If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To investigate factors that cause variation in the mechanical properties of flexor tendon repairs.
One surgeon repaired 50 homogeneous absorbent sticks and 40 porcine flexor tendons with a simple loop, an Adelaide repair, a peripheral over-and-over repair, or a combination of the latter 2 repairs. Ten hand surgeons repaired 1 porcine flexor tendon with the combined Adelaide core and over-and-over peripheral repair. We loaded the samples statically until failure and calculated the variations caused by the testing process, tendon substance, and surgical performance in terms of yield and ultimate load.
Tendon material and surgical performance both caused about half of the variation in the yield load of the combined repair. Surgical performance caused all variations observed in the ultimate load of the combined, peripheral-only, and core repairs. The effect of the tendon material was negligible in ultimate load. The intersurgeon variation was present only in yield load, and it represented one-tenth of the total variation.
The effect of tendon substance on variation of the ultimate load is minimal. In yield load, both tendon and surgical performance are responsible for the variation.
In clinical realm, variation caused by testing is not present, but intersurgeon variation may cause additional variation in yield load. A hand surgeon cannot change the variation due to tendon properties, but with a more meticulous surgical technique, the variation related to the surgical performance can probably be diminished.
In addition, surgeons should be able to perform the repair consistently and with low variability in strength. Two repair techniques having an identical mean strength in a laboratory setting are clinically different if one repair technique is highly variable and the other repair technique is highly reproducible. A repair technique with small variation is desirable because it lessens the probability of weak repairs that are prone to failure (Fig. 1). In other words, when there is variation in technique, there will be a greater risk that the repair will not perform, as it has been shown to in a laboratory setting. When there is a little variation, the repair should perform like it appears to in those same controlled settings.
The potential factors causing variation in the biomechanical performance of a repair are: (1) the structure of the tendon substance, (2) surgical performance (intrasurgeon and intersurgeon reproducibility), and (3) the testing procedure. The proportion of variation attributable to each of these factors is unknown. Understanding the sources and their relative contribution to variability could help optimize repair techniques.
The primary purpose of this study was to quantify the relative variation caused by tendon substance, surgical performance, and testing methodology. We designed a study, which can isolate these factors and identify the sources of variation in a flexor tendon repair model.
Materials and Methods
We tested 10 groups of 10 repairs to isolate the factors causing the variation in the tensile strength of the repairs. Each group had different factors producing variation and we used the differences to quantify the sources of variation (Figure 2, Figure 3). One author (LL) cut samples with a scalpel and performed all the repairs except the 10 repairs that tested the intersurgeon variability. We randomized the order of performing the repairs within the groups to minimize bias arising from a learning curve.
Absorbent sticks (BD Visisorb, Beaver-Visitec International, Inc., Waltham, MA) are homogeneous, soft, nonwoven cylinders with a diameter of 5 mm (Fig. 4). The sticks mimic tendons in shape without variation in the quality of tissue. Dental rolls that have a similar structure as absorbent sticks have been used in previous studies for both surgical training
We assumed that the source of variation in the stick model is negligible and all variation is caused by variation in the testing procedure or by surgical performance.
We used 10 samples to define the methodological variation related to the testing procedure (baseline variation). The investigator placed a 3-0 braided polyester (Ethibond Excel; Ethicon, San Lorenzo, Puerto Rico) simple loop (Fig. 5A) into sticks using a custom-made jig (Fig. 6), which standardized the repair. The rationale for this was to minimize variation caused by both substance and surgical performance. We considered this to represent the variation in the testing procedure. We subtracted this variation from the total variation of the repairs to quantify the technical and tendon-related variation (Table 1).
Next, the investigator placed 10 simple loops free-handed, without the jig, into 10 sticks to quantify the variation caused by surgical performance. We assumed that these samples had no variation due to tendon substance, and therefore, all variation was caused by variation in the surgical technique. We calculated the baseline variation by subtracting the variation of jig-made simple loops from the variation of free-handed simple loops (Appendix A, available on the Journal’s Web site at www.jhandsurg.org). This was followed by 10 core-only Adelaide repairs
(Fig. 5B), 10 peripheral-only repairs (Fig. 5C), and 10 combined repairs on sticks (Adelaide + peripheral repair) to measure the variation caused by surgical performance in these techniques using similar calculations (Table 1). We used a 5-0 polyamide monofilament (Ethilon; Ethicon) in all peripheral repairs.
We used porcine tendons to assess the effect of the tendon substance on the variation of the biomechanical properties. We dissected out 50 frozen thawed porcine flexor digitorum profundus tendons of the second ray of the trotter (FDP-II) and used them in 5 groups, 10 samples in each group. We measured the diameter with calipers. Then we calculated the cross-sectional areas of the tendons (A = π × ab, where a is the semi-minor axis and b the semi-major axis). After repair, the repaired tendons were kept moist in saline-soaked gauzes except when they were being measured. The investigator repaired the porcine tendons using identical repair methods (free-handed simple loop, core-only Adelaide repair, peripheral-only repair, and Adelaide + peripheral repair) to those used for the absorbent sticks.
Finally, 10 hand surgeons performed the combined Adelaide and peripheral repair to measure the intersurgeon performance-related variability. We gave schematic illustrations of the repair techniques (Fig. 5B, C) to the surgeons. Although they were all experienced in tendon surgery, none of them used the Adelaide repair in their clinical work. We chose this technique to minimize the effect of level of experience in the repair.
We performed biomechanical testing of the specimens using a material testing machine (LR 5 K; Lloyd Instruments Ltd, Hampshire, United Kingdom) connected to a computer with NEXYGEN software (Lloyd Materials Testing; AMETEK, Inc., Berwyn, PA). The author who performed the tests secured the samples to the testing machine with clamps 30 mm apart from each other, preloaded to 0.5 N, and then distracted at a constant rate of 20 mm/min velocity until they failed.
We measured yield load using a previously described technique (Lotz et al,
Appendix A, available on the Journal’s Web site at www.jhandsurg.org). We recorded the testing with 2 diametrically placed cameras (Canon EOS 550D and Canon EOS M, Tokyo, Japan) and used slow-motion videos to detect the mode of failure (suture pullout, suture rupture, or knot unraveling).
Analysis of data
The coefficient of variation—also known as relative standard deviation (SD)—was our primary variable representing the variation caused by each factor. The coefficient of variation allows a comparison of variations of datasets having considerably different magnitude (eg, simple loop vs combined Adelaide + peripheral repair). SD is dependent on the magnitude of the measurement: the higher the mean, the higher the SD, and therefore we did not use it in the primary analysis.
We calculated the coefficient of variation using the formula . Thus, throughout the results, the presented variations are percentages of the corresponding mean load (and not of total variation).
To isolate the variation caused by tendon substance, we subtracted the variation observed in sticks from the total variation observed in the corresponding tendons, that is, we assumed that the variation caused by surgical performance and measurement procedure was similar between the corresponding groups, and thus the difference was due to the variations in the tendon substance (Table 1).
To isolate the variation caused by surgical performance, we subtracted the variation caused by testing from the variation observed in sticks. That is, we assumed that the variation caused by stick substance was zero and all observed variance was due to the variation in the placement of the suture (Table 1).
To isolate the intersurgeon variation, we subtracted the variation observed in tendons repaired by a single surgeon from the variation observed in samples repaired by several surgeons (Table 1). Please see Appendix A (available on the Journal’s Web site at www.jhandsurg.org) for full details of the calculations.
One-way variance analysis was used to assess the differences in the cross-sectional areas of the tendons between the groups.
The mean of the tendon cross-sectional areas was similar between the groups (6.09 mm2, SD 0.76 mm2, P = .70).
In the stick repairs, sutures pulled out in all but one repair, which failed by core suture rupture after the pullout of the peripheral repair. In the tendon repairs, 3 peripheral repairs and 3 core repairs failed by suture rupture (in samples repaired by 10 different surgeons), whereas the others failed by pullout.
The means, SDs, and coefficients of variation of each group are presented in Tables 1 and 2.
Table 2Measured Loads, Standard Deviations, and Coefficients of Variation (CoV)
Total variation in the yield loads was the sum of the components of variation: variation caused by the surgical performance and variation caused by the tendon. In the combined repairs, both constituted about half of the total variation. In the core-only repairs, the technical performance constituted two-thirds, and the tendon one-third, of the total variation. In the peripheral-only repair, the technical performance caused all the variation. The total variation of the combined repair was the sum of the variations of its components (variation observed in the core and peripheral repairs). The variation caused by tendon was most pronounced in the simple loops.
Surgical performance caused a higher portion of the total variation in the peripheral-only repairs compared with the core-only or combined repairs, which showed comparable variations (Table 1). Of the total variation observed in the repairs performed by the 10 surgeons, only one-tenth was related to the intersurgeon performance (Appendix A, available on the Journal’s Web site at www.jhandsurg.org).
The coefficients of variation in ultimate load were consistently lower compared with the corresponding variation in yield loads. Surgical performance showed similar variations to those observed in yield load. The variation caused by surgical performance constituted all the observed total variation in the peripheral-only, core-only, and combined repairs. Therefore, the variation caused by the tendon was negligible in all but the simple loop. The intersurgeon variation in ultimate load was also negligible. Thus, the surgical-performance-dependent variation explained mostly all variations in the ultimate load (Table 1).
The variation in combined repairs was the sum of its components as in the yield load: the variation of the core repair constituted one-third of the total variation and the peripheral repair two-thirds.
We found that both the tendon material and the surgical performance constituted about half of the observed variation in the yield load. Only about one-tenth of the total variation was related to the intersurgeon variation in yield load, but in ultimate load, the intersurgeon variation was negligible. Whether we calculated the values from the components of the combined repair (core and peripheral repairs) or from factors causing the variation (surgical performance and tendon), the results were similar and concurred with the observed total variation.
suggested that combined repairs behave like 2 springs under load. The 2 components of the combined repair, the core and peripheral repairs, divide the load, as long as they stay intact, until the yield point. Our results suggest that the observed variation is the sum of the variations of its components, similar to load distribution. When the repair reaches its ultimate load, the peripheral repair is no longer intact but part of that can share loads up to the ultimate load. Therefore, the variation of the combined repair in ultimate load is higher than the variation of the core repair alone.
The yield load is the strength of the intact tendon repair.
Therefore, yield load may be a more important variable to be assessed than ultimate load. Our data show that irregularities in the placement of the suture cause half of the variation, and this may be decreased by standardizing the placement of suture. If the surgeons can lower the variation caused by their performance from 18% to, for example, 13%, the failure rate of 4% could decrease to half (2%) (see Appendix A, available on the Journal’s Web site at www.jhandsurg.org, for the details of this calculation). Moreover, we discovered that the variation increases if several surgeons, who are unfamiliar with the specific core suture, perform the repairs instead of a single surgeon who is experienced in performing the suture. That emphasizes the significance of the learning curve and suggests that surgeons consider using their previously well-practiced repair methods rather than necessarily adopting a new technique about which they might read.
In ultimate load, the variation caused by the tendon was negligible. Our model may overestimate the effect of the surgical performance as the stick model, which was used to assess the effect of surgical performance, may demonstrate magnified technical variation compared with porcine tendon because of different amounts of friction between suture and tendon/stick substance. However, the effect of the tendon must be relatively small because it was not measurable. Thus, we suggest that the variation observed in ultimate load is mostly variation caused by technical performance and the tendon-related variation plays little, if any, role.
The technical performance was also the greatest source of variation in the peripheral-only repairs. Peripheral repair has multiple loops, and it is plausible that it shows greater variance related to the placement of suture. Multiple loops also decrease the tendon-related variation as each loop constitutes only a fraction of the total pull-out resistance, and several loops offset the heterogeneity of the resistance of individual cross-links. Thus, tendon-related variance decreases in more complex repairs. This is supported by the observation that tendon substance was the greatest source of variation in the simple loop repair, where irregularities in tendon substance are more important because the simple loop depends on the individual cross-links between tendon fibrils.
The variation in yield load was greater compared with the variation in ultimate load. There are 2 plausible reasons for this. First, the analysis of the yield load is sensitive to inaccuracies related to the methodology, because small irregularities in the load-deflection curves can be interpreted as yield points. However, that is also a strength of the yield load measurement, because it is more sensitive to revealing early rupture of the repair. Secondly, based on the spring analogy,
the asynchrony in the load sharing of the core and peripheral repairs has a greater effect on yield load compared with ultimate load.
There are several limitations in this study. The format of this study does not allow a power calculation, because greater sample size does not affect the size of SD or the coefficient of variation. The group size of 10 samples was chosen based on the number of surgeons working at the clinic. Also, the assumption that the coefficient of variation in the baseline group consisted solely of the methodological variation related to the testing procedure is not entirely justified as, for example, the cutting of the stick and the placement of suture through the jig are also potential sources of variation. Thus, variations derived from surgical performance may appear lower than they actually are because the variation of the baseline group is subtracted from the variations of surgical performance groups (Table 1). In addition, handling of nonwoven material and tendon tissue may cause variation, which was not included in our model. Furthermore, only one core suture configuration was used in this study. The Adelaide core repair was used because of its biomechanical properties and tendency to fail by suture pullout instead of suture rupture. If suture rupture was more frequent, the variation would most likely decrease, as the testing would increasingly resemble testing of the suture material, which would obscure the contribution of technical variation. Finally, clinically tendons are loaded cyclically, and the results are applicable only to static testing. However, there is no clearly established method of carrying out cyclic testing, and cyclic testing provides a cycle count instead of a load with mean and SD.
To conclude, our study shows that the total variation in the yield load of the combined 4-strand Adelaide repair consists of similar variations caused by the tendon and the surgical performance. In ultimate load, the variation caused by the tendon is negligible and the variation is mostly due to the surgical performance. Because small variation is desirable, and the tendon properties cannot be affected, a focus on finding ways to reduce the variation in the surgical performance is important.
Matthew James, MSc, is acknowledged for the linguistic revision.
Determination of yield load and ultimate load
Yield load is the strength of the intact tendon repair and, thus, the beginning of irreversible deformation of the repair.
The difference of the yield load and the ultimate load is that a repair can usually sustain higher loads (ultimate load) even though the integrity of the repair has been lost at the yield point resulting in a slight gapping between the tendon ends and inevitable failure of the repair if the loading is continued.
with a 0.1 mm offset. We used custom computer software to determine yield load. We draw the offset line 0.1 mm under the steepest slope of the load-deflection curve. Yield load was at the intersection of the offset line and the load-deflection curve. If the load-deflection curve did not express an identifiable yield load, we used ultimate load instead. The computer software recorded ultimate load as the highest load monitored by the load cell.
Determination of coefficient of variation and isolation of partial variations
We calculated means and standard deviations for each group. On the basis of these, we calculated coefficients of variations (CoV) by . To isolate the variations of specific factors, we used the following equation: (Fig. E1). In the equation, equals the overall variation within the group with multiple variation-inducing factors. These include that is to be calculated, and that equals the overall variation within another group in which the same variation-inducing factors are present, except for the variation . For example, we isolated the variation due to heterogeneity in the pull-out resisting properties of porcine tendons by comparing simple loop repairs between tendons and sticks (Table 2). Because the aim of the study was to investigate clinically significant variation within the biological tissue (the tendon), the sum of the variations of specific factors cannot exceed the variation of the tendon repairs. Subsequently, if the value in the equation to be squared was negative (ie, the subtrahend is larger than the minuend), the calculation could not be fulfilled, and we assumed the variation of the specific factor to be equal to the overall variation. In addition, we also calculated the variation of the combined repair as the sum of the variation of the core repair and the variation of the peripheral repair. The equation is applicable if normal distributions of 2 or more independent, additional factors are assumed.
Determination of proportions of CoVs
Some CoVs are expressed as proportions in the text. For example, intersurgeon variation proportion of the total variation was calculated by dividing the CoV of the several surgeons (32%) by the subtraction of the latter and the CoV of the combined repairs (29%) resulting in 1/10. Calculation is based on the yield load of the combined repair.
Assumptions and calculations of clinical example
The variations reposted in the clinical example are variations of the combined repair in yield load with following assumptions that may not be entirely true: the variation of the porcine tendons is similar to the tendons of flexor tendon injury patients (1), the surgical performance of the tendon repair is similarly consistent in a laboratory and in an operating room (2), and the failures take place during the rehabilitation and not if the tendon repair is subjected to accidentally higher loads (3).
As was stated in the main text, the variation due to tendon properties was 17% and the variation due to the surgical performance was 18%, thus, yielding the total variation of 25%. The dataset was considered as a normally distributed sample with the mean of 62 N—the assumption of the average yield load of the used flexor tendon repair (4). The distribution describes the dispersion of yield loads with the number of samples bearing single yield load. Because forces up to 35 N were assumed to exist during the rehabilitation process
(5), a cutoff point could be set at 35 N. If the failure rate was 4%, 4% of samples would leave under the cutoff point. If surgical performance hypothetically improved, the variation would decrease from 18% to 13% and the total variation would be 21%. With the same cutoff point of 35 N, only 2% of samples would leave under the cutoff point. Thus, failure rate would decrease from 4% to 2%.
The search for the ideal tendon repair in zone 2: strand number, anchor points and suture thickness.
This study has been financially supported by the Vappu Uuspää Foundation . The sponsor had no role in the study design; the collection, analysis, or interpretation of the data; the writing of the manuscript; or the decision to submit the manuscript for publication.