Testing the Biomechanical Protection by Sacral Border Dressings in a Laboratory‐Based Model

ABSTRACT

Pressure mitigating dressings are one component of pressure ulcer/injury prevention strategies. There are many such devices on the market, but little data to compare them by. Herein we share our results of comparing sacral border dressings' capacity to mitigate deformations due to lateral forces. A modified version of a published digital image correlation‐based technique was used to monitor the distortions of a cast silicone sheet under varying loads. Four sacral border dressings were compared to no dressing with three replicates for each. Empirical displacements within the gel were quantified via image analysis and compared via two‐way ANOVA followed by Tukey's HSD test. Full field displacements were interpolated from the empirical data and the full field strains and shear were calculated and plotted. All four dressings were statistically significantly different from the control, but not among each other with magnitudes on the order of a hair's breadth. The maximum strains measured among the dressings were not significantly different in the direction of applied force, but two of the dressings were found to differ in the orthogonal direction, and those same dressings had computed strains equal to the control in the direction of the applied force. Our lab‐based data indicate that four commercially available sacral border dressings reduce lateral strain under most conditions and shear under all conditions compared to no dressing. While the absence of clear tissue failure criteria limits direct clinical translation, these findings provide a straightforward and quantitative foundation for pre‐clinical evaluation of sacral dressings.

Key Points

• Lateral forces are mitigated by sacral border dressings.

• The interplay between strain and shear is complex.

• Devices of similar construction share characteristic behaviors.

List of Abbreviations

2D‐DIC

two‐dimensional digital image correlation

ANOVA

analysis of variance

DoF

degrees of freedom

NaN

not a number

p‐adj

adjusted p‐value

Sum_sq

sum of squares

Tukey's HSD

Tukey's honestly significant difference

1. Introduction

When otherwise non‐load bearing skin bears a load for a prolonged time, injuries have long been known to form. The names used for these injuries have evolved with time but have most commonly and authoritatively been referred to as pressure ulcers [1], with pressure injuries being a newly emergent term. Colloquially known as ‘bedsores’, these tissue defects tend to occur on load‐bearing surfaces beneath bony prominences of the human anatomy. The initial insult is believed to be an ischemia‐inducing compression of the skin, which induces hypoxia and subsequent necrosis. The weakened tissue is then believed to be torn via a shearing mechanism [2, 3].

The prevalence of pressure ulcers in the United States of America is estimated to be between 2.5–3.0 million annually [4, 5], but this is believed to be substantially underestimated based on most studies being centred in acute care settings which neglect assisted living or nursing homes [6]. In a recent study, the prevalence of these injuries was estimated to increase from 2.03% to 2.22% between 2009 and 2019; suggesting an increase in incidence [5]. Over that same time period, the mean cost for an inpatient increased by about 48% [5]. An earlier estimate has the costs in the United States at $26.8B annually with the majority of those costs for the later stages of the injury [7]. As such, there is increased pressure on institutions and care providers to mitigate risks for the formation of pressure ulcers due to malpractice liability [8].

The most recent guidelines from the Wound Healing Society lead with patient positioning and support surface choices, but the precise details of when to reposition the patients are not yet clear and the results are not universal [6]. The guideline further calls for maintaining the head of the bed angle at the lowest angle possible [6]. While not cited by the guidelines, this comports with the findings of pressure measurements made in human studies for the sacrum and thigh, but stands in contrast for pressures on the coccyx and lateral side of the sacrum [9]. Guideline 1.3 calls for pressure reducing surfaces which reduce shear and provide microclimate control, which somewhat mirrors Guideline 8.11 which recommends prophylactic dressings comprised of silicone adhesives and multilayer foams in certain circumstances [6, 10, 11, 12]. These dressings tend to distribute pressures, reduce friction, have advanced capillary networks among the foams, and have vapour‐permeable surfaces which are expected to manage surface moisture [13, 14]. After the initial patent expired for the first of these dressings, many newly designed ones have become available. Herein, three of these recently commercially available sacrum‐protecting devices will be compared with the predicate dressing in a lab‐based strain and shear model.

2. Materials and Methods

2.1. Model Setup and Experimental Design

In order to measure the displacements communicated to the skin due to lateral forces, a variation of our previously reported two‐dimensional digital image correlation (2D‐DIC) system [15] was used with a bead‐loaded, custom cast, silicone (Ecoflex 0030, SmoothOn Inc., Macungie, PA, USA) sheet to monitor the strains in the silicone under physiological loads [9]. This was chosen based on its near‐to‐skin softness and its relative transparency. In other models, this type of silicone gel was an effective model to assess deep muscle tissue injuries [16], and the material properties were measured and compared to those reported in the literature. The Ecoflex 0030 had a reported shear modulus G of 22.081 kPa, which was reported to be in the range of values for transverse, active, and relaxed muscle; skin; and fat. The silicone was treated with corn starch to keep it from being tacky, and one edge was fixed to the table (Figure 1A). The rest of the silicone was allowed to slide along the device's imaging stage in a model of subcutaneous fat (Figure 1B). A range of normal pressures were generated by loading a custom 3D printed sled (Figure 1C) with Olympic‐style ‘micro‐weight’ plates ranging from 0 to 2.27 kg in 113.63 g increments which are sold in imperial units in increments of 0.25 lb up to 1 lb. Based on the geometry of the sled, this led to a pressure range from 0 to 172.38 mmHg in increments of ~8.62 mmHg. The interface was wrapped in cotton‐polyester blend sheet material. A stepper motor capable of up to 216 N of lateral force was used to generate in 0.625 mm stepwise displacement for 40 steps at a total of 25 mm of displacement in the x direction.

FIGURE 1.

The experiment setup. (A) A top view of the table and the beaded loaded gel. (B) The mechanical model of the experiment. The gel is fixed on its left‐most edge and is allowed to ‘roll’ along its length in the direction of the lateral force. (C) The 3D printed sled model which is designed to carry weights to generate pressure. (D) The strain and shear model for full field distributions in continuous materials.

2.2. Experimental Protocol

Three separate dressings were used as replicates for each trial and three trials without dressings were used as a control. The dressings used are commercially available and include: (1) Allevyn Life Sacrum (Smith & Nephew plc, Watford, England), Mepilex Border Sacrum (Mölnlycke Health Care, Gothenburg, Sweden), Optifoam Gentle EX (Medline Industries, LP, Northfield, IL, USA), and OptiView (Medline Industries, LP, Northfield, IL, USA). Abbreviated versions of these names are used from hereon.

Each dressing, in turn, was affixed beneath the restraint and the bottom surface was treated with corn starch. The experiment was begun with the sled unloaded for the first trial (0 mmHg). After each pressure level, the sled was removed, and the gel lifted to relax the gel and mitigate the risk of hysteresis (residual stresses in the gel). The gel was laid down again and the sled replaced with the next increment of weight. The next trial was then begun and this process of deloading and reloading was continued until all dressings and pressures were completed.

2.3. Data Processing and Analysis

The raw images were all processed using the same Python script and scikit‐image [17] whereby they were stacked in groups of 40 (the input steps) cropped to remove surrounding structures. The red channel was then isolated and the images were thresholded using scikit‐image's filters.threshold_local with a parameter of 71 and an offset of 10. The image was then converted to binary prior to using scikit‐image to remove small holes and then small objects. Scikit‐image was then used to identify the particles and their position and size. A frame‐to‐frame linking of each particle was done by choosing the particles with the least change from one frame to the next. Particles exceeding the maximum input displacement were ignored and removed. The dataset was then filtered for only particles with a continuous presence throughout all of the frames. One particle of ~1400 px representing a static sticker dot on the acrylic table was utilised to estimate the systemic noise of the system.

2.4. Software and Statistical Methods

The primary goal of this model is compare displacement reduction as it is the only truly accessible empirical variable. Full field displacements, strains, and shear will be estimated for completeness. Coordinate data was processed and cleaned in Python version 3.12.1 using Pandas version 2.2 [18]. Empirical displacements in 2D were calculated and compared step‐by‐step and cumulatively for each set of conditions using Jupyter Notebook version 7.1.0 [19], NumPy version 1.26.3 [20], and matplotlib version 3.8.2 [21]. The negative hypothesis was that no differences in maximum displacements exist among the groups. The maximums were chosen due to that being where failure will happen and the central tendency of the bead displacements not having a relevant physical meaning since they are not replicates of one another, but are instead spatially distinct measures. The hypothesis was tested via a two‐way ANOVA (n = 3, α = 0.05) for treatment (n = 5) and pressure (n = 20) at the maximum input displacement (Step 40, 25 mm) followed by a pairwise‐Tukey HSD post hoc test using Statsmodels version 0.14.2 [22] for the empirical displacement datasets. While the full combination of pairs was calculated using the post hoc test, only pairings within each pressure group were investigated; leading to an over‐correction in the p‐values due to the number of extraneous comparisons made. Displacement values were analysed as absolute values as negative displacements represent a directional component that would not necessarily cancel out an equivalent positive displacement.

The empirical data were then used to estimate full displacement fields using a radial basis function interpolation via RBFInterpolator from scipy.interpolate. Strains and shear in the 2D plane were calculated from the interpolated displacement data using the Green Lagrange formulas (Figure 1D), with the derivative terms calculated discretely by the NumPy library's gradient function. Interpolated maximum displacements, strains, and shear were plotted versus input pressure. The interpolated field replicates (n = 3) at the highest input pressure were then spatially averaged and used to generate a heatmap image to show the spatial distributions of values. Due to there being 1000‐fold more interpolation‐generated points than empirical points, interpolated data is only provided descriptively and not statistically tested.

3. Results

3.1. Gross Observations

The binary image stacks for a subset of the pressure inputs were processed and flattened to provide a rough image of what occurred during the experiment. The outlines of the beads form ‘smears’ where there is strain and/or shear (Figure 2). From the gross observation, it is clear that all four dressings are reducing the lateral displacements communicated to the sub‐dressing silicone surface when compared to the Nil control. At worst, the dressings under 172.4 mmHg appear to be equivalent to the nil under 34.5 mmHg. The visible differences among the dressings appear to be rather small and will require statistical analysis with the numerical data.

FIGURE 2.

Particle ‘Smear’ analysis. At 0 mmHg (top row), there is no discernible difference even in the control. As the pressure increases, there are more ‘smears’ in the control, but still not much discernible among the dressings.

3.2. Empirical Displacements

The empirical displacements in the x‐direction (‘u’) and y‐direction (‘v’) were first explored on an input displacement step‐by‐step manner as both averaged maximums and individual replicate trajectories. For the cumulative x displacements (Figure 3), the drastic difference between the Nil control and the rest of the dressings is clearly visible. A pair of groups not seen in the displacement smears (Figure 2) becomes apparent with the Gentle and Mepilex paired and the Allevyn and Optiview close, but not overlapping much. More interestingly, the replicate loading curves have characteristic shapes which are roughly shared by three of the dressings with the Gentle and Mepilex being nearly indistinguishable and the Allevyn sharing the rough profile, but attenuated. The Optiview and Nil have more unique profiles. A similar pattern, but with a different shape, was seen for the y‐direction displacements (Figure 4). The degree of overlap of the replicates differs among the groups and evidences the variance within each group. The cumulative y displacements showed a similar pattern, but there was more overlap in the averaged maximums, and the characteristic behaviour of the Optiview showed two different classes of behaviour.

FIGURE 3.

Stepwise empirical maximum displacements in the direction of the applied force (x‐direction). (A) Cumulative maximum displacements frame‐by‐frame at a pressure of 172.38 mmHg. (B–F) The empirical distribution of displacements vs. step in the x‐direction. The scales for the dressings are roughly one fourth that of the Nil so that the characteristic pattern can be seen.

FIGURE 4.

Stepwise empirical maximum displacements orthogonal to the direction of the applied force (y‐direction). (A) Cumulative maximum displacements frame‐by‐frame at a pressure of 172.38 mmHg. The error bars represent 1 standard deviation (n = 3). (B–F) The interpolated distribution of displacements in the y‐direction. The scales for the dressings are roughly one half that of the Nil so that the characteristic pattern can be seen.

Separate statistical hypothesis testing for each direction of displacement found extremely significant differences among dressings, pressure and an interaction between dressing and pressure for both directions (Figure 5). Post hoc testing of x‐direction displacements (Figure 5A) reveals that all dressings except Optiview (p = 0.0779) differed from Nil by 17.25 mmHg, while Optiview differed by the next increment at 25.87 mmHg (p = 0.0085). The differences compared to Nil remained significant throughout, and the dressings never differed from one another at any pressure (p _min = 0.9005). Conservative analysis use inter‐pressure comparisons focused on estimating the displacement magnitude difference in the x‐direction between groups that can be resolved at the level of variance present herein reveals a range between 1.9808 and 2.2007 mm with p‐values of 0.0779 and 0.0139, respectively. A post hoc power analysis was performed with G*Power for dressing, pressure and the interaction between the two and a similar pattern was seen in the y‐direction post hoc analysis (Figure 5B). All dressings but Optiview (p = 0.5255) differed from Nil by 25.87 mmHg, while Optiview required two more pressure increments by 43.12 mmHg to diverge from the Nil control (p = 0.0005). Again, the differences persisted for the remaining pressure inputs for all dressings. One major difference is that Optiview diverged from Gentle (p = 0.0026) and Mepilex (p = 0.0330) with lower displacements in the y‐direction at the maximum pressure input of 172.38 mmHg. While more clear in the y‐direction data (Figure 4B), Optiview has a unique pressure response profile in both directions. Both this unique profile and the slope and profile of the Nil control break the co‐linearity with the other groups explaining the pressure and dressing interaction seen in the ANOVA analysis (Table 1). This evidences that Optiview responds to pressure differently than the other dressings.

FIGURE 5.

The empirical pressure vs. maximum displacement behaviour in the (A) x‐direction and (B) y‐direction. The ANOVA indicated a strong interaction term (Table 1), which is evidenced here visually by the differences in slope (e.g., intersecting) among some of the test groups. The asterisk marks the location of the start of statistically significant differences (p < 0.05) and “ns” means “not significant”. Actual p‐values can be found in Table 2.

TABLE 1.

Two‐way ANOVA looking at the effect of dressing, pressure and the interaction between dressing and pressure on empirical displacements.

	Sum of squares	DoF	F	p
x‐Direction
C (pressure)	29.033	20	177.979	8.16 × 10−120
C (dressing)	95.050	4	2913.317	1.15 × 10−182
Pressure: C (dressing)	24.835	80	38.059	2.00 × 10−91
Residual	1.713	210
y‐Direction
C (pressure)	2.073	20	102.460	1.26 × 10−96
C (dressing)	4.415	4	1090.986	3.22 × 10−139
Pressure: C (dressing)	0.992	80	1225.250	1.11 × 10−47
Residual	0.212	210

TABLE 2.

A subset of the post hoc testing results and consideration of the magnitude of the differences amongst the dressings at the highest pressure.

Comparison	Direction	p	Mean difference (mm)
Allevyn vs. Gentle	x	0.9005	1.4011
Allevyn vs. Gentle	y	1.0000	0.3238
Allevyn vs. Mepilex	x	0.9885	1.2435
Allevyn vs. Mepilex	y	1.0000	0.2213
Allevyn vs. Optiview	x	1.0000	0.5297
Allevyn vs. Optiview	y	0.8187	−0.5167
Gentle vs. Mepilex	x	1.0000	−0.1576
Gentle vs. Mepilex	y	1.0000	−0.1025
Gentle vs. Optiview	x	1.0000	−0.8714
Gentle vs. Optiview	y	0.0026	−0.8405
Mepilex vs. Optiview	x	1.0000	−0.7138
Mepilex vs. Optiview	y	0.0330	−0.7380

3.3. Interpolated Displacements

The lateral/tangential displacement being input into this model leads to a spatial response in the silicone in the sub‐dressing space. While the smears in Figure 2 give some insight, the empirical data are sparse. The interpolated displacements were plotted versus pressure for the x‐ (Figure 6A) and y‐direction (Figure 7A) as was done for the empirical data (Figure 5). The plateauing of the Nil seen in the empirical data was less pronounced in the interpolation and the somewhat parabolic response of Optiview was recapitulated in the interpolated data (Figure 6A). The drastic differences between the dressings and Nil control are again apparent. Slight differences are seen with a similar sized loading region (green) for Allevyn, Gentle and Mepilex, with a lower intensity for Allevyn. The distribution of Optiview (green) was in a similar location, but appears to be a slightly lower level than Gentle and Mepilex and spread over a larger area than all dressings. Both Mepilex and Optiview also appear to share a well‐distributed trailing displacement (cyan) behind the local minima (dark blue/black).

FIGURE 6.

Interpolated values for x‐direction displacements. (A) The mean of the maximums among the three replicates plotted vs. pressure. (B–F) Heatmaps of the spatial distribution of the x‐displacements.

FIGURE 7.

Interpolated values for y‐direction displacements. (A) The mean of the maximums among the three replicates plotted vs. pressure. (B–F) Heatmaps of the spatial distribution of the y‐displacements.

The results in the y‐direction were similar with the differences in the empirical and interpolated data. The dressings' y‐displacement response curves to pressure (Figure 7A) similarly reflected the empirical data (Figure 5B), again with a less aggressive reduction in response in the Nil control at higher pressures. The drastic differences in the distribution between the dressings and Nil control (Figure 7B–F) are clear, with the differences among the dressings being slightly ‘hotter’ hotspots in nearly identical spatial distributions. The symmetric distribution of both positive and negative values of these distortions are the as the sled distorts and displaces the silicone in opposing directions and the silicone relaxing after the sled has passed. The persistence of the left side couplet represents y‐displacements that persist after the sled had passed. The couplets have consistently lower intensities/displacements than the right‐side couplets, evidencing at least some relaxation as the sled passed.

3.4. Interpolated Strain

The strain calculated from the interpolated displacement data in the x‐direction had some unexpected differences among the groups (Figure 8). In the pressure versus absolute value of the maximum strain chart (Figure 8A), the Optiview again had a parabolic‐like response across the pressures tested, but this time the Nil control did too. The declination of the strain at the highest pressure tested even led Gentle and Mepilex to have strain in the x‐direction on parity with the Nil control (Figure 8A); though the spatial distribution is narrower for those two dressings compared to Nil (Figure 8B,D,E).

FIGURE 8.

Interpolated values for x‐direction strain. (A) The mean of the maximums among the three replicates plotted vs. pressure. (B–F) Heatmaps of the spatial distribution of the x‐strain.

The differences in the x‐direction strain chart are seen in the spatially averaged heatmaps, where the ‘heat’ of the hottest spot is both smaller and ‘hotter’ in the Gentle than the Nil control (Figure 8B,D). Another pattern seen is that Allevyn, Gentle, and Mepilex have three visually distinct zones of loading whereas the Nil control and Optiview have two (Figure 8B–F). By the final input pressure, Optiview had the lowest strain, but its strains throughout the rest of the pressure levels either exceeded or equalled those of Allevyn. Finally, the zone of greatest strain for the Nil control is more centralised in the field, while the highest zones for all of the dressings are biassed towards the front of the strain input slide's trajectory.

The pressure response of the treatments of strain in the y‐direction was more in keeping with previously seen patterns (Figure 9), with less distinction among the dressings, with one exception. The Optiview again began diverging towards lower values at higher pressures (Figure 9A). The generalised pattern of y‐direction strains was likewise consistently a six‐zone pattern (Figure 9B–E) with an exception for Optiview having a four‐zone pattern (Figure 9F) similar to the y‐displacement data (Figure 6F).

FIGURE 9.

Interpolated values for y‐direction strain. (A) The mean of the maximums among the three replicates plotted vs. pressure. (B–F) Heatmaps of the spatial distribution of the y‐strain.

3.5. Interpolated Shear

The interpolated shear data presents a stark difference between the dressings and Nil, but intermingling of the dressings' values along the pressures tested (Figure 10A). As was the case in most metrics, the Allevyn tended to be the lowest, with an occasional crossing by Optiview. A stark and unexpected difference between the Nil and dressings was seen in that the ‘X’ pattern distribution had vertical symmetry with respect to the direction of shear whereas the dressings all had symmetry about an approximately 45° angle. This 45° symmetry was well developed in the Allevyn, Gentle and Mepilex (Figure 10C–E), but not in the Optiview which truly wasn't very symmetric at all (Figure 10F).

FIGURE 10.

Interpolated values for xy‐direction shear. (A) The mean of the maximums among the three replicates plotted vs. pressure. (B–F) Heatmaps of the spatial distribution of the xy‐shear.

4. Discussion

In side‐by‐side testing, all dressings appeared to largely mitigate lateral displacements, in both directions, in comparison to using no dressing at all. Three of the dressings had a roughly similar multilayer construction while one of them appears to be more monolithic (Optiview). When observing the experimental loading and application of lateral force, it is clear that the sled was sliding across the surface of the dressings. The sliding is evidence that the primary mode of action mitigating lateral forces is due to reduced friction; a property that all dressings tested shared in varying degrees. However, there were differences observed among the dressings.

The cumulative displacement behaviour was characteristic based on device design. The multilayer dressings had similarly shaped curves for displacements in the x‐ and y‐direction (Figures 3 and 4). The x‐direction profiles had an early point of inflection and a later one. Allevyn's curve was attenuated in its magnitude compared to Gentle and Mepilex, while Gentle and Mepilex were nearly indistinguishable. Optiview had an initial ‘hump’ where the others had a point of inflection and point of inflection later on. The y‐direction profiles were similarly distinct. It is possible that after further input displacement the Optiview would have the feature, or that the characteristic behaviour arose due to the independent slippage possible among the layers of the multilayer dressings. Though given the drastic differences in the y‐direction profile, the internal differences are more likely the source of the observed differences.

The primary outcome sought is the reduction in empirical displacements, as displacements are what generate strain and shear. Statistical testing revealed differences among the groups tested, with the primary difference being the dressings versus Nil. Post hoc testing did not find any statistical differences among the dressings at any pressure for displacements in the direction of the force (x‐direction); though all of the dressings differed from Nil one pressure increment sooner than Optiview did (Figure 5A). By appearance of the chart, it seems that this was due to variations in the plotted profile for Optiview which may be due to more noise in Optiview's performance based on the increased size in error bars at these oscillations. A similar pattern was seen in the y‐direction with Optiview departing from Nil 2 pressure increments later than the rest. However, the y‐direction profile appears to have a major point of inflection at or around 94.8 mmHg and the y‐direction displacements begin decreasing with increased input pressure. The downward trend resulted in statistical differences between Optiview and Mepilex and Gentle. While statistical differences were found among the dressings, the magnitude of the differences warrant more attention. For the point of greatest difference in the x‐direction, between Allevyn and Optiview (60.33 mmHg), the magnitude of the difference was 0.1571 mm (p = 1.0) while in the y‐direction, again between Allevyn and Optiview (77.57 mmHg), the difference was 0.102 mm (p = 0.211). Ignoring the p‐values and assuming that the differences are true, the magnitude is on the order of the diameter of a hair and are therefore not substantial.

Material failure and rupture occur due to strain and shear, but these properties are not empirically accessible. Strains and shear were calculated from interpolated whole field displacements in the x‐ and y‐direction. The interpolated maximum displacement data matched the gross appearance of the empirical data including Optiview's unique profile evidencing that the interpolation maintained the characteristics of the distribution of displacements. For calculated strain, both Optiview and Nil had what appeared to be either a noisy asymptote or a point where strain and shear began decreasing. The data in the y‐direction support an asymptote. Unexpectedly, both Mepilex and Gentle had strain in the x‐direction on parity with Nil at the highest pressures. The performance for other computed deformations was similar to the empirical displacements with noisy oscillations of relative behaviour that were drastically different from Nil as the pressure increased. We attribute this to the displacements and computer distortion values being at the limit of the system's discernability. As stated above, the empirical differences are on the order of a hair's breadth; the differences in the computed values are likewise not expected to be relevant. A review of the literature did reveal elastic and shear moduli, but no strain or shear failure criteria. The lack of primary evidence for ranges of tolerable strain and shear on skin precludes a solid judgement of whether the magnitude of remaining strain and shear, when compared to Nil, is clinically relevant.

Aside from the performance of the dressings, there is a major conceptual discontinuity between the clinical literature and basic mechanics of materials concepts. When forces are imparted on a restrained material, the material deforms, and if it deforms beyond its capacity to recover, the material fails. The concepts explaining this physical process are divided into the empirical applied forces and displacements, and the inferred quantities of stresses and strains. The forces arise from the patient's body weight and repositioning by the patient and/or caregiver, while the skin's distortion is governed by the nature of the skin and how much of the imparted force is borne by the skin. The inferred quantities of stress and strain are a means to understand material failure—tissue rupture or delamination in the clinical context. Mechanically, there are two forms of strain: normal strain and shear strain (typically used as strain and shear, respectively). Ultimately it is strain and shear that explain material failure where the material deforms beyond its ability to recover. Deformations that would make a rectangle stretch to a longer rectangle are strains while those that would make the rectangle look more like a rhombus or a non‐normal parallelogram would be shear. It is here that the current clinical understanding departs from established mechanics of materials fundamentals. A recent review highlighted the need for mitigating what the authors called displacements due to ‘shear forces’, which they define as loads parallel to the skin [2, 3]. The direction of the force doesn't determine whether the load will induce shear or strain. In the context of the skin, the extent of subcutaneous fat and degree of skin immobilisation by the pinch point between the patient's boney prominences and the support surface will largely govern whether the skin glides on the subcutaneous fat (predominantly strain) or does not (predominantly shear). We say ‘predominantly’ because it is also important to know that a real material under load will always have both strain and shear distributed throughout as reviewed in Zhang and Roberts [3]. At present, there is a paucity of clinical evidence of the extent of strain versus shear deformations and the mode of skin failure in load bearing skin. To ensure that these crucial gaps in the literature are addressed in future studies, empirical forces and displacements should be measured and reported [2] in as many dimensions as is feasible, and those displacements should be used to compute both normal and shear strains. The key consensus from the literature reviewed is that distortions and forces acting along the skin, not just the crushing forces, should also be investigated and targeted for reduction. The results from our empirical analysis with lateral forces demonstrate substantial reductions in empirical lateral displacements in the presence of sacral dressings, but with only minor insubstantial differences among the dressings being found.

The performance of some of our replicates was less reproducible than others. Whether this is a characteristic of the dressing or the apparatus is not known at present. The performance of the Nil is evidence that some of the variance is in the device itself as there was no dressing present. In early trials, hysteresis (due to moistening of the dry lubricant) was found in the silicone sheet which is now managed by lifting and replacing the gel on the table and re‐applying the dry lubricant. It is nearly impossible to get the gel in the same location and with the same initial strain (due to straining under its own weight) during this process, so we consider this intrinsic noise of the system. With this setup, a difference between 1.9 and 2.0 mm was necessary for a p < 0.05 by Tukey's HSD with three replicates; the extent that this needs to be improved is waiting for primary clinical evidence of relevant displacement magnitudes. While the statistics did reveal a specific resolvable difference among the groups, the number of replicates is limiting in another sense. Sources of variance, such as dressing batch variance, that occur less frequently than in 1‐in‐3 cases could have been missed altogether. Another limitation is that after the experimentation, literature about good practices in digital image correlation was found indicating that the labelled area (e.g., beads) should occupy ~50% of the field of view [23]. Future iterations will include a gel with a higher degree of labelling. Finally, as designed, the method used herein is a more benchtop‐oriented assessment. While some of the performance characteristics (like surface friction reduction) are not expected to change much in clinical use, the distributions of displacement and distortions, and the localization of motion constraint (e.g., pinch points) are expected to. Future laboratory iterations including anatomically correct casting of the clear gel with embedded bone structures is expected to make the model more clinically relevant. Additionally, given Optiview's transparency, trans‐dressing imaging enables in‐patient studies with labelled patients' sacra. The other non‐transparent dressings may be transparent in the near infrared II region, which may also enable their use in empirically measuring deformations in the clinic.

Despite the distance yet to traverse for clinical alignment of this laboratory model, the data herein support a high degree of similarity among the currently tested devices, with the measured difference among them being on the order of a hair's breadth. The tools and approaches developed herein enable more complex analysis in models which approach more clinically relevant models, and when paired with one of the dressings, may even allow clinical study of in situ deformations in covered sacra. Our next steps will be to adapt this system to 3D displacements and to be able to image anatomical geometry. These future efforts are expected to aid in the development of novel interventions and to provide guiding insights into studying in vivo tissue mechanics.

Funding

The funding for the initial series of these experiments and the dressings tested was provided by Medline Industries Inc.

Ethics Statement

The authors have nothing to report.

Conflicts of Interest

Dr. Gibson was contracted by Medline Industries to perform this work and has in the past been a consultant for Medline Industries. Jack G. Sherry declares no conflicts of interest.

Data Availability Statement

All data and code are available upon request.