Accuracy of interface pressure measurement systems

Communications

Accuracy of interface pressure measurement systems V. Allen, D. W. Ryan*, N. Lomax and A. Murray Regional

Medical Physics Department; Newcastle-upon-Tyne NE7 7DN, UK

and *Intensive

Therapy

Unit, Freeman

Hospital,

Received February 1992, accepted September 1992

ABSTRACT Interface pressure measurement is needed to assess beds designed to prevent pressure sores, so it is therefore important to establish the accurq of intnface pressuremeasuring systems. In this study, the Talky SASOOpressure evaluator (with 28mm and 700mm sensor pad), the DIPE (with 700mm sensorpad), and a water-@led blaader system (with 0.1 ml and 0.3 ml water) were assessed, Measurement errors were evaluated using a loading system with pressures up to 7.4 hPa (55mm Hg) in steps of 0.9hPa (6.9 mm Hg). All systems tested over-measured interface pressure, the error being approximately linearly proportional to the loading pressure. The repeatability for a given system was approximately constant. The mean error (zk SD) (96) and repeatability &Pa) for the systems were: 28mm Tally 72 + 196, + 0.07hPa; 1Omm Talley 75 + 196, f 0.07hPa; DIPE 27 zk 396, f 0.12hPa; 0.1 ml water bladder 17 k 196, f 0.13 hPa; 0.3 ml water bladder 26 + 3oh, + 0.07hPa. Dzflerent intelfaces affected accuracy marhedly, and repeatability was affected when an inhomogeneous interface was used. The study shows that the errors associated with interjk pressure measurement systems can be substantial, and can vary from one system to another. Keywords: Transducers, interface pressure, measurement errors

IT’ITRODUCTION Immobile patients may develop pressures sores’*2. These do not heal easily and steps should be taken to prevent them occurring. Pressure sores develop on areas of the body subjected to deformation by prolonged and unrelieved high pressure. When the pressure between the patient’s skin and the bed surface (the ‘interface pressure’) is higher than the mean capillary blood pressure, the capillary vessels tend to collapse with the result that blood cannot reach the high pressure area. If the pressure is sustained, then the area becomes ischaemic and eventually necrotic. Reduced levels of oxygen in the tissue below areas of high pressure have been clearly demonstrated by Bader and Gant”. Factors other than ressure influence the development of pressure sores. !b ey include humidity and temperature. Shear stress is also a causative factor4.“, though the measurement of shear stress at the skin/bed interface is at present a formidable task”. Capillary pressures tend to lie between 4 and 5.3 kPa (30 and 40mmHg), so to prevent pressure sores it has been suggested that interface pressures less than 4 kPa (30mm Hg) are desirable7. To achieve low interface pressures, beds and mattress overlays have been designed to reduce the support ressure by distributing the pressure more evenly. %xamples include the water bed, the low air Correspondence and reprint requests to: Dr V. Allen 0 1993 Butterworth-Heinemann 0141-5425193104344-05 344

for BES

J. Biomed. Eng. 1993, Vol. 15, July

loss (LAL) bed, the air-fluidized (AF) bed, and the ripple bed which works by regularly alternating the parts of the body at high pressuresx. It is important to evaluate quantitatively the pressure-relieving capability of any bed used by patients most at risk from pressure sores. Several studies have been carried out, measuring the interface pressure between body and bed at key sites. Thomson et al.“, Boorman et al.“‘, and ester and Weaver’i have assessed AF beds. Jeneidi’ J evaluated the LAL and Ring’s Fund beds, whilst Wild’” assessed the water and Ring’s Fund beds. Ryan and ByrneI have compared water, LAL, and AF beds with a standard hospital bed. Ten support surfaces including mattresses, pillows, operating tables and a linoleum-covered floor, as well as beds, were assessed by Redfem et aZ.15. It is inherent1 difficult to measure interface presY sures accurately s*17, since the presence of a pressure measurement system at the interface can introduce a ‘perturbation error’ and influence the pressure measurements obtained. O’Leary et aLI noted that the bed/body interface can also influence accuracy. Several systems have been developed to measure interface pressure. The Oxford Pressure Measurement System has been described by Bader and Hawken’s, and a capacitive transducer by FergusonPell et aZ.lg. Robertson et al. Z) described an electroneumatic device now marketed by the Talle RI edical Group. Barbenel and Sockalingham J altered a Talley sensor pad for use in their interface

htnjiie

pressure monitor. Other systems that have been used include a matrix system2’ and the Sub-Capillary Pressure Monitor 23. Although many systems have been described, few reports provide either a scientific assessment of errors or comparisons between devices. Patterson and Fisherz4 investigated the accuracy of five electrical transducers for the measurement of pressure applied to the skin. They found that when the transducers were used to measure pressure between a lower leg and an inflated pressure cuff, errors of between 1.3 to 6.7kPa (10 to 50mmHg) resulted. Other studies have made qualitative assessments of measurement errors. Robertson et ~1.” discussed in detail, potential sources of error in their electro-pneumatic device. Most published reports do not attempt to assess the accuracy of the device being used to measure interface pressure. Since the study by Patterson and Fisherz4 was confined to electrical transducers, it would be of value to investigate the accuracy of other types of interface pressure measuring system using a simple, easily standardized technique. The aim of this work was to determine the accuracy of some commonly available interface pressure monitors and to investigate factors influencing their accuracy. METHODS Interface pressure measurement systems A general interface pressure measurement system consists of a sensor, a pressure transducer, a pressure display, and in systems requiring inflation an audio or visual indicator showing when the pressure in the sensor is e ual to ap lied pressure. The first system investigate 1 was the J?alley Pressure Evaluator SA500 system (Talley Medical Group Ltd, Hams, UK). This is an electro-pneumatic system with an inflatable, flexible bladder which has electrical contacts on its ressure inside the bladder at inside surfaces. The which the contacts g reak represents the external ressure on the bladder. The system has a handset ousing a pressure transducer, an analogue display, a visual indicator, and a hand-operated inflator. There are two sizes of sensor: the smaller (28mm diameter) has one electrical contact to break, the larger (100 mm diameter) has a matrix of contacts. The Digital Interface Pressure Evaluator (DIPE) system (Next Generation Co Inc, CA, USA) was also studied. This also is electro-pneumatic, and is similar in operation to the Talle system, but it has a single size of sensor. It has a 1:andset housing a pressure transducer, a digital display, an audio indicator, and a detachable hand-o erated inflator. The water blad Ber system described by Ryan and B e14 was also investigated. This has a plastic bEd er sensor containing a small, known amount of water, which was connected to a Novatrans electronic strain gauge ressure transducer (Medex Medical Inc, Lanes, UK) g y a lastic tube with walls much more rigid than the bYadder walls. The output of the pressure transducer was amplified and displayed as a voltage on a digital voltmeter. This system required no inflation; an indicator was not therefore necessary. Since the volume of water in the bladder could be

R

pressure measurement mm:

V. Ah

et al.

varied, and since two sensors were available for the Talley system, several specific, variant systems were studied: Talley SA.500 with 28 mm diameter sensor Talley SA500 with 100mm diameter sensor DIPE with its own 100 mm diameter sensor 19mm diameter water bladder system with 0.1 ml of water in bladder 19 mm diameter water bladder system with 0.3 ml of water in bladder. Note that a 50ml air reservoir was linked in series with the 28mm sensor pad when used in the Talley system. This allowed greater control over inflation and deflation of the pad. Calibration of pressure transducers and sensors The calibration of the DIPE and Talley pressure transducers was checked against a reference ressure transducer (Novatrans), which was calibrate B against a head of water (up to 8 kPa or 60mm Hg in nine steps). Any mean calibration offset error was allowed for in subsequent readings. The calibration of the sensors was then checked, to verify whether the pressure within the sensor equalled the external pressure. This was done hydrostatically so that there was an even pressure distribution across the sensors. Each sensor in turn was connected to the reference pressure transducer and submerged in a tank of water such that the plane of the sensor was horizontal. The head of water above the sensor was varied (up to 4 kPa in four 1 kPa steps) to vary the pressure external to the sensor, and the displayed pressure was compared with the calculated pressure. Measuring the accura.cy of interface pressure measurement systems

Figure 1 shows the measurement arrangement used in this study. The sensor of the system being tested was laced centrally and between two layers of foam ‘combustion modified HG40’ foam - Carobel ManuP facturing Co Ltd, Tyne and Wear, UK). An aluminium baseplate of mass 6.2 kg and lead bricks (each of mass 6.3 kg) were used to load the foam, the force exerted b the load being distributed over the area of the foam r6.63 x lo-* m2>. The reference pressure was the calculated pressure (I’& due to the baseplate and bricks without the sensor in place. With the sensor in place, the foam was loaded with the baseplate and then with seven bricks, one by one

Lead

bricks

FoT~?r$q$y-; Measurement system Hard

Figure 1

bench

Experimental

arrangement

J. Biomed.

for testing system accuracy

Eng. 1993, Vol. 15, July

345

Int.e$rocc pressure measurement errors: V. Allen et al. 2.5,

(up to 7.4 kPa or 55 mm Hg). The baseplate was kept horizontal by monitoring it with a spirit level, and the pressure measured by the system was recorded at each step. When an eighth brick was added, no reading was taken. The bricks were then removed one by one, followed by the baseplate, with a measurement of ressure being taken at each step. This accounte dp for any possible load/unload hysteresis in the measurements. The foam was then lifted and the sensor repositioned in approximately the same position. The whole process was repeated four times.

o.o-

Assessing the influence of the interface on accuracy The perturbation

error caused by placing a sensor at an interface is not exclusively dependent upon the sensor; the interface itself is also likely to influence the accuracy. By varying the foam type and the position of the sensor within the measurement arrangement, a range of different interfaces could be produced. The procedure described in the revious section was used with a reference system (Ta Pley with 28 mm sensor) to assess the influence on accuracy of the following eight interfaces: between 2 layers of combustion modified HG40 foam between 2 layers of combustion modified HG24F foam between 2 layers of Dunlop Care foam 35 between 2 layers of Dunlop Care foam 67 between 2 layers of chip foam R3 between 2 layers of chip foam R6 between combustion modified HG40 foam (double layer) and the baseplate between combustion modified HG40 foam (single layer) and the baseplate

The last two arrangements were used to investigate the effect of one side of the interface being rigid.

RESULTS Calibration of pressure transducers and sensors

012345678012345678

Ptalc

a

(kPa)

’

’

b

’

’

’

’

“cafe (kPal

Figure 2 System error ucrsas PcnlCfor a, the DIPE system and Talky system variants (A = DIPE (IOOmm), m=Talley (IOOmm), l = Talley (28 mm)], and b, the water bladder system variants (H = 0.3 ml water, 0 = 0.1 ml water). The same interface was used in each case. Each point is the mean of eight measurements, and the error bars are +lSD

system since its anal0 e scale starts at 20mm Hg. However, note how tiYe DIPE and water bladder systems show the error increasing as Pcarc decreases below 2.7 kPa (20 mm Hg). Since system error increased linearly with Pcalc (above 2.7 kPa or 20 mm Hg), it can be expressed as a percentage of &,ro the loading pressure. Table 7 shows values of the mean s stem error expressed as a percentage (+SD), for eat K system. Since the Talle system could not measure below 2.7 kPa (20 mm Hg r , only values of system error at loading pressures eater than 2.7 kPa (20 mm Hg) were used to calcuffate the mean for any system. In each case this corresponded to six values. The standard deviation of the system error, also expressed as a percentage of P calo reflects how the system error changed over the range of loading pressures used. The error bars in Figure 2 represent repeatability and are approximately constant for all the loading ressures used. In Table I, the repeatability is there Pore expressed in absolute terms rather than as a percentage of Pcalc.

The DIPE pressure transducer

was found to have a negligible mean calibration offset error of 0.03 z? 0.04 kPa (0.2 z!z0.3 mm Hg). The Talley pressure transducer was found to have a mean calibration offset error of 0.28 + 0.04 kPa (2.1 + 0.3 mm Hg), which was subtracted from all subsequent measurements using the Talley s stem. For each sensor the J ifference between the ressure measured within the sensor and the calcu Pated external hydrostatic pressure was negligible.

Accuracy of interface pressure measurement systems Figure 2 shows plots of ‘system error’ versus Pcatc for all systems studied. System error is defined as (P,,, PC& where PsYs, is the pressure measured by the system and Pcarc is the calculated pressure. Each point is the mean of eight readings (four loading, four unloading), so the error bars (4 1 SD) embrace any loading/unloading hysteresis. Readings below 2.7 kPa (20 mm Hg) were not possible with the Talley

346

J. Biomed.

Eng. 1993, Vol. 1.5,July

’

Effect of interface on accuracy Figure 3 shows plots of system error versus Pcalcfor the Talley system (with 28mm sensor) when used at different interfaces. The results show that the interface does affect accuracy. In the case of a chip foam interface, repeatability is also affected, giving a much larger spread of readings at a given value of ear=. Also, if one side of the interface is rigid, the error increases, as shown in Figure 3d. Table

1

System error and system repeatability

System

System error + SD (l/o of PC.3

Repeatability

28 mm sensor lOOmm sensor

12f I 15 k 1

f 0.07 f 0.07

DIPE

IOOmm sensor

27 + 3

f0.12

Water bladder

0.1 ml 0.3 ml

17+ I 26 f 3

20.13 + 0.07

Talley pressure evaluator SA500

&Pa)

Inurfcc pressuremea.surcmcnterrors: V. Alien et al. 2.5 =Combustion . -Combustion

l

L1IIlILI,II’~‘II’,LIIII111,III1IIII,

0

1

a

2

3 P

4

. = Dunlop = Dunlop

modifified HC40 modified HG24F

5

talc ’ kPa)

6

7

8

0

b

n

1

care care

35 67

= Chip 6 5: Chip

n

234567801

‘car,

IkPa)

foam R6 foam R3

2345678

C

P

talc

1

01 ‘kPa)

d

2345678 Pcalc IkPa)

Figure 3

System error versus f’,,Ic for different interfaces: a-c, show results for foanxfoam interfaces whilst d, shows results for a foam:foam interface compared with foam:base-plate interfaces. The Talley system (28mm) was used in each case. Each point is the mean of eight measurements. Error bars are t 1SD

DI!XXJSSION Accuracy of systems Cross-calibration of the Talley and DIPE pressure transducers with a reference transducer revealed a calibration offset error in the Talley transducer which was taken into account in subsequent readings. Figure 2 shows that the 28mm sensor was more accurate than the 100mm sensor for the Talley system. This could be because the smaller sensor has only one contact to break, whereas the larger one has many more. These are not all likely to break at the same pressure. Most might break at the same pressure as the 28mm sensor contacts, with a few breaking either side of this pressure. It is, however, the pressure at which the last contact breaks that is recorded. The DIPE system was not as accurate as any of the Talley system variants. A ossible reason for this is related to the way the DIP l? system works. The DIPE system dis lays pressure digitally as the sensor is inflated. & en the contacts within the sensor break, the display ‘freezes’, making for easy reading of the pressure. It is not possible to take a reading of ressure as the contacts re-make upon slowly de&ting th e sensor, preventing the operator from homing in on the ‘correct’ pressure. This is possible with the Talley system, though it takes more skill to control the inflation/deflation of the sensor (especially the 28mm one). The water bladder systems were the easiest to use, requiring no inflation, though they were not as accurate as the Talley systems. The volume of water within the bladder affected the accuracy markedly: the 0.1 ml bladder was much more accurate than the 0.3 ml bladder. This is likely to be due to the greater volume of the 0.3ml bladder making the sensor thicker, which leads to a greater perturbation at the interface. A practical problem with water bladder sensors is that it is hard to expel all air bubbles from the system. Also, if the system had a leak, then low pressures would be recorded. In their study of the accuracy of several electrical transducers, Patterson and Fishers4 found that system errors were in some cases very large (ranging from

about 1.3kPa to 6.7 kPa). Although these results are important and show the size of errors which can be encountered in a more clinically realistic situation, a more standardized measurement technique for measuring accuracy is needed so that the accuracy of different types of interface pressure measuring system can be compared. The Talley s stem could not measure below 2.7 kPa (20 mm x g), though results for the DIPE and the water bladder s stems indicated a problem in this ay: at tKecreasing pressures below 2.7 kPa mm g) the s stem errors increased. This could /L!!Z3Z!Jfect. &I e bulk modulus of the foam may be higher at low compressive stresses due to drying out of the foam. At low pressures this would make the foam appear more rigid, less accommodating, and consequently increasing the perturbation at the interface by the sensor. Effect of interface on system accuracy Figure 3 illustrates how the accuracy of a s stem can be affected simpl by using different inte x aces. This is most likely cyue to different materials having different bulk moduli. In general, the more elastic and yielding an interface is, the less the presence of a sensor is noticed, resulting in a lower perturbation error. The interface can also affect repeatability, as the results for the chip foam indicate. Chip foam is not as homogeneous as the other foams used. Repositioning the sensor can result in it being in contact with a hard chip, a soft chip, or the matrix part of the foam. This results in a greater spread of pressure readings compared with there being no chips. When one side of the interface was rigid, errors were lar er than if both were yielding. This is to be expecte cf, and has implications for using a rigid surface as a reference support as a comparison for pressure-relieving beds. Any interface pressure measurement made between a patient and a rigid support surface will be false1 high due to the increased perturbation error, an d so should be treated with caution.

J. Biomed. Eng. lil93. Vol. 15, July

347

Inurfnccpwssuremcar~rement errors: V. Al&n et al.

CONCLUSIONS It is important to assess the accuracy of any device used for measuring interface pressure, since it has been shown in this study that different devices have different accuracies. This, along with the fact that the interface itself affects accuracy, should be borne in mind when making measurements of the pressurerelieving capabilities of specialist beds.

ACKNOWLEDGEMENTS We thank Support Systems International (SSI) who supported this study, Kinetic Concepts Inc (KCI) for the loan of the DIPE system, and Dr P.J. Lowe of the Regional Medical Physics Department, Newcastle General Hospital, for helpful discussions.

IUXRENCES 1. Wyllie FJ, McLean NR, McGregor JC. The problem of pressure sores in a regional plastic surgery unit. JRoyal Cell Surgs of Edinburgh 1984; 29: 38-43. 2. Versluysen M. Pressure sores in elderly patients. J Bone Joint Surg 198.5; 67-B: 10-13. 3. Bader DL, Gant CA. Changes in transcutaneous oxygen tension as a result of prolonged pressures at the sacrum. Clin Phys Physiol Mear 1988; 9: 33-40. 4. Dinsdale SM. Decubitus ulcers: role of pressure and friction in causation. Arch Phys Med Rehab 1974; 55: 147-52. 5. Bennett L, Kauner D, Lee BK, Trainor FA. Shear versus pressure as causative factors in skin blood flow occlusion. Arch Phys Med Rehab 1979; 60: 309-14. 6. Beebe D. Accuracy of pressure and shear measurements. In: Webster JG, ed. Prevention of Pressure Sores - Engineering and Clinical Aspects. Bristol, Philadelphia, New York: Adam Hilger, 1991, 155-74. 7. Garhn SR, Pye SA, Hargens AR, Akeson WH. Surface pressure distribution of the human body in the recumbent position. Arch Phys Med Rehub 1980; 61: 409-13. 8. Bliss MR, McLaren R, Exton-Smith AN. Preventing pressure sores in hospital: controlled trial of a large celled Ripple mattress. Brit MedJ 1967; 1: 394-7. 9. Thomson CW, Ryan DW, Dunkin LI, Smith M, Marshall M. Fluidised-bead bed in the intensive therapy unit.

348 J. Biomed. Eng. 1993, Vol. 15,July

Lancet 1980; 1: 568-70. 10. Boorman JG, Carr S, Kemble JVH. A clinical evaluation of the air-fluidised bed in a general plastic surgery unit. Brit J Ptkst Surg 198 1; 34: 165-8. ll., Jester J and Weaver V. A report of clinical investigation of various tissue support systems used for the prevention, early intervention and management of pressure sores. Ostomy/WoundManage 1990; Jan/Feb: 39-45. 12. Jeneid P. Static and dynamic support systems - pressure differences on the bodv. In: Kenedi RM. Cowdon IM. Scales JT, eds. Be&ore kiomechanics. London: Macmilan Press, 1976, 288-99. 13. Wild D. Body pressures and bed surfaces. NUTSStandard 1991; 5: 23-7. 14. Ryan DW and Byrne PO. A study of contact pressure points in specialised beds. Clin Phys PhysiolMea-s1989; 10: 331-5. 15. Redfem SJ, Jeneid PA, Gillingham ME, Lunn HF. Local pressure with ten types of patient/support systems. Lancet 1973; 2: 277-80. 16. Ferguson-Pell MW. Design criteria for the measurement of pressure at body/support interfaces. Eng Med 1980; 9: 209-14. 17. O’Leary JP, Lyddy JE, Little AD. A non-distorting transducer for measuring pressure under load-bearing tissue. Proc 5th Ann Conf System-s&Devices for the Disabled, 1978, Baylor College of Medicine, Houston, TX, 239-42. 18. Bader DL and Hawken MB. Pressure distribution under the ischium of normal subjects. J Biomed Eng 1986; 8: 353-7. 19. Ferguson-Pell MW, Bell F, Evans JH. Interface pressure sensors: existing devices, their suitability and limitations. In: Kenedi RM, Cowdon JM. Scales JT, eds. Be&ore Biomechanics. London: Macmillan Press, 1976, 189-97. 20. Robertson JC, Shah J, Amos H, Druett JE, Gisby J. An interface pressure sensor for routine clinical use. Eng Med 1980; 9: 151-6. 21. Barbenel JC and Sockalingham S. Device for measuring soft tissue interface pressures. J Biomed Eng 1990; 12: 519-22. 22. Seymour RJ and Lacefield WE. Wheelchair cushion effects on pressure and skin temperature. Arch Phys Med Rehab 1985; 66: 103-8. 23. Clark M. Continuous interface pressure measurement using an electro-pneumatic sensor: the SCP monitor. Care SC Pratt 1987; June: 5-8. 24. Patterson RP and Fisher SV. The accuracy of electrical transducers for the measurement of pressure applied to the skin. I’ Trans Biomed Eng 1979; 26: 450-6.