FormalPara Take-home message

Pressure injuries are common in adult ICU patients and ICU-acquired pressure injuries are associated with mainly intrinsic factors, and mortality. Increased clinical awareness, appropriate resource allocation, and further investigations into the pathophysiology of pressure injuries in critical illness and optimal prevention strategies for ICU patients are pivotal to tackle this important patient safety threat.

Introduction

Pressure injuries are localised lesions to the skin and/or underlying tissues due to pressure or pressure combined with shear [1, 2]. Often occurring at bony prominences, they can develop anywhere on the body. Predisposing factors include limitations in activity/mobility, deficiencies in nutrition and skin moisture, inadequate perfusion, and the use of mechanical devices that exert pressure on the skin [3, 4]. Frequently incorrectly considered a specific problem of long-term residents, they may develop as quickly as between the first hour and 4–6 h after sustained loading [5]. An international classification categorises the injuries into stages I–IV, Unstageable, and Suspected Deep Tissue Injury according to the extent of the tissue damage (Online Resource_2) [1, 2].

Pressure injuries cause pain and disability, compromise the quality of life [6], and extend the length of hospital stay by an average of 5–8 days per pressure injury [7]. By increasing the need for care resources they are a major economic burden for healthcare systems worldwide [8,9,10]. In the United States, the incremental hospital cost per patient of treating hospital-acquired pressure injuries is estimated at about US$10,708 and might exceed US$26.8 billion at the national level [11].

Patients residing in the intensive care unit (ICU) are extremely prone to developing pressure injuries due to their inherent immobility, haemodynamic instability, poor tissue perfusion and oxygenation, and to a plethora of complexly interacting intrinsic and extrinsic risk factors [12,13,14]. Additionally, they are highly exposed to medical devices [15]. Finally, medical and technological advances have generated a substantial ICU population of geriatric patients and long-term residents whose risk of developing pressure injuries might even be higher [16,17,18].

Despite the severity of the problem and the considerable unfavourable impact of these lesions on patient outcomes, patient care, and health economics, research interest in pressure injuries in the ICU population has remained restricted.

As a result, clear insight into the global epidemiology of pressure injuries in ICUs is still lacking [19]. A recent systematic review and meta-analysis on their occurrence in adult ICU patients found 10 studies published between 2002 and mid-2017 reporting cumulative incidences, and 12 providing prevalence data only [20]. Moreover, the included studies’ outcomes showed large variability. Cumulative incidence ranged from 3 to 39.3%, prevalence from 11.5 to 32.7%. These large differences cannot currently be explained due to a lack of large study cohorts capable of dealing with the clinical heterogeneity that is typical for the ICU setting, and with variations in the availability of healthcare resources worldwide.

The objective of this study was to provide an up-to-date picture of the extent and factors associated with pressure injuries in a large, geographically diverse cohort of adult ICU patients. More specifically, we aimed to identify the overall and ICU-acquired prevalence according to geographic region and anatomical location; risk factors associated with ICU-acquired pressure injuries; and the association of pressure injuries with hospital mortality. We hypothesised that a number of the individual patient and ICU contextual factors will be associated with the development of pressure injuries in adult ICU patients.

Methods

A full description is in Online Resource_3.

Study design and subjects

The Decubitus in Intensive Care Units study (DecubICUs) was a worldwide prospective, observational, 1-day point-prevalence study of pressure injuries among adult ICU patients with 12-weeks follow for survival status and length of hospital stay. All patients ≥ 18 years in ICU from 0:00 to 23:59:59 h on the study day were eligible; there were no exclusion criteria. DecubICUs was registered at ClinicalTrials.gov (NCT03270345).

Ethical approval

Overall, approval by established national, regional or local ethics committees and/or institutional review boards was granted.

Data collection

Data were collected on 15 May 2018. Alternative dates were set for Nigeria, Brazil and Libya due to delayed ethics approval. Anonymous patient data were collected by case report form. They encompassed demographic and admission data, and physiological data pertaining to the study day, including the severity of disease assessment by the Simplified Acute Physiology Score II (SAPS II) [21]. Pressure injury occurrence was measured by direct observation according to the international staging definitions [1, 2]. Pressure injury risk was assessed by the Braden scale that combines 6 subscales: mobility, activity, sensory perception, skin moisture, nutritional state, and friction/shear, with lower scores reflecting higher risk [22]. Follow-up data gathered were survival status, and length of ICU and hospital stay until hospital discharge or at 12 weeks following the study day (7 August 2018). The study protocol, including case and center report forms, is in Online Resource_4 and at https://www.esicm.org/research/trials/trials-group-2/decubicus/.

To maximise uniformity in reporting, we developed a training module with self-test on pressure injury staging (Online Resource_5) [1, 2] that was validated for content by 3 experts and published on the study website prior to study initiation. Registered participants were repeatedly encouraged to familiarise themselves with the module before data collection.

Data management

Quality and integrity of the reported data were checked. Missing, extreme or implausible values were returned to the local data collectors for review. Where data remained questionable, the primary investigators (SOL and SIB) made a final adjudication about study inclusion in mutual agreement. Missing values mutually judged eligible for inclusion were imputed with median values or deduced from other variables reported. Remaining missings were omitted from the analyses.

Statistical analyses

Analyses were performed at the patient level. Overall pressure injury prevalence was calculated as the proportion of the sample with at least one pressure injury on the study day, ICU-acquired prevalence as the proportion with at least one pressure injury acquired in ICU on the study day. Prevalence is reported as numbers (n) and percentages with 95% confidence intervals (CI). Continuous data are summarised by a median with interquartile range (IQR), categorical data as n (%). Univariate analyses used Chi square, Mann–Whitney U, and Kruskal–Wallis tests, as appropriate. Survival analysis was performed by Kaplan–Meier procedure (log-rank test). Associations with ICU-acquired pressure injuries were examined by generalized linear mixed-effects regression analysis with logit link function and a random effect for country. All variables were included following an exploratory approach, irrespective of univariate analyses results. As analyses did not focus on a prediction but on the identification of associations, feature selection was not applied, particularly as the risk of overfitting was minimised given the limited number of covariates (n = 24 for pressure injury occurrence, n = 22 for hospital mortality) and the adequate dataset size (n = 13,254). Results are reported as odds ratios (OR) with 95% CIs.

Statistical analysis was performed using IBM SPSS 24.0 (IBM Corp., NY, US) and R statistical software 3.6.1 [23].

Results

Hospitals and patients

We recruited 1117 ICUs in 90 countries (6 continents). Most were mixed medical-surgical units (n = 729; 65.2%) and in university hospitals (n = 675; 60.4%). Median (IQR) hospital and ICU capacities were 600 (329–1035) and 13 (8–20) beds, respectively; 1005 (89.9%) data collectors had studied a training module on pressure injury staging, of which 920 (82.3%) the module developed for this project. Participation rates and ICU characteristics are in Online Resources_6 and 7, respectively.

Data from 13,254 patients were eligible for analysis. Their demographic characteristics are in Table 1, completeness of data in Online Resource_8.

Table 1 Characteristics of included patients
Full size table

Prevalence

We identified 6747 pressure injuries in 3526 patients, of which 3997 were ICU-acquired (59.2%; 2145 patients). Overall, 2081 patients had 1 pressure injury, 653 patients had 2, 411 had 3, and 381 had > 3 pressure injuries; and 1284 patients had 1, 398 had 2, 243 had 3, and 220 had > 3 ICU-acquired pressure injuries. Injuries were acquired before ICU admission in 1381 patients; developed in the ICU in 1922; and 233 patients developed injuries both before and during ICU stay.

Table 2 reports the overall and ICU-acquired prevalence across the 6 continents. A detailed breakdown per Stages and continents is in Online Resource_9. The overall prevalence was 26.6% (95% CI 25.9–27.3) with 18.0% (95% CI 17.3–18.6; n = 2383/13,254) of stage II or worse. Overall stage II prevalence was 11.4% (95% CI 10.9–11.9), stage III prevalence 4.2% (95% CI 3.9–4.6), and stage IV prevalence 2.0% (95% CI 1.7–2.2). Prevalence of Unstageable and Suspected Deep Tissue Injuries was 2.1% (95% CI 1.9–2.4) and 2.3% (95% CI 2.1–2.6), respectively.

Table 2 Overall and ICU-acquired pressure injury prevalence according to continents
Full size table

ICU-acquired prevalence was 16.2% (95% CI 15.6–16.8), with 11.0% (95% CI 10.5–11.5) of stage II or worse. ICU-acquired stage II prevalence was 7.5% (95% CI 7.1–8); stage III prevalence 3.2% (95% CI 2.9–3.5), and stage IV prevalence 1.7% (95% CI 1.5–1.9). ICU-acquired prevalence of Unstageable and Suspected Deep Tissue Injuries was 2% (95% CI 1.7–2.2) and 2% (95% CI 1.8–2.3), respectively.

ICUs from low and lower-middle-income economies, where the mean percentage of gross national income spent on healthcare is least, reported the highest overall prevalence of pressure injuries (40.7%, 95% CI 36.7–44.8) and of ICU-acquired pressure injuries (27.7%, 95% CI 24.1–31.5; Online Resource_10).

The sacral region and heels were the most affected anatomical sites, accounting for 37 and 19.5% of all pressure injuries, respectively. Figure 1 shows numbers (percentages) of overall and ICU-acquired pressure injuries at the most affected body locations. A comprehensive overview of all body locations according to pressure injury staging is in Online Resource_11.

Fig. 1
figure 1

Anatomical locations of pressure injuries (most affected body sites). Left: Numbers (percentages) of overall pressure injuries − total number of pressure injuries = 6764. Right: Numbers (percentages) of intensive care unit-acquired pressure injuries − total number of pressure injuries = 3997

Full size image

Factors associated with ICU-acquired pressure injuries

Generalized linear mixed-effects regression analysis identified the following factors as independently associated with ICU-acquired pressure injuries: older age, male sex, being underweight, admission due to emergency surgery, decreasing Braden scores, increasing ICU stay, chronic obstructive pulmonary disease, immunodeficiency, renal replacement therapy, mechanical ventilation on ICU admission, higher SAPS II score, and being in a low or lower-middle-income economy, with strongest, gradually increasing associations with worsening Braden scores and increasing length of ICU stay before the study day, respectively (n = 12,533; Table 3).

Table 3 Factors independently associated with ICU-acquired pressure injury
Full size table

Hospital mortality

Overall hospital mortality was 22.5% (95% CI 21.8–23.3; n = 2929/12 989). Following adjustment for demographics and morbidity data, severity of pressure injury showed a gradually increased association with hospital mortality: OR 1.31 (95% CI 1.1–1.55) for stage I, OR 1.66 (95% CI 1.41–1.95) for stage II, and OR 2.31 (95% CI 1.96–2.71) for stage III or worse, i.e. stage IV, Unstageable, or Suspected Deep Tissue Injury (n = 11 889; Online Resource_12). Figure 2 reports survival distribution for patients with increasing severity of pressure injuries (i.e., no pressure injury, stage I, stage II, and stage III or worse; Log-rank test: p < 0.001).

Fig. 2
figure 2

Kaplan–Meier estimates of overall survival according to pressure injury status on the study day among adult intensive care unit patients. Green line indicates patients without pressure injuries; yellow line indicates patients whose worst pressure injury is of stage I; orange line indicates patients whose worst pressure injury is stage II; red line indicates patients whose worst pressure injury is stage III or worse (i.e. stage IV or Unstageable or Suspected Deep Tissue Injury). Tick marks indicate censored data (hospital discharge before 12 weeks). Log-rank test: p < 0.001

Full size image

Discussion

In this point-prevalence study encompassing 1117 ICUs in 90 countries across 6 continents and involving 13,254 adult patients, we found an overall pressure injury prevalence of 26.6% and an ICU-acquired prevalence of 16.2%. Although the prevalence was highest in low and lower-middle-income economies, our findings suggest that pressure injuries remain a considerable burden for healthcare systems worldwide, and highlight the necessity of additional efforts in patient safety initiatives.

These observational data confirm and reinforce previous findings resulting from meta-analysis [20]. Additionally, they are complementary to findings from systematic reviews aiming at determining risk factors for pressure injuries in ICU patients [12, 24,25,26]. These identified a broad range of factors including age, length of ICU stay, diabetes, mechanical ventilation, vasopressor support, hypotension, and cardiovascular disease, and suggest that an interplay of these factors increases the probability of pressure injury development. Our data, albeit resulting from cross-sectional research and thereby only representing the study day, are suggestive for associating pressure injury in ICU with a patient profile characterised by high vulnerability, as evidenced by the following findings. First, the occurrence of pressure injuries was associated with the Braden score, which summarises essential conditions that gradually contribute to a high-risk profile characterised by being bedridden, malnourished, incontinent, and with limited ability to react on or sense pain [22]. These conditions are characteristic for a majority of ICU patients and mirror an overall vulnerability level. Second, older age was independently associated with pressure injury occurrence. The steadily increasing proportion of very old ICU residents constitutes an overt influx of high-risk patients given the accumulation of chronic comorbidities, nutritional deficiencies, immobility, and aging skin [17, 18]. Third, an association was found with organ support (mechanical ventilation and renal replacement therapy), which implies a high severity of acute illness. Finally, this high-vulnerability profile is completed by the finding that patients who resided > 12 days in ICU before the study day had a 7.5-fold increased risk of ICU-acquired pressure injury compared to patients with a short ICU stay (≤ 3 days).

As such, our data suggest that the large majority of factors associated with pressure injury in ICU patients appear to be intrinsic or unmodifiable. This is in line with the unanimous agreement of experts that pressure injuries can be unavoidable in haemodynamically unstable or critically ill/injured individuals [27]. Our findings need validation, preferably in longitudinal studies. Prospective high-resolution data from smaller samples might also identify additional modifiable factors not sought in this study. A hint that these may exist is the lower prevalence reported in Asia where increased awareness may have been prompted by previous large-scale initiatives on this topic. Until such data is generated, the variables we identified can at least be used to flag patients who might benefit from greater vigilance for pressure injuries. Also research into pressure injury pathophysiology and prevention specifically directed towards the heterogeneous ICU population is recommended.

Another factor independently associated with pressure injury was being in a low or middle-low income economy ICU. Limited availability of human and material resources may contribute to this finding, as the mean percentage of gross national income spent on healthcare in these economies is less than half as compared with high-income economies (4.9% versus 10.3%). Additionally, pressure injury prevention might not be a healthcare priority in developing countries.

Manzano and co-workers [28] identified pressure injury as a significant independent predictor of mortality in mechanically ventilated patients (adjusted hazard ratio 1.28; 95% CI 1.003–1.65; p = 0.047). The mortality associated with pressure injuries remains however unclear. As their occurrence often mirrors a generally debilitated condition and high severity of acute illness, an association with mortality seems reasonable. However, our regression analysis demonstrated a gradual increase in mortality with increasing severity of pressure injuries despite adjustment for these covariates. Even though this does not imply causality, this observation calls for clinical concern towards patients presenting with pressure injuries or those at high-risk for developing such complications.

Stage I pressure injuries are generally considered reversible if promptly identified and appropriately managed [13] and, therefore, often excluded from scientific reports [19]. They were nevertheless shown to be prone to deterioration, as in 6 Dutch acute care hospitals where 22.1% worsened to a deeper lesion [29]. In line with several earlier prevalence reports [29], the majority of pressure injuries in our study were of stage I (38.1%). These currently often underreported injuries, however, emerged from our analyses as independently associated with hospital mortality, which calls for considering these lesions as full quality indicators and for the standardized recording of this data in institutional and research reports.

This study has limitations. The cross-sectional design might have resulted in bias toward patients who have long ICU stays [30]. Since the length of stay is associated with pressure injury risk, the reported prevalence might not be representative for the entire ICU population. Our data only represents a snapshot at the study day and cannot account for potentially influencing factors such as staffing levels. Data on pressure injuries on mucosal surfaces have not been collected as these are not staged by the international staging system [4]. Not all geographic regions are well-represented, thus impeding globally generalized results. As pressure injuries might be considered as a result of suboptimal care, fear of criticism or institutional censure may have hampered objective reporting. If so, the actual prevalence might be higher than the rates identified. Accurate assessment of pressure injury staging is challenging and data collectors were not required to be qualified tissue viability experts. Despite our efforts to obtain consistency in reporting using a well-documented data collection procedure and providing a training module, variability and errors in staging may have occurred. Given the scale of the study, it was however not feasible to assess the validity of the data using digital photographs. Nevertheless, the error resulting from our approach will if anything have led to random error in estimations, rather than a systematic error. We were unable to doublecheck the self-reported number of participants who indicated having studied the training module, which may be prone to social desirability bias. As we requested to report the number of ‘nurses’ on the study day, without further definition, we do not know whether assistant-nurses were also reported and included in the calculation of the number of patients-per-nurse. The unexpected association of this variable with pressure injury also needs further exploration. There is a view that Suspected Deep Tissue Injuries should not be included in epidemiological studies because it is unclear how many are actual deep tissue injuries that convert to pressure injuries. Their number was however small and unlikely to have substantial impact, if any, on the estimated prevalence. Finally, our study may be prone to random observer errors as data collectors depended on the reliability of patient files to determine whether a pressure injury was ICU-acquired.

The major strength is that it is the first to present a worldwide picture of the epidemiology of pressure injuries in adult ICU patients and to map a high-risk profile based on a large global sample. It may act as an incentive for tackling this patient safety issue and provide local and regional baseline data for quality improvement programmes. Furthermore, pressure injuries staging was assessed by the gold standard of skin inspection by trained outcome assessors, and the study used a rigorous protocol with clear attention to detail in standardising the data collection process.

Conclusions

This observational study identified a quarter of ICU patients with pressure injuries albeit with considerable regional variation in prevalence. However, approximately 60% of the patients developed these lesions in ICU irrespective of the regional prevalence. As pressure injuries are a common complication and a substantial burden for healthcare systems worldwide, their prevention deserves increased clinical awareness and appropriate resource allocation. Besides, further investigations into the pathophysiology of pressure injuries in critical illness and into optimal prevention strategies for ICU patients are pivotal to tackle this important patient safety threat.