Reliability and Repeatability of the Portable EPS-Platform Digital Pressure-Plate System

Abstract

Background: Abnormal plantar pressures are the hallmark characteristic of several conditions and pathologic abnormalities. Pressure platforms allow for quick and accurate screening of patients and help guide clinical treatment. However, it is essential to evaluate the reliability and repeatability of these devices before making clinical decisions. The purpose of this study was to determine the reliability of the EPS-Platform during static and dynamic activities. Methods: Fifty-six healthy individuals stood and walked onto the pressure platform. Five trials were performed during two separate testing sessions to determine intrasession and intersession reliability. Pressure data were obtained and several variables of interest were calculated for intrasession and intersession reliability using intraclass correlation coefficients (ICCs), SEM, percent error, and coefficient of variation. Results: Static and dynamic intrasession and intersession reliability produced moderate-to-excellent ICCs, low SEMs, low percent errors, and low coefficients of variation. Static trials had higher ICCs, lower percent errors, and lower coefficients of variation compared with dynamic trials. Intersession reliability also had higher ICCs, lower percent errors, and lower coefficients of variation compared with intrasession reliability. Conclusions: This study demonstrates that the EPS-Platform is a reliable device for collecting gait plantar pressures. Static trials produce better reliability, most likely owing to the large inherent variability during dynamic gait. Intersession reliability was higher than intrasession reliability owing to the intersession measures being calculated with an average of five trials. By averaging the trials, the variability of gait is decreased, and this improves the accuracy of the results. These results can be used as the basis for future studies and to determine a priori sample sizes for investigations that use the EPS-Platform.

Although the foot can withstand the enormous pressures generated by everyday dynamic activities, there are a variety of conditions and abnormalities that can lead to abnormal plantar pressures. These increased plantar pressures can cause the underly-ing skin and bone to degrade, leading to ulcers and stress fractures. Pressure platforms are commonly incorporated in gait laboratories to allow clinicians to quickly and effectively screen patients for possible pressure abnormalities that may be the underlying mechanism or end result of many common foot abnormalities.

Pressure platforms provide objective data that surgeons can use to direct surgical treatment or evaluate the efficacy of surgical intervention. Recently, these devices have been used to evaluate the preoperative and postoperative plantar loading characteristics of the hallux valgus in Chevron-Akin osteotomies.[1] Pressure platforms have also been useful in the preoperative surgical decision making and postoperative objective assessment of patients with clubfeet.[2,3] Thometz et al[3] found that radio-graphs used in conjunction with dynamic plantar pressure analysis provided a complete assessment of the corrected clubfoot. In addition, these authors concluded that preoperative use of the pressure mat to direct surgical methods and postoperative as-sessment using pressure mats improved the ability to measure the success of the surgery. Pressure platforms are also a useful clinical tool for evaluating high-risk populations, particularly per-sons with diabetes who experience reduced foot sensation, abnormal movement patterns, and a propensity for ulcerations due to arterial and venous insufficiencies.[4] Because pressure platforms are being used more frequently in the clinical setting as diagnostic criteria for corrective footwear and surgical interventions, evaluating the reliability of these devices is necessary to help guide clinicians and surgeons in providing optimal outcomes for patients.

Although pressure platforms may provide high-resolution pressure data, portable in-shoe pressure systems are also commonly used in clinical prac-tice. Despite the portability and relatively low cost of in-shoe pressure-sensing devices, in-shoe sys-tems alter normal movement patterns[5] and may have substantial drift in the pressure magnitudes when worn for extended periods.[6] Because these pressure systems are located in the shoe, accurate and repeatable placement of the sensor and movement of the sensors in the shoe over multiple trials remains an issue. Recently, a new digital portable pressure sensor system (EPS-Platform; Loran Engineering, Castel Maggiore de Bologna, Italy), was developed to measure the distribution of plantar pressures during static stance and dynamic gait. The digital portable pressure platform has been marketed as a clinical tool for the evaluation of barefoot pressure measurement in humans. Al-though this system is not an in-shoe system, it is relatively small, lightweight, and portable, making it an ideal system for laboratories or clinics that do not have a designated space to conduct gait analyses.

As the application of foot pressure measurement systems widens, it becomes necessary to determine the reliability and repeatability of these devices. No previous publications, to our knowledge, have addressed the repeatability of the EPS-Platform system. Therefore, the aim of this study was to evaluate the reliability and repeatability of this pressure system in healthy adults during static and dynamic conditions.

Materials and Methods

Fifty-six volunteers (36 men and 20 women) aged 19 to 81 years participated in the study. Participants were excluded if they reported pain in their feet within the previous 6 months, had any previous foot surgery, presented with congenital or acquired deformities of the feet on clinical examination, or had any other disability that might affect their gait (eg, flat feet, use of walking aids, visual or hearing impairments, or any problems in the lower limb or spine that might affect normal gait). Age, height, and weight were recorded (

Table 1. Patient Characteristics.

Characteristic	Mean	SD	Range
Age (years)	47.7	16.7	19–81
Weight (kg)	71.6	13.5	50–107
Height (cm)	167	8.1	153–186

). The study was approved by the Official and Professional College of Nurses at Principado de Asturias (podiatry section) ethics committee, and all of the participants gave their informed consent.

The platform was calibrated by the manufacturer before the study. The platform measured 70 3 50 cm, with a thickness of 5 mm; had a total weight of 7 kg; and comprised 2,304 resistive sensors. Measure-ments were made to the nearest 0.01 kPa for each sensor. Vertical force was recorded at a frequency of 60 Hz. The platform was linked via an interface unit to a personal computer containing the data collection software program Foot Checker, version 4.0 for Windows (Loran Engineering). Pressure maps were created for each foot. All of the participants underwent two testing sessions a mean ± SD of 9.90 ± 4.11 days (range, 3–20 days) apart. At both testing sessions, static and dynamic trials were recorded. Five trials were recorded at both sessions. All of the measures were recorded by one of two independent observers.

Static Data Collection

The volunteer stood on the pressure platform and simulated gait by walking in place for 15 seconds. After 15 seconds, the volunteer stopped moving his or her feet and stood still in a natural manner. With the participant’s feet entirely on the mat and at a comfortable angle, the participant looked straight ahead and maintained the arms close to the body. Plantar foot pressure measurements for both feet were simultaneously recorded during a 30-second period. If the individual moved during this time, the data were discarded and the trial was repeated until successfully performed. No practice trials were conducted. The static variables included 1) percent-age of body weight supported by each limb, 2) average pressure under each foot, 3) plantar surface area of each foot in contact with the platform, 4) percentage of body weight supported by the bilateral forefoot, 5) percentage of body weight supported by the bilateral rearfoot, 6) percentage of body weight supported by the left forefoot, 7) percentage of body weight supported by the right forefoot, 8) percentage of body weight supported by the left rearfoot, and 9) percentage of body weight supported by the right rearfoot.

Dynamic Data Collection

The platform was mounted in the center of a flat 5-m walkway at ground level. The long path allowed measurements to be recorded during steady state gait and, thus, ensured that the effect of accelera-tion and deceleration at the start and end of each walk was minimized. Volunteers walked at a self-selected speed for all of the dynamic trials. One step with each foot was collected per walking trial, and five steps with each foot were collected per session on the same walkway. The same procedures were followed for both testing sessions. The dynamic variables measured in each foot included 1) average pressure, 2) pressure-time integral, 3) contact time, 4) force-time integral, and 5) peak pressure. Measurements were calculated for both feet and were analyzed separately for reliability and repeat-ability.

Data Analysis

All of the data were evaluated for outliers and normal distribution. Samples were removed on an individual basis if found to be greater than 3 SD from the group mean. Descriptive data are present-ed as mean ± SD for the static and dynamic trials. Intratrial reliability was established by using the five trials of each dependent variable for static and dynamic conditions at the first and second testing sessions. Intraclass correlation coefficients (ICCs) using the [1,1] and the [1,k] models were calculated to first determine reliability between trials. The SEM was calculated from the ICCs and SDs for each of the five measurements.

Intersession reliability was established by retest-ing all of the participants at least 3 days after the initial measurements were taken. The average of five trials for each test session on each participant was used to calculate intersession reliability using an ICC[1,k] model. The SEM was calculated between sessions from the ICCs and SDs.

The coefficient of variation was calculated for intrasession reliability, and the percent error was calculated for intrasession and intersession reliabil-ity. The coefficient of variation is calculated as the mean normalized to the SD. This value represents the amount of variation between trials and sessions, normalized to the mean for each variable. Similarly, the percent error is calculated as the SEM divided by the mean and provides an estimate of the inherent error or variability normalized to the mean. A statistical software program (SPSS for Windows, version 17.0; SPSS Inc, Chicago, Illinois) was used for all of the data analysis.

Results

Static Intrasession Reliability

The reliability data (represented by ICC, SEM, percent error, and coefficient of variation) and normative data (represented by mean ± SD) for the static reliability trials are presented in

Table 2. Repeatability of Static Variables.

	First Session
Variable	Mean ± SD	CoV (%)	ICCa	ICCb	SEMa	% Errora	SEMb	% Errorb
% BW left	45.9 ± 3.8	4.8	0.684	0.915	2.1	4.7	1.1	2.4
% BW right	54.1 ± 3.8	4.0	0.684	0.915	2.1	3.9	1.1	2.0
Mean pressure left (kPa)	82 ± 16.9	8.5	0.783	0.948	7.9	9.6	3.9	4.7
Mean pressure right (kPa)	89.1 ± 16.4	6.9	0.853	0.967	6.3	7.1	3.0	3.3
Surface area left (cm2)	108.6 ± 34.6	12.2	0.673	0.911	19.8	18.2	10.3	9.5
Surface area right (cm2)	115.8 ± 30.2	12.3	0.566	0.867	19.9	17.2	11.0	9.5
% BW bilateral forefoot	49.7 ± 7.7	7.7	0.754	0.939	3.8	7.7	1.9	3.8
% BW bilateral rearfoot	50.3 ± 7.8	8.1	0.738	0.934	4.0	7.9	2.0	4.0
% BW left forefoot	21.6 ± 5.5	12.9	0.789	0.949	2.5	11.7	1.2	5.8
% BW right forefoot	28.1 ± 3.7	9.0	0.633	0.896	2.2	8.0	1.2	4.2
% BW left rearfoot	24.1 ± 5.8	13.8	0.741	0.935	3.0	12.2	1.5	6.1
% BW left rearfoot	26.1 ± 5.2	11.2	0.737	0.933	2.7	10.2	1.3	5.2
	Second Session
Variable	Mean ± SD	CoV (%)	ICCa	ICCb	SEMa	% Errora	SEMb	% Errorb
% BW left	46.3 ± 4.0	4.9	0.703	0.922	2.2	4.7	1.1	2.4
% BW right	53.7 ± 4.0	4.2	0.703	0.922	2.2	4.1	1.1	2.1
Mean pressure left (kPa)	82.1 ± 16.6	7.1	0.865	0.97	6.1	7.4	2.9	3.5
Mean pressure right (kPa)	89.3 ± 17.1	6.4	0.877	0.973	6.0	6.7	2.8	3.1
Surface area left (cm2)	111 ± 38.0	10.3	0.741	0.935	19.3	17.4	9.7	8.7
Surface area right (cm2)	117.1 ± 35.9	9.1	0.741	0.935	18.3	15.6	9.2	7.8
% BW bilateral forefoot	50.4 ± 7.7	6.3	0.827	0.96	3.2	6.4	1.5	3.1
% BW bilateral rearfoot	49.5 ± 7.6	6.8	0.781	0.947	3.6	7.2	1.7	3.5
% BW left forefoot	21.9 ± 5.8	10.9	0.835	0.962	2.4	10.8	1.1	5.2
% BW right forefoot	28.4 ± 3.7	7.8	0.689	0.917	2.1	7.3	1.1	3.8
% BW left rearfoot	24.1 ± 5.9	13.5	0.745	0.936	3.0	12.4	1.5	6.2
% BW left rearfoot	25.4 ± 5.6	10.0	0.819	0.958	2.4	9.4	1.1	4.5

Abbreviations: % BW, percentage of body weight supported; CoV, coefficient of variation; % error, percent error; ICC, intraclass correlation coefficient. ^aCalculated from ICC[1,1]. ^bCalculated from ICC[1,k].

. The ICCs for single-trial intrasession reliability ranged from 0.566 to 0.877, and the percent errors (calculated from the SEMs) ranged from 3.9% to 18.2%. The results of the average trial intrasession reliability produced good-to-moderate ICCs and low SEMs. The ICCs for average trial intrasession reliability ranged from 0.867 to 0.973, and the percent errors ranged from 2.0% to 9.5%. The coefficients of variation for the dependent variables ranged from 4.0% to 13.8% of the mean. This represents a small error that may occur within trials.

Dynamic Intrasession Reliability

The reliability data (represented by ICC, SEM, percent error, and coefficient of variation) and normative data (represented by mean ± SD) for the dynamic reliability trials are presented in

Table 3. Repeatability of Dynamic Variables.

				First Session
Variable	Mean ± SD	CoV (%)	ICCa	ICCb	SEMa	% Errora	SEMb	% Errorb
Mean pressure left (kPa)	139.9 ± 9.6	4.4	0.655	0.905	5.6	4.0	3.0	2.1
Mean pressure right (kPa)	142.3 ± 10.4	4.4	0.667	0.909	6.0	4.2	3.1	2.2
Pressure-time integral left	125,715.9 ± 28,424.1	8.6	0.817	0.957	12,159.4	9.7	5894.1	4.7
Pressure-time integral right	131,667.1 ± 36,053.0	10.4	0.772	0.944	17,215.1	13.1	8531.7	6.5
Contact time left (ms)	3,366 ± 272.9	4.2	0.664	0.908	158.2	4.7	82.8	2.5
Contact time right (ms)	3,419.4 ± 280.6	4.8	0.6	0.882	177.5	5.2	96.4	2.8
Force-time integral left	104.6 ± 12.9	5.5	0.765	0.942	6.3	6.0	3.1	3.0
Force-time integral right	107.1 ± 18.0	6.8	0.808	0.955	7.9	7.4	3.8	3.6
Peak pressure left (kPa)	250 ± 5.1	1.5	0.359	0.737	4.1	1.6	2.6	1.0
Peak pressure right (kPa)	249.9 ± 4.0	1.1	0.456	0.807	3.0	1.2	1.8	0.7
	Second Session
Variable	Mean ± SD	CoV (%)	ICCa	ICCb	SEMa	% Errora	SEMb	% Errorb
Mean pressure left (kPa)	139.9 ± 9.2	3.8	0.723	0.929	4.8	3.5	2.5	1.8
Mean pressure right (kPa)	142.6 ± 9.7	4.3	0.677	0.913	5.5	3.9	2.9	2.0
Pressure-time integral left	121,908.9 ± 27,357.2	6.9	0.871	0.971	9,825.8	8.1	4,658.8	3.8
Pressure-time integral right	128,453.9 ± 33,171.2	8.1	0.874	0.972	11,774.6	9.2	5,550.6	4.3
Contact time left (ms)	3,319.8 ± 240.4	3.3	0.754	0.939	119.2	3.6	59.4	1.8
Contact time right (ms)	3,351 ± 274.5	3.2	0.831	0.961	112.8	3.4	54.2	1.6
Force-time integral left	104.5 ± 12.9	5.3	0.767	0.943	6.2	6.0	3.1	2.9
Force-time integral right	108.7 ± 17.0	6.1	0.812	0.956	7.4	6.8	3.6	3.3
Peak pressure left (kPa)	250.7 ± 4.0	1.6	0.325	0.706	3.3	1.3	2.2	0.9
Peak pressure right (kPa)	250.3 ± 4.0	1.4	0.359	0.737	3.2	1.3	2.1	0.8

Abbreviations: % BW, percentage of body weight supported; CoV, coefficient of variation; ICC, intraclass correlation coefficient; % error, percent error. ^aCalculated from ICC[1,1]. ^bCalculated from ICC[1,k].

. The ICCs for single-trial intrasession reliability ranged from 0.325 to 0.874, and the percent errors ranged from 1.2% to 13.1%. The ICCs for average trial intrasession reliability ranged from 0.706 to 0.972, and the percent errors ranged from 0.7% to 6.5%. The coefficients of variation for the dependent variables ranged from 1.1% to 10.4% of the mean.

This represents a small error that may occur within trials.

Static Intersession Reliability

The reliability data (represented by ICC, SEM, and percent error) for static reliability trials are pre-sented in

Table 4. Intersession Measures.

	Overall Mean ± SD	ICC	SEM	% Error
Static Trials
% BW left	46.1 ± 3.9	0.948	0.9	3.9
% BW right	53.9 ± 3.9	0.948	0.9	3.3
Mean pressure left (kPa)	82.05 ± 16.9	0.979	2.4	6.0
Mean pressure right (kPa)	89.2 ± 16.7	0.977	2.5	5.7
Surface area left (cm2)	109.8 ± 36.2	0.949	8.2	14.9
Surface area right (cm2)	116.45 ± 33.0	0.932	8.6	14.8
% BW bilateral forefoot	50.05 ± 7.7	0.968	1.4	5.5
% BW bilateral rearfoot	49.9 ± 7.7	0.964	1.5	5.9
% BW left forefoot	21.75 ± 5.6	0.97	1.0	8.9
% BW right forefoot	28.25 ± 3.7	0.938	0.9	6.5
% BW left rearfoot	24.1 ± 5.7	0.971	1.0	8.1
% BW left rearfoot	25.75 ± 5.5	0.964	1.0	8.1
Dynamic Trials
Mean pressure left (kPa)	139.9 ± 9.3	0.958	1.9	2.7
Mean pressure right (kPa)	142.45 ± 10.0	0.95	2.2	3.1
Pressure-time integral left	123,812.4 ± 27,892.0	0.974	4497.4	7.3
Pressure-time integral right	130,060.5 ± 34,501.2	0.964	6546.1	10.1
Contact time left (ms)	3,342.9 ± 257.0	0.937	64.5	3.9
Contact time right (ms)	3,385.2 ± 270.3	0.923	75.0	4.4
Force-time integral left	104.55 ± 12.8	0.921	3.6	6.9
Force-time integral right	107.9 ± 17.4	0.965	3.3	6.0
Peak pressure left (kPa)	250.35 ± 4.0	0.822	1.7	1.3
Peak pressure right (kPa)	250.1 ± 4.0	0.884	1.4	1.1

Abbreviations: % BW, % of body weight supported; % error, percent error; ICC, intraclass correlation coefficient.

. The results of the average trial intersession reliability produced high ICCs and low SEMs. The ICCs for average trial intersession reliability ranged from 0.932 to 0.979, and the percent errors calculated from the SEM ranged from 3.3% to 14.9%.

Dynamic Intersession Reliability

The reliability data (represented by ICC, SEM, and percent error) for dynamic reliability trials are presented in Table 4. The ICCs for average trial intersession reliability ranged from 0.822 to 0.974, and the percent errors calculated from the SEM ranged from 1.1% to 10.1%.

Discussion

The EPS-Platform provided consistent results that were repeatable between trials and between ses-sions. Most ICCs were greater than 0.75, and nearly all of the percent error was lower than 10%. Together, these findings suggest that pressure measures collected from the EPS-Platform can be used to evaluate differences between participant groups and differences between sessions. The SEM and percent error are important variables that should be considered when formulating research protocols that use the EPS-Platform. The sample size required to determine significant changes can be based directly on these measures of intrasession and intersession repeatability.

The ICCs are a mathematical determination of the replication between multiple sets of numbers and are commonly used for scientific measurements to represent the repeatability of the measurement.[7,8] It has been suggested that ICCs greater than 0.75 are considered good reliability, and values of 0.75 or less are considered moderate-to-poor reliability.[8] In the present study, intersession variability was extremely low, and the ICCs for all of the measures except peak pressure during the dynamic trials were greater than 0.90. Peak pressures in the left and right feet during the dynamic trials were still greater than 0.75, suggesting that even the lowest of the intersession variables have good repeatability.

One important finding is that ICC values for single measures were lower than average intrasession and intersession test values. Physiologic data are inher-ently variable, and small changes in body position, walking speed, or muscle activity can dramatically influence the magnitude and spatial characteristics of pressure and force at the foot-ground interface. This inherent variability explains the fact that single-measure intrasession variability using the ICC[1,1] model was greater (lower ICC values) than the averaged measure ICC[1,k] and intersession repeatability ICC[1,k]. Physiologic fluctuations be-tween trials are to be expected, so using a single trial to capture a patient’s static or dynamic variables is not sufficient, and multiple trials should be averaged when making comparisons between time points or between groups.

All of the intrasession ICC[1,k] scores were greater than 0.75 except peak pressure during dynamic trials. Of the variables examined in this study, peak pressure during gait is most likely to be affected by small fluctuations in walking speed or muscular activity; therefore, it is expected that peak pressure during dynamic trials would be the most variable measure between trials. The ICC values are greatly influenced by the range of values for a given group. Peak pressure was relatively homogenous across all of the participants, so it is more likely that this variable would have the lowest ICCs. In comparison, the SEM, percent error, and coefficient of variation for peak dynamic pressure were extremely low. Collectively, these findings suggest that peak pressure is a reliable measure to use when evaluating differences across time points or between participant groups despite having only moderate repeatability measured using the ICC.

Although ICCs provide a numeric value for the repeatability of a measurement device or individual, they do not describe the amount of error or inherent variability that is expected each time the measure-ment is performed.[8] Assessing the error or variation each time a trial is performed is extremely important when capturing physiologic data where small differences between trials are expected. The SEM is another mathematical formula that uses the ICC and SD values to calculate the amount of expected error for the measurement device or individual.[8] The SEMs and percent errors for all of the variables in this study were relatively low, even with only moderate ICC values. A low SEM and percent error suggest that the variables are accept-able to use when assessing change before and after intervention or when measuring difference between participant groups. The SEMs provided in this analysis will allow future researchers to make clinical judgments about what degree of change is due to factors beyond errors associated with the normal variability of measuring between trials or between sessions.

The average ICC and coefficient of variation values found in the present study are very low, similar to previously reported values. Martinez-Nova et al[9] found ICC values for intrasession peak pressure to range from from 0.77 to 0.96 for an in-shoe pressure system, comparable with the results found in this study for intrasession variability. One limitation of this study is that the pressures were evaluated for the entire foot. Previous reports have found greater variability between trials when the data are compartmentalized into separate regions of the foot.[10,11] Future work with the EPS-Platform should evaluate the repeatability and normative values for regions of the foot that are known to be susceptible to high pressures in high-risk patients, including the area under the heel, hallux, and metatarsal heads.

Conclusions

The EPS-Platform is a reliable system for measur-ing static and dynamic foot pressures. Researchers and clinicians should feel confident incorporating this testing device into trials where repeatable measures of foot pressure are required for clinical decision making, evaluating comparative efficacy, or examining differences between pathologic groups.