Development of a method for fabricating polypropylene non-articulated dorsiflexion assist ankle foot orthoses with predetermined stiffness

Abstract

Background: A non-articulated plantarflexion resist ankle foot orthosis (AFO), commonly known as a posterior leaf spring AFO, is indicated for patients with motor impairment to the dorsiflexors. The AFO is often custom molded to a patient's lower limb anatomy and fabricated from polypropylene. There are no established guidelines for fabricating this type of AFO with predetermined stiffness of the ankle region for normal walking speeds. Therefore an AFO may not meet the biomechanical needs of the patient.

Objectives: Quantify the biomechanical ankle stiffness requirement for an individual with complete dorsiflexor impairment and develop a method for fabricating an AFO with ankle stiffness to meet that requirement.

Study Design: Experimental, bench research.

Methods: The literature on sagittal biomechanics of non-pathological adults was reviewed to derive the stiffness of the ankle during loading response. Computer models of 144 AFOs were created with geometric variations to account for differences in human anthropometrics. Computer-based finite element analysis was employed to determine the stiffness and safety factor of the models.

Results: Stiffness of the AFOs ranged from 0.04 to 1.8 Nm/deg. This ample range is expected to account for the stiffness required for most adults with complete dorsiflexor impairment. At 5° deflection the factor of safety (ratio of strength to stress) ranged from 2.8 to 9.1. A computer program was generated that computes AFO stiffness from user-input variables of AFO geometry. The stiffness is compared to a theoretically appropriate stiffness based on the patient mass. The geometric variables can be modified until there is a close match, resulting in AFO design specification that is appropriate for the patient.

Conclusion: Through validation on human subjects, this method may benefit patient outcomes in clinical practice by avoiding the current uncertainty surrounding AFO performance and reducing the labor and time involved in rectifying a custom AFO post-fabrication.

Clinical relevance

This method provides an avenue for improving patient outcomes by avoiding the current uncertainty surrounding non-articulated plantarflexion resist ankle foot orthosis performance. The ability to quantify the biomechanical ankle stiffness requirement for an individual with complete dorsiflexor impairment provides insight into how other AFO types should be designed as well.

Background

A non-articulated dorsiflexion assist ankle foot orthosis (AFO) is indicated for patients with flaccid equinus (foot drop) secondary to motor impairment to the ankle dorsiflexors. This AFO is often custom-molded to a patient’s lower limb anatomy and composed of a thermoplastic material such as polypropylene. The region posterior to the malleoli typically displays a narrow profile that transitions into the footplate distally and the calf band proximally. The flexibility of this region allows limited motion at the ankle joint. When weight is applied to the AFO footplate, the stiffness of the ankle region and its tendency to return to its thermoformed shape generates a resistance to plantarflexion of the ankle. ¹ The AFO serves to maintain the forefoot clear of the ground during swing and to simulate eccentric activity of the dorsiflexors during loading response. From mid to terminal stance the AFO permits ankle dorsiflexion with limited resistance. ^2,3 The ankle stiffness of the dorsiflexion assist AFO is dependent upon several factors: material properties, plastic thickness, cross-sectional shape of the flexible region and configuration of the trimlines. ^3,4

There are no established methods for custom fabricating this type of AFO with predetermined ankle stiffness. An orthotic practitioner relies on trial and error due to an inability to anticipate the stress distribution in the orthosis. ⁵ The stiffness is not known in advance of fabrication and is typically not measured afterwards. Because many orthoses are over- or under-engineered, they may not meet the biomechanical needs of the patient. There is the potential for the AFO to have negative effects on knee biomechanics. Excessively stiff dorsiflexion assist AFOs behave more like solid ankle AFOs by delaying loading response. This subjects the user to an increased knee flexion moment that may threaten stance stability if the patient lacks inadequate knee or hip extensor strength. ^6,7,8 Lehmann concluded that there is a positive correlation between the amount of plantarflexion resist for swing phase and the amount of knee instability induced after heelstrike. ⁶ A study on non-pathological subjects demonstrated that a solid ankle AFO may require 20% more activity of the quadriceps than an appropriate dorsiflexion assist AFO. ⁹

Studies suggest that the AFO should provide a level of plantarflexion resist that simulates eccentric contraction of the dorsiflexors, thereby allowing a limited amount of loading response plantarflexion to occur. ^7,8,10,11 For this reason a dorsiflexion assist AFO is contraindicated for individuals presenting with plantarflexor spasticity that would overpower the stiffness control. ¹² A study by Geboers et al. involving immobilization of the ankle and its effect on dorsiflexor strength concluded that people using an AFO after neurological damage risk further strength loss due to disuse. ¹³ A study by Hesse et al. concluded that reduced dorsiflexor activity may lead to disuse atrophy and long-term dependence on the orthosis. ¹⁴ Thus an AFO that provides excessive ankle stiffness could delay rehabilitation if the patient were expected to recover from the neurological damage. Patients should be fitted with an AFO having the minimal ankle stiffness required to promote a functional gait pattern. However, if the AFO stiffness is insufficient there will be inadequate biomechanical control of ankle motion during gait. Loading response may be too brief, which may create an excessive knee extension moment. Considering these consequences, it would be beneficial if orthotists had the ability to design a custom thermoplastic dorsiflexion assist AFO with predetermined stiffness that meets the needs of a patient. The purpose of this research was to develop a method for accomplishing this.

Materials and methods

In order to arrive at a method for achieving AFO ankle stiffness, it was first necessary to understand the biomechanical control necessary to restore function in an individual with complete dorsiflexor impairment. Thus the literature on the sagittal biomechanics of the non-pathological ankle was reviewed. Research has shown that the eccentric activity of the dorsiflexors during loading response can be modelled as a torsional spring with a linear spring rate. ¹⁵ Published gait data from non-pathological subjects was used to determine the AFO spring stiffness. Winter reported instantaneous values of internal ankle joint moment normalized for body weight throughout the walking gait cycle, with cadences ranging from approximately 85 to 125 steps per minute. ¹⁶ It was possible to calculate the average stiffness of the ankle over the first 6% of the gait cycle that corresponds to the controlled plantarflexion beginning at heel strike (Appendix A1). This stiffness was in units of torque per angular change per body mass (Nm/deg/kg). There was a positive correlation between walking speed and calculated ankle joint stiffness. Palmer provided stiffness data for 10 non-pathological subjects, which was converted into these units (Appendix A2). ¹⁵ Palmer’s mean ankle joint stiffness values for slow, normal and fast walking speeds of approximately 0.9 to 1.8 metres per second showed no such correlation, yet the overall mean was of similar magnitude to Winter’s normal-speed walkers.

In order to create fabrication guidelines that consider differences in human anthropometrics and biomechanical requirements, a wide range of AFO geometries needed to be tested. However, the labor and materials involved in fabricating and experimentally testing many thermoplastic AFOs would have been cost-prohibitive. Therefore a more efficient method involving computer-based finite element modelling (FEM) was employed. FEM reduces a physical structure to smaller substructures known as finite elements. These substructures and their interrelationships are used to determine the magnitude and distribution of stress and deflection under loading conditions. Although FEM can be very accurate, it is recognized that the multiple AFO geometric variables measured in this study are all sources for error. Therefore it was necessary to first evaluate the accuracy of FEM when used for the purposes of this study. A pilot study compared 10 polypropylene dorsiflexion assist computer models against real AFOs. ¹⁷ The real AFOs were fabricated to have geometry that closely matched their respective computer models at the level of the ankle where the majority of stress and strain was expected to occur. For each AFO, theoretical FEM and empirical testing resulted in less than 10% difference in stiffness.

One variable affecting the stiffness of a dorsiflexion assist AFO is the cross-sectional shape of the posterior extension. This shape is based on patient anatomy. A transverse sectional view of the anatomy at the level of the ankle demonstrates an irregular posterior contour (Figure 1). This contour is unique for every individual. Quantifying the expected stiffness of an AFO that matches the posterior anatomy would require prior acquisition of a 3D patient model and the use of analytical software. This technology is available but has not been universally applied to orthotic clinical practice. To be useful and promote acceptance, a new fabrication technique should be based upon current practice. Today this generally involves generating a positive model of the patient’s anatomy via a negative impression. The model is then rectified to ensure an AFO fabricated from thermoformed plastic will fit and function correctly. In order to base new guidelines on this practice and to avoid the issue of complex anatomy, it was decided that the region where the AFO bends should be of a generic shape with pre-determined dimensions (Figures 1 and 2). This was done by rectifying the posterior positive model to create a constant radius in the transverse plane that extends over a vertical range, beginning proximal to and ending distal to the malleolar apices. The other aspects of the AFO would remain custom molded to the anatomy to ensure an appropriate fit.

Figure 1.

Transverse ankle cross section with irregular posterior anatomy (——) and rectification radius (- - -).

Figure 2.

Vertical range of rectification (- - -) to create the posterior extension.

Computer models of 144 AFOs were created. Each AFO model had a unique flexural zone geometry to account for variations in human anatomy and to provide varying levels of stiffness. Geometric measurements that defined the flexural zone were: transverse plane inner radius, outside arc length, plastic thickness, sagittal zone height, and the radii of the transition from the flexural zone to the rest of the AFO shape (Figures 3 and 4). The inner radius was equal to the rectification radius and the rectification height was intended to be ample enough to span both the height of the zone and its transitional radii. Zone radius and height were based upon anthropometric values for bimalleolar width and malleolar height among adult males and females. ^18,19 All measurements that influenced the models are listed in Table 1. For the 5th percentile, mean and 95th percentile, bimalleolar diameters approximately matching transverse radii of 30.2, 36.5 and 44.5 mm, respectively, were used. These corresponded to the inner radii of Schedule 40 PVC pipe couplers with nominal dimensions of 2, 2.5 and 3 in. For each percentile and mean group, three different zone heights were used to ensure that the zone height and transitional radii were ample enough to avoid plastic impingement on the region of malleolar prominences as predicted by the anthropometric values for ankle height and distal malleolus height. The mean foot length, calf height and calf circumference values were used to define the remainder of the AFO model shape. Stature and body mass did not influence the models but are included for reference. For each of the small, medium and large inner radii, 48 models were created with combinations of four plastic thicknesses, four arc lengths and three zone heights. A summary of the combinations are listed in Table 2.

Figure 3.

Sagittal flexural zone geometry.

Figure 4.

Transverse flexural zone geometry.

Table 1.

Anthropometric data.

measure	female 5%ile	mean*	male 95%ile
stature (cm)	152.3	170.6	188.6
body mass (kg)	47.4	70.6	97.7
foot length (cm)	22.2	25.7	29.2
approximate foot torque (NM)	0.70	0.96	I.26
bimalleolar breadth (cm) †	6.0	7.0	7.8
ankle height prox. to malleoli (cm)	9.1	12.5	15.8
distal med. malleolus hight (cm) †	5.7	6.6	7.5
diff. of ankle height/malleolus height	3.4	5.9	8.3
calf height (cm)	28.7	34.5	39.6
calf circ. (cm)	30.7	35.9	41.3

*

average of male and female means.

†

female data estimated based on ratio of female/male foot length.

Table 2.

Flexural zone geometry combinations.

radius (mm)	arc length (mm)	zone height (mm)	thickness (mm)
30.2	30, 40, 50, 60	30, 40, 50	3, 4, 5, 6
36.5	40, 50, 60, 70	40, 50, 60	3, 4, 5, 6
44.5	50, 60, 70, 80	50, 60, 70	3, 4, 5, 6

Computer-assisted design software (SolidWorks®, Solidworks Corporation, Concord, MA) was used to create solid models (Figure 5). Regardless of flexural zone geometry, a 12.5 mm transitional radius was used to blend the proximal and distal flexural zones into the remainder of the structure (Figure 3). A finite element analysis tool (CosmosWorks®, Solidworks Corporation, Concord, MA) was used to determine AFO stiffness and peak stress. The mechanical properties of polypropylene from several orthotic and prosthetic plastic manufacturers were averaged and entered into the software as a linear elastic material (Table 3). Each model was restrained at the proximal shank and a normal force was applied to the footplate to represent the resultant of the distributed load from the foot. The load required to elicit 5° deflection in the direction of plantarflexion was determined, utilizing an iterative technique that accounted for geometric changes occurring during a large displacement (Figure 6). ²⁰ This amount of deflection approximates the amount of ankle plantarflexion that occurs during loading response at a non-pathological self-selected walking cadence. The required force and peak von Mises stress were recorded for each model. The moment arm was defined as the distance between the applied force and the centroid (centre of bending) in the transverse plane. The moment was calculated and converted to stiffness units of moment per angle (Nm/deg). From the peak von Mises stress, the factor of safety (ratio of strength to peak stress) at 5° plantarflexion was calculated.

Figure 5.

Computer-generated AFO model.

Figure 6.

AFO deflection post-analysis.

Table 3.

Mechanical properties of polypropylene homopolymer commonly used in orthotics. Poisson's ratio of 0.35 was used.

type and manufacturer	flexural modulus (MPa)	flexural strength (MPa)
OrthoformTM (S&L Plastics Inc., Nazereth PA USA) http://www.slpinc.cc/custom/Orthoform.pdf	1276	N/A
Proteus®, (Quadrant Plastics, Reading PA USA) http://americanplastics.net/Catalog/catalog_04.htm	1240	48.3
Polystone® (Rochling Engineering Plastics, Dallas NC USA) http://www.roechling-plastics.us/pdfs/P%20Rochling%20Grey.pdf	1241	N/A
ThermoLyn® (Otto Bock, Minneapolis, MN USA) Otto Bock Technical Information 7.1.1, p. 4	1300	N/A
Mean	1264	48.3

Results

Stiffness of the AFOs ranged from 0.04 to 1.8 Nm/deg. The results using the iterative loading technique were on average 11.2% greater than if geometric changes during deflection had not been considered. At 5° deflection the factor of safety (ratio of strength to stress) ranged from 2.8 to 9.1. For each zone height and plastic thickness combination (12 per radius size), stiffness was plotted against the four arc lengths (Figure 7). Using second order polynomial trendlines, it was possible to input arc length and determine the stiffness with a high level of accuracy in relation to the data points (coefficient of determination, R²> 0.99). The same was true for factor of safety versus arc length (Figure 8). The maximum differences between the stiffness data points and the trendlines were 3.53%, 1.82% and 0.76% for AFOs having small, medium and large transverse radii, respectively. The maximum differences for factor of safety were 5.08%, 1.31% and 1.83%. The medium AFO trendline in Figure 7 demonstrates that a large arc length and a small zone height result in the greatest stiffness. However, from Figure 8 it is seen that they also result in the lowest factor of safety. A large stiffness can be achieved without sacrificing flexural strength by using a thicker plastic, thereby making it possible to use a smaller arc length and/or larger zone height. From the results of finite element analysis, peak stresses always occurred along the inner edges of the flexural zone, particularly at the transitional radii (Figure 9). If an AFO were to become excessively stressed, its failure would begin in one of these locations.

Figure 7.

AFO stiffness vs. arc length for medium inner radius.

Figure 8.

AFO factor of safety vs. arc length for medium inner radius.

Figure 9.

Inner aspect of AFO with darkened regions representing areas of greatest stress under load.

Discussion

For each size of cross-sectional radius listed in Table 2, a calculation tool was created in Excel 2003 (Microsoft Corporation, Redmond WA) spreadsheet format. The calculation tool programming was based upon the data point trendline equations exemplified in Figures 7 and 8. Increased resolution of thickness was possible through linear interpolation between equations representing consecutive whole number thickness values listed in Table 2. The calculation tool determines the stiffness of an AFO based on flexural zone geometry and compares the stiffness to that recommended for a person with complete dorsiflexor impairment (sample in Appendix B). This tool serves as a guideline for an orthotist to fabricate an AFO.

A physiological stiffness factor of 0.01 Nm/deg/kg for a non-pathological person was used to represent the lower end of the range of values derived from the literature. ^15,16 This is a value normalized for body mass and when multiplied by a subject’s mass results in expected non-pathological ankle stiffness during loading response. It was chosen based on an expectation that the average patient treated with an AFO will ambulate within the slow to natural cadence/speed range of the non-pathological subjects studied by Palmer and Winter (Appendix A). The average stiffness factor of the slow and normal walking speeds from both Palmer’s and Winter’s studies was 0.0115 Nm/deg/kg. A low-magnitude factor was further presumed to be appropriate because a dorsiflexion assist AFO bends on an axis posterior to the anatomical ankle axis. ²¹ The reduced heel lever results in less plantarflexion moment that the ground reaction force can induce on the AFO structure as compared to the anatomical heel lever of a non-pathological subject. Additionally, any intrinsic resistance to plantarflexion remaining in the impaired ankle will further reduce the stiffness requirements of the AFO design.

The stiffness factor used in this study only accounts for patient body mass while other variables such as walking speed and limb length may also influence ankle stiffness. The stiffness of a passive dorsiflexion assist AFO cannot adapt to changing demands; therefore an ankle stiffness appropriate for a ‘slow’ to ‘normal’ speed was the basis for analysis. Further gait studies are necessary to determine the optimal physiological stiffness factor. This could be accomplished by repeating the research of Winter and Palmer with a larger sample size of non-pathological subjects and accounting for stature or limb length. It would be ideal to test the subjects at walking speeds that match those of the self-selected speeds for users of dorsiflexion assist AFOs. Presently, a practitioner has the ability to fabricate an AFO with excess initial stiffness and subsequently reduce stiffness through reconfiguration of trimlines to optimize patient gait. ³

A pilot study was performed that compared stiffness of actual dorsiflexion assist AFOs with computer models of similar AFOs, resulting in less than 10% difference. ¹⁷ Being able to provide an AFO with a predetermined stiffness that is 90% accurate should be acceptable to most practitioners. It is anticipated that a similar level of error would be present in AFOs fabricated using the technique described. Empirical testing would be beneficial since the theoretical results in this study encompassed a greater number of AFOs having a broader range of geometries. Sources of error include differences in thermoformed polypropylene flexural modulus and strength as compared to manufacturer data and how well the computer models reflect a fabricated AFO shape. The maximum difference between the average polypropylene flexural modulus and individual values was 2.8%, indicating consistency between the selected manufacturers. A thermoformed AFO with transverse flexural zone radius that is larger than the positive model due to ‘spreading’ from thermoforming insufficiency will be more flexible than intended. Techniques to avoid or rectify this phenomenon are necessary.

Regardless of initial accuracy, the results presented do not take time-dependent phenomena into account. Continuous cyclic loading on an AFO results in decreased stiffness, although original stiffness is gradually restored after cyclical loading ceases. ^3,22 Fatigue can lead to permanent stiffness reduction and failure. This is difficult to predict and is dependent upon excessive strain and surface imperfections. ²³ Thus it is important to design an AFO for a patient’s weight and activity and carefully finishing it to ensure smooth trimlines. Creep deformation is caused by long-term exposure to levels of stress below the yield strength. ^22,23 The gait cycle subjects an AFO to frequent reversed loads where creep is not an issue. However, prolonged static loading of an AFO, such as during sitting with the ankle plantarflexed, could eventually lead to permanent creep deformation of the AFO footplate in the direction of plantarflexion. Fabricating an AFO in dorsiflexion (accounting for the shoe last to create a shank to floor angle of less than 90°) will help avoid consequences of minor deformation, such as catching the toe during swing phase. There is no accepted standard for initial angle. ⁶ The goal is to prevent the combined masses of the foot and shoe from plantarflexing the AFO beyond a neutral sagittal ankle alignment during swing. An estimation of swing phase plantarflexion caused by the foot and a shoe having a mass equal to 50% of the foot mass was calculated for the 5th percentile, mean and 95th percentile anthropometric values (Table 4). Based on the results, a dorsiflexion angle of 2–3° is suggested to account for foot and shoe mass and future creep deformation. Exact values are beyond the scope of this research and are complicated by the varying masses of footwear a patient may wear.

Table 4.

Estimate of swing phase deflection based on anthropometric data and AFO stiffness.

measure	female 5%ile	mean*	male 95%ile
stature (cm)	152.3	170.6	188.6
body mass (kg)	47.4	70.6	97.7
foot length (cm)	22.2	25.7	29.2
foot mass (kg) †	0.79	0.93	1.1
post, heel to foot mass center (cm) †	9.1	10.5	11.7
approximate foot torque (Nm)	0.70	0.96	1.26
approximate foot + shoe torque (Nm)	1.05	1.43	1.83
AFO stiffness (Nm/deg)	0.474	0.706	0.977
AFO deflection during swing (deg)	1.5	1.4	1.3

*

average of male and female means.

†

female data estimated based on ratio of female/male foot length.

A single AFO was fabricated using the instructions described in Appendix C to assess the technical challenges and additional time involved when compared to fabricating a traditional dorsiflexion assist AFO. Fabrication instructions can be accessed at http://sites.google.com/site/afostiffness/. The additional step of creating a constant radius build-up posterior to the malleoli added a few minutes to the fabrication process. More time was spent carefully delineating the flexural zone and finishing the trimlines. The fabrication process took approximately 50% more time than would have been spent on a standard dorsiflexion assist AFO and no unexpected technical challenges were encountered. As with any new task, repetition of the process reduces the time involved. Stiffness of the AFO was not quantified and was not tested on a human subject. The AFO was loaded into plantarflexion and dorsiflexion. During dorsiflexion it was visually observed that the center of bending occurred in the distal flexural zone and that diametrical strain occurred in the heel section just distal to the flexural zone. In plantarflexion the center of bending occurred closer to the mid-level of the flexural zone and no diametrical strain was visually observed. Dorsiflexion of the AFO required less force than plantarflexion. This has been reported to occur in traditional dorsiflexion assist AFO designs as well. ²² It also supports the established contraindication of a dorsiflexion assist AFO for a patient that has significant plantarflexion weakness as it provides poor control of tibial advancement during stance. ^3,12,24

Conclusion

This research serves as a starting point for stiffness control of custom polypropylene dorsiflexion assist AFOs. The guidelines and calculation tools are intended for experimental use and validation through further research. If proven to be a reliable technique, orthotists would have a new method for AFO fabrication that avoids trial and error. Less time would be spent on modifications to achieve appropriate AFO performance. The practitioner would have greater confidence in the durability of an AFO. By recording the stiffness of a patient’s AFO it would be less difficult to reproduce it if it were to need replacement. The overall result would be improved patient outcomes and time savings for both patient and practitioner.