EUROPEAN PRESSURE ULCER ADVISORY PANEL

Over the past two years various members of the European Pressure Ulcer Advisory Panel have been working upon a draft guideline for the laboratory evaluation of pressure-redistributing support surfaces. The following document is the product of that endeavour - however, it must be borne in mind that this does not represent the 'finished' guideline - merely another step towards that eventual document. Please feel free to comment upon this draft, sending all submissions to the Working Group co-ordinator, Dr Alastair McLeod, at: alastair.mcleod@huntleigh-healthcare.com

I feel confident that all members of the Working Group would like to extend their thanks to Alastair for his commitment to this project.

Michael Clark

Purpose

This document is a collection of recommendations compiled from several meetings of interested parties facilitated by the EPUAP (see appendix 1). All were motivated by a common recognition that methodological and technical differences in the interface pressure (IP) measurement protocols used in published laboratory studies on support surfaces makes comparisons virtually impossible. Given the rising number of pressure redistributing (PR) products appearing on the market, there is much sense in future studies adopting a similar protocol and a common reported data set. Whilst this will not completely remove the problem of incomparability, it is an initial step towards a standardised approach which it is hoped will eventually be established.

Scope

Given the large number of possible purposes there are for conducting interface pressure measurements, it was necessary to limit the initial scope of these recommendations to the following:

1. Studies where the primary objective is to establish differences in the pressure redistributing properties of support surfaces relative to a 'standard' surface
2. Surfaces whose design intent is NOT to vary the interface pressure cyclically over time (e.g., foam mattresses, static air mattresses, NOT alternating or pulsating surfaces)

Limitations

A key limitation to the recommendations contained herein is that no link is established between the IP analysis outcome and likely patient outcomes. Nor is there any data presented which suggests how products may be allocated to different patients for pressure ulcer (PU) prevention or healing purposes based on IP performance. Such relationships can only be established with suitably powered randomised controlled clinical trials. However, if clinically significant differences between PR surfaces are measured in such trials, then adopting these guidelines for subsequent comparative IP testing will allow other surfaces to be compared to those used in such research. This will allow more rapid assessment of new products as they are introduced.

Additionally, implicit in this logic is the fundamental assumption that the primary effect of PR surfaces lies in their ability to modify soft tissue compression patterns in a positive way. Other effects, such as skin micro-climate control and frictional and shearing forces are not quantified in these guidelines.

Interface Pressure Perpendicular force exerted by a segment of skin on a support surface divided by the skin contact area (abbrev = IP)
Pressure Redistribution The beneficial modification of pressure patterns on human soft tissues to reduce pressure ulcer risk (abbrev = PR)
IP sensor A device placed between the skin and a support surface whose output is calibrated to represent the perpendicular pressure exerted over the sensor area
Repeatability The variation in pressure measured by a sensor when an identical calibrated, constant pressure is applied and removed several times
Reproducibility The variation in analysis outcomes when a calculation is performed on pressure data derived from a test subject who has repositioned several times
Accuracy The closeness of an IP sensor output to a known applied pressure
Resolution A representation of the smallest area over which each sensor is measuring interface pressure. In this context , 'high resolution' means more sensors per square centimetre
Hysteresis The difference in sensor output measured when the same force is applied gradually from a low and a high starting point
Creep A slow variation in subject position or sensor output over time
RCT Randomised Controlled Trial
Scan The acquisition of pressure data from an array of electronic IP sensors

Choice of test body

Most published work has hitherto employed human subjects for IP measurements. Whilst this is intuitively appealing, there are significant reasons why this approach limits the comparability of IP measurements, including:

1. Experimental repeatability will be poor, because joint angles and tissue properties will vary over time, thus introducing higher variations in pressure readings
2. Unless physically identical subjects are used by all researchers in this field, and cross-comparisons are conducted to measure actual differences in subjects, it is not known how comparable any product rankings will be using different human subjects in different test locations
3. The use of human subjects does not dispose the measurement process to a standardised approach that is eventually desired

It was recognised that for the purposes of exploring the effect of IP on soft tissue blood flow, the use of human subjects, and the use of sensors other than IP remains essential. However, for the limited purposes of comparative ranking of support surface PR quality, it is recommended that a human-like mannequin is employed. This follows a similar approach adopted by ISO committee TC173/SC1/WG11, who recommended the construction of a standard buttock mannequin to conduct IP testing of wheelchair cushions.

Test Mannequins

The functional specification for a test mannequin still requires definition. Ideally, these should be constructed to mimic important degrees of freedom found in humans, but should also minimise potential errors deriving from unplanned movements such as sagging or creep. Whilst not exhaustive, essential elements are thought to include:

1. Full body mannequin rather than partial mannequin
2. Representative of typical height and weight of patients using PR surfaces, with weight distributed in correct anthropomorphic proportions. Ideally, a spectrum of weights and heights should be tested, with both male and female skeletal forms
3. Jointed at knees, hips, shoulders and neck. Type and design of joint (planar or rotational) should be sufficient only to allow mannequin to contour with typical profiled hospital bed. High degrees of freedom in joints may reduce experimental repeatability.
4. Surface of mannequin to represent 3D shape of bony prominences and soft tissue coverage found in typical patient group. There is still debate over whether a coverage of artificial soft tissue is necessary in a mannequin. The only way to resolve this is to construct mannequins with and without artificial soft tissue and compare differences in measurements between 'standard' surfaces and 'good' (ideally via RCT study) surfaces. If the addition of artificial soft tissue increases measurement sensitivity, and brings results significantly closer to those measured on humans, then this should be adopted as a standard requirement

If different test mannequins are developed as a result of ongoing work in different research centres, then it is essential that cross-calibration data is generated and published to ensure future comparability of results. Such data could be derived using an internationally agreed specification of basic foam mattress.

There is also some debate over the usefulness or necessity of heating a mannequin, to achieve skin temperatures at the pressure interface. Some mattresses claim to change their PR properties

Experimental design

It is recommended that, as a minimum, the test protocol should incorporate the following elements:

1. Sensor calibration using reference pressures plus zero. In absence of recommendations from sensor manufacturers, daily calibration is suggested. Sensors must demonstrate a sufficient level of accuracy etc. before any tests are undertaken - see the 'Instrumentation' section later
2. The PR support surfaces, including a 'standard' surface, should be tested with the mannequin in at least three positions
   
3. Sensor arrays should either be attached to, embedded in or cover the following areas:
   i) Heel(s)
   ii) Ischial tuberosity (ies)
   iii) Greater Trochanter(s)
   iv) Sacrum (Centred 1cm above natal cleft)
   v) Occiput (posterior)

The minimum requirements for sensors are discussed in the next section.

4. At least 3 scans should be taken in each position and averaged, and the mannequin should be repositioned at least 6 times in each of the above postures. This is to ensure the maximum pressures are captured and will supply data for reproducibility calculations.
5. Either all mattresses use a standard cover provided by the manufacturer, or all are tested using a standard hospital cotton sheet draped loosely over the surface. If a product is sold with a special cover as standard, then this should NOT be removed from the product during any of the tests.
6. Test environment temperature should be controlled to ±2°C to avoid thermal drift problems with sensors. The test room temperature should be as close as possible to the temperature during calibration.

Accuracy and Calibration

Accuracy is important if absolute pressures are to be reported. However, it is recommended later in this guideline that only relative results are issued. Thus, provided the accuracy remains consistent throughout testing, then it is probably not necessary to specify this property. It is necessary for the sensor to be able to resolve sufficiently small areas to assess local pressures - this is discussed in the next section.

The system should be calibrated using positive air pressure or weights of known area whose accuracy is traceable to a nationally recognised standard.

Resolution

The sensor matrix must be able to detect changes in IP over small distances - e.g., heels. It is proposed that the technique cited in TC173/SC1/WG11 is adopted here. Thus, a mannequin of known mass is effectively 'weighed' by the sensor on a standard surface (sum of all IP's multiplied by total sensed area = weight on area). If, when three such 'weighings' have been performed and averaged, the error is greater than 10% of test weight, then the sensor in question is unlikely to be adequate for test purposes.

Repeatability/Reproducibility

The repeatability of an IP measurement system should be such that, when results are analysed, significant differences between the PR properties of surfaces are not swamped by larger levels of uncertainty caused by instrumentation error. For example, if it is judged that 5 mmHg represents a significant change in pressure in the range 0-40 mmHg, but the measuring instrument or experimental method has a repeatability / reproducibility of ±5 mmHg, then we cannot be certain that a 5 mmHg difference is truly significant. It is recommended that the coefficient of variation (SD/MEAN) of a set (minimum 10) of repeated 'weighings' as described previously is less than or equal to 10%.

Analysis Techniques

There was much discussion on which is the most appropriate or meaningful set analyses to perform. In common with the choice of mannequin, this can only be ascertained via correlation of differences in calculated parameters with outcomes of RCT's using identical products. Since it is comparatively simple for a computer to perform mathematical operations on a sequence of scanned pressure data sets, it is proposed that all the methods listed below are used and tabulated in a final report. A consistent performance by a product across many analysis methods should result in a higher ranking. However, techniques which result in high variation (e.g., peak pressure on heels) may need to be excluded from final ranking calculations. To estimate this, the Coefficient of variation of each analysis technique should be calculated from the six repeats performed. As in previous cases, CV > 10% may indicate poor repeatability. (See Table 1)

Table 1: Analysis Techniques

Statistical Considerations

Given the small sample size of scans per position, the opportunity for statistical analysis is limited. Differences between paired data sets for different products in different positions can be tested for significance using non-parametric tests such as Mann-Whitney (if data looks skewed) or Students t-test (if data reasonably well behaved). Significance should be set at the 5% level. When processing the data, differences in calculated parameters between different products and the standard mattress should be used. Thus, two questions are answered per test product:

1. How different is it from the standard mattress (in each position)?
2. How different is it from the other test products (in each position)?

As a minimum, the following information should be divulged:

1. Sensor array description, including: number of sensors, physical dimensions, calibration method & frequency of calibration.
2. Description of mannequin, including mass, joint numbers and types, and any areas of compliant material.
3. Results of 'weighing' test, with statement of error and repeatability.
4. Description of 'standard' mattress, ideally uniform foam design of 4'-6' (100-150mm) height. Inclusion of density and hardness data also useful, with description of cover type
5. Description of test products, with modes of operation and set up used during tests (if powered surface) together with any adjustments made to products between test positions.
6. Statement of number of scan repeats per position, along with indication of statistical methods used to compare results
7. Table of analysis results for each mannequin position, showing either absolute results of all products tested including the standard mattress, or results relative to the standard (except 95% CI)
8. Statistical analysis of (7), both comparing all test products to the standard mattress and between test products for each position

Key issues in this guideline that require further investigation include:

1. Development and specification of minimum complexity mannequin needed to conduct tests. The core question is: 'what specification of mannequin is required to achieve the same relative product rankings that would be established using a large pool of human subjects?' If, as is likely, different mannequins will be used in different parts of the world, will these differences result in different rankings when used to assess identical test products?
2. Verification of the '10% rule' for acceptance of accuracy and repeatability of pressure sensing arrays. For a given mannequin design, the core question is: 'what range of sensor accuracy and repeatability does not affect the ultimate ranking of identical test products?'
3. Whilst the majority view was that pressure sensing arrays provided larger volumes of useful data than single or small groups of sensors, current technology still has limited resolution. This has implications on their ability to assess accurately the pressure distribution around areas like heels and elbows. It may therefore be necessary to use a different sensing approach in these areas to provide accurate data.
4. Further research is needed to establish how much of this guideline can be applied to moving surfaces - i.e., alternating and pulsating designs. There was broad agreement that large sensor arrays are unsuitable for these surfaces because of the degree of hammocking imposed over deflated cells. In addition, analysis methods would require some form of time integration to assess the effect of cyclically varying interface pressures.

Appendix

1. Contributors

2. References

EPUAP WG1 is currently conducting a literature survey of all references relating to laboratory assessment of support surfaces. When all comments have been received regarding this guideline and a final draft agreed, a subset of WG1 references relating to interface pressure will be added here. All queries or suggestions for WG1 should be addressed to the chairman, Dr Michael Clark, at the email address given above. Some of these references have already been published in previous editions of EPUAP Review, available to all EPUAP members.

3. Comments

Your comments on this document are welcome. Please submit these via email only to the working group chairman, Alastair McLeod, at the following address:

alastair.mcleod@huntleigh-healthcare.com

Thank you for spending time reading this guideline, and I look forward to a lively exchange of ideas as we refine the wording.

Alastair McLeod