Skin Failure Clinical Indicator Scale: Proposal of a Tool for Distinguishing Skin Failure From a Pressure Injury

Skin failure is a concept that includes multiple similar phenomena described in the literature as Kennedy Terminal Ulcers, Skin Changes at Life’s End (SCALE), and Trombley-Brennan Terminal Tissue Injuries (TB-TTI).¹ These phenomena result in ulcerations that, although somewhat visually similar in appearance to pressure injuries, have a fundamentally different etiology despite the ability to occur concomitantly with a pressure injury.² Current consensus holds that skin failure may occur during an acute illness as a result of chronic illness or as part of the dying process.¹ Skin failure manifests with objective, observable characteristics such as a rapid onset, an acute progression, full-thickness tissue involvement, and ill-defined (often purpuric) borders.¹

Promotion and maintenance of skin integrity is a fundamental part of nursing. Pressure injuries seem to dominate the discussion in the literature, but this comes as no surprise when it is noted that pressure injuries alone carry a substantial cost burden to facilities, an estimated $9.1 to $11.6 billion annually.³ The Centers for Medicare and Medicaid Services⁴ modified its Inpatient Prospective Payment System in 2008 to reduce hospital reimbursements for pressure injuries not present-on-admission and, in 2015, began to penalize hospital reimbursements of the lowest performing quartile of hospitals. As a result of these reimbursement changes, misdiagnosis of skin failure as a facility-acquired Stage 3 or Stage 4 pressure injury can have a substantial financial impact.

Skin Failure Clinical Indicator Scale proposal
Currently, discernment of wound etiology is heavily reliant on visual analysis and patient history. A survey of 100 certified wound care nurses showed that, if limited to purely visual analysis, the correct etiology was determined 17% more often than by chance alone.⁵ To further complicate the issue of etiology, no validated assessment tools or clinical indicators are available that can assist in determining etiology or providing a more definitive diagnosis of skin failure.⁶

The purpose of this study was to develop a novel, objective scoring system, the Skin Failure Clinical Indicator Scale (SFCIS); this scale, in conjunction with pertinent history and readily available lab work, could assist in distinguishing between plausible instances of skin failure versus pressure injury.

Patients
This case-control study consisted of convenience sampling via retrospective review of the electronic medical records (EMRs) of deceased acute care patients who suffered acute skin breakdown prior to death between January 1, 2017, and March 1, 2019, in 2 hospitals in the United States. Patients were identified via computer-generated searches for all patients diagnosed with acquired, rapid-onset breakdown through each facility’s respective EMR department. The diagnoses of focus were the International Classification of Diseases 10th Revision (2004) categories L89 (pressure ulcer, mostly regarding L89.9, pressure ulcer of unspecified site) and L98.0 (disorder of the skin and subcutaneous tissue, unspecified).⁷

The locations of breakdown were recorded, and patients were divided into 2 groups, depending on if the breakdown occurred at locations that were typical or atypical (nonpressure) for pressure injury formation. Typical pressure injury locations were defined as overlying a bony prominence or those in contact with medical devices or other objects.⁸Atypical locations were defined as areas not exposed to pressure and/or shearing forces. Other causative etiologies (such as solely pressure, traumatic, moisture, infection, etc) were ruled out through assessment and root cause analysis performed by a certified wound care clinician. For the purposes of this study, atypical location was used as the main descriptor for skin failure.

Variables
After careful review of predictive factors for skin failure was substantiated via a comprehensive literature review (performed by authors RH and AP) (Table 1A, Table 1B),^9-34 the researchers chose to monitor 10 variables, including age,^{9,12,19,21,26-31} functional capacity (Karnofsky score),^{9,12,21,23,27,34} Braden score,^{9,12,14,17,20,30,31,34} body mass index (BMI),^12,19,22,26 serum albumin level,^{17,19,20,21,26} primary diagnosis,^{16,18,27,24,25,28,30} presence of sepsis/multiple organ dysfunction syndrome (MODS),^16,29,24,25 vasopressor/inotrope use,^13,16,27,29 and mechanical ventilation.^17,25,28,33 In several studies,^16,27,29 patients with skin failure were noted to have multiple statistically significant factors, including organ failure, sepsis, mechanical ventilation, and use of vasopressor/inotropes. Laboratory values (C-reactive protein, albumin, prealbumin, and erythrocyte sedimentation rate), weight, and functional indicator variable cutoffs were included in this study based on clinical experience and review of previous data. The chosen predictive factors were entered as categorical variables. After data collection, the functional capacity variable was removed from analysis, because 100% of patients were noted to be dependent; thus, analysis could not be performed. This finding was not unexpected, because it mirrors a prospective cohort study⁹ in which a majority (95.8%) of the observed patients with terminal illness and skin changes had more than 50% loss in mobility. Data were collected into a spreadsheet and then entered into a statistical analysis software.

Statistical analysis
Univariate data analysis was performed using SAS software, version 9.4 (SAS Institute Inc) (Table 2). Internal validation of the data set was done using the bootstrap resampling technique. Multivariate analysis was performed via backward stepwise logistic regression using SPSS software (version 26.0, IBM Corp). In univariate analysis, statistical significance is conventionally defined as P < .05, but it was decided to utilize P < .1 as a cutoff value for variable selection for multivariate analysis in order to avoid missing potential predictors.³⁵ In addition, after univariate analysis, the variables regarding diagnosis and mechanical ventilation were further modified to yes/no instances. The diagnosis variable was modified to include any primary diagnosis related to alteration of blood flow, and the ventilation variable was modified to include any length of time on mechanical ventilation. Subsequently, Pearson’s chi-squared testing was performed on all variables. After eliminating factors that were found not to be significant, analysis of variance (ANOVA) was performed on the remaining variables. Multivariate analysis included the Hosmer-Lemeshow Goodness of Fit Test³⁶ and the Omnibus Tests of Model Coefficients.³⁷

Of the 52 patients included in the present study, 16 had presentations in areas typical of pressure injury and 36 had atypical ulcer locations. 

Univariate analysis
Patient age, Braden score, and BMI were not significant predictors of skin failure. Analysis of variance was performed on the remaining variables and was noted to be statistically significant (P = .0175)(Table 3).³⁵ After univariate analysis, the variables regarding diagnosis and mechanical ventilation were further modified to yes/no instances. The diagnosis variable was modified to include any primary diagnosis related to alteration of blood flow. The ventilation variable was modified to include any length of time on mechanical ventilation. Subsequently, Pearson’s chi-squared testing was performed on all variables and was found to be significant in all cases except for albumin (P = .08); this is thought to be in part due to the established cutoff value of 3.5 g/dL in which 82.69% (n = 42) of patients fit.

Multivariate analysis
The full model containing the determined predictors was statistically significant (X² [5, N = 52] = 19.562; P = .002), indicating the model was able to distinguish between patients with and without atypical breakdown. The Hosmer-Lemeshow Goodness of Fit Test showed a value of .425, which indicated a good fit (if > .05).³⁴ The Omnibus Tests of Model Coefficients results were also significant at .001.³³ 

The model as a whole explained between 35.3% (Cox and Snell R²) and 49.9% (Nagelkerke's R²) of the variance of atypical breakdown and correctly identified 83.7% of cases. However, all variables were not significant as predictors, as defined by P < .05 (Table 4). In large part, this may be due to the sample size, as evidenced by the alarmingly high odds ratio of the vasopressor/inotrope variable and the seeming contradiction in findings otherwise. 

Multivariate analysis results prompted analysis of each variable as a single independent variable predicting the dependent. The results were significant in all cases except albumin at .09 (Table 5). The regression coefficients of each single predictive variable were converted into integers by multiplying by 2.5 and rounding to the nearest whole number. These values were used as the basis for the proposed scale (Table 6). The cutoff values were determined based on clinical findings but, upon validation testing of the scale, may be subject to change.

The literature review defined serum albumin as predictive for skin failure if less than 3.5 g/dL. The current analysis of the findings suggested that this level may be an overprediction. Future analysis may be more insightful if the cutoff level is lowered. Additionally, if a larger sample size is collected, the age ranges and BMI variables might need to be revisited as potential significant predictors.

Although the study population was sufficiently sized to create the tool, the sample was insufficiently sized to validate the instrument. In the future, it would be advisable to perform a multifacility validation of the SFCIS by utilizing the scale and analyzing the data from the resultant scoring. Analysis of these data should be able to determine the specificity and selectivity of the tool. 

Implications for practice
The implications of the development and success of the SFCIS are vast and encompass all aspects of the medical field. Although the phenomenon associated with the failure of skin and the associated signs are well known, a universal definition with definitive criteria has yet to be reached.¹ The creation of indicative diagnostic criteria will assist in the development of a uniform diagnostic tool. With a set definition of skin failure, more providers may acknowledge the validity of the diagnosis, which could ensure more accurate coding.³⁸ Diagnosis of skin failure through the use of the SFCIS could offer more accurate data concerning prevalence and incidence, which could then be utilized for future research.⁵ As previously noted, facility costs involved in misidentification of skin failure as an acquired pressure injury might be ameliorated.⁹

A predictive tool for skin failure, such as the SFCIS, could benefit the patient and the family as well. Education concerning the differences between skin failure and pressure injury could alleviate caregiver strain related to caring for a patient who develops skin breakdown despite appropriate interventions,²⁷ potentially perpetuating the understanding that not all factors that are associated with skin failure can be controlled or prevented.^6,9 The insight offered by identification of skin failure concerning the almost certain imminent mortality of the patient could open the door for the discussion of end-of-life care. Setting realistic goals and appropriate, quality treatment would ensure the comfort of the patient, could result in a more positive experience for the patient and family, and potentially decrease health care costs.⁶

Limitations of this study include an expectedly small sample size. Skin failure is an uncommon and also relatively underreported phenomenon, and this translates into a small patient pool. This reality is further complicated by the lack of an adequate, universal definition and diagnosis of skin failure. Patient recruitment in wound care research is notoriously generally suboptimal, with no recruitment enablers and a subsequent multitude of barriers.³⁹ Retrospectively, this study did not include race/ethnicity data, and this oversight may have inadvertently introduced a level of bias. Retrospective studies are prone to recall and misclassification bias and, although there were steps taken to mitigate bias (via root cause analysis and thorough review), there is no guarantee that this study was not victim to these oversights.

Another limitation was that the facilities tended to have insufficient and inadequate documentation in the desired population. These deficiencies were mostly related to missing laboratory values (especially albumin) and inadequate wound descriptions. These missing data resulted in the elimination of many potential participants. A large portion of skin and wound care is driven by nursing experiential learning and clinical ritualistic practice, and this can result in disparities in documentation.⁴⁰

Despite the study limitations, the analysis did prove fruitful enough to create a plausible scale that could be tested. The 83.7% prediction rate of the model shows promise that there may be a means of measuring the skin failure phenomenon. Although this tool requires further testing to be validated, it is a step in the direction of developing a much-needed predictive tool.

With the development of the SFCIS, identification of skin failure and a higher awareness could increase. This documentation could lead to inherent changes that include increased regulatory reimbursement, decreased pressure injury-related litigation, and decreased costs among facilities. 

Authors: Richard Hill, RN, CWCN, FACCWS¹; and Amy Petersen, RN, BSN²

Affiliations: ¹Natchitoches Regional Medical Center, Natchitoches, LA; and ²Cass County Health System, Atlantic, IA

Correspondence: Richard Hill, RN, CWCN, FACCWS, Wound Ostomy Nurse, Natchitoches Regional Medical Center, 501 Keyser Avenue, Natchitoches, LA 71457; hill.richardb@gmail.com

Disclosure: The authors disclose no financial or other conflicts of interest.