The initial search of the indices yielded 540 published reports; 358 publications remained after deleting duplications. When inclusion/exclusion criteria were applied, 28 were examined for possible inclusion; 12 studies were eliminated because they did not meet the predetermined inclusion criteria: no multivariate analysis (2); single risk factor (1); focus only on risk assessment scale (4); focus on nursing documentation (1); focus on PU healing (1); trauma population (1); conceptual model development (1); and comparison study (1). This exclusion yielded the 16 studies ultimately identified and included in this review.6-21 Across these 16 studies, 43 potential risk factors were identified in multivariate analyses. For ease of interpretation, these risk factors were grouped under 6 broad headings: demographic/patient characteristics, comorbidities, intrinsic factors, iatrogenic/care factors, PI risk assessment scales, and severity of illness/mortality risk. After examination of these risk factors, 7 emerged in 3 or more studies and included age, ICU length of admission, diabetes mellitus, cardiovascular disease, hypotension, prolonged mechanical ventilation, and vasopressor agents.
Of the 16 studies, 7 were conducted in the United States7,8,13,14,16,18,20 and 9 were conducted internationally, representing Brazil,9,17 Greece,10 Iran,11 Italy,15 Japan,21 Saudi Arabia,6 Spain,19 and Turkey.12 ICU settings represented included medical surgical ICUs (5),7,8,10,11,18 cardiovascular ICUs (3),7-9 and surgical ICUs (2)16,20; 2 studies included multiple types of ICUs7,8 and in 8 studies, the type of ICU was not specified.6,12-15,17,19,21 With regard to study design, 7 were retrospective7,8,13-16,18, and 9 were prospective6,9-12,17,19-21 (see Table 1A, 1B, 1C).
Study sample characteristics. Across the studies, the mean age of critical care patients ranged from 52 years6 to 71 years.8 The most common ICU admitting diagnoses were respiratory-10,12,17,18 and cardiac-related.8,9,15 Mean ICU length of admission was reported6-9,11,12,16-19,21 and ranged from 5 days18 to 17 days.12 A measure of patient severity of illness/mortality risk was reported in 10 of the studies6,8-10,12,14,17-19,21 using a variety of instruments including the Acute Physiologic and Chronic Health Evaluation (APACHE) II, APACHE III, Sequential Organ Failure Assessment (SOFA), Simplified Acute Physiologic Score (SAPS II),22 and All Patients Refined Diagnosis Related Group (APR-DRG). The APACHE II was the most common tool used to measure illness severity (6 studies8-10,12,18,21), with mean scores across these studies ranging from 9.521 (indicating an 8% mortality risk) to 229 (indicating a 40% mortality risk).23 The American Society of Anesthesiologist’s risk classification was used in 1 study to determine surgical risk, but no mean score for the study sample was reported13 (see Table 2A, 2B).
PI characteristics in studies. PU occurrence rates across the study populations varied from 9.4%16 to 39.3%.6 When stratified by stage, 5 studies identified Stage 2 PI,9,10,13,18,21 4 identified Stage 1,6,9,11,12 1 identified Stage 3,15 and 18 identified deep tissue injury (DTI) as most common. In 6 studies,7,14,16,17,19,20, the most common stage of PU was not recorded. With regard to PU location, 7 of the 9 studies that used this descriptor identified the sacrococcygeal region as the most common anatomical site.7-10,12,18,19 In 1 study,21 the ear, scapula, and heels were documented as the most common sites for PU development; in another,6 the heel was the most common location. With regard to the application of PU prevention strategies, 7 studies6,8,11,12,18,19,21 reported the presence of or consideration for PU prevention strategies as part of the design and implementation of the study (see Table 2A, 2B).
Risk factors. Numerous potential PU risk factors were examined in the studies in this review. Table 3 (3A, 3B, 3C, 3D) highlights the significant predictors found in each study according to its multivariate analysis in addition to risk factors found to be significant in univariate analysis, those found to be nonsignificant, and other risk factors considered in the study design.
Overall, 43 risk factors were found to be significant predictors across the studies in this review and were grouped into 6 broad categories as previously described; 7 were found significant in multivariate analysis in 3 or more studies (see Table 4). These risk factors included age, ICU length of admission, diabetes mellitus, cardiovascular disease, hypotension, prolonged mechanical ventilation, and vasopressor agents.
Demographic/patient characteristics. Age was the most frequently reported predictor and found to be significant in 6 studies.6,9,11,18-20 However, the mean age of patients in these 6 studies ranged from 55 to 69 years, representing middle-aged to younger elder adults. When the mean age of patients who developed a PU across these 6 studies was examined, ages ranged between 60 years and 73 years. In multivariate analysis, 1 prospective cohort study9 of 370 cardiopulmonary ICU patients found age >42.5 years to be a significant predictor of PU development. In a prospective 2-phase study of 369 surgical ICU patients, Slowikowski and Funk20 found age ≥70 years to be a significant predictor. Advancing age has long been considered a risk factor for PU development, as earlier studies24,25 in the ICU population noted.
Length of stay. Length of stay was the second most common predictor across these 16 studies and was found to be significant in 5 studies.6,9,11,17,18 In a prospective cohort study of 84 ICU patients in Saudi Arabia, Tayyib et al6 reported the average length of stay for patients who developed a PU was 13.3 ± 8.3 days as compared to the PU-free group with 9.1 ± 6.6 days. In multivariate analysis, length of stay was found to be a significant predictor for patients with any stage of PI (odds ratio [OR] 1.8, 95% confidence interval [CI] 1.01-3.30; P = .045) as well as for patients with Stage 2 through Stage 4 PI (OR 1.2, 95% CI 1.08-1.39).6 Campanili et al9 found an ICU admission of >9.5 days was a significant predictor of PU development. Patients with a PU were found to have a mean length of stay of 14 days as compared to patients who remained PU-free at 4.5 days (P <.001). A prospective study of 160 ICU patients in Brazil by Cremasco et al17 also found length of stay to be predictive for patients who developed PU (OR 1.12, 95% CI 1.04-1.20; P = .002), with a mean number of ICU days of 23.9 days as compared to those that remained PU-free at 9.4 days (P <.001). Similarly, a retrospective study of 347 medical/surgical ICU patients by Cox18 found length of stay to be significant in multivariate analysis for all stages of PI (OR 1.033, 95% CI 1.003-1.064; P = .03) and for Stage 2 or greater PIs (OR 1.008, 95% CI 1.004.-1.012; P <.001) and reported a mean length of stay of 281 hours (12 days) in patients that developed a PU compared to 81 hours (3 days) for patients who remained PU-free (P <.01). In their prospective study of 352 medical/surgical ICU patients, Nassaji et al11 also reported ICU length of stay as a significant predictor (OR 1.19, 95% CI 1.13-1.25; P <.001).
Comorbidities. Significant comorbidities found across these studies included diabetes mellitus, cardiovascular disease, renal disease/failure, pulmonary disease, trauma, and anemia. Of these, diabetes mellitus11,15,20 and cardiovascular disease13,15,18 were each found to be significant in multiple studies. These 2 comorbidities are known to be strongly associated with PU development across patient populations.3
Diabetes mellitus was identified as a significant predictor in 3 studies.11,15,20 Slowikowski and Funk20 found a history of diabetes was a significant predictor of PU development (OR 1.93, 95% CI 1.11-1.35; P = .019), as did Nassaji et al.11 The latter study found PU development was more than 5 times more likely in patients with a history of diabetes (OR 5.58, 95% CI 1.83-18.7; P = .003). In a retrospective study of 610 critical care patients, Serra et al15 found both diabetes and congestive heart failure to be significant predictors, (OR 2.25, 95% CI not reported. P = .047; OR 2.25, 95% CI: not reported; P = .0047, respectively [in women]). Results of a retrospective, observational study by O’Brien et al13 showed a preoperative history of congestive heart failure to be predictive of ICU PU development (OR 1.78, 95% CI 1.27-2.49; P = .001), and a history of cardiovascular disease was found to be predictive of PU development in a retrospective, correlational study18 among medical-surgical ICU patients (OR 2.9, 95% CI 1.3-6.4; P = .007). One (1) retrospective study8 of 306 cardiovascular and medical surgical ICU patients found cardiovascular disease protective of PU development (OR 0.035, 95% CI 0.002-0.764; P = .03), which may have been a reflection of the sampling methodology.
Intrinsic factors. Hypotension is defined by the Society for Critical Care Medicine Sepsis 3 guidelines26 as a systolic blood pressure <90 mm Hg, mean arterial pressure (MAP) <60 mm Hg or 70 mm Hg, or a drop in systolic blood pressure >40 mm Hg. Hypotension denotes that perfusion is impaired.27 Blood is shunted from the periphery in an ongoing effort to preserve vital organ function, affecting tissue tolerance as capillaries close at lower tissue interface pressures.28 Hypotension was found to be a significant predictor of PUs in 3 studies.7,8,16 In their retrospective, 345 ICU-patient study, Bly et al7 reported a systolic blood pressure <90 mm Hg was predictive of PU development (OR 3.5, 95% CI 1.24-9.91: P <.05), and results of the retrospective cohort study by Wilczewski et al16 reported prolonged periods of MAP <70 mm Hg as the only predictor associated with PU development in a sample of 95 spinal cord injured patients in a surgical ICU. Cox and Roche8 found longer mean hours of MAP <60 mm Hg while on vasopressor agents to be predictive of PU development (OR 1.096, 95% CI 1.020-1.178; P = .01). Cox18 found MAP <60 mm Hg was significant in univariate analysis but not in multivariate analysis.
Iatrogenic/care factors. Prolonged mechanical ventilation was found to be a significant predictor in 3 studies.8,10,19 Cox and Roche8 found patients who required mechanical ventilation for >72 hours were 23 times more likely to develop a PU (OR 23.604, 95% CI 6.4-86.6; P <.001), and in a prospective study of 216 ventilated patients in 2 medical/surgical ICUs, Apostolopulou et al10 reported mechanical ventilation for >20 days was a significant predictor of PU development (OR 7.225, 95% CI 2.46-21.201; P <.001). In their prospective cohort study of 299 ventilated medical/surgical ICU patients, Manzano et al19 found the time on mechanical ventilation before PU development was an independent risk factor (OR 1.042, 95% CI 1.005-1.080; P =.024), with PU risk increasing by 4.2% for each day on mechanical ventilation. O’Brien et al13 identified the presence of an endotracheal or tracheostomy airway in an ICU patient before the surgical procedure was a significant predictor of postoperative PU development (OR 1.663, 95% CI 3.63-7.67; P <.001).
Vasopressor agents. Vasopressors are potent vasoconstricting agents used to raise MAP in patients with profound hypotension unresponsive to fluid resuscitation29; administration commonly is confined to critically ill patients in the ICU setting. Common agents include norepinephrine, epinephrine, phenylephrine, vasopressin, and dopamine. In this review, 4 studies7,8,14,18 found vasopressor agents to be significant predictors of PU development. Bly et al7 found patients receiving more than 1 vasopressor (type not specified) were 3.3 times more likely to develop a PI (OR 3.71; 95% CI 1.65-6.62; P <.05). Cox and Roche8 found vasopressin administration to be independently associated with PU development (OR 4.816, 95% CI 1.66-13.92; P = .004). Vasopressin is usually a second-line agent used to manage shock states refractory to a single vasopressor agent. Cox and Roche8 postulate the addition of a second vasopressor agent (usually vasopressin) was considered the tipping point for PU risk. The retrospective cohort study by Tschannen et al14 reported patients who received vasopressor agents (type not specified) were 33 times more likely to develop a PU (OR 1.33, 95% CI 1.03-1.73; P = .03). Cox18 found norepinephrine to be an independent predictor of Stage 2 or greater PUs (Stage 3, Stage 4, DTI, unstageable). Apostolopoulou et al10 found vasopressor agents to be significant in univariate analysis only; however, in this study the variable inotropic agents (medications primarily indicated to improve cardiac contractility) was used. Although 4 studies9,13,16,21 did not find significant associations between these agents and PU development, in most of these studies the vasopressor agents under investigation were not stated and as such the ability to draw meaningful conclusions was impaired. Additionally, changes in treatment protocols such as the SEPSIS 3 guidelines26 have recently modified the hierarchy of vasopressor agent selection, potentially compromising knowledge of their effect on PU development.
Pressure ulcer risk assessment scales. The Braden Scale is the most well-known PU risk assessment scale and is used across the care continuum, including ICUs. The Braden Scale measures PU risk on 6 subscales (sensory perception, activity, mobility, moisture, friction/shear, and nutrition). Cumulative scores range from 6 to 23, with lower scores indicating greater PU risk. Used mostly abroad, the Jackson-Cubbin Scale30 is a tool designed for use in ICUs and measures risk based on 12 risk factors (age, weight, past medical history, general skin condition, mental condition, mobility, hemodynamics, respirations, oxygen requirements, nutrition, incontinent, hygiene). Cumulative scores range from 9 to 48, with lower scores indicating greater risk.
The total Braden Scale score was found to be a significant predictor in 2 studies.14,20 Tschannen et al14 found that a lower Braden score on admission was a significant risk factor in multivariate analysis (OR 0.89, 95% CI 0.86-0.93; P <.001); Slowikowski and Funk20 also cited the total Braden Scale score (time of measurement not specified) as an independent risk factor (OR 1.30, 95% CI 1.15-1.47; P <.001). In the study by Cox18 that measured the total Braden score and the individual subscale scores on admission to the ICU, the total Braden score was found to be significant in univariate analysis only (r = -0.276, P ≤.01), and the subscales friction/shear and mobility were independent predictors of PU development (OR 5.715, 95% CI 1.423-22.95; P = .01; and OR 0.439, 95% CI 0.21-0.95; P = .04, respectively). In other studies that included the Braden Scale score in data analysis, 1 found it to be significant in univariate analysis only,8 while another found no relationship between the Braden Scale score and PU development.9
In the 1 study that employed the Jackson-Cubbin Scale to measure PU risk, Apostolopoulou et al10 found a Jackson-Cubbin score ≤29 to be a significant predictor of PU development (OR 0.015, 95% CI 0.005-0.050; P <.001). A comparative review31 of PU risk assessment scales found the Jackson-Cubbin superior to the Braden Scale in the ICU population. A systematic review and meta-analysis32 of the predictive validity of PU risk assessment scales used in the critical care population concluded the Braden scale was the best PU risk assessment scale for this population due to a lack of large studies using alternative PU risk assessment scales including the Jackson-Cubbin Scale. To date, no critical care-specific PU risk assessment scale has been validated in multiple empirical investigations using larger sample sizes in the critical care population. As such, a need for a PU risk assessment tool in this population persists.
Severity of illness. Severity of illness/mortality risk measures were reported in 10 of the studies in this review.6,8-10,12,14,17-19,21 Scales used to measure severity of illness included the APACHE II,8-10,12,18,21 APACHE III,19 SOFA,19 and SAPS II.17 In addition, APR-DRG14 and ASA Risk Classification13,33 were used as indicators of severity of illness; these are not tools specific to the ICU population. Of these scales, 4 were found in individual studies to be significant predictors of PU development.13,14,17,19 Cremasco et al17 found the SAPS II to be a significant predictor of PU (OR 1.058, 95% CI 1.004-1.114; P =.035); using the SOFA scale. Manzano et al19 found the fourth day cardiovascular SOFA score and the first day respiratory SOFA scores to be predictive (OR 1.33 95% CI 1.066-1.664; P = .12; and OR 1.56, 95% CI 1.026-2.360; P = .37, respectively). The APACHE III score also was measured in this study and found to be nonsignificant in univariate analysis. The ASA risk classification was used in a study of intraoperative risk factors in 2695 critical care patients; higher anesthesia risk classes of 4 or 5 as compared to lower risk classes of 1, 2, or 3 were found to be predictive of PU development (OR1.63, 95% CI 1.19-2.23; P = .003).13 Tschannen et al14 measured risk for mortality using the APR-DRG and found patients in the highest mortality risk categories were 11 times more likely to develop a PU (OR 11.15, 95% CI 7.1-17.5; P <.001). Higher APACHE II scores were found to be significantly associated to PU development in univariate analysis in 4 studies,8,12,18,21 while in 2 studies9,10 no relationship between PUs and the APACHE II score emerged. The use of multiple severity of illness/mortality indices across these studies makes it difficult to determine if any of these tools offers some insight into the relationship between severity of illness measures and PI risk. Although validated tools to measure severity of illness and mortality risk may help describe the burden of illness experienced in the ICU population, these may not translate to valid indicators of PI risk.
PU characteristics. In 6 studies, PU stage was not reported7,16,17,19-21 and in 7 studies PU location was not reported.11,13-17,20 Of the studies that reported PU stage, Stage 2 was the most common,9,10,13,18,21 with the sacrococcygeal region cited as the most common location in 6 studies,7-9,11,18,19 which is consistent with the literature.34
Interestingly, only 1 study8 (conducted in the United States) reported DTI as the most common stage. This finding may reflect the difference in staging classification utilized in the United States and abroad. In the 2009 NPUAP/European Pressure Ulcer Advisory Panel (EPUAP) International Pressure Ulcer guidelines,35 the categories termed DTI (now termed DTPI) and unstageable PU were only recognized in the United States. DTPI begins at the muscle-bone interface and occurs as a result of tissue deformation due to a short period of intense pressure and/or as a result of tissue ischemia due to prolonged periods of immobility.36,37 Critical illness coupled with prolonged immobility may render a patient vulnerable to the effects of pressure; thus, it would be logical this population would be at higher risk for more severe PIs. In fact, in a large United States prevalence study,34 critical care patients had a higher prevalence of severe PUs (Stage 3, Stage 4, DTPI, or unstageable) as compared to the general hospital population.