The correlation between sub‐epidermal moisture measurement and other early indicators of pressure ulcer development—A prospective cohort observational study. Part 1. The correlation between sub‐epidermal moisture measurement and ultrasound

Abstract

The correlation between sub‐epidermal moisture (SEM) and other early indicators of pressure ulcer (PU) development is yet to be determined. This three‐part series aims to bridge this knowledge gap, through investigating SEM and its correlation with evidence‐based technologies and assessments. This article focuses on the correlation between SEM and ultrasound. A prospective cohort observational study was undertaken between February and November 2021. Patients undergoing three surgery types were consecutively enrolled to the study following informed consent. Assessments were performed prior to and following surgery for 3 days at the sacrum, both heels and a control site, using a SEM scanner and high‐frequency ultrasound scanner (5–15 MHz). Spearman's rank (r _s ) explored the correlation between SEM and ultrasound. A total of 60 participants were included; 50% were male with a mean age of 58 years (±13.46). A statistically significant low to moderately positive correlation was observed between SEM and ultrasound across all anatomical sites (r _s range = 0.39–0.54, p < 0.05). The only exception was a correlation between SEM and ultrasound on day 0 at the right heel (r _s  = 0.23, p = 0.09). These results indicate that SEM and ultrasound agreed in the presence of injury; however, SEM was able to identify abnormalities before ultrasound.

1. INTRODUCTION

Pressure ulcers (PU) remain an important clinical challenge which place an extensive burden on patients and on the healthcare system. ¹ , ² Given the high risk of PUs among surgical patients, ³ prevention of PUs in the surgical setting is imperative to reduce the complex interplay of risk factors that accompany a surgical operation. ⁴ Prolonged periods of immobility during surgery is a key risk factor at play, which heightens the localised pressure and/or shearing forces at the bony prominence. ⁵ PUs that develop early in the postoperative period link back to the pressure and shear that the individual is exposed to during the intraoperative period. ⁶ The challenge, however, lies upon the detection of injury, as surgery‐related PUs can develop up to 72 h after the surgical operation. ⁶ , ⁷ Thus, considering PUs are largely a preventable problem, current methods of detection should focus on identifying injury at the very early stages, as this is when the damage is reversible if corrective methods are applied to manage the early inflammatory processes. ⁸

PUs range in severity from superficial tissue damage to a cascade of cell and tissue damage involving multiple interacting pathways. ⁷ This range in severity represents the two predominant pathways of PU development described by Gefen. ⁹ Superficial damage is caused by friction and/or shearing forces, which occurs at the interface between the skin and supporting surface at the level of the epidermis. ⁹ The second development pathway, however, occurs chronologically in the deeper tissues, below the epidermis, and results from concentrated pressure and/or shearing forces at the surface overlying bony prominences. ⁹ Internal strain and stress next to the bony prominence are considered to be substantially higher when compared to the superficial skin surface. ¹⁰ Thus, current clinical practice assessments are fraught with challenges, as visual skin assessment (VSA) relies on visible changes at the skin surface. ⁷ A growing body of evidence suggests that PUs commonly originate from the deeper tissues ⁹ , ¹⁰ , ¹¹ , ¹² ; thus, a visible PU is only present once the early stages of microscopic inflammation have advanced to involve a macroscopic cascade of injury. ¹⁰ Therefore, a PU that has migrated up to the visible skin surface is a sign that irreversible damage has taken place. ⁸ This assessment is further challenged in patients with dark skin tones, as visible changes cannot readily be seen on the surface of the skin. ¹³ This renders VSA alone an unreliable assessment method that is not inclusive for detecting PUs within all patient populations. ¹⁴

Sub‐epidermal moisture (SEM) offers a solution to this problem, in that the early inflammatory response associated with PU development can be detected. ⁸ The SEM scanner™ is a CE‐marked medical device designed for the early detection of PUs, which can penetrate to a depth of 3–4 mm, extending through the epidermis, dermis and reaches the subcutaneous fat tissue. ¹⁵ SEM is a product of plasma that has leaked from the interstitial space in response to inflammation, having increased the local tissue permeability. ⁸ The trigger point for this onset of inflammation in PU pathophysiology is direct cell deformation at the bony prominence. ¹⁰ Subsequent to compressive forces that have distorted the soft tissue, the cytoskeleton of cells loose structural integrity and increased permeability causes abnormal transport of material to and from the cell. ¹⁶ Eventually, loss of homeostasis causes cell death by apoptosis, which is followed by an inflammatory response within the surrounding tissues. ¹² Apoptosis generates localised oedema at the damaged site, and this is the body's natural response to injury. ¹⁰ It assists the migration of inflammatory cells to the injured area to recover the damage that has taken place. The oedema causes further increase to interstitial pressures, which in turn results in further cell distortion and further cell deformation, as mechanical loads on cells and tissues increase. ¹⁰

A recent study by Martins de Oliveira et al. ⁵ supports SEM as an early inflammatory biomarker, as a high PU incidence (51%, n = 116) according to measures of SEM was observed within a surgical population. An important difference in PU incidence was observed, however, when relying on VSA alone (3%, n = 7). In consideration of research evidence to date, SEM has demonstrated the ability to detect injury an average of 5 days prior to a visual PU developing. ¹⁷ However, as abnormal SEM delta values do not always result in a visual PU, ¹⁷ this has led to questions pertaining to the significance of abnormal SEM delta values in absence of a visual skin injury. ¹⁸ A high proportion of abnormal SEM delta values have also been observed across other studies, ¹⁹ , ²⁰ , ²¹ which may unearth questions regarding the positive prediction value of SEM. However, the literature posits various explanations for this finding, one of which points to SEM's ability to detect damage in a premature form. In this instance, there is the ability to reverse the early inflammation through the use of prevention strategies as identified by Byrne et al. ²² Furthermore, many years ago, Bouten et al. ²³ identified that damage to the muscle may have been developing for some time before any evidence of this may manifest itself visually on the skin. This is because the deep damage emerges from the underlying structures and develops outwards towards the skin. Therefore, identification of an abnormal SEM reading, in the absence of a visual manifestation of the damage, is not necessarily an overprediction by SEM measurement, rather it is because that the true assessment of predictive ability is greatly influenced by the prevention strategies employed.

There is a paucity of evidence assessing SEM in combination with other assessment measures, which show promise in the early detection of PUs. ¹⁷ , ²⁰ , ²⁴ Gefen and Gershon ²⁴ conducted an observational pilot study to investigate SEM and ultrasound, whereby both assessments had consistently agreed in the presence of deep tissue injury (DTI). However, this study had included participants with existing grade 1 PUs and suspected DTI at baseline, which limits the ability to draw firm conclusions on the correlation between SEM and ultrasound for detecting early PU development. ²⁴ Furthermore, a recent systematic review has indicated the requirement for future research to assess SEM in populations free of PUs at baseline, so that the development of injury can be detected with a combination of prognostic measures. ²⁵

Another study investigating the correlation between SEM and Interleukin‐1α in an intensive care unit (ICU) setting uncovered weak correlations between these measures. ²⁰ As this study was conducted among critically unwell patients, a rationale for these findings is that SEM is measuring localised inflammation, whereas Interleukin‐1a could also be elevated in the presence of a systemic inflammatory response. ²⁰ Ultrasound has demonstrated the ability to visualise the severity or extent of the soft tissue damage subjected to pressure loading or shear forces in the deeper tissues. ²⁶ , ²⁷ , ²⁸ Thus, assessing the correlation between SEM and ultrasound will strengthen knowledge pertaining to the significance of abnormal SEM delta values in absence of a visual skin injury.

2. METHODS

2.1. Study aim and primary outcome

The overarching aim of this study was to explore the correlation between SEM, VSA epidermal hydration, temperature, pain and ultrasound, to establish whether SEM alone is viable as an instrument to use in practice for the detection of early PU development. SEM was the dependant variable, and VSA was the secondary dependant variable. While all measures were collected in this study, part 1 of this article series focuses on the results for SEM, VSA and ultrasound.

2.1.1. Secondary outcome

To report the incidence of PU development among the study participants, a visual PU was reported. To supplement this, the abnormal SEM delta threshold incidence was reported.

1. A visual PU identified through VSA was defined as a grade 1–4, unstageable and suspected DTI as categorised by the 2014 NPUAP, EPUAP and PIPPA PU classification system.

2. The SEM delta threshold was defined as abnormal when two sustained abnormal SEM delta values (≥0.6) over two consecutive days were observed on the same anatomical site succeeding a normal SEM measurement.

2.2. Study design

This study employed an observational, prospective cohort study design, following the STROBE guidelines. ²⁹

2.3. Study setting

This single‐centre study was conducted in an acute hospital setting within the Republic of Ireland.

2.4. Ethical considerations

Ethical approval was granted from the local hospital Institutional Review Board (1/378/2168) prior to study commencement, and written informed consent was obtained from all participants. Participant recruitment and data collection took place from February to November 2021.

2.5. Sample size

G*power software (version 3.1.9.4; Heinrich‐Heine‐Universität Düsseldorf, Düsseldorf, Germany; http://www.gpower.hhu.de) was used to calculate the sample size for a correlation analysis. A moderate correlation (±0.50–0.70) was hypothesized between the variables. To identify a correlation of 0.5 using a two‐sided test with a significance level of 0.05 and 95% power, the sample size was calculated as 46 participants. The sample size aim was 60 participants to account for drop out.

2.6. Study population

The study population of interest was adult inpatients scheduled for elective surgery in an acute hospital setting who were at‐risk of developing PUs.

2.7. Eligibility criteria

A convenience sample of patients admitted for an elective cardiothoracic, orthopaedic or general surgery were included. The inclusion criteria were adult patients (1) ≥18 years (2) with a planned 3‐day length of hospital stay (3) who provided written informed consent. The exclusion criteria were (1) patients who had a visual PU at baseline across all anatomical locations assessed (i.e. sacrum and both heels) and (2) patients undergoing emergency surgery.

2.8. Study assessments

2.8.1. VSA

A visual assessment of the skin was conducted using the 2014 NPUAP, EPUAP and PIPPA PU classification system. ³⁰ A transparent disc was used to distinguish blanching from non‐blanching erythema, and clinical photographs were obtained for validation by a second researcher.

2.8.2. SEM

SEM delta values were obtained using the Provizio SEM scanner™ (Bruin Biometrics), which works by measuring biocapacitance, a measurement of the electrical property of the tissue. ¹² Material properties of the tissues are sensitive to water content; therefore, as plasma leaks from the interstitial space during inflammation, this changes the electrical property of the tissue. ¹² Six measurements were collected from the sacral area, and four measurements were collected from the heels and control site. A singular delta value is then calculated which differentiates between the highest and lowest measurements around an anatomical site to determine tissue status. ³¹ In line with international literature and U.S Food and Drug Administration, a SEM delta threshold of ≥0.6 was considered abnormal. ²⁴ , ³² , ³³

2.8.3. Ultrasound

Ultrasound images were collected using a high‐frequency linear scanner 5–15 MHz, and the transmit frequency for all observations was 14 MHz (Clarius L15). This was connected wirelessly to a research dedicated tablet device to display ultrasound images. The musculoskeletal application and spatial compounding mode were selected as default for consistent imaging. This mode compounds imaging frames acquired at different angles to produce a single image. This has been shown to generate a higher echo amplitude, thus reducing anisotropic effects and improving overall image quality. ³⁴ The researcher, trained in musculoskeletal ultrasound, performed all assessments for consistency.

The transducer was held at the base between the thumb, index and long fingers. The small and ring fingers, and ulnar aspect of the hand made contact with the participant. The transducer rested on a thick layer of ultrasound transmission gel (Aquasonic® 100), which was placed over the anatomical area and compression was avoided to limit superficial hypoechoic areas moving out of field of view. This firm grip provided stability and controlled the force applied to the tissue. The transducer was held at a perpendicular (90°) angle to the assessed tissue structure, and transducer toggling techniques were applied to avoid anisotropy. Once the bony landmark of interest was identified, the probe moved around the skin surface from medial to lateral and proximal to distal direction to capture all relevant images of the skins underlying tissue. Frequency and gain settings were adjusted and optimised to a balanced state whereby all tissues could be clearly visualised. The depth was altered to increase field of view, and the region of interest was centred on the screen. The ultrasound images were observing an average depth of 2 cm at the heel and 3 cm at the sacrum.

Longitudinal and transverse image views were obtained for validation by a qualified radiographer and second researcher (validation reviewers). The ultrasound abnormality was visualised in both longitudinal and transverse planes. Videos were obtained when both normal and abnormal tissues were identified to allow for extensive review. A second screen was implemented for side to side comparison when reviewing ultrasound images and to identify subtle abnormalities compared with the previous day. Validation was conducted separately by both reviewers on a random sample of 10% of the total participants. The ultrasound data were examined on the clarius cloud drive in isolation of any of the other study data. Abnormal ultrasound images were placed into four categories to reflect severity following previous research methodology. ²⁴ , ³⁵ , ³⁶ An unclassified category was added to reflect poor imaging quality, resulting in a total of six ultrasound categories including normal assessment (Table 1). As gel was applied to the skin, ultrasound was consistently assessed after SEM, epidermal hydration and temperature.

TABLE 1.

Ultrasound categories with the associated definitions.

Ultrasound category	Explanation
Normal	No abnormalities detected; tissue structures appear normal on examination
Unclear layered structure	Tissue structures appear blurred or cloudy in appearance with low contrast and often do not show clearly defined tissue layers
Discontinuous fascia/fat pad	An interruption to the clear fibrillar hyperechoic lines associated with deep or superficial fascia tissue/fat pad
Hypoechoic area(s)	Hypoechoic area(s) with a clear shape and clearly defined margins
Heterogeneous hypoechoic area(s)	Hypoechoic area(s) with unclear boarders that disrupt the deep tissue layered structure
Unclassified	Unable to classify due to poor imaging quality despite the same technique applied over the anatomical area

2.9. Data collection

Baseline characteristics including age, sex, surgery type, anaesthetic type, surgery duration, intraoperative positioning and intraoperative protection were collected from each participant. ⁵ , ³⁷ , ³⁸ All participants had daily VSA, SEM measurements, epidermal hydration, temperature, ultrasound imaging and pain assessments at the sacrum, both heels and a control site before surgery (day 0) and after surgery for 3 days (day 1–3). The skin was free from cleansing or barrier cream for ≥2 h prior to conducting all skin assessments. The control site was the anterior aspect of the head of the humorous, which was not exposed to pressure/shear.

2.9.1. Positioning

For the data collection, participants were positioned onto their side and supported into a comfortable lateral 90° position. A small pillow was placed between the knees and lower limbs to enhance visibility of the heels. An alternative position was offered for assessing the heels, whereby the heels were offloaded by extending the end of the bed, elevating the foot of the bed and placing a towel or pillow under the lower limbs to elevate the heels. The control site was assessed when the participant was positioned on their back with the arms lying to the side in a relaxed manner. These collective assessments took approximately 15 minutes to complete.

2.10. Data analysis

Categorical variables are presented as percentages (%) and numbers (n=), while continuous data are presented as mean ± standard deviation (SD) or median (Interquartile range [IQR]), depending on the distribution of the analysed variables. Differences between categorical variables were assessed using the chi‐squared test. The kappa statistic was used to test inter‐rater reliability between the reviewers performing validation. As SEM and ultrasound represent continuous and ordinal data scales, respectively, a Spearman's rank (r _s ) test explored the correlation between SEM and ultrasound. This was supplemented with a Pearson correlation to explore ultrasound as a dichotomous variable. A spearman's correlation assessed the correlation between SEM and VSA. A correlation result is not presented when the analysis was unable to be performed such as when all VSA or ultrasound outcomes were categorised as normal. A correlation between 0.00 to 0.30/−0.30 was defined as a negligible correlation; a correlation >0.30/−0.30 to 0.50/−0.50 was defined as a low positive or low negative correlation, respectively; a moderate positive or moderate negative correlation was defined as a correlation >0.50/−0.50 to 0.70/−0.70, respectively; and a correlation >0.9/−0.9 to 1.00/−1.00 was defined as a very high positive or very high negative correlation respectively. ³⁹ Correlations (±)0.00–0.01 were defined as no evidence of a correlation between the variables. Statistical significance was set at p < 0.05. Data analysis was performed using the statistical software SPSS version 9.0 (SPSS, Inc., Chicago IL).

3. RESULTS

3.1. Demographics

A total of 76 patients were assessed for eligibility and 60 participants were consented into the study between February and November 2021. Of these, 33.3% (n = 20) were equally recruited to a cardiothoracic, orthopaedic or general surgery cohort. The mean age of participants was 58 years (SD: 13.46), and 50% (n = 30) were male. The average surgery duration was 220.5 min (SD: 106.9) (Table 2). Following study commencement, one participant was lost to follow‐up due to surgery cancellation, and one participant was unable to complete day 3 of follow‐up due to clinical deterioration. Therefore, a total of 58 participants completed full study follow‐up (Figure 1).

TABLE 2.

Baseline characteristics.

Demographic variable	N = 60	Demographic variable	N = 60
Age, years		Surgery type, % (n)
Mean (SD)	58 (13.46)	Orthopaedic	33.3 (20)
Median (IQR)	57 (48, 69)	Cardiothoracic	33.3 (20)
Gender, male, % (n)	50 (30)	General surgery	33.3 (20)
Anaesthetic type, % (n)		Surgery duration, minutes
GA	96.67 (58)	Mean (SD)	220.52 (106.59)
GA and spinal	3.33 (2)	Median (IQR)	195 (130, 305)
Intraoperative protection, % (n)		Intraoperative positioning, % (n)
All bony prominences	96.67 (58)	Supine	68.33 (41)
Heels	3.33 (2)	Prone	26.67 (16)
		Lithotomy	5 (3)

Abbreviations: IQR, Interquartile range; SD, standard deviation.

FIGURE 1.

Participant recruitment flowchart. This figure outlines the number (n=) of patients assessed for eligibility and the flow of consented participants throughout the study follow‐up period.

3.2. SEM delta values

The mean SEM delta values over the weight‐bearing areas were abnormal across all anatomical sites and all days of follow‐up. Conversely, all SEM delta values at the control site were normal. To explore further trends in the data, abnormal SEM delta values were divided into two categories (borderline and high). A borderline SEM delta value was defined as a delta ranging from 0.6 to 0.8, and a high SEM delta value was defined as a delta ≥0.9. ²⁰

When assessing the sacrum, day 0 had the lowest mean delta value (0.68 ± 0.35) and percentage of abnormal SEM delta values across all days (63%, n = 38). The percentage of abnormal SEM delta values increased across days 1 and 2 (64%, n = 30, and 70%, n = 32, respectively). On day 3, the percentage of abnormal SEM delta values had reduced when compared to day 2 (64%, n = 35) (Figure 2).

FIGURE 2.

Sub‐epidermal moisture (SEM) (delta) trends assessed on the sacrum from baseline (day 0) to day 3. This figure shows the SEM delta values recorded throughout the study follow‐up period at the sacrum. Normal SEM delta values (<0.6) are presented in green. Abnormal SEM delta values (≥0.6) were placed into two categories (borderline and high) to explore further trends in the data. Borderline values are presented in yellow, indicating a SEM delta value ranging between 0.6 and 0.8. High values are presented in red, indicating a SEM delta value ≥0.9. IQR, Interquartile range; SD, standard deviation.

For the left heel, the mean SEM delta value was lowest on day 0 (0.57 ± 0.39), representing the lowest percentage of abnormalities across all days (42%, n = 25). A higher percentage of abnormalities were observed on day 2 (58%, n = 34) and day 3 (53%, n = 31) (Figure 3).

FIGURE 3.

Sub‐epidermal moisture (SEM) (delta) trends assessed on the left heel from baseline (day 0) to day 3. This figure shows the SEM delta values recorded throughout the study follow‐up period at the left heel. Normal SEM delta values (<0.6) are presented in green. Abnormal SEM delta values (≥0.6) were placed into two categories (borderline and high) to explore further trends in the data. Borderline values are presented in yellow, indicating a SEM delta value ranging between 0.6 and 0.8. High values are presented in red, indicating a SEM delta value ≥0.9. IQR, Interquartile range; SD, standard deviation.

When assessing the right heel, day 3 had the lowest mean delta value (0.58 ± 0.25) whereas, day 0 had the lowest percentage of abnormal SEM delta values across all days (47%, n = 28). Day 1 had the highest percentage of abnormal SEM delta values (60%, n = 34), followed by day 2 (53%, n = 31) (Figure 4).

FIGURE 4.

Sub‐epidermal moisture (SEM) (delta) trends assessed on the right heel from baseline (day 0) to day 3. This figure shows the SEM delta values recorded throughout the study follow‐up period at the right heel. Normal SEM delta values (<0.6) are presented in green. Abnormal SEM delta values (≥0.6) were placed into two categories (borderline and high) to explore further trends in the data. Borderline values are presented in yellow, indicating a SEM delta value ranging between 0.6 and 0.8. High values are presented in red, indicating a SEM delta value ≥0.9. IQR, Interquartile range; SD, standard deviation.

3.3. Ultrasound

A lower percentage of ultrasound abnormalities were detected at baseline across all anatomical sites, with all assessments considered normal at the control site. As there was 100% agreement between the reviewers performing validation, the kappa was 1. The percentage of ultrasound abnormalities increased across each day of follow‐up, with the highest number of abnormalities observed on day 3 across the sacrum, left and right heels (33% [n = 13]; 32% [n = 16]; 27% [n = 13], respectively) (Figure 5).

FIGURE 5.

Ultrasound results across all sites throughout the follow‐up period. This figure shows the ultrasound results throughout the study follow‐up period at the sacrum, left heel and right heel. Numbers (n=) represent the number of ultrasound assessments that had a normal or abnormal result across each given day. Pie charts present the percentage of normal (green) and abnormal (red) assessments across each day. The severity of abnormal ultrasound assessments are then presented in a bar chart to the right of the pie charts, which represent each of the four abnormal ultrasound categories.

The Figures 6, 7, 8, 9 provide a visual overview for each of the ultrasound abnormalities, which developed throughout the study period: (1) unclear layered structure (Figure 6), (2) discontinuous fat pad/fascia (Figure 7), (3) hypoechoic area(s) (Figure 8) and (4) heterogeneous hypoechoic area(s) (Figure 9). A high percentage of unclassified images were observed across days 0 to 3 at the sacrum (30% [n = 18]; 33% [n = 15]; 37% [n = 17]; 27% [n = 15]) left heel (13% [n = 8]; 14% [n = 8]; 12% [n = 7]; 14% [n = 8]) and right heel (11% [n = 6]; 9% [n = 5]; 16% [n = 9]; 17% [n = 10], respectively). Unclassified images were excluded from all data analysis.

FIGURE 6.

Ultrasound example of an unclear layered structure. This figure provides a visual overview of an unclear layered structure identified with ultrasound, which developed across the study follow‐up period. A transverse (top left) and longitudinal (bottom left) ultrasound image show normal images of the sacral tissue on day 0. The transverse (top right) and longitudinal (bottom right) ultrasound images show tissue structures (indicated with red arrows) with a blurred or cloudy appearance with low contrast, which do not show clearly defined tissue structures on day 3.

FIGURE 7.

Ultrasound example of discontinuous fat pad/fascia. This figure provides a visual overview of a discontinuous fat pad identified with ultrasound, which developed across the study follow‐up period. A transverse (top left) and longitudinal (bottom left) ultrasound image show normal images of the heel tissue on day 0. The transverse (top right) and longitudinal (bottom right) ultrasound images show interruption to the clear fibrillar hyperechoic lines associated with the heel fat pad structure (indicated with red arrows) on day 1.

FIGURE 8.

Ultrasound example of a hypoechoic area. Legend: This figure provides a visual overview of a hypoechoic area identified with ultrasound, which developed across the study follow‐up period. A transverse (top left) and longitudinal (bottom left) ultrasound image show normal images of the heel tissue on day 0. The transverse (top right) and longitudinal (bottom right) ultrasound images show a hypoechoic area with a clear shape and clearly defined margins (indicated with red arrows) on day 3.

FIGURE 9.

Ultrasound example of a heterogeneous hypoechoic area. Legend: This figure provides a visual overview of a heterogeneous hypoechoic area identified with ultrasound, which developed across the study follow‐up period. A transverse (top left) and longitudinal (bottom left) ultrasound image show normal images of the heel tissue on day 0. The transverse (top right) and longitudinal (bottom right) ultrasound images show a hypoechoic area with unclear boarders that disrupt the deep tissue layered structure (indicated with red arrows) on day 3.

3.4. Correlation analysis

The correlation between SEM and ultrasound revealed a statistically significant low to moderately positive correlation across all anatomical sites (r _s range = 0.39–0.54, p < 0.05; r range = 0.41–0.56, p < 0.05). The only exception was a correlation between SEM and ultrasound on day 0 at the right heel (r _s  = 0.23, p = 0.09; r = 0.26, p = 0.06) (Table 3). Scatterplots in Figures 10, 11, 12 provide a visual overview of the correlation between SEM and ultrasound. Cross‐tabulations outline the relationship between SEM delta values and the corresponding ultrasound categories in Tables 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15. It is important to highlight that there is one outlier in the data on day 3 at the left heel. The ultrasound has detected damage at 6.11 mm; however, the SEM delta value is normal at 0.1 (Table 11, Figure 11). The SEM delta value had been abnormal (0.7) on the previous day (day 2), and on the same day (day 2), ultrasound showed advanced injury (a hypoechoic area).

TABLE 3.

The correlation between sub‐epidermal moisture (SEM) and ultrasound across all sites and days of follow‐up.

SEM and ultrasound	Day 0	Day 1	Day 2	Day 3
Sacrum
	N = 42	N = 30	N = 29	N = 40
Spearman's correlation
r s	0.41 a	0.54 a	0.38 a	0.50 a
p value	0.01 a	<0.001 a	0.04 a	<0.001 a
Pearson correlation
r	0.46 a	0.41 a	0.44 a	0.44 a
p value	<0.001 a	0.02 a	0.02 a	0.01 a
Left heel
	N = 52	N = 49	N = 52	N = 50
Spearman's correlation
r s	0.39 a	0.40 a	0.41 a	0.48 a
p value	<0.001 a	<0.001 a	<0.001 a	<0.001 a
Pearson correlation
r	0.56 a	0.51 a	0.44 a	0.41 a
p value	<0.001 a	<0.001 a	<0.001 a	<0.001 a
Right heel
	N = 54	N = 52	N = 49	N = 48
Spearman's correlation
r s	0.23	0.45 a	0.47 a	0.52 a
p value	0.09	<0.001 a	<0.001 a	<0.001 a
Pearson correlation
r	0.26	0.50 a	0.52 a	0.47 a
p value	0.06	<0.001 a	<0.001 a	<0.001 a

^a Indicate statistical significance.

FIGURE 10.

Scatterplots exploring the relationship between sub‐epidermal moisture (SEM) and ultrasound at the sacrum on day 0 (A), day 1 (B), day 2 (C) and day 3 (D), respectively.

FIGURE 11.

Scatterplots exploring the relationship between sub‐epidermal moisture (SEM) and ultrasound at the left heel on day 0 (A), day 1 (B), day 2 (C) and day 3 (D), respectively.

FIGURE 12.

Scatterplots exploring the relationship between sub‐epidermal moisture (SEM) and ultrasound at the right heel on day 0 (A), day 1 (B), day 2 (C) and day 3 (D), respectively.

TABLE 4.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the sacrum on day 0.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	17	22	39
Unclear layered structure	0	1	1
Hypoechoic area(s)	0	1	1
Heterogeneous hypoechoic area(s)	0	1	1
Total	17	25	42
Chi‐squared	p = 0.53

TABLE 5.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the sacrum on day 1.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	14	10	24
Unclear layered structure	0	4	4
Discontinuous fascia	0	1	1
Heterogeneous hypoechoic area(s)	0	1	1
Total	14	16	30
Chi‐squared	p = 0.09

TABLE 6.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the sacrum on day 2.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	10	13	23
Unclear layered structure	0	6	6
Total	10	19	29
Chi‐squared	p = 0.045*

Note: The asterisks (*) indicate a statitsically significant result (p < 0.05).

TABLE 7.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the sacrum on day 3.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	12	15	27
Unclear layered structure	0	7	7
Discontinuous fascia	0	1	1
Hypoechoic area(s)	0	3	3
Heterogeneous hypoechoic area(s)	0	2	2
Total	12	28	40
Chi‐squared	p = 0.09

TABLE 8.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the left heel on day 0.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	28	19	47
Discontinuous fascia	0	4	4
Heterogeneous hypoechoic area(s)	0	1	1
	28	24	52
Chi‐squared	p = 0.04*

Note: The asterisks (*) indicate a statitsically significant result (p < 0.05).

TABLE 9.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the left heel on day 1.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	25	19	44
Unclear layered structure	0	3	3
Discontinuous fascia	0	1	1
Hypoechoic area(s)	0	1	1
Total	25	24	49
Chi‐squared	p = 0.12

TABLE 10.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the left heel on day 2.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	20	20	40
Unclear layered structure	1	6	7
Discontinuous fascia	0	1	1
Hypoechoic area(s)	0	4	4
Total	21	31	52
Chi‐squared	p = 0.08

TABLE 11.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the left heel on day 3.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	23	11	34
Unclear layered structure	0	4	4
Discontinuous fascia	1	4	5
Hypoechoic area(s)	0	5	5
Heterogeneous hypoechoic area(s)	0	2	2
Total	24	26	50
Chi‐squared	p = 0.002*

Note: The asterisks (*) indicate a statitsically significant result (p < 0.05).

TABLE 12.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the right heel on day 0.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	26	23	49
Unclear layered structure	0	1	1
Hypoechoic area(s)	1	2	3
Heterogeneous hypoechoic area(s)	0	1	1
Total	27	27	54
Chi‐squared	p = 0.47

TABLE 13.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the right heel on day 1.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	19	23	42
Unclear layered structure	0	4	4
Discontinuous fascia	0	2	2
Hypoechoic area(s)	0	2	2
Heterogeneous hypoechoic area(s)	0	2	2
Total	19	33	52
Chi‐squared	p = 0.13

TABLE 14.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the right heel on day 2.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	22	15	37
Unclear layered structure	0	4	4
Discontinuous fascia	0	2	2
Hypoechoic area(s)	1	2	3
Heterogeneous hypoechoic area(s)	0	3	3
Total	23	26	49
Chi‐squared	p = 0.03*

Note: The asterisks (*) indicate a statitsically significant result (p<〈 0.05).

TABLE 15.

Cross‐tabulation results for sub‐epidermal moisture (SEM) and ultrasound at the right heel on day 3.

Ultrasound categories	Normal SEM (<0.6)	Abnormal SEM (≥0.6)	Total
Normal	20	15	35
Unclear layered structure	0	6	6
Discontinuous fascia	0	4	4
Hypoechoic area(s)	0	1	1
Heterogeneous hypoechoic area(s)	0	2	2
Total	20	28	48
Chi‐squared	p = 0.01*

Note: The asterisks (*) indicate a statitsically significant result (p < 0.05).

The results from the correlation analysis revealed a negligible correlation between SEM and VSA at the sacrum across days 1–3; however, only the result on day 1 was statistically significant (r _s  = 0.29, p = 0.049). A negligible correlation was observed between SEM and VSA across both the left and right heels, with the result on day 1 at the right heel showing no evidence of a correlation (r _s  = 0.00, p = 0.98). These results were not statistically significant (Table 16).

TABLE 16.

The correlation between sub‐epidermal moisture (SEM) and visual skin assessment (VSA) across all sites and days of follow‐up.

SEM and VSA	Day 0	Day 1	Day 2	Day 3
Sacrum
	N = 60	N = 47	N = 46	N = 55
Spearman's correlation	/	0.29 a	0.10	0.15
p value	/	0.049 a	0.52	0.28
Left heel
	N = 60	N = 57	N = 59	N = 58
Spearman's correlation	0.21	0.08	−0.12	0.12
p value	0.12	0.54	0.36	0.36
Right heel
	N = 60	N = 57	N = 58	N = 58
Spearman's correlation	0.13	0.00	0.11	0.07
p value	0.32	0.98	0.42	0.63

^a Indicate statistical significance.

3.5. Pressure ulcer development

There was a 3.3% (n = 2) incidence of visual PU development, and all were assessed as grade 1 (non‐blanching erythema). All visual PUs developed at the sacrum, one of which was identified 2 h after surgery and the second was identified on day 3 of follow‐up.

For the first participant that developed a PU, on day 0, the SEM delta value was 1.2 and the ultrasound showed heterogeneous hypoechoic areas. On day 1, the SEM delta value was 1.3 and ultrasound showed heterogeneous hypoechoic areas. For the second participant developing a PU, on day 0, the SEM delta value was 0.6 and the ultrasound was categorised as normal. On day 1 and day 2, a postoperative spinal dressing was in place which extended over the sacrum and prevented data collection for SEM and ultrasound. On day 3, the SEM delta value was 0.7 and the ultrasound showed an unclear layered structure.

A total of 5% (n = 3) of participants developed sustained abnormal SEM delta values across all anatomical sites from baseline. Similarly, sustained abnormal SEM delta values were observed from baseline at the sacrum (36.7%, n = 22), left heel (23%, n = 14) and right heel (31.7%, n = 19). These sustained abnormal SEM delta values (≥2 days) had occurred predominately already at baseline which meant that this data was excluded from the abnormal SEM delta threshold incidence calculation, as these participants did not develop this abnormal SEM delta during hospital admission. Thus, the abnormal SEM delta threshold incidence was 50% (n = 28), of which 39.2% (n = 11) of these participants developed ≥1 abnormal SEM delta threshold. Of these, 48.3% (n = 14) developed on the sacrum, 35.5% (n = 16) on the left heel and 30% (n = 12) on the right heel of participants (Table 17).

TABLE 17.

Abnormal sub‐epidermal moisture (SEM) delta threshold incidence across the study follow‐up period.

Abnormal SEM delta threshold incidence (n = 56)
Overall abnormal SEM delta threshold incidence, % (n)	50 (28)
>1 abnormal SEM delta threshold, % (n)	39.2 (11)
Excluded (reasons)
Sustained abnormal SEM delta value from baseline	5 (3)
Missing data	1.6 (1)
Sacrum (n = 29)
Abnormal SEM delta threshold (sacrum), % (n)	48.3 (14)
Excluded (reasons), % (n)
Sustained abnormal SEM delta value from baseline	36.7 (22)
Missing data	15 (9)
Left heel (n = 45)
Abnormal SEM delta threshold (left heel), % (n)	35.5 (16)
Excluded (reasons), % (n)
Sustained abnormal SEM delta value from baseline	23 (14)
Missing data	1.6 (1)
Right heel (n = 40)
Abnormal SEM delta threshold (right heel), % (n)	30 (12)
Excluded (reasons), % (n)
Sustained abnormal SEM delta value from baseline	31.7 (19)
Missing data	1.6 (1)

4. DISCUSSION

The study findings have revealed a low to moderately positive correlation between SEM and ultrasound across all anatomical sites, with results showing ultrasound abnormalities increased across day 0 to day 3 of follow‐up. Negligible correlations were observed between SEM and VSA.

A statistically significant low to moderately positive correlation existed between SEM and ultrasound (US) across all anatomical sites. The only exception was a correlation between SEM and US on day 0 at the right heel (r _s  = 0.23, p = 0.09). The consistent correlation pattern indicates that SEM and US both agreed in the presence of injury; however, the localised oedema measured with SEM, preceded and predicted the later stage tissue damage detected with US. These results compare with an observational pilot study comparing SEM and US for the early detection of PUs in 15 elderly participants, which showed a consistent agreement between SEM and US when a suspected DTI existed. ²⁴ SEM was able to detect an abnormality 3 days prior to US imaging when a DTI developed at the heel in one participant. The present study builds on this initial pilot study to further corroborate whether SEM and US agree in terms of the assessment of the presence of early PU development. It is worthy of note, however, that Gefen and Gershon ²⁴ included participants with existing grade 1 PUs and suspected DTI at baseline. The present study brings a new perspective to this as participants had no evidence of an existing PU at baseline, and thus, if a PU or DTI developed, there was a greater possibility to capture this abnormality at the earliest possible stage of injury. The present study is the first of its kind to explore SEM and US in a cohort of surgical participants with no evidence of PU at baseline.

It is not an unexpected finding that SEM and US correlated but did not have a perfect correlation. US and SEM assess PU development beneath the skin surface; however, they represent two differing methods of PU identification. For example, SEM carries the ability to detect inflammation at the inception of PU injury. ⁴⁰ This is because SEM measures biocapacitance, the electrical property of the tissue, which is known to increase as the extracellular/interstitial water content increases. ¹² This process of increasing extracellular water represents the presence of microscopic, localised oedema that occurs during the early inflammatory damage cascade. ¹² On the contrary, US transmits sound waves into the tissue and depending on the tissue consistency, a proportion of this energy is reflected through echoes. ⁴¹ In turn, these echoes dictate the variation in tissue density that appears on screen in bright to dark patterns. ⁴¹ , ⁴² These brighter and darker patterns can correlate with the presence and severity of injury associated with PU development. ²⁴ , ³⁵ , ³⁶ Indeed, US carries the potential to detect changes in the deep tissue such as oedema; however, US is unable to detect this level of injury until it has progressed to a macroscopic scale beneath the skin tissue. ²⁴ Therefore, an understanding of how these two technologies differ in their assessment of PU development provides rationale for the correlation between SEM and US in the present study. Considering both SEM and US can detect and agree in the presence of injury beneath the skin surface, these correlation findings show further promise in SEM's ability to detect early PU development.

A possible explanation for the findings of a normal SEM delta value, but an abnormal ultrasound image is that although damage appears to have progressed to the underlying tissue structures, the SEM delta value has improved when compared to the previous day. Indeed, the patient's skin was intact and there were no visible or tactile signs of damage that had manifested. This indicates that the interventions provided to the participant have been effective, as evidenced by the reducing SEM delta value. This is consistent with other research evidence, wherein the odds of localised oedema returning to normal were six times higher (p = 0.0001) among participants receiving enhanced targeted PU interventions based on SEM delta values, versus participants receiving usual care. ²²

There was a high incidence of participants exceeding the abnormal SEM delta threshold associated with surgery (50%, n = 28) combined with a low incidence of visual PU development across the study period (3.3%, n = 2). The incidence of PUs within this study, combined with the negligible correlation analysis between VSA and SEM, provides further evidence to suggest that PUs are more likely to develop in the deeper tissues. These results compare with Martins de Oliveira et al., ⁵ who found that regardless of a low incidence of PUs identified using VSA following surgery (3%, n = 7), there was a high incidence of persistent abnormal SEM delta values (51%, n = 116). Surgical populations are exposed to prolonged periods of external pressure during the surgical operation. ⁴ Sustained external pressure is known to increase pressure and damage at the deeper tissue near bony prominences before damaging the superficial skin tissue. ¹⁰ , ¹¹ , ¹² Therefore, abnormal SEM delta values are indicating that this damage has started to occur within the deeper tissues. ¹² As SEM detects early signs of tissue damage before reaching the superficial skin surface, one would expect SEM to have detected a greater number of abnormalities when compared to VSA. This is consistent with the existing research evidence investigating SEM and its ability to detect PU prior to VSA. ⁵ , ¹⁹ , ²⁰ , ²⁴ , ³³ , ⁴³

Considering the difference between the abnormal SEM delta threshold incidence and visual PU development, it does not come as a surprise that negligible correlations were observed between these variables. O'Brien et al. ⁴³ reported a stronger correlation between SEM and VSA at the sacrum (r = 0.65), left heel (r = 0.23) and right heel (r = 0.43) among 47 medical inpatients that developed 21 grade 1 PUs (40%, n = 19) over a 4‐week follow‐up period. Irrespective of this, SEM had identified the development of PUs an average of 4 days earlier than VSA. ⁴³ Thus, this study provides further evidence to suggest that when a longer follow‐up is implemented, SEM and VSA remain dissociated in terms of their ability to correlate in the presence of tissue damage. ⁴³

It is also worth pointing out that not all subdermal tissue abnormalities, such as those detected with SEM, correspond to a visual PU. Previous studies investigating SEM for a longer follow‐up have concluded that not all abnormal SEM delta values correspond with an eventual PU. ¹⁹ , ²⁰ , ²¹ , ³³ , ⁴³ The SEM delta value is indicative of the early inflammatory process associated with PU development, and thus, if interventions were put in place, there would likely be a corresponding decrease in the SEM delta value. Pressure redistributing interventions were carried out for all participants within this study, as standard of care PU prevention protocols were ongoing at the hospital site. As a result of these pressure redistributing interventions, tissue damage associated with early PU development may have been reversed and a visual PU subsequently prevented. Alternatively, if a participant has undergone surgery and was immobile for the first few days, one would likely anticipate the SEM delta value decreasing as the level of mobility increases. ⁴⁴ , ⁴⁵

When implementing an assessment into clinical practice, it is important to consider the practicality of performing these assessments. A higher number of assessments on the sacrum were reduced due to pain from the surgical operation. A comprehensive ultrasound assessment required participants to be positioned on their side for 1–2 minutes, which was challenging to achieve in an ICU setting. SEM offers a valuable approach whereby a real time value is generated within seconds, which guides clinical decision making at the bedside. This quick and simple approach benefits both the healthcare team and the patient, as preventative measures can be initiated simultaneously with the result. This contrasts to ultrasound, whereby for a more precise assessment to be made, a trained researcher must analyse the findings in further detail once leaving the bedside. In clinical practice, this could cause a delay to an applied intervention, in which SEM offers a superior approach. SEM has demonstrated superiority in terms of its feasibility when compared to other assessment measures in previous studies. ²⁰ Therefore, irrespective of ultrasound showing promise from a research perspective to visually confirm the presence of injury, the translation of this evidence into practice would be challenging.

5. LIMITATIONS

A limitation arises from these findings due to a 3‐day follow‐up after the surgical procedure. Considering most participants were discharged on day 3, community follow‐up would be required if a longer observational period was implemented. Nonetheless, evidence points to a PU that has developed within 72 hours following surgery as linking back to the surgical operation. ⁶ , ⁷ , ⁴⁶ Gefen et al. ⁴ argues that these later diagnosed PUs are likely representing PUs that have started as DTIs under intact skin. Considering this evidence, a three‐day follow‐up with SEM for the early detection of PUs could be considered a sufficient period to detect damage occurring within the subdermal layers that links back to the surgical operation.

Ultrasound is a highly subjective assessment technique, and this may exist as a limitation. The researcher applied consistent assessment techniques, followed a validation protocol, and undertook extensive training; however, the percentage of abnormal ultrasound images may have been underestimated. Ultrasound imaging on the sacrum was challenging to conduct and resulted in a lower number of assessments when compared to other anatomical sites assessed. In the orthopaedic group, several participants had a sacral dressing postoperatively, which prevented a comprehensive assessment from taking place on each day of follow‐up. A higher number of assessments on the sacral site were also reduced due to pain from the surgical operation, limiting compliance with repositioning or due to clinical instability following major surgery. A large majority of participants were assessed in the ICU, particularly on day 1, which represented the lowest number of US assessments conducted. Previous authors examining ultrasound for the detection of PU development have considered ultrasound a challenging assessment to conduct within clinical practice. ²⁷ , ³⁶ , ⁴⁷ , ⁴⁸ , ⁴⁹

The number of unclassified images also exists as a limitation within this study, which reduced the number of observations included in the correlation analysis. Similar ultrasound limitations have been reported across previous studies, whereby ‘uninterpretable’ categories were added to reflect poor penetration of ultrasound waves ⁴⁷ or attenuation due to increased adipose tissue volume. ⁵⁰ Poor image quality prevents accurate assessment and thus, ultrasound is not an inclusive assessment for the early detection of PUs in all patient populations.

6. CONCLUSION

The findings from this study provide evidence that a correlation exists between SEM and US. This correlation pattern indicates that SEM and US agreed in the presence of injury; however, SEM could detect abnormalities prior to US. Considering both measures assess the inflammatory process beneath the skin tissue, these results shed new light on the significance of abnormal SEM delta values in absence of a visible skin injury, confirming the ability of SEM to detect early PU development. In this surgical cohort of participants, there was a high incidence of participants exceeding the abnormal SEM delta threshold and a low incidence of visual PUs. Future research is warranted to validate or refute these findings. The next article of this series will focus on the correlation between SEM, temperature, epidermal hydration and pain.

FUNDING INFORMATION

This research was funded by the Royal College of Surgeons in Ireland, University of Medicine, and Health Sciences, School of Nursing and Midwifery.

CONFLICT OF INTEREST STATEMENT

The RCSI School of Nursing and Midwifery has a research collaboration with Bruin Biometrics. The SEM scanner was donated by Bruin Biometrics for the duration of the study.

Wilson HJE, Patton D, Budri AMV, et al. The correlation between sub‐epidermal moisture measurement and other early indicators of pressure ulcer development—A prospective cohort observational study. Part 1. The correlation between sub‐epidermal moisture measurement and ultrasound. Int Wound J. 2024;21(3):e14732. doi: 10.1111/iwj.14732

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.