)	NR	Flexion: 73°, extension: 66°; DASH: 10.2	4.4 months	12/14 (85.7%); X-ray
Hegazy[10], 2015, case series	21 (17:4)	23	Waist	6.8 months	PC (volar)	25 months	Flexion: 74°, extension: 75°; DASH: 8	4 months	21/21 (100%); X-ray, CT
Capo et al[7], 2012, retrospective study	12 (11:1)	24	Waist	8.7 months	PC (volar/dorsal)	35 months	Flexion: 71°, extension: 66°; DASH: 6	4 months	11/12 (91.6%); X-ray
Mahmoud and Koptan[11], 2011, prospective study	27 (26:1)	28.3	Waist	18.2 months	PC (volar)	24.6 months	Flexion: 65.7°, extension: 60.7°; MMW: Excellent (n = 25) and good (n = 2)	2.9 months	27/27 (100%); X-ray, CT
Slade et al[12], 2003, prospective study	15 (14:1)	20	Waist/PP	9 months	PC (dorsal)	35 months	Flexion: 61°, extension: 49°; MMW: Excellent (n = 12) and good (n = 3)	3.5 months	15/15 (100%); X-ray
Graft-augmented fixation of scaphoid nonunion
Zondervan et al[27], 2024, retrospective	8 (8:0)	R: 16-28	Waist/PP	13 months	Cancellous DR BG (dorsal)	88.9 months	Flexion-extension arc: 143.7° ± 13.9°; median DASH: 2.9	12 weeks	85.7%; X-ray
Bhat et al[21], 2023, retrospective	38 (36:2)	28.7 ± 7.9	Waist/PP/distal pole	18.7 ± 4.9 weeks	Cancellous DR BG (volar)	22.2 months	Clinical outcome: NR; complications: Screw prominence (n = 1), grade 1 SNAC (n = 1)	15.7 ± 3 weeks	100%; X-ray
Ma et al[23], 2023, retrospective study	21 (18:3)	29.8 (R:18-46)	Waist/PP	38.3 months	IC BG (volar)	30.6 months	Flexion-extension arc: 118° ± 46°; PRWE: 14 ± 11, VAS: 3 ± 1, grip strength: 83% ± 26%	2.7 months (R: 2-4)	100%; X-ray
Schäfer et al[26], 2023, retrospective study	16 (16:0)	29 (R: 18-41)	Waist/PP	17 months (R: 6-60)	IC NVBG (volar/dorsal)	54 months	Flexion: 55°, extension: 50.1°; DASH: 2.36, grip strength: 99% (R: 72%-134%)	12.9 weeks (R: 11-15)	73%; X-ray, CT
Papatheodorou et al[25], 2021, retrospective study	64 (45:19)	27 (R: 19-42)	PP	23 months (R: 12-47)	DR VBG (dorsal)	32 months (R: 24-56)	Extension: 70°, flexion: 47°; MMW: 86, grip strength: 83%; complications: Persistent nonunion (n = 8), fibrous union (n = 1)	12 weeks	55/64 (86%); X-ray, CT
Manako et al[24], 2023, retrospective study	5 (5:0)	Med: 30; R: 18-41	Waist/PP	≥ 6 months	Zaidemberg’s DR VBG (volar/dorsal)	Med: 24 months; R: 8-30 months	MMW score: 95 points (R: 75-100)	4.25 months	4/5 (80%); X-ray
Keller et al[22], 2020, retrospective study	37 (36:1)	26.2	Waist/PP	51 months	MFC VG (volar ± dorsal assisted)	16 months	Flexion: 44° ± 16°, extension: 50° ± 15°; DASH: 19, PRWE: 30; grip strength: 44 kg, VAS: 5; complications: Hardware irritation necessitated implant removal (n = 4), late infection (n = 1), venous bleeding (n = 1), postoperative boutonniere-like deformity of thumb MCPJ (n = 1), delayed wound healing of volar approach (n = 1)	16 weeks (R: 12-22)	95%; X-ray, CT

M:F: Male-female ratio; NU: Nonunion; PP: Proximal pole scaphoid; NR: Not reported; CT: Computed tomography; R: Range; PC: Percutaneous; DASH score: Disabilities of Arm, Shoulder and Hand score; PWRE: Patient-rated Wrist Evaluation; MMW score: Mayo modified wrist score; DR: Distal radius; BG: Bone graft; SNAC: Scaphoid nonunion advanced collapse; IC: Iliac crest; VAS: Visual analogue scale; NVBG: Non-vascularized bone graft; VBG: Vascularized bone graft; Med: Median; MFC: Medial femoral condyle; VG: Vascularized graft; MCPJ: Metacarpo-phalangeal joint.

Functionally, the graft-less group outperformed the grafted group at the final FU (27.5 ± 3.1 months). Grip strength ratio (P < 0.0001), as well as radial and ulnar deviations (P = 0.004 and P = 0.0001), were significantly better, while flexion, extension, and pain levels were comparable (P > 0.05). Prior comparative studies focused on the MMW and pain scores, rather than ROM and grip strength[19,20,28]. This limits direct comparison but highlights a unique aspect of our findings. The superior function in the graft-less group may reflect reduced soft tissue dissection and postoperative stiffness. While grafting appears to accelerate radiographic healing, graft-less fixation may favor selective motion recovery and grip strength. This trade-off should guide surgical decisions, particularly for patients prioritizing early return of function. Table 5 compares previous comparative studies of both techniques.

Table 5 Comparative studies evaluating graft-less vs graft-augmented fixation in scaphoid nonunion treatment.

Ref.	Technique	Fixation device	Follow-up period	UR	HT	Clinical results	Complications
Graft-less vs graft-augmented fixation approaches for scaphoid nonunion
Current study	Mini-open screw fixation vs open curettage-grafting and screw fixation	Screw	27.52 ± 3.1 months	Graft-less fixation: 76.2%; fixation with BG: 87%	Graft-less fixation: 16.2 ± 2 weeks; fixation with BG: 12.5 ± 1.4 weeks (P < 0.001)	Graft-less fixation: Radial and ulnar tilt were greater (P = 0.004, P < 0.001, respectively). The grip strength ratio was also higher (P < 0.001). The MMW score was significantly better (P = 0.01). No relevant difference was noted between both groups regarding all other clinical outcomes	Nonunion: 5 patients (23.8%) after graft-less fixation and 3 patients (13%) with BG utilization. Graft-less group: Superficial wound infection (n = 1). Graft-augmented fixation: Mechanical radial sided wrist pain (n = 1)
Zengin N et al[20], 2023, retrospective	PC screw fixation (n = 16) vs open curettage-grafting and screw fixation (n = 17)	Screw	12 months	CRIF: 12/16 (75%); ORIF: 15/17 (88%) (P = 0.34)	CRIF: 11.8 weeks (8-17 weeks); ORIF: 12.5 weeks (9-17 weeks) (P = 0.41)	No statistically significant difference between both groups regarding postoperative flexion, extension ROMs, grip strength, and DASH score	Nonunion: 4 patients (25%) in the CRIF group and 2 patients (11%) in ORIF group
Zhong et al[28], 2023, retrospective	ORIF (group A) vs PC fixation (group B) vs PC fixation and PRP injection (group C)	Screw	12 months	Grade III SNU: No significant difference among the three groups (P = 0.52). URs were 90.9%, 100%, and 100%, respectively; grade IV SNU: UR was lower in group B than other groups. URs were 90%, 71.4%, and 90.9%, respectively	NR	Grade III SNU: No significant differences in VAS and MMW among the three groups. Grade IV SNU: Groups A and C had significantly lower VAS score and significantly higher MMW score, compared to group B. However, there was no significant difference between groups A and C	Grade III SNU: Nonunion in 9% of patients of group A. Grade IV SNU: Nonunion in 10%, 28.6%, and 9% of groups A, B, and C, respectively
Atilgan et al[19], 2022, retrospective	PC screw fixation (n = 19) vs NVBG and screw fixation (n = 24)	Screw	CRIF: 18.9 ± 1.8 months; ORIF: 16 ± 1.95 months	CRIF: 84.2% (n = 16); ORIF: 87.5% (n = 21)	CRIF: 17.19 ± 2.19 months; ORIF: 14.14 ± 1.82 months (P < 0.001)	MMW score: No statistically significant difference (P = 0.930). VAS score: Statistically significantly higher after CRIF than in ORIF (P = 0.012)	Nonunion: Three patients (15.7%) in the CRIF group, and 3 (12.5%) in the ORIF group
RCTs comparing employment of VBGs vs NVBGs in fixation of scaphoid nonunion
Fan et al[42], 2023	DR pedicled VBG (n = 20) vs IC NVBG (n = 18)	K-wires	One year	VBG: 19/20; NVBG: 16/18	HT was similar (VBG: 16 ± 22 weeks; NVBG: 17 ± 33 weeks)	There were no significant differences between the two groups in ROM, grip strength or functional scores	Nonunion: VBG group (3/20) following K-wire migration, and NVBG group (2/18)
Caporrino et al[41], 2014	DR VBG (n = 35) vs DR NVBG (n = 38)	K-wires	VBG: 29.4 months; NVBG: 28.6 months	No significant difference in UR (88% for VBG vs 80% for NVBG; P = 0.312)	VBG: Bony union was earlier by 12 days compared to NVBG (P = 0.002)	VBG group: Ulnar deviation was lower (P = 0.003), and radial deviation as well (P = 0.064). No significant differences were found for the remaining clinical outcomes	Nonunion rate in VBG group (11%) and NVBG group (20%)
Goyal et al[35], 2013	DR NVBG (n = 42) vs IC NVBG (n = 46)	Herbert screw	Group 1: 4.8 years; group 2: 5.2 years	No significant difference in UR (87.1% vs 86.5%)	Group 1: 4.2 months; group 2: 4.5 months	No significant difference regarding ROM, functional scores. Mean pain VAS scores were significantly different (group 2: 7.1, group 1: 4.2)	Iliac NVBG (local): Local hematoma required drainage (n = 1). Chronic pain for 3 months (n = 6); ASIS avulsion fracture (n = 1); nonunion in group 1 (12.8%), and group 2 (13.5%). Both groups (last visit): Tenderness over anatomical snuffbox in group 1 (n = 5), and group 2 (n = 6); degenerative changes in group 1 (n = 15), and group 2 (n = 14)
Ribak et al[40], 2010	DR VBG (n = 46) vs DR NVBG (n = 40)	Three K-wires	VBG: 24.4 months; NVBG: 21.7 months	Higher UR after VBG (89% vs 72%; P = 0.024). PP nonunion: Slightly higher UR with VBGs (90% vs 69%; P = 0.09)	VBG: Significantly earlier healing compared to NVBG (9.7 weeks vs 12 weeks; P = 0.0001)	VBG: Significantly higher functional outcome scores (P = 0.007)	Nonunion: 5/46 (11%) in VBG group and 11/40 (28%) in NVBG group
Braga-Silva et al[36], 2008	DR VBG (n = 35) vs IC NVBG (n = 45)	K-wires or screws	2.8 ± 1.4 years	Union occurred in all NVBG group and 32/35 of the VBG group (P = 0.135)	NVBG: 8.89 ± 2.26 months; VBG: 7.97 ± 3.06 months	There was no significant difference with respect to ROM or grip strength between the two groups	Nonunion in 3/35 (8.5%) of the VBG group with graft osteonecrosis

UR: Union rate; HT: Healing time; BG: Bone graft; MMW score: Mayo modified wrist score; PC: Percutaneous; CRIF: Closed reduction and internal fixation; ORIF: Open reduction and internal fixation; ROM: Range of motion; DASH score: Disabilities of Arm, Shoulder and Hand score; PRP: Platelet-rich plasma; SNU: Scaphoid nonunion; VAS: Visual analogue score; NVBG: Non-vascularized bone graft; K-wires: Kirschner wires; RCTs: Randomized control trials; VBG: Vascularized bone graft; DR: Distal radius; IC: Iliac crest; ASIS: Anterior superior iliac spine; PP: Proximal pole scaphoid.

Scaphoid NUs are classified as linear, cystic, or displaced, as outlined by Ikeda et al[29]. In our cohort, displaced NUs were excluded based on preoperative MRI or intraoperative absence of an intact cartilaginous shell. This approach aligns with the “peanut fracture” concept by Maudsley and Chen[30], where a cartilaginous envelope signifies fragment stability. To ensure accurate assessment, the NU site was visualized in all patients, including those undergoing graft-less fixation, using a mini-dorsal approach.

For linear SNUs, the UR reached 92.3%, with a mean HT of 14.41 ± 2.99 weeks. Graft use significantly reduced HT (P = 0.013), though graft-less fixation achieved 100% union at the cost of slower healing. These results corroborate earlier findings by Tada et al[14], Capo et al[7], and Slade et al[12], highlighting the suitability of graft-less fixation in stable linear NUs, provided accurate preoperative imaging confirms NU stability. In contrast, cystic SNUs - characterized by bone resorption and cyst formation - demonstrated an overall UR of 77.4%. Graft-augmented fixation improved UR (87.5%) and shortened HT (P = 0.001), while graft-less fixation showed lower UR (66.7%) and longer HT. These findings align with Zengin N et al[20] and support the use of BGs in cystic NUs with significant resorption (Slade grades IV-V), enhancing osteogenesis and structural support.

We further evaluated the impact of bone resorption on healing. NUs with < 5 mm resorption showed no significant difference in UR or HT compared to those with ≥ 5 mm resorption (P = 0.74). However, grafting consistently improved outcomes in both groups. In < 5 mm resorption, grafting led to shorter HT and higher URs (P = 0.01); in ≥ 5 mm resorption, the benefit was even more pronounced (P = 0.003). These results support grafting as beneficial in resorptive NUs, especially with greater bone loss, aligning with Mani and Acharya’s report[31] of 93.3% union using cancellous grafts.

Earlier studies reflect ongoing debate over graft necessity in cystic NUs. Mahmoud and Koptan[11] achieved good union with percutaneous (PC) fixation alone, even with significant bone loss. However, their cohort was limited to young patients with waist NUs and preserved vascularity, these conditions were not always applicable to cystic PP NUs. In our study, although graft-less fixation was effective in cystic NUs, it showed longer HTs and lower URs. Grafting was especially beneficial in high-risk patients (e.g., elderly, diabetic, delayed presentation), supporting its broader use in cystic NUs.

Anatomical sites of SNU also influenced healing. Despite prior studies highlighting poor vascularity and lower union in PP NUs[15], we found no significant difference in HT or UR compared to waist NUs. However, preoperative vascularity was not consistently assessed in our investigation, limiting AVN-related conclusions. Literature still suggests superior outcomes in waist NUs due to better blood supply[5,15,22].

In PP SNUs, grafting significantly reduced HTs (P = 0.0009), though URs were similar across both groups (85.7%). Previous research showed inconsistent outcomes with graft-less fixation of PP NUs[6,9,12-14]. For instance, Belloti et al[6] and Slade et al[12] reported a 100% UR following PC fixation. Conversely, studies by Gurger et al[9] and Somerson et al[13] found NUs rates of 8.4% and 14.3%, respectively, with PP NUs representing a significant portion of their cohorts. Tada et al[14] also reported a 10.6% NU rate following open graft-less fixation. These variations may be partly due to the challenge of accurately assessing vascularity in PP during PC fixation techniques, leading some studies to exclude cases with suspected AVN. Given these concerns, open fixation procedures are often favored for PP NUs. The reliance on grafting in these cases may reflect the need to mitigate compromised blood flow, though vascular status was not systematically evaluated in our cohort. Various studies[21-27] have reported on the outcomes of graft-augmented fixation in PP NUs, with Bhat et al[21] and Ma et al[23] both demonstrate 100% URs. However, other reports have noted NU rates ranging from 5% to 27%, regardless of the utilized graft[22,24-27].

In waist SNUs, grafting also significantly accelerated healing (P = 0.01). While previous studies report high URs (92%-100%) with graft-less fixation for waist NUs[6-12], success is typically linked to minimal resorption and absence of deformity. Our results diverge; graft-less fixation showed lower UR (71.4%) compared to grafting (88.8%). It is suggested to consider other cofactors such as NU-duration, bone vascularity, and amount of resorption when selecting the appropriate surgical technique.

The pivotal influence of SNU-duration on healing outcome following fixation was highlighted in earlier reports[11-13,32,33]. Graft utilization outperformed graft-less fixation in NUs ≥ 1 year, achieving a significantly higher UR (92.3% vs 70%, P = 0.002). Grafting appeared to buffer the adverse effects of chronicity, with outcomes unaffected by duration (P = 0.24). In contrast, URs declined with longer duration in the graft-less group. Thus, grafting is especially advisable in prolonged NUs. Though some studies report acceptable outcomes in chronic NUs ≥ 1 year using graft-less fixation[11-13,32-34], success is contingent on specific factors; intact cartilaginous envelope, minimal resorption, and absence of degenerative changes, humpback deformity, carpal instability, or ligamentous injury. Nonetheless, healing tends to be slower, and union less predictable, potentially impacting functional recovery. Graft use may help mitigate these limitations in delayed presentations.

Multiple studies have compared VBGs and NVBGs in SNUs with largely consistent findings[35,36], echoed by our results. Early research by Boyer et al[37] and Harpf et al[38] reported modest VBG success (URs 60%-80%), highlighting technical challenges. Later studies, such as Munk and Larsen[39], showed higher URs for VBGs (91%) vs NVBGs (80%-84%), though complications remained a concern. Ribak et al[40] also favored VBGs (UR: 89% vs 72%, P = 0.024; HT: 9.7 weeks vs 12 weeks, P = 0.0001).

In contrast, our study found no significant difference in UR (NVBG: 91.7%, VBG: 81.8%, P = 0.5) or HT (12.2 weeks vs 13 weeks, P = 0.2). Caporrino et al[41] similarly found no UR difference (VBG: 88%, NVBG: 80%, P = 0.312), though VBGs showed slightly faster union (P = 0.002). Zhang et al’s meta-analysis[5] confirmed a slight edge for VBGs in UR (1.13-fold higher) and HT (1.73 weeks earlier), but functional outcomes remained comparable, aligning with our findings (grip strength and MMW scores slightly favored NVBG, P = 0.6 and P = 0.08). Duncumb et al[15], in a meta-analysis, reported high and near-equivalent URs following DR BG (VBG: 87%, NVBG: 86.6%, combined average 86.9%) and following IC BG (VBG: 91.7%, NVBG: 87.1%, combined average 87.6%). Specifically, in our cohort, IC NVBGs yielded 80% UR in 12.7 weeks; pedicled DR VBGs had 81.8% UR in 13 weeks; cancellous impaction DR NVBGs achieved 100% UR in 11.8 weeks (P = 0.2). Recent report by Fan et al[42] showed no significant difference between VBG and NVBG in HT (16 ± 22 weeks vs 17 ± 33 weeks) or functional outcomes, such as ROM and grip strength; consistent with our findings. Although earlier studies suggested VBGs might accelerate healing, current evidence, including ours, supports NVBG as equally effective, particularly in functional recovery[15,39,42]. Variability in outcomes across studies likely reflects differences in patient selection, NU-duration, and graft techniques. In our cohort, both VBG and NVBG promoted union in stable cases, though small sample sizes limit definitive comparisons. Prior comparative studies are summarized in Table 5.

Our results in PP SNUs - where VBGs are commonly preferred - were noteworthy. Although 90.9% of VBGs were used for PP NUs (P = 0.005), URs and clinical outcomes were not significantly better than NVBGs. This questions the assumed biological advantage of vascularization and suggests bone quality may be more critical. Rancy et al[43] similarly reported high URs (97% within 12 weeks) despite impaired vascularity in 60% of cases, finding no significant link between vascular status and healing success. This aligns with our observations: Despite disrupted vascularity on MRI in some patients, healing proceeded well with either graft-less or NVBG fixation. However, as MRI was not routinely performed, the true role of preoperative vascularity remains uncertain.

The comparable union and clinical outcomes observed between VBGs and NVBGs in our study challenge the prevailing assumption that the enhanced perfusion provided by VBGs inherently yields faster or more reliable healing. However, these findings should be interpreted with caution, given the limited sample size, which may underpower the detection of subtle yet clinically meaningful differences. Moreover, the increased technical complexity and longer operative times associated with VBG harvesting (P = 0.1) may influence surgical decision-making, particularly when considering patient recovery trajectories and resource utilization. Larger-scale studies are warranted to better delineate the relative advantages of these grafting techniques.

Capsulotomy design is a critical yet often underappreciated factor in wrist surgery, as it directly impacts both surgical exposure and postoperative functional recovery[44-46]. In this study, different capsulotomy techniques were employed based on the chosen approach: Z-shaped and L-shaped capsulotomies for the dorsal and mini-dorsal approaches, and an oblique capsulotomy for the volar approach. The configuration of the capsulotomy is especially relevant in preserving dorsal capsular vascularity and proprioceptive innervation - particularly when a transverse limb lies adjacent to the DR, which poses a risk of vascular compromise and subsequent capsular fibrosis[47]. While the Z-shaped arthrotomy offers excellent exposure of the carpus, it has been associated with potential drawbacks, including disruption of dorsal wrist ligaments, injury to the posterior interosseous nerve (PIN), and compromised vascular supply[48,49]. Although this investigation did not directly evaluate vascularity or proprioception, these concerns are supported by previous reports. For instance, Garcia-Elias et al[50] observed increased carpal collapse, evidenced by greater scapholunate and LC angles, following a volar approach compared to dorsal exposure. Nevertheless, a recent meta-analysis reported no significant differences in clinical outcomes between volar and dorsal approaches[34].

Compared to the Z-shaped capsulotomy in the standard dorsal approach, the inverted L-shaped incision used in the mini-dorsal approach resulted in less disruption of surrounding soft tissues and ligaments. This likely contributed to the observed trend toward improved functional outcomes, including higher grip strength ratio (P = 0.05), greater ulnar (P = 0.07) and radial tilt (P = 0.29). Interestingly, this approach was associated with a longer HT (16.3 ± 1.7 weeks vs 13.9 ± 2.4 weeks, P = 0.01), suggesting that while capsulotomy design may affect function, its impact on union is more likely secondary to factors such as fixation strategy and patient characteristics

The selection of a fixation device plays an influential role in SNU healing. Duncumb et al’s meta-analysis[15], which examined 78 studies, reported URs of 85.8% with compression screws, 90.8% with Association for the Study of Internal Fixation screws, and slightly lower rates with K-wires (87.1%) and plates (84.4%), with an overall mean UR of 88.2%. Various studies - ours included - demonstrated stable fixation construct via a single screw that promoted early mobilization, and avoided K-wire related complications such as delayed union, hardware irritation, infection, and stiffness[11,15,34]. Double screw fixation has been shown to improve rotational control and angular stability in SNU treatment[51], with reported URs ranging from 90.5% to 100% in two case series[52,53]. It is particularly useful for unstable PP fractures and small scaphoid bones[54]. Offering flexibility using screws ranging from 1.5 mm to 3.5 mm in diameter[54,55]. Early mobilization - starting as early as 10 days to 14 days - was also feasible with this method, without compromising healing[53]. In our study, we focused exclusively on stable NUs with intact cartilage, which provided inherent stability. We excluded displaced or unstable NUs and used single screw fixation (antegrade or retrograde), achieving a UR of 81.8%. This suggests that in stable cases, single screw fixation offers adequate compression and stability, without the added complexity of dual screws. However, we cannot determine whether single screw fixation is superior. Comparative studies are warranted to establish the optimal fixation method for different SNU types.

Our analyses identified age and smoking as relevant variables influencing HT. Age showed a weak negative correlation with HT (r = -0.32, P = 0.05), indicating a tendency for faster healing in younger patients, consistent with existing literature[56], though not statistically significant within individual groups. Smoking significantly prolonged HT in grafted cases (r = -0.45, P = 0.04), highlighting its detrimental effect on bone healing[57]. This aligns with established evidence that smoking impairs angiogenesis and osteogenesis[58]. Interestingly, smoking had no significant impact on HT in the graft-less group, suggesting that graft-less techniques may be more resilient to systemic risk factors - an observation that merits further investigation.

This study presents several strengths. Strict inclusion criteria focusing solely on stable NUs enhanced sample homogeneity and internal validity. Moreover, the comprehensive evaluation of both clinical and radiological outcomes over a minimum 24-month FU provided robust insight into HTs, URs, and functional recovery, allowing for a thorough assessment of the two surgical techniques.

Limitations

Several limitations must be acknowledged. The retrospective design introduces potential recall bias, particularly regarding fixation choice, which may have been influenced by patient factors and NU characteristics. The overall small sample size, and further subgrouping into VBG and NVBG, reduced statistical power and limited generalizability. Inconsistent use of advanced imaging, particularly preoperative MRI, restricted our ability to comprehensively assess scaphoid vascularity, especially in PP NUs. Furthermore, postoperative AVN could not be systematically evaluated, as FU MRI was not routinely performed. Consequently, the influence of surgical approach, grafting, or graft type on the risk of AVN remains uncertain.

Additional confounders might have influenced outcomes. Smoking was associated with a significant adverse effect on healing in grafted cases, and future studies should stratify participants accordingly. The variability in bone resorption, though partially addressed, warrants more detailed analysis in larger cohorts. Finally, prospective randomized controlled trials with standardized imaging protocols are needed to better control confounders and provide higher-quality evidence.

CONCLUSION

Graft-less fixation in stable scaphoid NUs offers comparable URs to graft-augmented techniques and may provide superior functional outcomes, particularly in grip strength and ROM. However, it is associated with a longer HT. Surgical decisions should be individualized, balancing healing goals with functional demands and considering factors such as NU duration, bone loss, and smoking status. Further prospective studies are needed to validate these findings.