Analysis of Localized Erythema using Clinical Indicators and Spectroscopy

Abstract

  Localized erythema is regularly used as an indicator of post-ischemic events, including reactive hyperemia and Stage I pressure ulcers. The National Pressure Ulcer Advisory Panel definition of a Stage I ulcer includes both visual and nonvisual indicators — in part to improve identification in darkly pigmented skin.

A prospective, repeated-measures design was used to collect information on pressure-induced erythema that includes reactive hyperemia and Stage I pressure ulcers with an emphasis on distinguishing indicators in light and dark skin The relationships among clinical indicators (skin assessments) and results from tissue reflectance spectroscopy, as well as the clinical utility of spectroscopy for discerning tissue blanching status, were examined in a convenience sample of 76 inpatients and outpatients (95 test/control site pairs). Chi-square analysis and generalized logistic models were used to identify relationships and distinguishing characteristics of erythema. Analysis of variance was used to analyze blanching using spectroscopy. Nonblanching sites were more likely to be persistent erythema (χ2=5.3; P = 0.021) but exhibited no relationships to temperature, tissue resilience, or disability. Erythema in subjects with dark skin was more likely to be nonblanching and have poor resilience. Spectrographic analysis of blanching found significant differences across skin pigmentation (P = 0.0001) and blanching status (P = 0.019). These results reinforce the belief that dark skin must be assessed differently than light skin and indicate that clinicians should use persistence of erythema rather than blanching status to judge incipient pressure ulcers. These results validate the use of visual and nonvisual indicators included in the National Pressure Ulcer Advisory Panel Stage I pressure ulcer definition.

  Localized erythema is regularly used as an indicator of a post-ischemic event. Transient erythema is typical of reactive hyperemia, a physiological response to tissue oxygen debt. Persistent erythema is deemed more significant and is one indicator of a Stage I pressure ulcer. The National Pressure Ulcer Advisory Panel (NPUAP)¹ defines a Stage I pressure ulcer as "an observable pressure-related alteration of intact skin whose indicators, as compared to the adjacent or opposite area on the body, may include changes in one or more of the following: skin temperature (warmth or coolness), tissue consistency (firm or boggy feel), and/or sensation (pain, itching). The ulcer appears as a defined area of persistent redness in lightly pigmented skin; whereas, in darker skin tones, the ulcer may appear with persistent red, blue, or purple hues."

  This definition combines visual with nonvisual indicators and specifically addresses the differences in visual indicators across skin pigmentation.² Lyder³ compared the definitions of Stage I ulcers and found a wide discrepancy between classification systems. While all the definitions limited Stage I ulcers to superficial involvement, the descriptors varied and included persistence of erythema (20 minutes to 24 hours), warmer temperature, blanching, and nonblanching tissue, and elevated capillary pressure.

  Stage I ulcers represented 37% of all ulcers in a recent prevalence study of acute care patients.⁴ Stage I ulcers also accounted for 48% of all ulcers in Caucasians patients but only 20% in African American patients. This discrepancy corroborates additional findings^5,6 and illustrates problems of identifying erythema in people with darkly pigmented skin. Despite its importance in assessing skin and in the prevention of pressure ulcers, research on clinical erythema is limited.

  Nixon and colleagues⁷ tracked erythema and pressure ulcers in postoperative patients. They distinguished between blanching and nonblanching erythema and also tracked different levels of open ulceration. Of the 398 observations, 65 exhibited erythema or a pressure ulcer. Most common was blanching erythema (56 out of 65) followed by nonblanching erythema (four out of 65) and superficial damage to the epidermis (five out of 65). Of the 56 occurrences of blanching erythema, 20 had resolved by the second postoperative day, 33 resolved by the eighth day, two progressed to nonblanching erythema, and one progressed to superficial epidermal damage. The four sites of nonblanching erythema resolved by day 8 and none progressed to higher stage ulceration.

  To meet the objectives of this study, the number of characteristics used to describe pressure-induced erythema was increased from those used by Nixon and colleagues, but the areas are not followed over time. All occurrences of erythema, including reactive hyperemia and Stage I pressure ulcers, are described with an emphasis on distinguishing indicators in light and dark skin. The primary objective was to study the relationships among clinical indicators (characteristics) obtained from a skin assessment and results from tissue reflectance spectroscopy to improve the understanding of erythema across skin pigmentation and validate the visual and nonvisual indicators of the NPUAP Stage I definition. A second objective was to determine the clinical utility of using tissue reflectance spectrometry (TRS) to discern blanching versus nonblanching erythema.

Methodology

  The protocol followed a prospective, repeated-measures design using a convenience sample of subjects. Test sites were areas of erythema or skin discoloration at anatomical sites that are at risk for pressure ulcer development. For subjects with darkly pigmented skin, erythema could not always be visualized. Test sites in these subjects consisted of discolored areas located at weight-bearing bony prominences or other high-risk sites. Control sites in all subjects were non-discolored areas of skin adjacent to the test site. Adjacent control sites rather than contralateral sites were used to better accommodate variations in skin pigmentation and tissue characteristics and to permit relative measures of temperature and tissue stiffness variables.

  Inpatients and outpatients of an acute rehabilitation hospital qualified for the study. Research staff were notified of potential subjects by hospital staff, typically nurses or physical therapists. Patients were excluded if they were 1) unable to provide consent, 2) under 18 years of age, and 3) had skin conditions that precluded assessment of localized erythema. The hospital's Institutional Review Board approved the protocol and consent form.

  After informed consent was obtained, subjects were positioned comfortably on a mat or bed and a skin assessment was performed to identify all potentially erythematic sites. Two members of the research team performed test site inspections and described these sites using seven categorical clinical indicators. The blanching status of the site was determined using a curved lens (see Figure 1) to facilitate visualization. Two investigators, experienced in pressure ulcers, categorized the clinical indicators. This approach was used to enhance reliability over the use of two investigators assessing tissue independently, although no formal analysis was done.

  Before spectrographic evaluation, test sites were cleaned and shaved, if necessary. Tissue reflectance spectrometry data were taken during an initial resting period, during pressure application of 150 mm Hg, and for 2 minutes post-blanch. Pressure application was designed to mimic the clinical practice of applying finger pressure to an erythematic site to determine if it blanches. Following TRS data collection, the temperature of each site was taken using Dermatherm Perfusion Monitors (Sharn, Inc, Tampa, Fla.). Dermatherm Monitors are single-use liquid crystal thermometers mounted to a nonlatex-based paper tape. The thermometers have a precision and accuracy of 0.5 degrees F over the temperature range from 80 degrees F to 100 degrees F, with a response time of less than 15 seconds.⁸

  The clinical assessment classified erythematic or discolored "at risk" sites using seven categorical groups. Two categories - tissue resilience and temperature - were relative to the state of the control site, with the remaining groups being independent measures of test sites. The seven categorical groupings with associated definitions included:
  1. Transient or persistent erythema. Transient erythema was defined as erythema that resolved or was resolving within the 20- to 30-minute clinical and spectroscopic examination. Persistent erythema was, therefore, defined as erythema that did not or was not resolving over this time period.
  2. Blanching or nonblanching. A test site was defined as blanching if it visibly transitioned from rubor to pallor to rubor during application and release of pressure; in some people with darkly pigmented skin, blanching might not have been visualized due to skin pigmentation.
  3. Superficial insult or deep tissue injury (DTI). Test sites where erythema or discoloration appeared limited to the epidermis and dermis were categorized as a superficial insult. Most areas of reactive hyperemia and erythema seen by clinicians fit this description. Deep tissue injury described areas that had the appearance of deep trauma or bruising. In people with light skin pigment, DTI often appears with purple or blue hues.
  4. Normal or poor tissue resilience. Tissue resilience was compared between test and control sites using digital palpation. Sites with poor tissue resilience were deemed less resilient or less stiff compared to adjacent tissue. This characteristic also has been described as "mushy" or "boggy."
  5. Bony or soft tissue. Sites were categorized anatomically if they were located over a bony prominence (eg, ischial tuberosity, greater trochanter) or soft tissue (eg, posterior thigh).
  6. Skin temperature. Test site temperatures were compared to control site temperatures and categorized as having the same temperature or being warmer or cooler. The latter groups were defined as having >1 degrees F difference in skin temperature compared to control sites.
  7. Skin pigmentation. Skin was categorized as lightly pigmented or darkly pigmented. Darkly pigmented skin was assigned if blanching of a control site was not visible.⁶

Tissue Reflectance Spectrometry

  Tissue reflectance spectrometry, a common noninvasive measurement for inspecting skin color,^9-12 uses the characteristic absorption of light by the skin constituents to measure the amount of constituents present. A white light is shone on the skin while detectors measure the returning light. This light is divided into spectral components by a monochromator and detected by a photomultiplier tube. Absorption of light by tissue is calculated by comparing the measured reflectance with that of a white standard. The theory of TRS is based upon a simple anatomical model.¹² Light passes through the epidermis (melanin layer) and a plexus of blood vessels in the dermis (hemoglobin layer) before being reflected off collagen in the lower dermis. From the spectral characteristics of the light returned, the relative amounts of each skin chromophore can be determined. Three human tissue chromophores within the visible spectrum influence the measurement of erythema: hemoglobin, oxyhemoglobin, and melanin.

  Equipment. A Monolight 6800 series spectrophotometer (Rees Instruments, now Imaging and Sensing Technologies, Horseheads, NY) consisting of a scanning monochromator, photodetector, and light source with a fiber optic cable was used for TRS data collection. This system operates in the spectral range from 350 nm to1100 nm and provides a resolution of 1 nm in the visible spectrum. The photodetectors are calibrated before use with a standard white source. Light is transmitted and received through a fiber optic cable. The end of the cable was fit with a probe that had a 3-mm diameter optic aperture. The aperture side of the probe had a slightly convex shape to minimize edge effects during tissue pressure application, and a merosphere was affixed to the top of the probe to interface with the pressure application system.

  A pneumatic indentor applied 150 mm Hg pressure orthogonally to the skin surface to mimic the pressure clinicians use to investigate whether erythema blanches. The pressure application system consisted of a computer-controlled, pneumatically driven piston. The piston is equipped with a load cell (Entran ELF-C1000-10, Entran, Fairfield, NJ) which provided feedback to the controller and maintained 150 mm Hg of pressure application while accommodating for slight body movements during monitoring.

  Erythema detection algorithm. The algorithm selected for this study had a sensitivity of 0.9%, a specificity of 85%, and reliability of 0.812.¹³ This algorithm was selected because it had equally high sensitivity and specificity in people with lightly and darkly pigmented skin. The algorithm first applies a melanin compensation routine to the reflected light spectrum. After melanin compensation, the spectrum is regressed using a standard, concentration-independent absorption curve of hemoglobin (in vitro Hb response) as the regressor. The coefficient (beta) of this regression is a measure of hemoglobin concentration. The hemoglobin concentration can be used as an erythema index, which varies directly with the amount of hemoglobin or "redness" in the skin. The average and standard deviation of the indices for a test/control site pair was used to calculate Z-scores using the equation

:       Z = (µ test site - µ control site/(s control site)

  The empirically determined optimal threshold to detect erythema was Z ≥ 4, meaning that erythema is defined whenever the test site mean had a hemoglobin content that was at least four standard deviations above that of the control site.

Data Analysis

  Clinical indicators. The binary clinical indicators assigned to test sites were converted into dummy variables. Chi-square analysis was performed to determine relationships between clinical indicators. A generalized logistic model was used to determine distinguishing characteristics across descriptor variables. Additional analyses investigated the relationships between TRS determination of erythema and clinical indicators. Statistical significance was determined at the 0.05 level, but results approaching that level also were reported.

  Detection algorithm sensitivity. For the subjects with light skin in whom erythema could be easily visualized, data were analyzed to determine algorithm sensitivity (sensitivity = [true positives]/[true positives + false negatives]) - that is, the ability of the algorithm to correctly identify tissue with visible erythema.

  Spectrographic analysis of blanching. For the sites determined to be erythematic by TRS, data were analyzed to determine whether the resting spectrum was different from the blanched spectrum. Z scores of spectra from scanning and blanched measurements were calculated using the formula:

      Z = (µ resting - µ blanch)/(s blanch)   This calculation represents how far the mean Hb content from a test site is from the distribution of Hb quantity from the same site while blanched. The underlying premise is that the scanning spectra of nonblanching sites would be much closer to its blanched spectra than blanching sites, since nonblanching sites, by definition, retain hemoglobin upon pressure application. Sites were categorized according to their clinical descriptors: light or dark skin and blanching or nonblanching tissue. A two-factor ANOVA analyzed differences in the four groups with significance determined at the P <0.05 level.

Results

  Data collection. Eighty-two subjects were recruited from patients presenting clinically with erythema or those with deeply pigmented skin that had discoloration at high-risk sites. Data from six subjects was excluded due to equipment error and early protocol changes, resulting in analysis of data from 76 subjects who exhibited 95 sites of erythema or localized discoloration or 95 test site/control pairs. Of these subjects, 45 were male and 31 were female. Subjects ranged in age from 19 to 81 years with an average age of 47.6 years (SD = 16.4). Twenty subjects had multiple sclerosis, 17 had spinal cord injury, 27 had a transtibial amputation, and the remaining 12 had other disabilities including stroke, head injury, and orthopedic injuries. Self-reported ethnicity divided the group into one Asian, one Indian, four Hispanic, 24 African American, and 46 Caucasian subjects.

  Clinical indicators. Occurrences of the categorical clinical indicators and relationships with erythema duration and blanching status are tabulated in Table 1. A near-equal distribution was achieved for transient-versus-persistent and blanching-versus-nonblanching erythema. The chi-square of these variables was significant (chi-squared = 5.3, P = 0.021); more nonblanching sites showed persistent erythema. Both blanching and nonblanching sites exhibited a range of temperatures relative to control sites (chi-squared = 1.7, P = 0.43).   Sites identified as DTI or bruises were more likely to be nonblanching but the difference was not statistically significant (chi-squared = 3.2, P = 0.071). Deep injury sites also exhibited a range of temperature and no significant difference was found (chi-squared = .80, P = 0.67) (see Table 2).

  The model used to identify the distinguishing characteristics of erythema within the different skin pigmentation groups used duration, resilience, blanching, and temperature as the regressors. Controlling for other characteristics, blanching and resilience emerged as significant predictors of skin type (model chi-squared, P ≤ .0016). The odds of blanching were 87% lower in dark skin compared with light skin (odds ratio = .13, 95% confidence interval [CI], 0.04 to 0.44). Additionally, light skin sites were substantially more likely to have normal resilience, showing a nearly seven-fold increase in the odds of resilience than dark skin (odds ratio = 6.95, 95% CI, 1.34 to 35.9). The wide confidence interval on the estimated odds ratio for resilience reflects imprecision in the estimate due to small sample size.

  Spectrographic determination of erythema. Fifty-seven test/control site pairs were taken from subjects with light skin. Using these sites, sensitivity of the algorithm and technique was found to be 0.79. No relationship was found between disability type and the TRS determination of erythema (chi-squared = 3.2, P = 0.36) (see Table 3). Proportionally more sites in patients with dark skin were deemed nonerythematic by TRS than in light skin (chi-squared = 2.3; P = 0.131) (see Table 4).

  The model performed to identify distinguishing characteristics of sites deemed erythema by TRS compared to nonerythematic sites used duration, blanching, and resilience as the regressors (model chi-squared, P = 0.607). None of these regressors approached significance (all P >.30).

  Spectrographic analysis of blanching. Mean and dispersion values of the 69 sites identified via spectroscopy as exhibiting erythema are shown in Table 5. ANOVA showed that significant differences existed between light and dark skin (P = .0001) and blanching/nonblanching (P = .019). The interaction between these two factors was not significant (P = .345).

Discussion

  The overall finding of the clinical indicator analysis was that erythematic sites possess a variety of characteristics and that absolute relationships across descriptors should not be anticipated by clinicians.

  The finding that persistent erythema was more likely to be nonblanching was expected, but this pairing was not absolute. Persistent redness was found at blanching and nonblanching sites, a result that corroborates the work of Nixon.⁷ Relatedly, relative temperatures varied across blanching and nonblanching sites, indicating that temperature also is not an absolute indicator. In related work, Sprigle et al¹⁴ concluded that both increased and decreased temperature differences can be used to indicate a potential skin integrity problem, but a problem might still exist despite the absence of a temperature difference. The authors believe all sites of persistent redness deserve attention, regardless of blanching status or temperature, because persistence indicates a pathological response to ischemia in contradistinction to reactive hyperemia, which is a transient event.

  The model designed to identify characteristics within the different skin pigmentation groups had two significant clinical indicators: blanching and resilience. The finding that nonblanching sites occurred more often in subjects with dark skin was not surprising because some sites, by definition, do not visibly blanch due to skin pigmentation alone. This result corroborates Bennett⁶ and Lyder et al,¹⁵ who stated that skin pigmentation can mask blanching. The finding that dark skin sites tended to be less resilient than their respective control sites was interesting. Localized changes in resilience or stiffness have been mentioned with respect to pressure ulcers, using terms such as mushy, boggy, and induration. However, these results indicate that a localized loss of resilience in dark skin may be a useful nonvisual indicator of erythema or ischemic damage.

  Fourteen sites were categorized as bruises or indicated DTI. This type of injury is well known clinically but poorly defined in the literature. Clinicians recognize that certain pressure ulcers begin in the deep tissues and by the time ulceration occurs in the epidermis, significant damage has already occurred. Barton and Barton¹⁶ described these as Type II pressure sores that were characterized by localized endothelial cell damage. The results of this study indicate that sites with deep injury exhibit persistent erythema are typically (but not always) nonblanching and exhibit a range of temperature compared to adjacent sites. Purple hues often indicate deep trauma in lightly pigmented skin; in dark skin, such coloration can indicate superficial damage. Therefore, color should not be used as a descriptor because DTI appears differently in dark and light skin. Better understanding of this entity is needed to appropriately document and acknowledge what is, and is not, a Stage I ulcer.

  Two uses of TRS were investigated for clinical utility: erythema and blanching detection. The algorithm had a slightly lower sensitivity in detecting erythema in light-skinned subjects compared to previously published controlled tests (0.79 versus 0.857). This suggests that the nature of the clinical erythema has a broader range of erythema variation than the induced hyperemia and the challenges of using TRS instrumentation are magnified when dealing with many anatomical sites and the varying lengths of time required to expose those sites (eg, buttock sites of wheelchair users take longer to expose than limbs of prosthesis wearers). The algorithm threshold may need to be modified to eliminate some of the Type I errors encountered; however, this will increase Type II errors. In fact, if the Z-score threshold to determine erythema was changed to Z = 3, reliability would have increased to 0.86, nearly identical to the 0.875 sensitivity found in the controlled experiment at this detection threshold.13 The protocol used in this study did not allow calculation of specificity because multiple control sites were not measured on each subject.

  The model configured to identify distinguishing characteristics of sites categorized as having erythema by TRS showed no significant or near-significant regressors. In addition, no relationship was found between disability type and the TRS determination of erythema. This result is important because it indicates that TRS results are not influenced by clinical blanching status, disability type, persistence of erythema, or the resilience of the tissue. Since TRS can "see" erythema independent of skin pigmentation, it appears to have the potential to identify incipient pressure ulcers in people with darkly pigmented skin.

  Relatedly, proportionally more sites in subjects with dark skin were deemed nonerythematic by TRS. This result was expected because the nonerythematic sites in people with light skin were false positives; whereas, in dark skin the sites were the sum of false positives and true negatives. This study assessed 37 risk sites on people with dark skin, with 24 of the sites identified as erythema. Some of these risk sites were not identified by clinical staff but were checked by research staff because of discoloration. This disparity in identifying risk sites reinforces the belief that dark skin must be assessed differently than light skin.

  Using TRS, sites deemed "nonblanching" via clinical inspection blanched differently than those clinically deemed "blanching." This result reflects a physiological difference between the sites. As expected, hemoglobin - and therefore, blood - is evacuated by pressure on tissue in blanching risk sites more than in nonblanching sites, and TRS is able to quantify this difference in Hb content. However, while TRS demonstrates the blanching phenomenon across multiple erythematic sites, its use as a diagnostic instrument is less definitive. The standard deviations of the Z scores in the TRS blanching analysis indicate a wide variation, meaning that some nonblanching sites blanched more like blanching sites and vice versa.

  Explanation of this finding reflects error in both the human determination of blanching and limitations of spectroscopy in this application. As stated, research staff used convex blanching tools to better view the blanching phenomenon during pressure application and refill as pressure is relieved. These blanching lenses not only allow visualization during pressure but also magnify the tissue in question. This technique is not without error because human judgment was still involved and determination is especially hindered in the case of darkly pigmented skin. The limitation of TRS in determining blanching is due to the potential optical changes when compressing tissue. Although pressure was applied consistently using feedback control on a pneumatic system, the effects on optical coupling during blanching pressure are not known. Deformation almost certainly changes the optical characteristics of tissue, and given the variations in skin elasticity and resilience, these changes are probably quite disparate.

Conclusions

  Compiling characteristics of clinical erythema offers insight into the variability of reactive hyperemia and Stage I ulcers. Clinicians should not disregard blanching erythema because some blanching sites exhibited persistent erythema and skin pigment can mask blanching as a visual indicator. The persistence of erythema should be the construct used by clinicians to identify incipient pressure ulcers. Other visual and tactile indicators such as temperature and resilience can be helpful in characterizing sites at risk for pressure ulcer development. The finding that dark skin exhibited localized resilience loss may be significant and deserves further investigation. Collectively, these findings support the NPUAP Stage I definition that utilizes both visual and nonvisual indicators.

  Although limited, the occurrences of DTI led to an important clinical conclusion: neither purple hue nor nonblanching status should be used as a singular defining characteristic of DTI because color is significantly affected by skin pigmentation and most, not all, DTI sites were nonblanching. Only persistence of erythema was an absolute characteristic of all DTI sites.

  The use of TRS appears to have clinical utility in assessing the skin of people with darkly pigmented skin. Because erythema cannot always be visualized, pressure ulcer risk assessment can be hindered; research indicates that Stage I ulcers are underreported in patients with dark skin. Tissue reflectance spectrometry can be used to check areas of discoloration located at weight-bearing bony prominences or other typically high-risk sites to determine the presence of erythema or discoloration from some other cause such as hyperpigmentation of the skin. Although TRS should not replace good clinical assessment, it has utility in providing additional information in assessing darkly pigmented skin and assisting in addressing the disparity of skin assessments identified by the aforementioned researchers. The use of TRS as an indicator for blanching is less clear. Further research is needed to quantify the effects of optical coupling on the quantification of Hb in tissue.

Dr. Sprigle is the Director of the Center for Assistive Technology and Environmental Access, Georgia Institute of Technology, Atlanta, Ga. Ms. Linden is the Manager of the Assistive Technology Program, Goodwill Industries, Cincinnati, Ohio. Dr. Riordan is currently completing his internship before pursuing a Physical Medicine and Rehabilitation residency at Mount Sinai Hospital, New York, NY. Please address correspondence to: Stephen Sprigle, PhD, PT, Center for Assistive Technology and Environmental Access, Georgia Institute of Technology, 490 Tenth Street, Atlanta, GA 30332; email: sprigle@arch.gatech.edu. This material is based upon work undertaken at Helen Hayes Hospital, West Haverstraw, NY, and supported by the National Institute on Disability and Rehabilitation Research, Award #H133G50018, and the New York State Department of Health.

1. National Pressure Ulcer Advisory Panel, Reston, Va. Available at www.npuap.org. Accessed August 1, 2002.

2. Henderson C, Ayello EA, Sussman C, et al. Draft definition of Stage I pressure ulcers: inclusion of persons with darkly pigmented skin. Advances in Skin and Wound Care. 1997;10:16-19.

3. Lyder CH. Conceptualization of the stage I pressure ulcer. Journal of ET Nursing. 1991;18(5):162-165.

4. Ambung SR, Miller WL, Bosley LM. The 1999 National Pressure Ulcer Prevalence Survey: a benchmarking approach. Advances in Skin and Wound Care. 2001;14:297-301.

5. Lyder C, Preston J, Scinto J, Grady J, Ahern D. Medicare quality indicator system: pressure ulcer prediction and prevention module. Final Report. Washington, DC: US Health Care Financing Administration; November 1998.

6. Bennett MA. Report of the task force on the implications for darkly pigmented skin in the prediction and prevention of pressure ulcers. Advances in Wound Care. 1995;8(6):34-35.

7. Nixon J, Brown J, McElvenny D, Mason S, Bond S. Prognostic factors associated with pressure sore development in the immediate post-operative period. Int J Nurs Stud. 2000;37:279-289.

8. CITECH. Temperature Indicator Comparison. CITECH Test Report #490-074. Plymouth Meeting, Pa; 1993.

9. Feather JW, Ryatt KS, Dawson JB. Reflectance spectrophotometric quantification of skin colour changes induced by topical corticosteroid preparations. Br J Dermatol. 1982;106:436-440.

10. Wan S, Parrish JA, Jaenicke KF. Quantitative evaluation of ultraviolet induced erythema. Photochem Photobiol. 1983;37:643-648.

11. Ferguson-Pell M, Hagisawa S. An empirical technique to compensate for melanin when monitoring skin microcirculation using reflectance spectrophotometry. Medical Engineering and Physics. 1995;17(2):104-110.

12. Takiwaki H, Serup J. Measurement of erythema and melanin indices. In: Serup J, Jemee GBE, eds. Handbook of Non-invasive Methods and the Skin. Boca Raton, Fla.: CRC Press; 1995:377-397.

13. Riordan B, Sprigle S, Linden M. Testing the validity of erythema detection algorithms. J Rehabil Res Dev. 2001;38(1):13-22.

14. Sprigle S, Linden M, McKenna D, Davis K, Riordan B. Clinical skin temperature measurement to predict incipient pressure ulcers. Advances in Skin Wound Care. 2001;14(3):133-138

15. Lyder CH, Yu C, Emerling J, Mangat R, Stevenson D, Empelo-Frazier O, McKay J. The Braden scale for pressure ulcer risk: evaluating the predictive validity in Black and Latino/Hispanic elders. Appl Nurs Res. 1999;12(2):60-68.

16. Barton A, Barton M. The Management and Prevention of Pressure Sores. London, UK: Faber and Faber; 1981.