Single versus double knot tension band wiring: Biomechanical analysis of compression forces in chevron olecranon osteotomies

Abstract

Purpose

Olecranon fractures and osteotomies worldwide are commonly stabilized using tension band wiring (TBW), traditionally tightened with two knots to achieve optimal compression. However, soft tissue irritation and technical issues have led to the adoption of a simplified single-knot technique. This study aimed to compare compression forces at the medial, lateral, and ventral osteotomy surface between single- (SKO) and double-knot osteosynthesis (DKO) using ultra-thin custom pressure sensors.

Methods

Standardized V-shaped olecranon osteotomies were performed on composite ulnas and fixed with TBW using either SKO or DKO, each performed 10 times. Three pre-calibrated pressure sensors were placed medially (PM), laterally (PL), and ventrally (PV) on each specimen. A 1.25 mm cerclage wire was tightened using a 0.45 Nm torque limiter. Compression forces were compared between SKO and DKO using t-tests, and intra-group differences assessed compression symmetry.

Results

No significant differences in PM, PL, or PV were found between SKO and DKO (p = .62, .89, .20). In both groups, ventral pressure was significantly lower than medial and lateral (p < .001), and lateral pressure exceeded medial in both SKO (p = .003) and DKO (p = .004).

Conclusion

Single-knot TBW provides comparable compression to the traditional double-knot technique, challenging the standard approach. Given its simplicity and similar biomechanical performance, SKO may offer a viable alternative, warranting further clinical evaluation regarding outcomes and complication rates.

Purpose

Worldwide, displaced simple olecranon fractures and osteotomies are most commonly treated with tension band wiring (TBW), resulting in good clinical outcomes and high consolidation rates.^1–3 The tension band principle theoretically converts the triceps tendon’s pulling force into compression at the fracture site, although this has never been biomechanically proven.^4–6

However, the tightening of the figure of eight cerclage wire during application results in compression at the fracture or osteotomy site and is achieved by twisting the wire ends evenly around each other. According to the teaching opinion of leading institutions such as the AO ASIF (Association for the Study of Internal Fixation), equal-sided compression can only be achieved with two knots, one on the lateral side and one on the medial side. ⁷ This has also been listed as one of the quality criteria of a correctly performed TBW construct. ⁸

On the other hand, several complications have been described following TBW of olecranon fractures or osteotomies, including soft tissue irritation, hardware migration and construct failure.^9–11 The latter has been shown to include the knot of the figure of eight configuration, resulting in loosening or breakage of the twisted cerclage wire.^12,13 Furthermore, application pearls include inadequate knots and osteosynthesis with two equally twisted knots is even more challenging. ⁸

Although the use of two knots may still be doctrine, the biomechanical superiority of equal compression forces to the lateral and medial fracture sites has never been biomechanically proven. Following the development of ultra-thin custom high-resolution pressure sensors, ¹⁴ the aim of the present study was to biomechanically compare the differences in compression symmetry of TBW constructs in chevron osteotomies of the olecranon tightened with one or two knots. The second outcome parameter was to monitor the static compression effect near the articular surface compared to the lateral and medial osteotomy planes.

Methods

This in vitro study was performed on biomechanically validated synthetic ulnas (large left 4th Generation Composite Ulna, Sawbones^® Europe AG, Malmo, Sweden). An osteotomy jig was used to create a chevron osteotomy with an apex angle of 135° using an oscillating saw with a blade thickness of 0.6 mm.

The custom sensors used for pressure measurements consisted of a dielectric made of a conductive polydimethylsiloxane made of a base polymer (Neukasil RTV-23, Altropol Kunststoff GmbH, Stockelsdorf, Germany) containing 10.6 wt% carbon black particles (ENASCO^® 250 P, TIMCAL Ltd, Bodio, Switzerland). The sensors had a diameter of 5 mm and a total thickness of 180 µm (see Figure 1(a)). The pressure was converted into an electrical signal that was recorded by the measuring instrument (Agilent^® 4263B LCR-Meter, Agilent Technologies^®, Inc., Santa Clara, CA, USA). During the calibration process, the high sensitivity and accuracy of approximately 3 kPa was determined, as previously published along with a detailed description of the sensor fabrication process. ¹⁴ The ultra-thin design enables the placement within the osteotomy without relevant interference with fragment reduction.

Figure 1.

Pressure sensors and osteosynthesis. (a) Custom pressure sensors with a diameter of 5 mm and a total thickness of 180 µm. (b) Illustration of the positioning of three pressure sensors at the medial (PM), lateral (PL), and ventral (PV) aspects of the V-shaped chevron osteotomy. (c, d) Illustrations of a TBW construct, comprising two 1.8 mm Kirschner wires and a 1.25 mm cerclage wire, tightened with a single knot.

Osteosynthesis was performed with TBW tied with either one knot (single knot osteosynthesis group, SKO) or two knots (double knot osteosynthesis group, DKO). Each fixation technique was tested with a total of 10 independent measurements per group (N = 10). To ensure balanced data collection and reduce potential bias, a randomization protocol was applied, assigning five measurements of each fixation method to each of two artificial bones used. The sequence of fixation methods was randomized using a block randomization scheme to minimize order effects. Between measurements, specimens were carefully repositioned and retightened to maintain consistency and reduce inter-test dependency.

Two parallel 1.8 mm Kirschner wires were inserted from the proximal dorsal side, crossing the osteotomy site near the articular surface in a distal and ventral direction, and finally protruding through the ventral cortex. After pre-positioning of the Kirschner wires, three of the custom-made pressure sensors were placed at standardized positions in the cross-section of the osteotomy. Position 1 (PM) was at the medial aspect, position 2 (PL) at the lateral aspect and position 3 (PV) at the ventral aspect of the osteotomy (see Figure 1(b)).

Subsequently, a 1.25 mm cerclage wire was placed in a figure-of-eight configuration, including the proximal ends of the Kirschner wires, through a 2 mm drill hole 3 cm distal to the olecranon tip. According to the randomization protocol, the cerclage wire was tied with one knot (SKO) or two knots (DKO) by twisting the ends of the cerclage wire evenly around each other. Final tightening of all knots was controlled with a 0.45 Nm torque limiter (Garant 0.45 Nm, Hoffmann GmbH Qualitätswerkzeuge, Munich, Germany). The knot of the SKO constructs was always placed medially as shown in Figure 1(c) and (d).

The final outcome parameter was the maximum pressure at PM, PL and PV as recorded by the measurement device. In addition, the mean compression per sample was calculated using the formula (PM + PL + PV)/3. After confirming normal distribution with the Shapiro-Wilk test, group comparisons were made with the Student's t-test. The significance level was set at p < .05. In addition, Pearson’s correlations were drawn between the different measurement points and a calculated dorsal ((PM + PL)/2) to ventral (PV) compression. Data were processed using SPSS software for Mac OS (version 24; IBM, Amonk, NY, USA).

Results

Sensor application was performed without interfering with the Kirschner wires crossing the osteotomy site. All results were normally distributed, as confirmed by Shapiro-Wilk tests. Overall, medial pressure (PM) was 401.2 kPa (SD = 166.4) after SKO and 389.6 kPa (SD = 145.2) after DKO. The difference between groups was not significant (p = .62). Laterally (PL), pressure was 487.4 kPa (SD = 192.9) after SKO and 483.4 kPa (SD = 149.0) after DKO, with no significant difference (p = .89). Ventrally (PV), 175.5 kPa (SD = 80.3) after SKO and 239.9 kPa (SD = 37.8) after DKO were measured, which was also not significantly different (p = .20).

Ventral pressure was significantly lower than both lateral and medial pressure for SKO and DKO constructs (p < .001 for all comparisons).

When compression symmetry was analyzed, lateral pressure was significantly higher than medial pressure in both the SKO (p = .003) and DKO (p = .004) constructs.

The calculated mean compression was 354.7 kPa (SD = 49.0) after SKO and 371.0 kPa (SD = 68.0) after DKO. This difference was not significant (p = .55). These results are visualized in Figure 2.

Figure 2.

Results. Pressure distribution is visualized using box plots. Blue boxes represent Single Knot Osteosynthesis (SKO), and red boxes represent Double Knot Osteosynthesis (DKO). No significant differences were observed between SKO and DKO at PM (medial pressure sensor), PL (lateral pressure sensor), PV (ventral pressure sensor), or the mean calculated pressure, determined using the formula (PM + PL + PV)/3. Within both SKO and DKO, pressure differed significantly among PM, PL, and PV. Black p-values represent calculations for both SKO and DKO.

A moderate positive correlation was observed between PM and PV at both SKO (r (8) = 0.490, p = .150) and DKO (r (8) = 0.475, p = .166); however, neither reached statistical significance. Similarly, a moderate to strong positive correlation was found between PL and PV in both SKO (r (8) = 0.452, p = .190) and DKO (r (8) = 0.514, p = .128), but again these results were not statistically significant. In contrast, a strong positive correlation was found between calculated dorsal compression ((PM + PL)/2) and PV in both SKO (r (8) = 0.593, p = .071) and DKO (r (8) = 0.639, p = .047), with the latter reaching statistical significance. Finally, a small positive correlation was found between PM and PL in both SKO (r (8) = 0.264, p = .461) and DKO (r (8) = 0.172, p = .635), neither of which reached statistical significance.

Discussion

The main finding of the present study is that stabilizing olecranon osteotomies with tension band wiring tightened with either one or two knots results in similar compression forces on both the lateral (p = .89) and medial (p = .62) osteotomy surfaces. The calculated mean compression forces were also not significantly different (p = .55). Secondary observation was a significantly lower compression force recorded on the ventral side of the osteotomy compared to the lateral or medial side in both single and double knot constructs (all p < .001). We used customized ultra-thin pressure sensors with a total thickness of 180 µm to minimize the influence on fragment positioning. In light of these results, the use of two knots in an olecranon tension band construct must be critically questioned, as the cerclage wire knot is prone to failure^12,13 and often imperfectly performed.^8,15

Tightening the cerclage wire with two knots was described by Deliyannis in 1973 ¹⁶ and biomechanically evaluated by Fyfe et al. in 1985. ¹⁷ To date, it is the technique taught by leading institutions such as the AO ASIF in worldwide courses and online manuals.^18,19

Fyfe et al. performed a biomechanical evaluation on 10 embalmed human cadaveric ulnas with standardized transverse osteotomies. Among other techniques, they compared TBW constructs with one or two knots by loading the dissected ulnas with the fixed triceps’ tendon up to 350 N in a 90° position of the elbow. Single-knot constructs were tensioned to 0.80 Nm, while each knot of the double-knot constructs was tensioned to 0.40 Nm. Stability was monitored with two displacement transducers placed across the osteotomy on the medial and lateral sides of the olecranon. Interfragmentary compression was not monitored. The double-knot constructs resulted in significantly less displacement than the single-knot constructs (p < .001), leading to the aforementioned recommendation to use two knots.

Our biomechanical setting differs in various aspects from the study conducted by Fyfe et al. First, we analyzed chevron osteotomies compared to transverse osteotomies. Second, no loading protocol was added to our study, as we only examined the static primary compression achieved after tension band wiring, and third, we tightened each knot by 0.45 Nm. The latter was chosen because we wanted to prove that the defined torque applied to one side of the wire would be transferred to the entire system, resulting in the tightening of both sides. Increasing the torque in the SKO group could have resulted in over-tightening of the wire, causing posterior over-compression, possibly resulting in a more ventral than bilateral fracture gap. This phenomenon can be observed in clinical practice (see Figure 3) and may explain the different results of Fyfe et al. where SKO constructs were fixed with twice the torque of DKO constructs, possibly resulting in more displacement ventrally and bilaterally, the region where they placed their displacement transducers. As shown in the clinical case in Figure 3, overtightening of the cerclage wire must be strictly avoided to prevent a ventral fracture gap. Therefore, the ideal torque value for twist tightening should be the subject of future research, which can be easily performed with the present test setup.

Figure 3.

Pitfall of Over-compression. (a) Clinical case of a 42-year-old male patient with a Mayo type IIa olecranon fracture. (b) Intraoperative fluoroscopy image showing dorsal over-compression and a resulting fracture gap at the articular surface (highlighted by white arrows) during tightening of the 1.25 mm cerclage wire. Note the slight bending of the 1.8 mm Kirschner wires near the articular surface, indicating over-compression. (c, d) Postoperative X-rays of the TBW construct tightened with two knots, still demonstrating asymmetric compression with a slight gap at the articular surface.

Fyfe et al. did not report construct failure in the single knot group because they used various fixation techniques per specimen. Therefore, the only alternative explanation for the poor outcome is that the single knot constructs in their cohort resulted in a loose figure-of-eight configuration that allowed displacement despite being tightened by 0.80 Nm. A loose figure-of-eight configuration following SKO was not observed in our study, as evidenced by comparable interfragmentary compression between the groups. Loose figure-of-eight configurations are one of the pitfalls of the TBW procedure and must be carefully avoided. ⁸

A well-published and frequently cited study was published by Brink et al. in 2013, ⁵ refuting the tension band principle of dynamic compression at the fracture site after olecranon osteosynthesis. The visibility of their results is represented by the revision of the AO ASIF manual, ¹⁸ changing the term ‘tension band wiring’ to ‘cerclage compression wiring’. ²⁰ Brink et al. investigated the compression forces of transverse olecranon osteotomies created in six fresh-frozen human cadaver elbows. The wires were tightened to 0.50 Nm in a figure-of-eight configuration with one knot each. During active flexion from 10° to 90°, simulated by pulling the brachialis tendon on a hanging arm, they could not observe dynamic compression at the fracture site. This is not surprising since their test setup fails to simulate the increasing counterforce to the olecranon process by increasing passive tension of the triceps tendon during flexion. Hence, the biomechanical setup of Brink et al. fundamentally fails to analyze the tension band principle, which is defined as the conversion of the triceps tendon’s traction force into compression at the fracture site and, therefore, needs to be re-evaluated.

Nevertheless, the cerclage compression wiring construct of Brink et al. ⁵ with an applied tension of 0.5 Nm resulted in an initial compression force of 410 kPa posteriorly and 200 kPa anteriorly, which is in the same range as the results of our study. We also found significantly less compression near the articular surface compared to the posterior aspect. Since we report a strong positive correlation between dorsal and ventral compression, meaning that more compression at the posterior aspect of the osteotomy was accompanied by more compression near the articular surface, the phenomenon of dorsal over-compression as visualized in Figure 3 was not observed in our cohort at a torque of 0.45 Nm.

Another interesting observation of our study was that in both single- and double-knot osteosynthesis, the lateral compression recorded was significantly higher than the medial compression (see Figure 2). This could be explained by the configuration of the V-shaped chevron osteotomy, which was created with an oscillating saw as used in clinical practice. The 0.6 mm blade thickness inevitably resulted in a small bone defect, which may have led to a slight incongruence between the fragments caused by the osteotomy apex, possibly forcing the proximal fragment to tilt to one side. In theory, this construct could be even more susceptible to asymmetric compression from a single knot. Since the knots of all our SKO constructs were positioned on the medial side, where less compression was recorded, and the medial-to-lateral asymmetry was similar compared to DKO constructs, this theory may even strengthen the statement that one knot transfers tension equally to the entire cerclage wire. Another explanation for the recorded asymmetric compression could be the configuration of the pressure sensors with two sensors (PM and PV) on the medial osteotomy plane and one sensor (PL) on the lateral plane (compare Figure 1(b)). Similar results for both single and double knot osteosynthesis still indicate similar force transmission, as this configuration was consistent across all measurements.

The choice of a 1.25 mm stainless-steel tension-band wire was guided by the need to balance construct conformity, ease of handling, and mechanical strength. Thinner wires (≤1.0 mm) are easier to contour but tend to deform or fail under cyclic loading,^21,22 whereas thicker wires (≥1.5 mm) increase construct stiffness at the expense of malleability and may cause soft-tissue irritation or hardware prominence and are subject to other indications e.g. cerclage wiring in femoral shaft fractures. ²³ Our selection therefore represents a compromise that supports both surgical practicality and biomechanical stability and is in line with prior biomechanical research on olecranon TBW.^12,24

In another previous biomechanical investigation, Wilson et al. ⁶ used a plane osteotomy of the olecranon in synthetic bone models to compare the compression after TBW and plate osteosynthesis. Their loading protocol consisted of dynamic motion between 75° and 125° of flexion, generated by a simulated triceps pull transmitted to a screw attached to the olecranon process. Counterforce from the brachialis muscle was provided by a spring connected by a wire to the ulnar tubercle. Compression was recorded with a film pressure sensor. They reported superior interfragmentary compression after plate osteosynthesis compared to TBW, both overall and near the articular surface. They also questioned the TBW principle because they were unable to record increasing interfragmentary compression near the articular surface during motion. From our own previous biomechanical research, ²⁵ we are well aware that it is challenging to simulate tensile forces that are physiologically transmitted by tendons on artificial bone models that lack soft tissue attachments. Analyzing the Wilson et al. test setup image in relation to the simulated triceps pull vector, the maximum flexion was limited to 90° rather than 125° as reported. In addition, the simulated triceps insertion was placed on the extension of the ulnar shaft axis as opposed to the physiological dorsolateral aspect of the ulna. Unfortunately, they did not specify whether TBW was performed with one or two knots, nor did they report the exact amount of tension applied. Although their loading protocol differed from our static protocol and absolute compression values cannot be compared, they stated that TBW resulted in less ventral compared to dorsal compression, which is consistent with our results in olecranon osteotomies. An experimental setup using a plane osteotomy with four pressure sensors positioned anteriorly, posteriorly, medially, and laterally comparing TBW constructs with one and two knots tightened with a defined torque may be of interest for future research.

Subsequent investigations should apply the ultra-thin pressure sensor design used in our study to cadaveric specimens instead of artificial bone. This would allow for the inclusion of physiological muscle attachments and enable a more realistic evaluation of the dynamic behavior of TBW constructs. By capturing joint motion under both active and passive conditions, this approach would provide more accurate counterforces acting on the olecranon and thus a more representative assessment of dynamic compression. Such investigations are necessary to expand our understanding of TBW biomechanics under physiological conditions and to address the inherent limitations of current experimental models.

Another implication for pending biomechanical research is the comparison of different screw configurations in anatomically preshaped plate osteosynthesis, where nowadays mostly angular stable screws are used without the possibility to generate serious interfragmentary compression. In addition to Chevron osteotomies, the evaluated pressure sensors should also be used to analyze simple and complex fracture patterns at the olecranon, as they can be positioned without affecting the overall reduction quality.

One limitation is the in vitro design with the lack of surrounding soft tissue and muscular influences on the bone fragments. The use of artificial bone allows the comparison of identical specimens and facilitates the ethical feasibility of an experimental study without losing biomechanical significance, as the 4th generation sawbones used have been validated to optimally mimic human bone in terms of fracture toughness, tensile strength, compressive strength, fatigue crack resistance, and implant subsidence. ²⁶ In addition, the advantage of using a 4th generation composite bone model is that it provides uniformity that cannot be achieved in cadaver studies.

Another drawback is that only static primary compression at the osteotomy site was investigated. The lack of muscular attachments precluded dynamic analysis of compression forces during active joint motion, which was not included in this study.

Furthermore, the group size of n = 10 may have missed significant differences that would be revealed in a larger cohort, although our sample size is consistent with comparable biomechanical research.^4–6,17

Finally, the present study was performed on chevron osteotomies of the olecranon, which means that our results are not fully applicable to TBW after olecranon fractures.

Conclusions

Contrary to the widely accepted doctrine advocating for the double-knot technique in olecranon tension band wiring, our findings demonstrate that a single knot achieves comparable interfragmentary compression. As it simplifies the procedure and may reduce intraoperative errors without compromising biomechanical performance, the single-knot technique represents a valid alternative for routine surgical use.