In-shoe pressure measurements with a viscoelastic heel orthosis

In-Shoe Pressure Measurements With a Viscoelastic Heel Orthosis Wei-Li Hsi, MD, PhD, Jin-Shin Lui, MD, Pey-Yu Yang, MD ABSTRACT. Hsi W-L, Lai J-S, Yang P-Y. In-shoe pressure measurements with a viscoelastic heel orthosis. Arch Phys Med Rehabil 1999;80:805-10. Objective: To detect the mechanical effect of a viscoelastic heel orthosis. Design: Two-factor analysis of variance with interactions between the orthosis and the subjects. The number of subjects was determined by presuming the effect of the orthosis to be twice as large as the error-free standard deviation (SD) of the interactions, the step-to-step SD four times as large as the error-free SD of the interactions, type 1 error probability equal to .05, and type 2 error probability equal to .20. Setting: A gait laboratory in a university hospital. Subjects: Twenty-two consecutive patients with treated heel pain. Main Outcome Measures: Peak pressure (PP), pressuretime integral (PTI), and foot-to-sensor contact time (COT) measured for five steps at 24 discrete sensors of predetermined positions in the foot with treated heel pain. Results: The orthosis reduced PPs,PTIs, and COT (p < .05) in the median midfoot and lateral midfoot; reduced PPs and PTIs (p < .05) in the posterior heel and medial midfoot; increased PP and PTI (p < .05) in the anterior part of the first metatarsal head; and increased PTI (p < .05) in the lateral part of the hallux. The ratios of the estimated step-to-step SDSto the estimated error-free SDS of the interactions of PPs, PTIs, and COT were less than four at all the sensors. Conclusion: Proper design and estimation of the variations ensured that there was sufficient power to detect the effect of an a priori specified size as statistically significant: the orthosis reduced the mechanical loads in the posterior heel and the midfoot and increased the mechanical loads in the anterior part of the first metatarsal head and the lateral part of the hallux during walking. 0 1999 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation UBCALCANEAL HEEL PAIN is a very common foot S problem in clinical practice. It can be caused by inflammation of the plantar fascia at its insertion on the calcaneus, heel spur, fracture, calcaneal periositis, subcalcaneal bursitis, nerve entrapment, and intrinsic heel pad patho1ogy.l Its treatment includes relative rest, nonsteroid anti-inflammatory medication, surgery, and heel 0rthoses.l” From the Department of Rehabilitation, National Taiwan University Hospital, Taipei, Taiwan. Submitted for publication October 26, 1998. Accepted in revised form December 31, 1998. No commercial uartv having a direct financial interest in the results of the research supporting this at&k has or-will confer a benefit upon the authors or upon any organization with which the authors are associated. Reprints requests to Jin-Shin Lai, MD, Department of Rehabilitation, National Taiwan University Hospital, 7 Chung-Shan South Road, Taipei, Taiwan. 0 1999 by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation 0003-9993/99/8007-5269$3,00/O

The heel orthoses can be either hard or soft. The soft orthoses are wedge-shaped inserts made of viscoelastic materials, such as foam rubber, latex, silicone, polyethylene, and polyurethane foams. They presumably unload the painful area by redistributing the pressure over the heel and shifting the ground reaction force forward. is5Early techniques to measure load such as force plate and pressure platform only determined the mechanical loads of isolated barefoot steps, but could not measure the pressure-relieving effects of the viscoelastic heel orthoses directly.5,6 New techniques provide flexible insoles with multiple transducers that measure the pressure distribution at the interface between the foot and the shoe to assessthe mechanical effects of the foot orthoses.7 The acquisition of data from multiple steps also allows the variations of the in-shoe pressure measurements to be examined. Zhu and colleagues* used the coefficients of variation and the intersubject standard deviations (SDS)to indicate the step-to-step variations and the intersubject variations of the measurements, respectively. The intersubject SD contained both step-to-step variation and error-free intersubject variation.g Because large numbers of steps were measured by Zhu,* the component of the step-to-step variation in the intersubject SD was minimized. Kernozek and colleagueslO used the coefficient of reliability to indicate the extent to which the component of the step-to-step variation is reduced. As more steps are measured, the coefficient of reliability becomes larger. Excellent reliability is achieved when the coefficient of reliability is larger than .90. Only eight steps need to be measured for the coefficients of reliability to be larger than .90 in all regions of the foot in healthy subjects.1° When the in-shoe pressure measurements are compared within subjects to evaluate the effect of a foot orthosis,11,12the variation of the effect, not the variation among the subjects, is the major concern.13The variation of the effect contained both step-to-step variation and the error-free variation of the interactions between the orthosis and the subjects.13Large numbers of steps were measured to minimize the component of the step-to-step variation. 11.12 Although collecting data from a large number of steps is useful in general for improved statistical power,14 it is not efficient to measure more steps than are necessary to detect the effect of the orthosis. Because patients with pathologies cannot ambulate repeatedly down a walkway due to increased trauma or fatigue, it may be important that the measurement of a few steps be representative of the condition assessed.1°In this study, the mechanical effect of a viscoelastic heel orthosis was detected with the measurements of five steps, and the variations of the measurements were estimated to examine the presumption by which the number of subjects was determined. METHODS Subjects Twenty-two consecutive patients who visited our clinic and had unilateral pain and tenderness localized at the plantar Arch

Phys

Med

Rehabil

VolttO,

July

1999

806

IN-SHOE

PRESSURES

WITH

HEEL ORTHOSIS,

Hsi

surface of the heel with no pain along the plantar fascia by passive toe dorsiflexion, no foot deformity, and no heel pad atrophy were included in the study.lF5All patients had been treated with relative rest, anti-inflammatory agents, and viscoelastic heel orthoses and did not have heel pain during the study. There were 8 men and 14 women, with ages ranging from 19 to 74 years (48 + 14yrs, mean + SD), heights from 1.42 to 1.77m (1.62 ? .OS),and weights from 48 to 82kg (63 +- 9). Measurement of In-shoe Pressures The procedures followed were in accordance with the Helsinki Declaration of 1975, as revised in 1983. Informed consent was obtained from all patients before measurement of in-shoe pressures. In-shoe pressures were measured using the Parotec System,a which consists of the following components: (1) a controller that receives signals from the measuring insoles; (2) four pairs of measuring insoles of different sizes: 35 to 36 (US men 3 to 4% or women 4 to 5), 37 to 38 (US men 5 to 6 or women 51/2to 7), 39 to 40 (US men 61/2to 7% or women 7% to 9), and 41 to 42 (US men 8 to 9 or women 9% to 10); (3) software for analysis of data; and (4) an IBM-compatible personal computer. The measuring insoles are constructed using a 2.5-mm thick sheet of polyvinyl chloride and contain 24 conductive transducers embedded in hydrocells with a resolution of 2.5kPa and a range of 6OOkPa.The transducer consists of a membrane on a mounting ring that produces bending in a silicone beam when deflected by applied pressure; the deflected beam then alters resistance in the transducer and produces a measurable deviation in current proportional to the mechanical displacement of the beam.15 The positions of the 24 sensors in the measuring insole are shown in figure 1. These positions were selected at the sites where the maximum loads most frequently occur during walking, based on a previous study of the load distributions of 350 subjects.l6 Sensors 1 through 6 are intended to measure the mechanical loads in the heel; sensors 7 though 12 are intended to measure the mechanical loads in the midfoot; sensors 13 though 20 are intended to measure the mechanical loads in the metatarsal heads; and sensors 21 though 24 are intended to measure the mechanical loads in the toe region. The surface area of sensors 1 through 20 are identical and range from 2.8 to 3.8cm2 each, depending on the insole size; the surface area of the sensors21 through 24 are identical, from 1.7 to 2.9cm2 each, depending on the insole size. The total measuring area of the 24 sensorsis approximately 46% of the plantar area.16The relative positions of all of the 24 sensorsare maintained in all four sizes of the measuring insoles. The sensorsare calibrated once in the factory and the calibration data are stored in the software for pressure analysis. The subjects wore their own flat-soled shoes and hosiery during testing. The measuring insoles were selected according to foot size and inserted into the shoes. Each subject was measured both while wearing and not wearing a pair of Silicone Heel Cushions,b in a randomly assigned sequence.The viscoelastic heel orthosis is 91mm long, 58mm wide, and 6mm thick at the posterior end and gradually becomes thinner until reaching nearly Omm at the anterior end. The edge of the orthosis is gradually raised to 20mm high posteriorly, and gradually decreasesto nearly Omm anteriorly. The orthoses were placed in the rear area of shoes, and the measuring insoles were placed on top of the orthoses. Each measuring insole was connected by a thin cable to the controller which, fastened around the patient’s

Arch

Phys

Med

Rehabil

Vol 80, July

1999

1 00 L 2

Fig 1. The measuring

predetermined insole.

positions

of the 24 discrete

sensors

in a

waist, recorded in-shoe pressures on a 64Kb SREM memory card at 1OOHzas the patient walked. There was no connection between the subjects and the computer during walking. The subjects walked at a self-chosen, comfortable speed along a walkway 12m long after two practice sessions. The in-shoe pressures were measured for more than five sequential steps in each foot starting from the third step; the first two steps and the last four steps were not recorded to reduce the stand-to-walk and walk-to-stop transition factor.16 The patients were instructed to maintain the same speed at all times during the tests. If walking speeds when wearing and not wearing the orthoses differed by more than lo%, in-shoe pressures were remeasured.l2 When the subjects were done walking, the pressure data were downloaded from the controller to the computer. Because of software limitations, only the pressure data of the first five steps in each foot were used in the analysis. Data Analysis The in-shoe pressure measurements, including peak pressures (PPs), pressure-time integrals (PTIs), and foot-to-sensor contact time (COT) at the 24 sensorsin the foot with treated heel pain, were analyzed.8 The PP was the maximal pressure measured by a sensor per step. The PTI was the sum of all the pressures at a sensor multiplied by the time intervals (.Olsec)

IN-SHOE

PRESSURES

WITH

HEEL

ORTHOSIS,

807

Hsi

per step. The COT was the total time during which pressure was larger than 2SkPa at a sensor per step. Let Yijk represent a pressure measurement of step (k) by subject (i) in condition (i) at a sensor.The model for this experiment is:

study, the effect size of the orthosis on a pressure measurement at a sensor was calculated as follows:

Yijk = H + OLi+ pj + (ap)ij + Eijk

and 22 subjects were required to reach a conclusion with the probability of type 1 error equal to .05 and the probability of type 2 error equal to .20 by Cohen’s power tablesI Two-factor analysis of variance with interactions between the orthosis and the subjects was applied to evaluate the changes of all the pressure measurements by the orthosis.t3 Statistical significance was determined at p < .05 level by F test. The estimated step-to-step SD, estimated error-free intersubject SDS, estimated error-free SDS of the interactions between the orthosis and the subjects, coefficients of reliability, coefficients of variation (CV), mean when not wearing the orthoses, changes by the orthosis, and standard errors of changes (SE) in all the pressure measurements were calculated.10~r3

i = 1,2 j = 1,2, . . , 22 k= 1,2;..,5 where p is a constant; oi is the effect of the orthosis; l3j is the effect of the subject, and a random variable with mean equal to 0 and SD equal to era, which represents the error-free intersubject variation; (o@)ij is the effect of the interactions between the orthosis and the subjects, and a random variable with mean equal to 0 and SD equal to (~,a, which represents the error-free variation of the interactions between the orthosis and the subjects; and Eijk is the error of measurement, and a random variable with mean equal to 0 and SD equal to o,, which represents the step-to-step variation.13 The number of subjects was determined a priori. A formula for the effect size index (d) was derived as follows:

where 6 is the change of a pressure measurement by the orthosis at a sensor, a,a is the error-free SD of the interactions between the orthosis and the subjects, u, is the step-to-step SD, and n is the number of steps measured.13s17As 6 was presumed to be twice as large as uola, cr, was presumed four times as large as a+ and five steps of measurements were available in this

Table SellSOr

c?‘. (kPa)

GB (kPa)

bp Wa)

IF&p

18

55

9

2

22

49

12

3

18 12

36 40

16 13

1.9 1.9 1.1 0.9

10 8

20 20

6 7

1.5 1.2

12 4

17 12

1.5 0.7

9 10

15

12 14

8 5 2 9

2.3 1.7

11 12

3 5

10 6

4 2

0.7 2.0

13 14 15

30 24 15

43 51 43

12 14 13

2.5 1.7 1.2

16

27

45

17 18

27 31

29 59

13 13

19 20 21

22 50 14

55 70 33

22 23 24

16 17 45

34 46 109

4 5 6 7 8

1: Estimated Reliability

RESULTS

The average step length (.61 -t .12m), step time (.54 t .07sec), and walking speed (1.16 f .30m/sec) when wearing the orthoses were not significantly different from those (.61 ? .lOm, .55 2 .06sec, and 1.14 t .26m/sec, respectively) when not wearing the orthoses (p > .05). Tables 1 through 3 show the estimated SDS, coefficients of reliability, means when not wearing the orthoses, CV, mean changes by the orthosis, SE, andp values of F test in PPs,PTIs, and COT at the 24 sensors, respectively. In general, the estimated step-to-step SDS (I?,), the estimated error-free intersubject SDS (&a), and the estimated error-free SDS of the interactions (B,p) of PPs and PTIs were larger at sensors 1 through 3 in the heel, sensors 13 through 20 in the metatarsal heads, and sensors23 and 24 in the hallux than other parts of the

SD, CV, and SE of PPs Mean

(kPa)

cv (%)

Change

(kPa)

197

9

182

12

156 131

11 9

1.6 7.6

72 22

14 35

-0.8 4.7

54 22

22 17

.82

9 47

52 31

.99 .91

12 7

21 65

.91 .96

92 96

2.1

.97 .93

85 96

32 25 18 28

-7.0 1.9

2.1 1.5

.85 .95

103 225

26 14

-7.7 -8.9

15 5

1.6 3.2 2.6

.97 .91 .97

220 150 60

10 33 23

-0.1 26.3

8 17 32

1.9 1.0 1.4

.96 .98 .97

68 108 214

23 15 21

21 14

.98 .96 .96 .98 .95 .97 .91 .98 .97

d = ~‘4cr& + (2,:s + 160$/5) = .877

SE (kPa)

-32.9 -22.9

.000001 .OOl

7.1

.82

5.8 3.1

.21 .79

-18.4

3.0 3.8

.I3 .OOOl

-9.4 -3.4

2.2 1.1

.0004 ,006

~16.4 -6.7

4.3 1.5

,001 .0003

-3.3 -11.5 -8.5

1.1 6.5

,008 .09

6.8 5.8

.23 .24

6.5 6.6

.77 .26 .37

9.7 6.6 9.3

-2.8 2.2 13.6 -1.5

Arch

P

4.7 5.7

Phys

Med

.99 .Ol

2.9 4.1 7.5

.34 .60 .08

15.0

.92

Rehabil

Vol 80, July

1999

808

IN-SHOE Table

PRESSURES

WITH

2: Estimated

HEEL

(kPa

1 2

Mean (kPa set)

set)

Change (kPa set)

SE (kPa

set)

5.2

11.3

3.2

1.6

.96

42.4

12

-7.1

1.5

6.0 4.8

8.4 9.1

2.4 3.9

2.5 1.2

.91 .95

39.0

15 13

-4.0

4.5

11.0 6.0

3.1 2.2

1.4

.97

36.0 30.0

5.7

1.5

1.7 1.5

.93 .97

19.1 5.5

5.7 4.0

2.9 1.7

1.6

.89

16.9

8

4.4 1.1

9 10

1.3 5.1

2.8 5.6

0.5 3.6

0.6 2.6 1.4

.99 .96

5.9 1.8

11 12

0.6 1.2

2.6 1.7

0.9 0.7

.86 .99

16.2 3.1

13 14

9.3 6.1

13.7 14.3

4.4 5.2

.91 .91

1.6 32.1

15 16

4.0

12.7 13.6

3.5

.96 .98

29.1 23.2

2.0

.93 .93

6.1 4.7

1.5 1.5

10.2

4.8 1.6

7.8 12.1 24.7

3 4 5

3.6 2.3

6 7

17

8.1 9.3

18 19

9.1 6.8

20 21

14.6 3.9 4.3

22 23

5.8 13.4

24

1.3

.005

15

1.8 1.5

.89 .07

19 41

-1.1 1.4

1.0 0.7

.31

26 19

-6.7 -2.6

1.4

.06 .OOOl

74 31

-0.6

0.8 0.3

,002 -04

-6.0 -1.8

1.7 0.4

.002 -0002

29 21

-0.7 -3.5 -1.4

0.3 2.2

,048 .I3

17

-1.3

2.4 1.6

-57 -41

26.7 35.8

30 26

0.4 -1.4

1.9 2.3

.85 .56

.96 .97

64.5 58.8

14 12

-0.6 0.1

2.9 2.2

.83 .98

3.0

.92

37.2

2.9

.03

1.3 3.2

.97 .94

15.8 14.5 25.0

39 25

6.5

2.4 3.5 1.8

30 23

0.3 0.8 4.0

0.9 0.8 1.6

.72 .35

6.9

2.0

30

0.9

3.4

0.7 1.8 2.1 1.2 1.1 2.2

3.8 4.5

15.0 20.3 18.4 22.2

Table

.96 .94

.02 .80

SD, CV, and SE of COT Change (msec)

SE (msec)

1.0

.97

541

8

-35

21

.I1

30 43 48

1.7 0.8 0.8

.97 .98 .97

519 522 515

10

-8

15

.61

7 8

-3 23

19 21

.87 .29

96 175

59 64

0.8 1.1

.96 .97

566 388

1.0 0.7 0.7

.92 .98 .97

573 475 218

8 18 9

-39 -23 -48

26 29 23

.I5 .44 .046

75 162 167

51 88 94

12 29

-169 -30

38 41

.0002 .47

77 186

39 78

1.5 0.6

.90 .99

608 358

10 14

-55 -127

18 34

,007 ,001

163 85

103 30

0.8 1.2

.95 .97

228 652

85 78

38 53

1.0 0.9

.96 .93

634 563

37 5 6

-21 -17 -24

46 14 17

.66 .22 .I8

77 94 95

29 29 32

2.2 0.9 1.1

.87 .98 .97

592 645 645

8 11 4

-21 -9

24 15 13

.84 .I9 .47

91 106 102 100 128

32 24 29 48 39

0.9 2.3 1.4 1.1 1.3

.98 .95 .97 .95 .97

634 588 589 535 585

6 5 9

-8 -7 -4

15 14 12

.57 .63 .77

102

43

1.2

.95

613

13 21 18 20

.80 .68 .71 .34

(m%c,

1

44

122

46

2 3

51 35

4 5

40 46

132 116 111

6 7 8

70 51 57

9 10

64 58

11 12

49 85

13 14

36 39

15 16 17

47 65 27

18 19 20

37 30 55

21 22 23 24

40 51 50 50

Rehabil

21 74

PTIs, and COT were 3.2, 3.5, and 2.3 respectively (tables 1 through 3), not larger than the presumed ratio of four, even though the step-to-step SDS of PPs, PTIs, and COT were presumed to be four times as large as the error-free SDS of the interactions in determining the number of subjects.

Mean (msec)

(m%,

Med

3: Estimated

44.9

Reliability

Sensor

Phys

P .OOOl

0.3 2.8

foot (tables 1 and 2); while 6,, BP,and 6,s of COT were larger at sensor 6 in the heel and sensors7 through 12 in the midfoot than other parts of the foot (table 3). The maximal ratios of the estimated step-to-step SDS to the estimated error-free SDS of the interactions (I!?$&,,$ of PPs,

Arch

Hsi

SD, CV, and SE of PTls

+‘e Sensor

ORTHOSIS,

Vol 80, July

(nzc,

1999

a&,

7 10 9 8

5

3 9 -7 -19

P

IN-SHOE

PRESSURES

WITH

The coefficient of reliability was calculated as the intraclass correlation coefficients of the mean measurement in five consecutive steps.9 The coefficients of reliability of the PPs were larger than .90 at all sensors except sensors 10 and 17; those of the PTIs were larger than .90 at all sensors except sensors7 and 10; and those of COT were larger than .90 at all sensorsexcept sensors 10 and 16. While 6s was usually much larger than 6,, 6~ was close to or smaller than 6, at the sensors with coefficients of reliability less than .90 (tables 1 through 3). In general, the mean PPs and PTIs when not wearing the orthoses were larger in the posterior and middle heel (sensors 1 through 4) metatarsal heads (sensors 13 through 20) and hallux (sensors 23 and 24) than other parts of the foot (tables 1 and 2). The mean COT when not wearing the orthoses was larger in the metatarsal heads (sensors 13 through 20) and hallux (sensors 23 and 24) than other parts of the foot (table 3). The CV was calculated as 6, divided by the mean. The CV of PPs, PTIs, and COT were larger at sensor 6 in the heel and sensors7 through 12 in the midfoot than other parts of the foot (tables 1 through 3). The change by the orthosis was calculated as the mean measurement when wearing the orthoses minus the mean measurement when not wearing the orthoses. Negative change means reduction, and positive change means increase in the measurement by the orthosis. The SE was related to both &,a and 6,. For example, the SE of PP at sensor 24 was the largest of the 24 sensors, as 6,~ and 6, of PP at sensor 24 were the largest and the second largest of the 24 sensors,respectively (table 1). The SE of the PTI at sensor 24 was the largest of the 24 sensors, as 6,p and 6, of PTI at sensor 24 were the largest and the second largest of the 24 sensors,respectively (table 2). The SE of COT at sensor 12 was the largest of the 24 sensors, as were both 6,~ and 6, (table 3). The orthoses significantly reduced PPs, PTIs, and COT (p < .05) in the median midfoot (sensors 7 and 10) and lateral midfoot (sensors 8 and 11); significantly reduced PPs and PTIs (p < .05) in the posterior heel (sensors 1 and 2) and medial midfoot (sensors 9 and 12); significantly increased PP and PTI (p < .05) in the anterior part of the first metatarsal head (sensor 20); and significantly increased PTI (p < .05) in the lateral part of hallux (sensor 23) (tables 1 through 3). DISCUSSION In this study, the positions of the 24 discrete sensors were selected at the sites where the maximal loads most frequently occur during walking, so that the pressures they measured adequately represented the actual distribution of the maximal loads on the plantar surface. The mean PPs and PTIs when not wearing the orthoses in this study were close to those reported by Kemozek using a matrix system, but smaller than those reported by Zhu8 and Soams18 using discrete sensors. The discrepancy may be caused by differences in the sensor size, dynamic range, sampling rate, spatial resolution, frequency response, linearity, hysteresis, temperature sensitivity, and reproducibility among the pressure measuring devices.14 Although pressures were not measured in the midfoot by ZhuX the CV in other parts of the foot distributed similarly to ours. The largest CV of PPs and PTIs found by Zhu8 at seven sites of the plantar surface were 22% and 24%, respectively, in the first metatarsal head, and that of COT was 13% in the hallux in healthy subjects. In our study, the CV of PPs, PTIs, and COT were predominantly larger in the midfoot than other parts of the foot, and the mean PPs, PTIs, and COT were smaller in the midfoot than other parts of the foot. Five steps of PPs, PTIs, and COT achieved excellent reliability at most sensors in this study. More steps than five would

HEEL

ORTHOSIS,

809

Hsi

need to be measured to achieve excellent reliability at those sensors with coefficients of reliability lower than .90. By calculation, to achieve excellent reliability at a given sensor, 10 steps need to be measured for PP at sensor 10, eight steps need to be measured for PTI at sensor 10, and seven steps need to be measured for COT at sensor 16. A maximum of 10 steps would be needed for PPs, PTIs, and COT to achieve excellent reliability at all of the sensorsin this study, while a maximum of eight steps was needed by Kemozek.‘O The discrepancy may be from the differences in the pressure measuring device (matrix system versus discrete sensors), subjects (healthy subjects versus patients with treated heel pain), and type of test (treadmill versus floor). The use of discrete sensors has drawbacks due to their placement in predetermined positions.14 The proper performance of sensor placement requires an a priori hypothesis regarding loading of the foot, instead of using the device to determine the load relative to the plantar surface of the foot.‘O In this study, the effect of the variability related to the placement of discrete sensors in predetermined positions was treated as part of the variations of the in-shoe pressure measurements. Because there is minute movement between the foot and the measuring insole during walking, the same anatomic sitesof the foot cannot be always in exact contact with the same sensorsin every step.l4 The consequence of the relative movement is embedded in the step-to-step variations of the measurements. Though the predetermined positions of the sensorsare based on the load distributions of 350 subjects,16there is intersubject variability in the load distributions relative to the positions of the sensors.14The consequence of the variability is embedded in the error-free intersubject variations of the measurements. Wearing the orthoses may change the positions of the anatomic sites of the foot relative to the sensors,and there is intersubject variability in this change. The consequence of the variability is embedded in the error-free interaction variations of the measurements. Though custom discrete sensor arrays are created for each subject and groups of elements or masks are established in the matrix system, so that the the same region of the foot is always monitored, there will still be step-to-step variations, error-free intersubject variations, and error-free interaction variations in the in-shoe pressure measurements from other causes.8,10J2J4 The statistical model used in this study, however, will be applicable in estimating these variations. The distribution of the ground reaction forces on the plantar surface is very uneven during walking. As the body moves forward while walking, it is only supported by part of the plantar surface. In the early contact phase of gait, the heel sustainsmost of the ground reaction force. In the middle contact phase of gait, both the heel and the metatarsal heads sustain most of the ground reaction force. In the late contact phase of gait, the metatarsal heads and toes sustain most of the ground reaction force. The median and lateral midfoot are lifted by the orthosis; thus PP representing maximal transient load, PTI representing accumulated load, and COT representing time of load acting are significantly reduced in these regions. The PPs and PTls are significantly reduced by the orthosis in the posterior heel, as the load in the posterior heel is redistributed to other parts of the heel. Though PPs and PTIs may be increased in other parts of the heel, the changes are not significant. The medial midfoot is an arch and sustains the least load in the foot. Lifting has less effect on the load in the medial midfoot than the median and lateral midfoot; thus PPs and PTIs, but not COT, are reduced significantly in the medial midfoot by the orthosis. The fourth (sensor 14 and 18) and fifth (sensor 13 and 17) Arch

Phys

Med

Rehabil

Vol 80, July

1999

810

IN-SHOE

PRESSURES

WITH

metatarsal heads are adjacent to the median and lateral midfoot. They are lifted by the orthosis to a lesser extent than the median and lateral midfoot. Though PPs,PTIs, and COT are reduced in the fourth and fifth metatarsal heads, the changes are not significant. The wedge-shaped orthosis shifts the center of body mass forward, and increases the plantarflexion angle of the ankle as do high-heel shoes,rg thus increasing the load in the first metatarsal head and the lateral part of the hallux, but not increasing the time of load acting. A major part of the increased load is distributed to the anterior part of the first metatarsal head; thus both PP and PTI are increased significantly in the anterior part of the first metatarsal head. A lesser part of the increased load is distributed to the lateral part of the hallux; thus, only PTIs are increased significantly in the lateral part of the hallux. Previous studies also reported that the ground reaction force and in-shoe pressures are decreased in the lateral forefoot, and increased in the medial forefoot and hallux by high-heel shoes.20-22 The interrelation among the change, the step-to-step SD, the error-free SD of the interactions, the number of steps measured, and the number of subjects is delineated by our formula for the effect size index and Cohen’s power tables.17If 150 steps were measured in this study as in the study by Shaw and associates,12 and all the other presumptions remained the same, by calculation only 10 subjects would be required to reach the same conclusion. When other orthoses, subjects, and measuring devices are tested, the variations and the changes of the in-shoe pressure measurements may be different, but the number of subjects can still be determined using our formula for the effect size index and Cohen’s power tables.17For example, in a study where the step-to-step SDSof the measurements are presumed as six times as large as the error-free SDS of the interactions, and the changes by the orthosis are presumed twice as large as the error-free SDS of the interactions, it will require 41 subjects if five steps are measured, or 22 subjects if 12 steps are measured, to reach a conclusion with the probability of type 1 error equal to .05 and the probability of type 2 error equal to .20. As more steps are measured, however, it may become increasingly difficult for subjects to maintain a consistent speed, and the resulting fluctuation of speed during walking may further increase the step-to-step variations of the in-shoe pressure measurements. It is therefore important to estimate the variations of the in-shoe pressure measurements, and optimize the test protocol accordingly. CONCLUSION Although the in-shoe pressures measurements at 24 discrete sensors in predetermined positions for five consecutive steps did not all achieve excellent reliability in the subjects with treated heel pain, proper design and estimation of the variations ensured that there was sufficient power to detect the effect of an a priori specified size as statistically significant. The orthosis reduced the mechanical loads in the posterior heel and the midfoot, and increased the mechanical loads in the anterior part of the first metatarsal head and the lateral hallux during walking. References 1. Karr SD. Subcalcaneal heel pain. Orthop Clin North Am 1994;25: 161-75.

Arch

Phys

Med

Rehabil

Vol 80, July

1999

HEEL

ORTHOSIS,

Hsi

2. Davis PF, Severud E, Baxter DE. Painful heel syndrome: results of nonoperative treatment. Foot Ankle Int 1994;15:531-5. 3. MacLellan GE, Vyvyan B. Management of pain beneath the heel and achilles tendinitis with visco-elastic heel inserts. Br J Sports Med 1981;15:117-21. 4. Levitz SJ, Dykyj D. Improvements in the design of viscoelastic heel orthoses. J Am Pod&t Med Assn 1992;82:412-6. 5. Katoh Y, Chao EYS, Morrey BF, Laughman RK. Objective technique for evaluating painful heel syndrome and its treatment. Foot Ankle 1983;3:227-37. 6. Boulton AJ, Franks CI, Betts RP, Duckworth T, Ward JD. Reduction of abnormal foot pressures in diabetic neuropathy using a new polymer insole material. Diabetes Care 1984;7:42-6. 7. Lord M, Hosein R. Pressure redistribution by molded inserts in diabetic footwear: a pilot study. J Rehabil Res Dev 1994;31: 214-21. 8. Zhu H, Wertsch JJ, Harris GF, Alba HM, Price MB. Sensate and insensate in-shoe plantar pressures. Arch Phys Med Rehabil 1993;74:1362-8. 9. Fleiss JL. The design and analysis of clinical experiments. New York: John Wiley & Sons; 1986. 10. Kernozek TW, Lamott EE, Dancisek MJ. Reliability of an in-shoe pressure measurement system during treadmill walking. Foot Ankle Int 1996;17:204-9. 11. Wertsch JJ, Frank LW, Zhu H, Price MB, Harris GF, Alba HM. Plantar pressure with total contact casting. J Rehabil Res Dev 1995;32:205-9. 12. Shaw JE, Hsi WL, Ulbrecht JS, Norkitis A, Becker MB, Cavanagh PR. The mechanism of plantar unloading in total contact casts: implications for design and clinical use. Foot Ankle Int 1997;18: 809-17. 13. Neter J, Wasserman W, Kunter MH. Applied linear statistical models: regression, analysis of variance, and experimental designs. 3rd ed. Homewood (IL): Irwin; 1990. 14. Cavanagh PR, Hewitt FG, Perry JE. In-shoe plantar pressure measurement: a review. Foot 1992;2:185-94. 15. Lereim P, Serck-Hanssen F. A method of recording pressure distribution under the sole of the foot. Bull Pro&et Res 1973;20: 118-25. 16. Kraemer FW. Parotec system instruction manual. Neubeum, Germany: Paromed; 1995. 17. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillsdale (NJ): Lawrence Erlbaum; 1988. 18. Soames RW. Foot pressure patterns during gait. J Biomed Eng 1985;7: 120-6. 19. Snow RE, Williams KR. High heeled shoes: their effect on center of mass position, posture, three-dimensional kinematics, rearfoot motion, and ground reaction forces. Arch Phys Med Rehabil 1994;75:568-76. 20. Schwartz RP, Heath AL, Morgan DW, Towns RC. A quantitative analysis of recorded variables in the walking pattern of normal adults. J Bone Joint Surg 1964;46A:321-34. 21. McBride ID, Wyss UP, Cooke TDV, Murphy L, Phillips J, Olney SJ. First metatarsophalangeal joint reaction forces during highheelgait. FootAnkle 1991;11:282-8. 22. Nyska M, McCabe C, Linge K, Klenerman L. Plantar foot pressures during treadmill walking with high-heel and low-heel shoes. Foot Ankle Int 1996;17:662-6. Suppliers a. Paromed Medizintechnik GmbH, D-83 115 Markt Neubeuem, Germany. b. F.W. Kraemer, D-42828 Remscheid, Germany.