Assessing the Validity and Reliability of a New Pressure Ulcer Risk Assessment Scale for Patients in Intensive Care Units

The high incidence of pressure ulcers/injuries (PU/Is) among patients in intensive care units (ICUs) suggests a need for improved risk assessment. Purpose: The study aimed to develop and assess the validity and reliability of a new PU/I risk assessment scale. Methods: The authors developed the Efteli Günes (EFGU) Pressure Ulcer Risk Assessment Scale based on a conceptual framework of risk factors developed by Coleman et al. These factors comprised direct (immobility, skin/PU status, poor perfusion) and indirect (poor sensory perception and response, diabetes, moisture, poor nutrition, low albumin) factors, as well as factors that could potentially influence risk (older age, medications, pitting edema, chronic wound infection, acute illness, increased body temperature. These factors were operationalized into 8 scale variables: skin status in areas exposed to pressure, discomfort and pain sensation in areas exposed to pressure, incontinence, diastolic blood pressure, age, diabetes, ability to make small position shifts in areas exposed to pressure, and skin tolerance test. The presence and/or extent of each factor was assigned a value; the total score ranged from 0 to 15, with higher values indicating increased risk. Intraclass correlation (ICC) was used to assess interrater agreement. To test the instrument’s validity and reliability, a prospective, methodological study was conducted from September 1, 2015 to November 1, 2016, in the Neurology, Internal Medicine, Neurosurgery, Orthopedics, and Traumatology ICUs of a university hospital in Turkey. Eligible participants had to be bedbound ICU patients at least 18 years old, without a PU/I on admission, not receiving inotropic and/or vasopressor medications, and with a minimum ICU stay of 6 days. Demographic and clinical data were collected upon admission and daily thereafter until ICU discharge (maximum stay 12 weeks) or death. Descriptive statistics and Student’s t and chi-squared tests were used to analyze the data. Reliability was determined using Cronbach’s alpha. The Kaiser-Meyer-Olkin coefficient was used to determine validity, and the diagnostic and Youden indices were used to establish the cutoff value for risk. Results: Of the 207 patients included in this study 117 [56.5%] were male, mean age was 60.85 ± 16.45 years, the majority of participants (88 [42.5%]) were in the Neurology ICU), and 56 (27.1%) developed a PU/I. The presence of diabetes was found not to be a risk factor (r = 0.18), but the inability to make small position shifts (r = 0.79) was found to be a significant risk factor. After removing the diabetes variable (maximum score 14), 97.1% of patients with a score of 6 or greater on the EFGU scale score developed a PU/I. The Cronbach alpha coefficient for reliability was 0.81, sensitivity of the scale was 0.97, specificity was 0.83, positive predictive value was 0.69, and negative predictive value was 0.99. The ICC coefficient was 0.99. Conclusions: The validity and reliability of the EFGU Scale seem to indicate a high predictive value for PU/I occurrence among ICU patients involved in the study. Multicenter studies involving larger samples of ICU patients are needed to validate the results.

Pressure ulcers/injuries (PU/Is) are an important and life-threatening problem in health care institutions all over the world. A case-control¹ and a prospective study² have shown PU/Is affect patient quality of life, increase the cost of health care, and prolong hospital stay. Uzun and Tan³ found an 11.6% PU/I prevalence among a group of 344 medical, surgical, and intensive care unit (ICU) patients in Turkey. Based on a prospective, analytic, descriptive study⁴ among 84 surgical patients in Turkey, PU incidence was 54.8%. In a cohort study by Ahtiala et al⁵ in Finland conducted among 1629 high-dependency care patients in a medical-surgical ICU, PU/I prevalence was 11.8%. According to a prevalence/incidence study by Davis et al⁶ in 2 long-term health care institutions in Canada, PU/I prevalence was 36.8% and 53.2%, respectively.

PU/Is are particularly concerning in the context of critical care, where their incidence is high (range 7%–38%].⁷ The high incidence of PU/Is among patients in ICUs indicates the need for improved risk assessment and the use of preventive measures.⁸ According to a descriptive study by Gunningberg et al,⁹ assessing and identifying PU/I risk on a regular basis and initiating preventive interventions as early as possible may help reduce the incidence of PU/Is.

Interventions used to address PU/I risk often are expensive, and because health care resources are limited, accurately detecting patients who need preventive actions is crucial. According to a systematic review by Moore and Cowman,¹⁰ risk assessment is the first step in planning PU/I preventive measures. Early identification of patients at risk may be possible only through the use of a risk assessment tool with high validity and reliability; many clinical guidelines support the use of these tools.^11,12 However, despite the fact that existing PU/I risk assessment tools have high reliability values, a systematic review¹³ has shown that results (especially regarding the predictive validity of these tools) vary.^14-19 The purpose of this study was to develop and evaluate the predictive validity and accuracy of a new PU/I risk assessment scale for use among patients admitted to ICUs. 

Cohort studies^16,20 have found patients admitted to ICUs are at a higher risk of developing PU/Is than patients admitted to general care; ICU stays represent the greatest chance of occurrence of adverse events due to the clinical instability of patients and the high number of interventions. PU/Is are common adverse events with high prevalence and incidence in the ICU²¹; while community prevalence rates vary from 3% to 22% in hospitalized patients, prevalence rates in the ICU vary from 13.6% to 82%. Incidence rates in hospitalized patients range from 1% to 11%²²; incidence rates in ICU vary from 7% to 38%.⁷ The risk factors important in PU/I development are becoming better understood owing to a remarkable increase in epidemiological studies in recent years. Evaluation of PU/I risk can be achieved only through a well-structured approach that combines comprehensive skin assessment that utilizes a risk assessment scale with clinical decision-making.^12,15,16,18

Several extrinsic and intrinsic risk factors are involved in assessing PU/I risk. According to the European Pressure Ulcer Advisory Panel (EPUAP) and the National Pressure Ulcer Advisory Panel (NPUAP),²³ risk scales should assess the most important and high-evidence risk factors to accurately determine a patient’s risk. Scale accuracy can be measured in terms of its sensitivity, specificity, and predictive values. Sensitivity is defined as the proportion of patients classified as at risk who have PU/I. Specificity is the proportion of persons classified as not being at risk that did not develop a PU/I.²³ Test sensitivity and specificity often are inversely related.²⁴ Tools with low sensitivity facilitate use of PU/I prevention interventions even if patients do not need them or nonuse among patients whose need for preventive interventions is ignored.¹⁵ Positive and negative predictive values of the scale also should be high; a positive predictive value indicates the possibility of PU/I development in patients in the high-risk group, and a negative predictive value indicates the likelihood of patients not in the risk group to develop a PU/I.²⁵ However, a scale may not always have both high sensitivity and high specificity. In that case, a value with the best balance between sensitivity and specificity is usually considered the cutoff point. The first criterion for the best cutoff point was acceptable values for sensitivity and specificity of at least 70%; the second was to choose the cutoff point with the highest sensitivity. If the first criterion could not be reached, the cutoff point for an acceptable sensitivity of at least 70% was chosen, and the corresponding specificity was determined. The same was performed separately for a specificity of at least 70%.^15,16 However, the results regarding the sensitivity and specificity of the risk assessment scales in the literature are inconsistent,^14,18 and not all assessment scales are fast and easy to administer.^11,15 

Among the Braden, Norton, Waterlow, Knoll, and Gosnell scales (the most commonly used risk assessment scales in the literature),¹⁸ the Braden scale is the 1 most widely used in Turkey and all over the world.²³ When analyzing studies related to the predictive validity of the scales,^20,26,27,28 no scale with a balanced sensitivity and specificity values was found. The evaluation of the results obtained from these studies showed the highest sensitivity was determined by the Waterlow Scale, the highest specificity by the Douglas Scale, the highest positive predictive value by the Fragmment Scale,¹³ and the highest negative predictive value by the Douglas scale.¹⁶ In their reliability study, Kottner and Dassen²⁸ did not recommend using the Braden and Waterlow Scales for the risk assessment in ICU units. The prospective cohort study by Schoonhoven et al²⁹ reported that some of the recommended risk assessment scales might be ineffective in identifying surgical, internal, neurological, or geriatric ward patients at risk and implementing preventive interventions. 

Development of the Efteli and Güneş (EFGU) Pressure Ulcer Risk Assessment Scale. A systematic review of risk factors for PU/I development identified that PU/I risk assessment scales should be developed on the basis of multivariable analyses to identify factors that are independently associated with a PU/I.³⁰ In addition, a conceptual framework to establish a theoretical structure is needed to determine factors that are critical to the development of PU/Is.³¹ Several conceptual frameworks have been proposed in this regard over the past 30 years.^23,32 The PU/I risk factor assessment tool developed by the current authors was based on the conceptual framework of Coleman et al,³¹ where characteristics are classified as key direct, key indirect, and other potential indirect causal risk factors³¹ (see Figure 1). The risk assessment scale was intended to comprise risk factors with high level of evidence not found in the current risk assessment scales and that could distinguish individuals according to their PU/I risk. A total of 10 draft scale items that have a high level of evidence was chosen from the conceptual framework of Coleman et al³¹ and evaluated by a panel of 10 specialists with professional expertise in PU/Is. The items were not divided into sections or weighted equally. Each item had a score range of 0 to 3 in accordance with the 3 response categories.

Direct risk factors (ie, immobility, skin/pressure ulcer status, and poor perfusion). Physical inactivity is a primary risk factor in PU/I development. Strong evidence³¹ indicates that inadequate perfusion and the skin’s status (ie, status factors that may indicate the skin is more vulnerable to PU/I development such as redness, blanching erythema, dryness) reduce patient tolerance of pressure and accelerate PU/I development. These critical factors were incorporated into the new risk assessment scale. PU/I prevention guidelines^23,33,34 encourage the individual to make small shifts in body position every hour to relieve the pressure on the bony prominences. Therefore, the current authors assessed the ability of the patient to shift positions using the following scale: 0 = being able to make repositioning shifts of at least 15˚ to 20˚ in approximately 60 seconds once an hour; 1 = making these shifts once in 2 hours; 2 = making these shifts once in 4 hours; and 3 = no ability to shift. Skin status in the areas exposed to pressure, especially on sacrum and great trochanter, were scored as 0 = healthy skin; 1 = thin-sensitive skin, edema, moist-cold skin, dry-cracked skin; and 2 = hyperemia. 

Diastolic blood pressure and a skin tolerance test were used to evaluate perfusion status. Low diastolic blood pressure reduces perfusion in peripheral tissues and has been shown in a cohort study^35,36 to increase PU risk. The average diastolic blood pressure per day was obtained and scored as 0 = 60 mm Hg or more, and 1 = less than 60 mm Hg. The skin tolerance test showed how much pressure an individual can tolerate without damage and whether the circulation in the region deteriorated.²³ In order to evaluate skin tolerance, a finger was pressed over the reddened area for 10 seconds. If the area blanched within 20 seconds of the finger being lifted, the score was 0 (normal response); if it took more than 20 seconds, the score was 1 (late response); if the area stayed red, the score was 2. 

Key indirect risk factors (poor sensory perception and response, diabetes, moisture, poor nutrition, and low albumin). These factors indirectly influence PU/I risk and include discomfort/pain sensation, diabetes, and incontinence. As shown in a systematic review,³³ the patient’s perception of pain and discomfort in the area under pressure after inadequate perfusion and the change in position in this direction are extremely important to what can be accomplished to prevent PU/Is³³; on the new scale, this item was scored as 0 = no discomfort on the sacrum or great trochanter (areas typically exposed to the greatest pressure), 1 = discomfort, and 2 = no sensory perception in the aformentioned areas. Diabetes is thought to be a contributing factor in PU/I development, despite the lack of strong evidence.³⁷ The presence of diabetes was scored as 0 = does not have diabetes and 1 = has diabetes. Moisture accelerates PU/I development by reducing epidermal resistance.²³ Extrinsic moisture from perspiration, urine, and feces can macerate the skin surface and was scored as 0 = no incontinence or the presence of a urinary catheter, 1 = urinary incontinence, 2 = fecal incontinence, and 3 = urinary and fecal incontinence. 

Other potential indirect risk factors (older age, medication, presence of pitting edema, chronic wound infection, acute illness, and/or raised body temperature). Although scientific evidence is weak or limited regarding these indirect factors, they are thought to influence key direct or indirect risk factors.³¹ In a systematic review³⁰ of PU/I risk factors, older age emerged as an important predictor of PU/I development. Therefore, age was considered a variable for the new scale and was assigned scores of 0 = 65 years of age or under and 1 = older than 65 years. 

The total score on the EFGU scale ranged from 0 to 15, with higher scores indicating a higher risk of PU/I development (see Table 1). The total score was obtained by calculating the highest scores for each scale item. 

Assessing the EFGU scale.

Design and setting. A prospective, methodological study was conducted between September 1, 2015, and November 1, 2016, in the of Neurology, Internal Medicine, Neurosurgery, Orthopedics, and Traumatology ICUs of Ege University Hospital, Izmir, Turkey.

Participants. The sample comprised patients newly hospitalized in the units where the study was conducted. The inclusion criteria stipulated participants must be patients older than 18 years, without PU/Is on admission, bedbound, who did not take inotropic and vasopressor medications (which affect peripheral vascular resistance), and were expected to stay in the hospital for a minimum of 6 days (see Figure 2). Previous studies on the subject have been used to determine the sample size.^1,38 The sample volume was determined at α = 0.05 and power 90% using the standardized effect sizes determined by Cohen.³⁹ The sample size should be at least 5 to 10 times greater than the number of items in the scale to achieve meaningful and reliable results in the scale development process.^40,41 The scale consisted of 8 items; 207 persons participated in this study. The sample size was 25 times greater than the number of items.

Instruments. Data were collected using a 3-part instrument that comprised 1) patient demographic data, 2) the EFGU Pressure Ulcer Risk Assessment Scale, and 3) ulcer staging information according to the NPUAP²³ Pressure Ulcer Classification System. The patient demographic data form included 7 questions regarding age, gender, hospital unit, diagnoses, follow-up time (in days), diabetes status, the state of consciousness as determined and recorded by the physician, and PU/I presence. Demographic data were collected within the first 24 hours of admission.

Data collection. Data were collected from medical records by a single investigator using a paper and pencil tool. Patients were evaluated during the first 24 hours of admission to the hospital and once a week after until a PU/I developed or the patient was discharged from the unit (maximum stay 12 weeks). All variables were assessed every day during hospital stay. Because patients’ diastolic blood pressure was measured manually 3 times a day, the average was taken and recorded (48; EAKA, Bad Tölz, Bavaria, Germany). Preventive measures for PU/I development, such as positioning, skin care, and nutrition, were provided to all patients. 

Data analysis. Data were analyzed by an expert in biostatistics using the Statistical Package for Social Science, version 16.0 (SPSS Inc, Chicago, IL). Means and percentages were calculated, and Student’s t test, chi-squared, and validity and reliability analyses were applied. 

Reliability was calculated using Cronbach’s alpha reliability coefficient (>0.80), inter-item correlation (>0.30 and <0.70), and corrected item-total correlation (>0.30) to evaluate the internal consistency of the EFGU Pressure Ulcer Risk Assessment Scale.^41,42 Intraclass correlation (ICC) was used to assess interrater agreement. The interrater agreement was examined by Cohen Kappa statistics analysis.⁴³ In this study, interrater reliability was assessed before data collection. Two (2) nurses in the ICU were trained and informed about the aim of the study and received instructions on how to use the EFGU Scale. After completing the training program, the nurses were asked to complete the EFGU Scale for patients newly admitted to the unit. The assessments were made at the same time and each patient was assessed independently by both nurses. The scale was applied to 30 patients to evaluate the interrater agreement of the EFGU Pressure Ulcer Risk Assessment Scale.

Construct validity was determined using explanatory factor analysis. The principal components analysis and varimax rotation method were used to examine the factor structure. An eigenvalue 1 or greater was used as a criterion in factor selection.^37,42 The sensitivity, specificity, and positive and negative predictive values were used to determine the predictive validity of the scale. The Kaiser-Meyer-Olkin coefficient and the Bartlett test were used to determine whether the scale is suitable for explanatory factor analysis. The diagnostic index (DI) and Youden index values were calculated at the end of the receiver operating characteristics (ROC) analysis for determining the cutoff value.^40,41The theoretical framework of the EFGU is presented in Figure 3.

Ethical consideration. Approval to perform the study was obtained from the ethics committee of Ege University Faculty of Nursing, and the Ege University Medical Faculty Hospital Neurology, Internal Medicine, Neurosurgery, and Orthopedics and Traumatology departments. Additionally, written informed consent was obtained from the patients and their relatives who agreed to participate in the study. Patient confidentially was maintained by separating the personal data page from the instrument. Only the researcher had access to these data. Patients were informed of the precautions that would be taken to protect the confidentiality of the data and who would or might have access. The research data were stored securely by using a USB drive. Patient anonymity was maintained using a code name.

Participant characteristics. Of a population of 312, 207 patients (117 [56.5%], male, 90 [43.5%] female; mean age 60.85 ± 16.45 years) participated in the study. The demographic and clinical characteristics of the patients are presented in Table 2. Of the 207 participants, 88 (42.5%) were in Neurology Intensive Care, 78 (37.7%) were in Neurosurgery Intensive Care, 21 (10.1%) were in the Orthopedics and Traumatology Unit, and 20 (9.7% ) were in the Internal Medicine ICU. The follow-up period of patients was 10.66 ± 5.52 days (see Table 2).

PU development. Among the 207 participants, 56 (27.1%) developed a PU/I; of those, 51 (91.1%) had a Stage 1 and 5 (8.9%) had a Stage 2 PU/I. The mean time between admission and PU/I development was 10.45 days. All patients had only 1 PU/I. The cumulative incidence during the 12-week follow-up was 22%. The most common site for a PU/I was the sacrum (50; 89.3%) (see Table 2).

The mean age of patients who did and did not develop a PU/I was 68.51 years and 58.01 years, respectively, a statistically significant difference (t = 4.24; P <.05). No statistically significant difference was noted in PU/I development between genders.

Reliability. Diabetic status was removed from the scale because its correlation coefficient was less than 0.25 (r = 0.18). Item-total correlation coefficients varied between 0.27 and 0.79. The correlation between small repositioning shifts and incontinence, discomfort, and pain sensation was more than 0.70 (see Table 3). Corrected item-total correlations were 0.18 for diabetic status, 0.79 for small position shifts, 0.74 for incontinence, and 0.67 for discomfort and pain. All correlations were statistically significant (P <.001) (see Table 4). After diabetic status was removed, the mean total correlation coefficient of the 7 items increased from 0.32 to 0.40, and the Cronbach alpha coefficient increased from 0.80 to 0.81. The maximum score was 14 when diabetic status was removed. The interrater agreement was 99%, and the ICC was 0.99 (n = 30) (see Table 5).

Validity. The Kaiser-Meyer-Olkin coefficient of EFGU scale was determined to be 0.80, and the Bartlett test result was 1107.57 (P <.001). These values underscored the fact that the scale was suitable for explanatory factor analysis, showing a single factor with an eigenvalue greater than 1 in the scale, which explained 50.41% of the variance.

The DI and Youden index values were calculated at the end of the ROC analysis for determining the cutoff value and are provided in Table 6. The score that corresponded to the point where the scale had the highest value in the DI and Youden index was determined as the cutoff point. Patients with 6 or more points on the EFGU Pressure Ulcer Risk Assessment Scale were considered to be at high risk of developing a PU. At the point where the cutoff score was 6, the sensitivity of the scale was 0.97, specificity 0.83, positive predictive value 0.69, and negative predictive value 0.99. The area under the curve (AUC) was found to be excellent (0.95; 95% CI: 0.93–0.97, P  <.001) (see Table 5 and Table 6). At a cutoff score of 6, the average score was 9.10 ± 2.02 for patients at high risk and 2.02 ± 1.36 for patients at low risk (t = 37.78; P  <.01). PU/I developed in 97.1% of the patients detected as high risk, and 2.9% occurred in persons at low risk (X²: 1.74; P  <.01). 

This study evaluated the validity and reliability of EFGU Pressure Ulcer Risk Assessment Scale. After removing diabetes and reducing the scale to 7 items, the scale had the most balanced sensitivity and specificity values when the cutoff point was taken as 6. The AUC value (0.95) also indicated that the scale had an excellent discrimination power (ie, it represented an outstanding discrimination between the PU/I and no PU/I groups). Most importantly, the new scale was able to accurately diagnose many patients who developed PU/Is at a cutoff point of 6 (sensitivity 0.97 and specificity 0.83). At this point, both the sensitivity and specificity of the scale were high. This could be attributed to the fact that each item in the scale was determined in the context of a current conceptual framework, and the contribution of each factor to the development of PUs was well defined. 

The available literature shows a limited number of scales developed for patients in the ICU, including the Decubitus Ulcer Potential Analyzer⁴⁴ (DUPA), the Jackson-Cubbin Scale,⁴⁴ and the Suriadi and Sanada (SS) Scale.^44,45 The DUPA scale is a modified version of the Gosnell, Norton, and Braden scales. The Jackson-Cubbin Scale also was adapted from the Norton Scale. Only the SS Scale⁴⁵ achieved balanced sensitivity and specificity values (sensitivity, 81%; specificity, 83%), which was consistent with the findings of the present study. However, when using the SS Scale, interface pressure between skin and bed must be measured using a multipad pressure evaluator, which may not be convenient when risk assessment scales should be completed quickly and not put extra workload on nurses.⁴⁶ The DUPA Scale had specificity and sensitivity of 68.8% and 64.9%, respectively,³² whereas the Jackson-Cubbin Scale had a sensitivity and specificity of 100% and 54%, respectively.⁴³ The results of these studies indicated these scales did not have optimal specificity and sensitivity values. 

The Braden Scale, developed by Bergstrom et al,³⁶ is commonly used in all hospitalized patients in Turkey.³ At a cutoff of 16, the scale achieved the most stable sensitivity (83%) and specificity (64%) values.³⁶ However, patients were evaluated only upon hospital admission in this study. These results may not be satisfactory because the condition of patients in the ICU can change rapidly. Hyun et al⁴⁸ also evaluated the validity of the Braden Scale in 7790 patients in the ICU. The scale was found to be insufficient in distinguishing patients in the ICU at risk of a PU/I because it had extremely low specificity and positive regression values at the cutoff points of 13, 16, and 18.⁴⁸ The study conducted by Dicle and Soyer¹⁶ in an ICU in Turkey among 120 patients demonstrated that the Braden Scale had high sensitivity (91.8%) and low specificity (62.7%) at a cutoff point of 13. The Waterlow scale also was found to have high sensitivity and low specificity in patients in the ICU.^47,49 The new scale was found to have high sensitivity and specificity (sensitivity, 0.97; specificity, 0.83) and was able to accurately diagnose many patients who developed PU/Is at a cutoff point of 6. 

In the present study, the positive and negative predictive values of the scale were 0.69 and 0.99, respectively. The fact that the positive predictive value in this study was not extremely high could be explained by the effect of PU/I preventive interventions applied in the units where the study was conducted and/or a result of improvement in the health status of the patients during their stay in the hospital. The positive and negative predictive values can change in different environments. Moreover, the PU/I prevalence can affect these values. A higher prevalence is associated with higher positive and lower negative predictive values^39,50 Studies also found high positive and low negative predictive values of PU/I risk assessment scales.^{13,15,18,27,51}

The Cohen Kappa statistical analysis of the EFGU Scale showed interobserver agreement was 99.4%. Previous studies^52,53 had shown the Norton and Waterlow scales had lower interobserver agreement scores.Watkinson⁵² demonstrated an interobserver agreement with the Waterlow scale was 55.6% among 9 patients and 12 observers. Wang et al⁵³ found a 59% the interobserver agreement using the Norton scale in a group of 23 patients and 5 observers. In other research, low compatibility between 2 independent observers indicated that the reliability of the scale was also low.^41,43 This situation was a result of the knowledge, experience, and judgments of the person making the evaluation. The fact that the number of items in the new scale was low and the items in the scale were understandable might have influenced the high correlation between the observers. 

The correlation coefficients were determined by performing Cronbach alpha reliability analyses on the 8 items in the EFGU scale in the present study. The item diabetic status showed an item correlation lower than 0.20 and was removed from the scale. Blood and oxygen transport to tissues put inpatients with diabetes at higher risk of PU/I occurrence due to microangiopathy and neuropathy.^23,37 Although cohort, prospective, and descriptive studies^20,35,54,55  have shown diabetes can affect PU/I development, a pressure ulcer development study by Berlowitz et al⁵⁴ using a Minimum Data Set to derive a risk-adjustment model demonstrated that this was not an effective method to assess PU/I development. Hence, more studies should be performed on this field to achieve definite conclusions.

After the number of items was reduced to 7, the item-total point correlations were re-calculated after Cronbach alpha reliability analyses of the EFGU Scale and found to be 0.81, supporting the instrument’s reliability and showing it was higher than the 0.80 noted for the Braden,^{15,18,17,28,51} Norton,^15,18,27,28 Waterlow,^18,28 Emina,^18,28,56 and Risk Assessment Pressure Sore scale developed by Lindgren.^{1,18,27,28,57} These results indicated that all scales had a high reliability score. 

The present study had some limitations. First, it was conducted in only 1 center. Hence, the results cannot be generalized. During the course of the study, the PU/I preventive interventions were applied continuously to the patients for ethical reasons and this may have affected the results. Another limitation was the lack of a measurement tool that could be considered the gold standard in determining PU/I risk for comparison purposes. Moreover, the study was conducted only on patients in the ICU. The high rates of PU/I development in this group might have affected the outcomes because these patients were generally elderly people with limited physical activity. Further, specific differences might exist within the Turkish population in terms of variables, such as preventive strategies, clinical environment, diseases, and patient status. Therefore, the study should be repeated in other units and among different populations.

This study evaluated the validity and reliability of EFGU Pressure Ulcer Risk Assessment Scale. The AUC value (0.95) also indicated the scale had an excellent discrimination power. Most importantly, at a cutoff point of 6 (sensitivity 0.97 and specificity 0.83) the new scale was able to accurately diagnose many patients who developed PU/Is. Identifying patients at high risk for PU/I development is important in order to target persons who need appropriate preventive interventions. The authors saw the need to develop a PU/I risk assessment scale for use in the ICU that had a better reliability and validity results than existing scales. The present study revealed that the EFGU Scale is reliable concerning internal consistency and has achieved a balance between sensitivity and specificity. Further studies that take into account the effect of PU/I preventive measures and include a larger sample size and multicentered populations are warranted.

Dr. Efteli is an Assistant Professor, Faculty of Health Sciences, Mehmet Akif Ersoy University, Burdur, Turkey. Dr. Günes is a Professor, Faculty of Nursing, Ege University, İzmir, Turkey. Please address correspondence to: Elçin Efteli, PhD, RN, Assistant Professor, Faculty of Health Sciences, Mehmet Akif Ersoy University, Burdur, Turkey; email: elcin.efteli@gmail.com.

Potential Conflicts of Interest: none disclosed