Influence of foot positions on the spine and pelvis




The management of knee osteoarthritis includes the use of wedged shoe insoles to unload the affected knee compartment. Although the biomechanical effects of shoe insoles on the knee joint are known and described, only little is known about their influence on the pelvis and spine. Therefore, the purpose of this study was to evaluate the effects of different foot positions, such as how they could be achieved by shoe insoles, on pelvic position and spinal posture.


A total of 51 test subjects were measured for this study. The different foot positions (inner and outer margin increase, positive and negative heel height) were simulated with a specially designed stand platform. A rasterstereographic device was used to measure the immediate effects of the simulated foot positions on the pelvic position and spinal posture.


Positive and negative heel heights as well as an increase of the outer margin of the platform led to significant changes of the pelvic tilt. The pelvic torsion also changed significantly during positive heel height changes of 10 and 15 mm and increases of the outer margin of the foot. No significant changes were found between foot position and spinal parameters.


The results of our study support the existence of a kinematic chain, where changes of foot position also led to significant alterations of the pelvic position. Whether these changes could lead to long-term pathologic alterations still needs to be evaluated. However, in our setting, no correlation between foot position and spinal posture changes was found.


Osteoarthritis (OA) of the knee occurs in approximately 30% of the population at age 65 years or older (1). Knee OA is associated with significant pain, disability, impaired quality of life, and a reduction of the mobility in the elderly more than any other disease (2). Management of OA includes pharmacologic treatment, surgery (osteotomy and arthroplasty) (3, 4), and physiotherapeutic treatments, which are playing an important role in the rehabilitation and prevention of OA (5). Osteoarthritic changes are 10 times more common in the medial than in the lateral compartment of the knee (6–8). The loss of cartilage tissue and joint space usually results in a varus misalignment of the knee, which leads to a higher knee adduction moment (9). For the first time, Sasaki and Yasuda (10, 11) reported in 1987 about the potential of laterally wedged shoe insoles for the treatment of medial knee OA. Since then, many authors have described the effects of shoe insoles on the biomechanics of the lower extremity. Kerrigan et al demonstrated that a 5° laterally wedged insole reduces the knee adduction moment of the knee by 6% in medial knee OA (12). In addition, Maly et al observed a significant varus to valgus shift in knee alignment on static radiographs (13). Other clinical applications of wedged insoles are the reduction of stress fractures and soft tissue injuries in runners (14, 15). Although the biomechanical changes were described by many authors, only little is known about the effects of different foot positions and heights, e.g., how they are used in the treatment of knee OA, on pelvic position and spinal posture. Artificial changes of foot positions by shoe insoles may lead unintentionally to secondary changes in the joints above the knee. Since the loading of the lower extremities depends on a nearby symmetric loading of the legs during bipedal stance as well as during locomotion, it must be assumed that especially a unilateral change of body scheme may also result in a compensatory change of posture to keep the center of gravity within the center of pressure. Therefore, the purpose of this study was to evaluate the effects of different foot positions, such as how they could be achieved by shoe insoles, on the pelvis and spine.

Significance & Innovations

  • In addition to changes in the joint position of the lower extremity, the pelvis and spine should also be investigated when evaluating for foot position changes.

  • Rasterstereography may be helpful to better understand changes in the kinematic chain of the body due to variations in foot position.


Healthy volunteers from our clinic were examined for this study. Included were test subjects without preexisting leg, pelvis, or spinal abnormalities. Exclusion criteria were a pelvic obliquity due to a functional leg length discrepancy greater than 1 cm and obesity with a body mass index >35 kg/m2, which could impede the automatic detection of the anatomic landmarks by the measuring system we used. Another exclusion criterion was back pain during the previous year for longer than 2 days. None of the subjects had any serious medical condition. A total of 51 subjects (33 women, 18 men) participated in this investigation (Table 1). All of the subjects were informed about this study and gave their written consent as well as were given the option to quit participation at any time. Additionally, the local ethics committee approved the study protocol.

Table 1. Epidemiologic data of the measured test subjects (n = 51)*
 Age, mean ± SD (range) yearsWeight, mean ± SD kgHeight, mean ± SD cm
  • *

    Altogether, a mean height of 175.75 cm and a mean weight of 71.65 kg were measured for this study. The mean age of the subjects was 27.49 years.

Women (n = 33)27.0 ± 6.5 (20–51)62.0 ± 10.6168.6 ± 5.9
Men (n = 18)27.9 ± 3.5 (22–34)80.9 ± 7.5182.9 ± 5.2

The 3-dimensional analysis of the spinal posture and the pelvis position was conducted with the rasterstereographic device formetric 4D (Diers International). Rasterstereography is a method for stereophotogrammetric surface measuring of the back, which was developed in the 1980s by Drerup and Hierholzer (16, 17). Based on the principle of triangulation, it allows a contact- and radiation-free determination of the body surface (18–20). Two cameras from 2 different angles thereby record the back shape. In rasterstereographic measurements, a slide projector used as an optical equivalent of an inverse camera replaces one of the cameras. Parallel white light lines are projected on the back surface of the patient by the slide projector. The 3-dimensional back shape leads to a deformation of the parallel light lines, which can be detected by the camera (21, 22). Anatomic landmarks are thereby automatically captured by assigning concave and convex areas to curved light pattern (23, 24). With these anatomic fix points, the system is able to calculate a 3-dimensional model of the human spine, and clinically relevant parameters such as surface rotation, kyphosis, or lordosis angle of the spine can be determined. The position and the 3-dimensional orientation of the pelvis can be measured by detecting the location of the 2 lumbar dimples with the rasterstereographic measuring system. Conclusions about the position of the pelvis can be drawn because the 2 lumbar dimples (left lumbar dimple [DL] and right lumbar dimple [DR]) are in close relation to the underlying posterior superior iliac spines of the pelvis (25). Therefore, it is possible to determine the pelvic obliquity from the position of the dimples to each other. From the orientation of the skin surface over the lumbar dimples, it is also possible to draw conclusions on pelvic torsion around the transverse axis (17). The high accuracy and reliability of this technique was shown in multiple studies (22, 26). In comparison to radiographic measurements, Hackenberg et al (20) showed a good correlation coefficient of 0.89 for frontal deviation of the spine in subjects with scoliosis. Good reliability was also found by Goh et al (27), who analyzed healthy subjects.

A special constructed stand platform (Diers International) was used to achieve variations in foot posture (Figure 1). The platform dimensions are 510 × 520 mm divided into 2 independent standing areas with a width of 250 mm and a length of 380 mm for each leg. Driven by an electric motor, the height as well as the tilt of the standing area could be controlled separately for the left or the right leg. The maximum possible vertical height is 40 mm with steps of 1 mm. Maximum increase of the medial edge of the platform is 40 mm, which results in a maximum tilt angle of the platform of 9.1°, and the maximum increase for the lateral edge is 30 mm with a respective angle of 6.8°. The maximum increase of the rear or front part is 30 mm, resulting in a maximum tilt of 3.3°. To avoid changes in foot positions between the different measured conditions, a coordinate system was printed on the simulating foot platform. With this coordinate system it could be assured that the feet were placed exactly in the same position of the platform between the different conditions. The different changes in foot position were not randomized. However, all measurements were conducted using the same protocol (as described below) to ensure standardized examinations in all test subjects. This protocol was repeated 3 times and the mean values were used for analysis.

Figure 1.

A test subject is shown standing on the constructed stand platform during a measurement. The dimensions of the stand platform are shown. Three spindles, driven by an electric motor, allow the simulation of different foot positions.

Before we started the actual measurement, the test subjects were placed for 60 seconds on the simulating platform to adapt to the simulated platform. One recording of a foot position took 7 seconds and the device was set at a recording frequency of 15 Hz during the measurements of the subjects. Therefore, with one measurement we had a total number of 315 single recordings for each condition.

The following different foot (left and right) positions were simulated, and then the subjects were rasterstereographically measured: inner margin (IM) increase of the foot (+5 mm, +10 mm, +15 mm), outer margin (OM) increase of the foot (+5 mm, +10 mm, +15 mm), positive heel (PH) height (+5 mm, +10 mm, +15 mm), and negative heel (NH) height (−5 mm, −10 mm, −15 mm). For purposes of this study, it is necessary to define certain terms regarding the parameters that were measured with the device, which are listed below.

Pelvic tilt (mm) is the different height of the 2 lumbar dimples (DL to DR) from each other in mm. For a positive value, the DR is higher than the DL based on a horizontal line of the surrounding system.

Pelvic torsion (degrees) is the torsion of the surface normal on the 2 lumbar dimples. A positive pelvic torsion signifies that the right hip bone is oriented further anterior than the left one.

Maximal lateral deviation of the spine (degrees) is the maximum deviation of the midline of the spine from the direct connection vertebra prominens (VP) and middle between the DL and DR (DM) in the frontal plane.

Lordotic angle (degrees) is the angle between the surface tangents on the twelfth thoracic vertebral body (T12) and DM points.

Kyphotic angle (degrees) is the angle between the surface tangents on the VP and T12 points.

Trunk imbalance (mm) is the lateral distance between the VP and the DM. A positive value signifies a shift of the VP to the right, and a negative value signifies a shift to the left.

Surface rotation (degrees) is the angle between the horizontal components of the surface normals on the symmetry line. A positive value of the surface rotation describes a surface rotation to the right (spinous process points to the right).

Statistical analysis.

All of the data were checked for Gaussian distribution by the chi-square test and are shown as the mean ± SD. Unifactorial analysis of variance (ANOVA) was calculated to check for changes in pelvic and spinal parameters between the left and the right legs. Repeated-measures ANOVAs were calculated to analyze changes in all defined variables across the 4 different height conditions of the platform. When global significance for each condition was determined, Bonferroni post hoc analysis was employed in order to determine changes across the heights. Differences between the left and right sides were tested using Student's t-test for paired samples. Two-factorial ANOVA with repeated measures on a single factor was used to check for changes between women and men. The level of significance was set at P values less than 0.05. Statistical analysis and graphic presentation were prepared using SPSS software, version 19.0. A post hoc power analysis was performed using the software G*Power 3.1 (Faul, Erdfelder, Lang, and Buchner).


The immediate effects of different wedged foot positions on spinal posture and pelvis position were evaluated in 51 subjects (33 women and 18 men) during this study. In the neutral position, the mean ± SD pelvic tilt was 0.4 ± 5.03 mm for all measured subjects. Figure 2 shows the mean pelvic tilt of all measured platform heights (5, 10, 15 mm) in all different simulated foot positions (IM, OM, PH, NH). A lift of the IM did not result in any significant changes in pelvic tilt, either if the left side or the right side was increased. If the OM is increased, significant (P < 0.05) changes can be shown between the neutral position and each of the 3 different height steps (5, 10, 15 mm). The observed increase in pelvic tilt is also significant when the heel is lifted (PH) or lowered (NH) as well between the neutral position and between the height steps (P < 0.05).

Figure 2.

The mean values of the pelvic tilt for the different simulated heights and foot positions are shown. The results of both sides, left (shaded bars) and right (solid bars), are included. Changes of foot position led to an increase of pelvic tilt in all simulated cases. The greatest changes of the pelvic tilt occurred when the lateral margin of the platform (OM) was increased. IM = inner margin; OM = outer margin; PH = positive heel; NH = negative heel; DL = left lumbar dimple; DR = right lumbar dimple.

A mean ± SD pelvic torsion of 0.51° ± 2.03° was found in the neutral position for all 51 measured subjects. Figure 3 shows the mean pelvic torsion for all simulated different foot positions and heights. Significant differences can be observed between the reference value and the height steps 10 mm (P = 0.003) and 15 mm (P = 0.001) lifting the OM, if the left side is increased. Lifting the OM or PH 10 mm or 15 mm on the right side showed no significant differences for the pelvic torsion (P > 0.05). Also, no changes were observed for all other conditions.

Figure 3.

The mean values of the pelvic torsion for the different simulated heights and foot positions are shown. The results of both sides, left (shaded bars) and right (solid bars), are included. Changes of foot position led to a significant increase of pelvic torsion when the lateral margin (OM) of the stand platform was increased. IM = inner margin; OM = outer margin; PH = positive heel; NH = negative heel.

The immediate effects of different foot positions and heights on the spinal posture were measured with a rasterstereographic device in this study. The results of all different foot positions on the spinal parameters are shown in Table 2. Regarding Table 2, only minor changes of the trunk imbalance, the surface rotation, and the lateral deviation were measured by changing the foot position and height. In addition, no significant changes between simulated foot position and kyphotic and lordotic angles were found.

Table 2. Changes of the spinal posture of the different simulated foot positions and heights*
 Trunk imbalance, mean ± SD mmSurface rotation, mean ± SD °Lateral deviation, mean ± SD mmKyphotic angle, mean ± SD °Lordotic angle, mean ± SD °
  • *

    Both simulated sides are shown. All differences found in spinal posture due to different foot positions were nonsignificant. OM = outer margin; IM = inner margin; PH = positive heel; NH = negative heel.

  5 mm8.11 ± 5.903.65 ± 1.5511.33 ± 4.9050.84 ± 9.4142.35 ± 9.19
  10 mm8.35 ± 6.253.87 ± 1.6011.63 ± 4.7650.49 ± 9.5142.79 ± 9.70
  15 mm8.04 ± 5.993.68 ± 1.4411.72 ± 4.2250.37 ± 9.6842.86 ± 9.05
  5 mm8.33 ± 5.784.02 ± 1.6312.15 ± 4.4150.95 ± 10.0042.89 ± 9.22
  10 mm8.67 ± 6.014.43 ± 2.0113.18 ± 5.1551.00 ± 9.8143.48 ± 9.94
  15 mm8.62 ± 6.654.91 ± 2.2814.96 ± 5.9851.52 ± 9.6543.06 ± 9.41
  5 mm7.84 ± 5.953.97 ± 1.6811.66 ± 5.2250.96 ± 9.2542.33 ± 10.00
  10 mm7.59 ± 5.963.99 ± 2.0012.33 ± 4.6051.39 ± 9.4042.48 ± 9.41
  15 mm8.05 ± 6.394.44 ± 2.1013.13 ± 5.6351.37 ± 9.3842.90 ± 9.97
  5 mm8.91 ± 6.333.79 ± 1.9411.65 ± 4.5750.57 ± 9.5542.47 ± 9.51
  10 mm8.41 ± 6.873.74 ± 1.7911.81 ± 4.6451.66 ± 9.4442.84 ± 9.47
  15 mm7.49 ± 6.264.12 ± 1.8312.44 ± 4.9050.96 ± 10.0742.46 ± 10.05
  5 mm8.33 ± 6.243.87 ± 1.5911.75 ± 4.5550.88 ± 9.2342.79 ± 9.56
  10 mm7.80 ± 6.074.11 ± 1.8111.89 ± 4.1251.12 ± 9.8042.31 ± 9.25
  15 mm8.12 ± 5.313.91 ± 1.5011.57 ± 4.4951.03 ± 9.8842.75 ± 9.63
  5 mm9.15 ± 7.093.96 ± 1.5411.51 ± 4.1051.71 ± 9.7743.36 ± 9.67
  10 mm8.23 ± 6.574.13 ± 1.5011.55 ± 4.9751.28 ± 10.0043.77 ± 9.77
  15 mm8.46 ± 6.454.28 ± 1.5011.46 ± 4.4751.70 ± 8.8943.47 ± 9.70
  5 mm8.02 ± 6.233.64 ± 1.5510.51 ± 3.9251.13 ± 9.4042.91 ± 9.39
  10 mm8.17 ± 6.003.67 ± 1.4710.47 ± 4.2050.70 ± 9.5342.51 ± 9.61
  15 mm8.28 ± 6.163.88 ± 1.4611.22 ± 4.6250.79 ± 9.5743.63 ± 10.19
  5 mm8.90 ± 7.423.81 ± 1.6310.94 ± 4.6250.66 ± 10.5742.30 ± 10.00
  10 mm8.17 ± 5.593.55 ± 1.4110.23 ± 3.8550.98 ± 10.0943.12 ± 9.56
  15 mm7.48 ± 5.783.81 ± 1.6010.80 ± 4.8850.80 ± 9.6442.71 ± 9.53

A comparison between the results of women and men was performed to check for sex differences. The mean pelvic tilt in the neutral position did not differ significantly (P > 0.05) between women (4.7 mm) and men (3.1 mm). We also found no differences in the response to different foot positions on the pelvic position and spinal posture between women and men.


The biomechanical effects of different shoe insoles on the knee joint are widely described (1, 9, 28). Specifically, laterally and medially wedged insoles are commonly used to treat early stages of knee OA by reducing the load on the affected arthritic knee compartment. This can be achieved by a reduction of the knee adduction moment during walking by 4–12% (2, 12, 29). In addition to the reduction of the knee adduction moment, insoles seem to be able to shift the center of pressure location of the ground reaction forces (30). Because of a lack of randomized long-term clinical studies, the benefit of wedged shoe insoles in the treatment of knee OA is still controversial. Nevertheless, shoe insoles are recommended in 13 of 14 guidelines for nonoperative treatment of knee OA (31, 32). In this present study we investigated the immediate effects of different foot positions, such as how they could be achieved by shoe insoles, on the pelvic position and the spinal posture. The variations of foot position were achieved by a specially constructed stand platform, where the subjects were placed during the measurements. Most of the commercially available shoe insoles led, due to their various compositions and designs, to various and complex changes in the position of the foot. Therefore, it must be noted that a shoe insole mostly does not create only one effect but leads to a summation effect in the foot position. Because each different shoe insole changes various parameters of the foot position, we have focused our investigation on the influence of 1 parameter change, i.e., the increase of the OM. Therefore, it is not possible to draw direct conclusions from the results of our study to the effects of different shoe insoles. However, we think that before evaluating the summation effects of shoe insoles on the foot, it is necessary to investigate in a first step, one parameter at a time, to better understand the influence of variations in foot positions on the kinematic chain of the lower extremity, pelvis, and spine. By using the above described stand platform, we were able to simulate the effects of insoles without actually having to use them. Besides an increase of the lateral and medial margin of the platform, we also have evaluated the influence of increased or decreased heel height on spinal posture and pelvis position. The maximum height change of the foot position that we have simulated was 15 mm. Greater height changes were not simulated with the platform, although they are possible, because they create a feeling of discomfort in patients (33) and are usually not reached by insoles. In this study, we measured the immediate effects of different foot positions and height on pelvis position and spinal posture with a rasterstereographic device. We have decided to use rasterstereography because it is contact free, accurate, and highly reliable (26, 34). A further advantage of rasterstereography is that it allows statements about both pelvic position and spinal posture. By measuring the position of the 2 lumbar dimples, it is possible to determine the position and orientation of the pelvis. Drerup and Hierholzer (17) showed that the 2 lumbar dimples can be localized rasterstereographically with an SD of ±1 mm. Furthermore, the 2 lumbar dimples can be taken as an indicator for the pelvic position and for pelvic movements because they are in close relation to the underlying posterior superior iliac spines. Drerup and Hierholzer confirmed this by a high radiologic correlation (r = 0.99) between lumbar dimple and pelvic movements (35). We have measured a mean pelvic tilt of 0.4 mm in this study. Our results show that different foot positions lead to significant changes of the pelvic tilt and pelvic torsion, as seen in Figures 2 and 3. These findings are in accordance with the present literature. Pinto et al described in their study an increase of pelvic anteversion by a consequent bilateral or unilateral calcaneal eversion (36). Khamis and Yizhar stated in 2007 that the pelvic alignment is influenced by foot alignment, irrespective of plane of motion (37).

Interestingly, in our study the greatest pelvic changes were found when the lateral margin of the foot was increased. It must be assumed that the biomechanical chain of the lower extremity can compensate a hyperpronation of the ankle joint at least. A possible explanation for the changes of the pelvic position is that by the increase of the lateral margin of the foot the calcaneus averts, so that the talus is forced to slide medially and inferiorly (38, 39). This movement may lead to an internal rotation of the shank, which is associated with a rotation of the thigh, but in lesser amplitude (37, 40). An internal rotation of the femoral head increases the pressure on the posterior portion of the acetabulum, which finally can lead to a tilt of the pelvis (36). So far only little is known about the effect of pelvic tilt and torsion on the hip joint. Therefore, it is possible that an unphysiologic hip joint position may increase the intraarticular pressure, resulting in secondary negative effects such as OA of the hip joint. The results of our study also showed that there seems to be a side difference in the response of the pelvic torsion in the OM condition. However, it is not clear why there is a difference in pelvic torsion values between the left and the right side lifting the OM. A possible explanation for this observation could be a dominant leg effect of the test subjects. As in the upper extremity a left or right hand domination is well documented in the current literature (41), it could be assumed that there does also exist such an effect on the lower extremity (42). This side difference might be responsible for the described phenomenon. It is therefore possible that muscle imbalances or differences in ranges of motion of the hip joint exist, which could have led to the found results (43, 44). Further studies will be necessary to address this topic.

The pelvic girdle is connected over a strong fibrous tissue to the lumbar spine at the sacroiliac joint (45). It is assumed that changes of the pelvis can possibly lead to alterations of the spinal posture, in particular of the lumbar lordosis (46, 47). This is explained by the fact that the pelvic position is highly correlated with the lumbar position (47).

Due to an increase of anteversion and tilt of the pelvis, Gurney and Legaye et al propose the presence of hyperlordosis and scoliosis of the spine (48, 49). This may lead to an increase of loads on the facet joints, which could result in lumbar back pain (50, 51). Based on this thesis, we were interested in the influence of different foot positions and heights on the spinal posture measured with rasterstereography. Although we measured significant changes of the pelvic position due to the different foot positions, no significant changes on the spinal posture occurred. Our findings are in part confirmed by Duval et al, who described no significant relationship between foot pronation and supination with pelvic tilt and lumbar lordosis (45). Previous studies were also not able to establish a link between pelvic tilt and spinal posture changes (52, 53). However, despite the high sample size of 153, it cannot be excluded that there are significant differences in foot position and spinal posture changes between the compared groups since the high SD of the measured spinal parameters resulted in a low statistical power (range 0.17–0.69 [1 − β error of probability]).

An explanation could be that our changes of foot position were too small to create significant changes of the spine (45). In addition, since we only focused on the immediate effects of different foot positions on spinal posture, long-term effects of different foot positions are still pending. It is also possible that the kinematic chain of the lower extremity and the pelvic girdle compensates the changes to a certain amount, so that alterations of the spine did not occur in this study. Therefore, it must be noted that individual reactions to different foot positions do exist, as expressed in the SDs of the spinal parameters. Further evaluation is needed to identify the amount of pelvic tilt at which the spinal posture is affected.

The current literature states that there might be a difference in lower extremity and spinopelvic alignment between women and men (54, 55). Our study does not support this, as we have measured no significant differences in the pelvic and spinal parameters between women and men. The results of our study have also shown that the effects of different simulated foot positions on pelvis and spine in both sexes are equal. It must therefore be concluded that in the setting of our study the physiologic response of various foot positions does not differ between men and women.

The results of our study support the existence of a kinematic chain in the measured collective, where changes of the foot position (increase of the lateral margin) led to an immediate significant change of the pelvic position. A correction of the foot alignment due to different foot positions may be used not just for the correct alignment of the lower extremities, but for the pelvis as well. Further studies will have to verify if our findings in healthy subjects can be transferred to patients with pathologic pelvic orientation. In addition, it must be evaluated if changes in pelvic position due to variations in foot position can lead to long-term pathologic alterations, such as OA of the hip.

In a collective of young healthy subjects, an increase of the lateral margin or heel height of the foot results in significant changes in pelvic tilt and torsion. No effects on the pelvic position were observed when increasing the IM of the foot. However, in this setting we were not able to detect significant changes of the spinal parameters due to foot position changes.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Schneppendahl had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Betsch, Dor, Jungbluth, Rapp, Wild.

Acquisition of data. Betsch, Schneppendahl, Grassmann, Hakimi, Rapp.

Analysis and interpretation of data. Betsch, Dor, Windolf, Thelen, Rapp, Wild.