Sequential Organ Failure Assessment (SOFA) to Predict Pressure Ulcer Risk in Intensive Care Patients: A Retrospective Cohort Study

Pressure ulcer (PU) risk assessment practices in adult intensive care unit (ICU) patients remain varied. Purpose: The authors assessed the performance of the Sequential Organ Failure Assessment (SOFA) scale and its subcategories in predicting the development of PUs. Methods: A retrospective cohort study was conducted of all adult patients admitted to the mixed medical-surgical ICU of a Finnish tertiary referral hospital between January 2010 and December 2012. Data (diagnoses, demographics, clinical information, treatments, and instrument scores) were retrieved from the ICU database. Wilcoxon and chi-squared tests were used to examine patient subgroup (medical or surgical ICU and intensive care or high-dependency care patients), length of ICU stay (LOS), modified Jackson/Cubbin (mJ/C) scores and SOFA subcategory variables, and first-day SOFA scores. PU association was determined by logistical regression. Results: Among the 4899 patients in the study population, the overall PU incidence of acquired PUs was 8.1%. Medical patients had significantly more PUs (145/1281; 11.3%) than surgical patients (212/3468; 6.1%) (P <.0001). In all subgroups, significantly more patients with PUs had higher SOFA scores (mean 8.24) than patients without PUs (mean 6.74) (P=.001). The difference persisted when patients with LOS ≥3 days in the ICU were considered. Among the SOFA subcategories, the Glasgow Coma score, renal and respiratory disorders, and hypotension were significantly (P <.0001) linked to PU development. First-day total SOFA score and its cardiovascular and respiratory subcategory scores were the most important predictors of PUs. Conclusion: The total SOFA score provides an additional tool to assess PU risk in ICUs and should be used together with the Braden or the mJ/C Scale.

According to cohort studies and systematic reviews,^1-5 patients in intensive care units (ICUs) are at a high risk of developing pressure ulcers (PU) because of their condition and limited ability to move due to critical illness and therapeutic interventions. The prevalence of PUs among ICU patients varies from 5% to 30%; these numbers have decreased during the last 2 decades.^1-4 PUs predispose the patient to a considerable risk of complications (eg, prolonged hospitalization and infection), and the deleterious consequences of PUs cause suffering to the patient, carry high societal costs, and place extra demands on the nursing staff.⁵ 

PUs have a multifactorial etiology, and more than 100 different risk factors have been identified^6-9; recently, 46 PU risk indicators in 6 broad categories were acknowledged as relevant to patients in ICUs.^6-8,10 Multifactorial analysis has identified an association between the development of PUs and the patient characteristics age and length of stay (LOS), the comorbidities diabetes and cardiovascular disease, the intrinsic factor hypotension, and the iatrogenic/care factors prolonged mechanical ventilation and use of vasopressor agents.¹⁰ Furthermore, PU risk scales such as the Braden Scale or Jackson/Cubbin Scale and severity of illness/mortality risk scales such as the Simplified Acute Physiology Score (SAPS) II calculator¹¹ or the Sequential Organ Failure Assessment (SOFA) scale¹² predicted PU development.¹⁰ 

The Braden Scale is the most widely used risk assessment instrument in intensive care, although it is not designed for use in an ICU setting.^4,13 The Jackson/Cubbin risk scale was specifically designed for the assessment of PU risk in intensive care, but it has not gained wide acceptance even though it has been shown in cohort and validation studies to more accurately identify patients at PU risk than the Braden Scale.^14-16 According to 2 retrospective cohort studies,^3,17 the modified Jackson/Cubbin (mJ/C) risk scale is a useful but not an optimal instrument for PU risk assessment in the ICU setting. The first study examined the suitability of the mJ/C risk scale for PU risk assessment among adult ICU patients³ and the second¹⁷ examined the specific accuracy of the 12 main subcategories of the mJ/C for predicting PU development.

Two (2) categories of the mJ/C — oxygen requirement and hemodynamics — were found in a retrospective cohort study¹⁷ to be important prognostic indicators for PU development. The Braden Scale does not take these variables into account.¹³ The SOFA score assesses consciousness using the Glasgow Coma score (GCS) and takes into account respiratory and hemodynamic variables, albeit differently from the Jackson/Cubbin scale.^3,14,18 

A prospective cohort study¹² involving 9 medical-surgical ICUs (N = 299 patients) assessed SOFA as a risk indicator for PU in intensive care patients. Average patient age was 60 ± 17 years, 68% were men, and all study participants were on mechanical ventilation ≥24 hours. The possible risk factors for PU development in this study were assessed; first-day respiratory and fourth-day cardiovascular SOFA scores, winter season, and duration of mechanical ventilation were found to significantly affect PU risk, but the total SOFA score was not associated with the risk of PU. 

Currently, only LOS and hypotension/use of vasopressor agents can be considered uniform and clinically meaningful risk indicators for PU development in ICUs. The aim of this retrospective study was to examine the roles of the SOFA scale, mJ/C risk score, and LOS in predicting PU development among different patient groups in an ICU. 

Hospital unit. The Turku University Hospital (Finland)  serves as a tertiary referral hospital for approximately 700 000 individuals from the cities of Turku and Vaasa and their rural areas. The adult ICU, staffed by 160 nurses, has 24 beds and serves as a national center for hyperbaric oxygen therapy. This ICU treats surgical and medical patients needing intensive care or high-dependency care (HDC). Patients are classified on admission by the treating physician according to treatment needs; patients with major burns and organ transplantation are treated elsewhere. The physician makes the admission diagnosis(es) to determine whether a patient requires medical or surgical care and enters these data into the electronic ICU database (Clinisoft, GE Healthcare, Buckinghamshire, UK). 

Nurses with special training in using the mJ/C scale for PU risk assessment³ and in wound identification and care enter relevant information into the database, including type and frequency of skin cleansing, skin integrity, PU presence, and National Pressure Ulcer Advisory Panel/ European Pressure Ulcer Advisory Panel (NPUAP/EPUAP) PU classification. The database has the capability to calculate the mJ/C and SOFA scores.^14,18

Study design. This retrospective cohort study included all adult patients (≥18 years of age) admitted to the ICU from January 2010 to December 2012 (see Table 1). The PUs included Stage 1 to Stage 4 and unstageable ulcers.¹⁹ Patients who had a PU on admission were excluded. The primary endpoint was the development of a PU during the ICU stay. The primary variable was the SOFA score and its subcategories. The secondary variables were the mJ/C score and LOS in the ICU and patient subgroup (ie, medical or surgical and ICU or HDC patients).

Risk assessment instruments.

SOFA. The SOFA scale includes 6 assessment categories: serum bilirubin concentration, platelet count, renal dysfunction, nervous system (GCS) status, hypotension, and presence of respiratory disorder.¹⁸ Each category is scored from 0 (low risk) to 4 (high risk), giving the SOFA score a range from 0 to 24. The SOFA score system is validated; the higher the score, the more severe the patient’s condition and the higher the mortality risk.^20,21

mJ/C. The original Jackson/Cubbin Risk Scale was created to assess PU risk in ICUs.¹⁴ Minor modifications have been introduced (ie, the mJ/C) to improve reproducibility of the original J/C scale in clinical use.^3,14,17 The categories weight/tissue viability, respiration, and incontinence, and the 3 subcategories of deduction points (transport to examinations or treatments, blood product, and hypothermia) were subject to minor modifications.^3,17 The mJ/C scale consists of 12 main categories graded from 1 (high risk) to 4 (low risk) to describe 12 specific variables of the ICU patient’s clinical situation. The minimum score is 9 and the maximum score is 48, with a lower score indicating a higher risk for PUs.^3,14 An electronic version of the mJ/C scale was introduced in 2009 into the clinical database of the ICU of Turku University Hospital for use by the ICU staff.

Patient management. ICU patients are observed according to clinical needs. One (1) nurse is responsible for 1 to 2 patients. Several physiological and laboratory variables (eg, hemodynamic, oxygenation, GCS, hemoglobin, electrolytes, C-reactive protein) are followed continuously every day to ensure proper management of these critically ill patients. The SOFA¹⁸ score is determined at admission and daily thereafter; mJ/C risk scale assessment follows a similar pattern. If the mJ/C score sum is <29 points, the PU risk is high or extremely high.^3,14 As soon as possible (ie, on admission or within 24 hours after risk assessment), high-risk patients are provided a dynamic pressure redistribution mattress if they do not already have one. Otherwise, PU prevention follows general care guidelines¹⁹ in terms of positioning therapy and skin assessment. 

Data extraction and statistical analysis. The predefined data for patient demographics (age, gender, body mass index [BMI]), ICD10 admission diagnoses, ICU type and medical or surgical patients as described in Table 2, LOS, SOFA and mJ/C scores and subscores, hemoglobin concentration, sedation status, body temperature, mattresses used, PU data, and ICU outcome (moved from the ICU as recovering or deceased) were retrieved from the ICU database by the database administrator. The analysis dataset was transferred to SAS version 9.4 (SAS Institute Inc, Cary, NC) for further analyses. SOFA and mJ/C scores were treated as discrete values, whereas LOS data were analyzed into 2 categories (<3 or ≥3 days after admission).²² The first-day SOFA or subcategory scores were used to test whether they predict the development of PUs. If the first-day SOFA score was not available for all subcategories, the subscore was given a value of 0 (ie, normal). If any other information was missing, no data from that patient were included in the analysis (N = 35, see Tables 2–5).

Patients with 1 or more PU, regardless of the stage, were included.¹⁹ The PU incidence (percent) was calculated by dividing the number of patients who developed 1 or more new PUs (irrespective of stage) during their ICU stay by the total number of patients in that group (times 100). The total mJ/C and SOFA scores of ICU and HDC patients and PU and non-PU patients were compared using Wilcoxon’s signed-rank test. The differences in proportions were determined with the χ² test. The association between the SOFA scores and PUs was determined by logistic regression.  

Ethical consideration. The study plan was approved by the Ethics Committee of Hospital District of Southwest Finland (T25/2011, 14.06.2011, §172).

Of the 4899 adult patients admitted in the study time frame, 115 had PUs on admission and were excluded from further analysis. The 4784 study participants included 3017 (63%) men and 1767 women, with a mean patient age of 61.8 ± 15.7 years. Mean BMI was 27.2 ± 5.5 kg/m². More men (264, 8.8%) than women (126, 7.1%) developed a PU (P = .0028). Average LOS in the ICU was 3.6 (range <1–≥ 60) days; the proportion of patients with LOS ≥3 days was 30.3%. PU patients tended to be older (62.2 ± 15.6 years with a PU versus 60.7 ± 15.7 years without; P = .084). Among all patients, 1264 were HDC (26.6%); 1281 (27.0%) of all patients were medical (see Table 2).  PU incidence was 11.2% (120/1071) among medical intensive care patients and 11.9% (25/210) among medical HDC patients. PU incidence was significantly lower among surgical intensive care (164/2414, 6.8%) and surgical HCD (48/1054, 4.6%) patients (P <.0001), respectively (see Table 2). Of the 49.6% of patients at high risk of PUs (mJ/C score ≤29), 59.9% developed PUs (P <.0001, χ² test, data not shown). 

SOFA scores. On admission, the intensive care patients had significantly higher SOFA scores than HDC patients (P <.0001) (see Table 2). Patients who developed PUs had a mean SOFA score of 8.24 ± 3.44; the mean SOFA score of patients who did not develop PUs was 6.74 ± 3.16 (P =.001, 2-sided, Wilcoxon’s signed-rank test). The total SOFA scores on admission predicted the development of PUs (Wald’s χ² test 34.3517, DF 19; P = .0167).

Among all patients (4749), 362 (7.6%) were in the high SOFA score group (≥12 points). The higher the SOFA score, the higher the PU incidence in all patient subgroups (see Table 2). PU incidence in the <6 SOFA score group was 4.5% (74/1635) and 18.2% (66/362) in the SOFA score group ≥12 (P <.0001). Significantly higher SOFA scores (≥12) were noted more among intensive care patients than HDC patients (9% [315/3485] versus 3.7% [47/1264]; P <.0001). Thus, the distribution of patients (PU versus non-PU) among different SOFA score groups and between medical and surgical patients within ICU or HDC groups was statistically different (see Table 2). 

Among all patients, 1441 (30.3%) were treated for ≥3 days. Of the 357 patients with PUs, 304 (85.2%) had a LOS of ≥3 days (P <.0001; see Table 3). The overall PU incidence in the group with a LOS ≥3 days was 21.1% (304/1441) versus 1.6% (53/3343) in LOS <3 days. In the population where LOS was ≥3 days, proportionately more patients with PUs had a higher SOFA score (≥12) (20.7% [63/304]) than those without PUs (13.5% [153/1137]; P =.0037; see Table 3). 

In terms of the SOFA subcategories, PU development was not statistically linked to serum bilirubin concentration (P =.1188; see Table 4), but it was related to the platelet count (P =.0042), the GCS, renal dysfunction, hypotension, and respiratory disorder (P <.0001 for each subcategory). The subcategories bilirubin concentration, platelet count, GSC, renal dysfunction, respiratory disorder, and hypotension were found to have 0 points in 97%, 70%, 66%, 40%, 21%, and 18% of patients, respectively (see Table 4). 

The distribution of patients within the respiratory category (see Table 5) was more even than in the hypotension category, where only 12 out of 4794 patients collected 2 points (data not shown). The worse the oxygenation in terms of the ratio of partial pressure of oxygen in the blood to the fractional inspired oxygen, the higher the incidence of PUs (P <.0001; see Table 5).

mJ/C scores. The mJ/C scores of different patient populations were divided into equal-size groups (median, Q1, and Q3). The intensive care and HDC populations differed regarding mJ/C score distributions (see Table 2), showing that the SOFA scores are valid over a wide range of mJ/C score points. The mJ/C scores of patients with LOS ≥3 days were not different between patients with and without PUs, showing that the various risk factors within the mJ/C score do not influence the interpretation of the importance of different SOFA score groups (see Table 3).

The population of this study represented the typical mixed ICU population with regard to age, male gender dominance, LOS, and SOFA scores.^{3,8,9,11,12,18,22,23} PU incidence was lower than in many previous studies,^{1,2,5,8,9,11,20,21} which is probably due to improved preventive measures introduced in recent years.⁵

PUs have a multifactorial etiology and although many potential risk factors have been identified,^5,7 only a few seem to be independent risk factors in intensive care patients.^6,10 Male gender dominates among ICU-admitted patients and among PU patients with some exceptions.^9,10,12 Older age is a generally accepted risk factor for PU development.^{6,10,11,12,21} Patients in the present study exhibited a trend toward higher age among PU patients, but the mean age differences were small, as has been reported previously in cohort studies.^11,12,22 Thus, it is questionable whether gender and age ultimately affect clinical decision-making related to PU risk. 

Confirming previous reports,^9,12,22,23 in this study medical patients also had a higher rate of ICU-acquired PUs than surgical patients. This was probably due to the fact that medical patients have more comorbidity and are more severely ill than surgical patents, as indicated by their SOFA and mJ/C scores (see Table 2). Irrespective of the patient’s subcategory, the higher the SOFA score, the higher the incidence of PUs. This is not surprising, given that a higher SOFA score indicates multiple organ dysfunction and is associated with increased mortality.^18,20,21 Manzano et al¹² did not identify an association between total SOFA score and PU development, probably due to the rather small number of ventilated patients in that study. In a review of intensive care risk factors by Cox,¹⁰ the association between the severity of illness/mortality risk scales such as SOFA based only on the Manzano et al¹²study or PU risk scales for PU development was not considered sufficient. The present study shows total SOFA scores are associated with PU development, regardless of patient subgroup. Although the specific role of the mJ/C score in the development of PUs was not examined in the current study, patients with PUs had significantly lower mJ/C scores than non-PU patients. This result supports the important role of PU risk assessment in the ICU setting.^4,14,15 

Severely ill patients require longer treatment in the ICU.^1,18,22,24 In the current study, proportionately more patients with an extended LOS (≥3 days) — a uniformly accepted risk factor for PUs^{1,6,8,10,12,22,24} — developed PUs when their SOFA scores were high. This suggests that in addition to prolonged ICU LOS, a high initial total SOFA score is related to the development of PUs. A retrospective cohort study²⁴ found that if the mJ/C scores did not differ markedly between the PU and non-PU patients in the group of patients with LOS ≥3 days, the mJ/C score and the LOS complemented each other when the risk of PU was assessed for ICU patients. 

To the authors’ knowledge, the association between platelet count and plasma bilirubin concentration and PU development has not been studied previously.^5,12 The ability to demonstrate only a trend favoring an association between these variables may have been due to normal platelet counts and bilirubin concentration values or a lack of these data on the first day of patient management at the ICU, because most patients had 0 points in these subcategories on admission.

Frankel et al²⁵ and Nijs et al² reported renal impairment is a risk factor for PU development, and the current study results concur. This contrasts with the study by Manzano et al¹² that did not detect an association between renal failure and PU development. However, in the current study, an association was found between GCS and the development of PUs. Neither Manzano et al¹² nor Cremasco et al¹¹reported GSC data in relation to PUs; still, GCS is part of both the SOFA and SAPS II scales. Neurological failure among non-PU patients in the study by Manzano et al¹² was more common than renal failure among PU patients, possibly related to the fact that the patients were mechanically ventilated and thus immobile.^5,17

Data from the cardiovascular and respiratory SOFA categories were evenly distributed among patients. These categories were the most reliable indicators of PU risk; they also contained the fewest 0-value data points at admission. This is in line with the report by Manzano et al¹² that identified the first-day respiratory SOFA and the fourth-day cardiovascular SOFA subcategories as risk factors for PU development in ventilated ICU patients.  

Tissue oxygenation is important for tissue viability. As such, it was not surprising that the SOFA cardiovascular and respiratory subcategories provided important data with regard to PU risk. Hypotension often is related to cardiovascular disease, and both are regarded as risk factors for PUs.¹⁰Cardiovascular disease represents a broad range of conditions; a 5-fold difference in the incidence of PUs recently was shown among the various groups of cardiovascular diseases.²⁶ Thus, the SOFA cardiovascular category score gives a better indication of the risk of PU than the broad categories of cardiovascular diseases. One of the subcategories of the mJ/C score – hemodynamics – has been significantly associated with PU development.¹⁷ The definitions of the cardiovascular/hemodynamics subcategory are different between the SOFA and mJ/C scales, but both take into account the use of vasopressor agents and the risk of PU.¹⁰  

Mechanical ventilation seems to be only a surrogate risk indicator of PU development in the Manzano et al study¹² because it is used to improve oxygenation.¹⁷ The respiratory subcategory of SOFA considers direct tissue oxygenation and is defined differently from that of the mJ/C scale, but both scales link the respiratory status to PU development.^12,17

The mJ/C risk scale includes 12 main categories, 5 of which correspond to the Braden categories.^3,13 Of the 12 subcategories, 8 (incontinence, mental condition, mobility, medical history, hygiene, oxygen requirements, hemodynamics, and general skin condition) appear to contribute to PU development.¹⁷The Braden Scale does not take into account respiratory/hemodynamic variables.¹³ The present study showed the SOFA score predicts PU development in an unselected cohort of intensive care patients. The total SOFA score is currently recommended to be used as an adjunctive risk scale in addition to either the Braden or the (m)J/C scales.

This study was retrospective by design. Some SOFA score values were not available for the day of admission (this was especially the case for platelets, bilirubin concentration, and the GCS), casting some uncertainty on the interpretation of the ultimate impact of these variables. The authors also cannot ascertain that PU preventive measures were provided as prescribed or that they remained the same throughout the 3-year period of the study. The large sample size of this study probably minimizes the potential impact of preventive measures on the outcomes of this study.

A retrospective examination of the role of the SOFA scale, mJ/C risk score, and LOS in predicting PU development among different patient groups in an ICU found the higher the first-day SOFA scores, the higher the PU incidence. This association was independent of the subgroup of patients treated in the ICU (medical or surgical) and included patients whose LOS was 3 or more days in the ICU. Among the SOFA subcategories, cardiovascular and respiratory scores were the most important for predicting the development of PUs (P <.0001). Therefore, the authors recommend the use of the total SOFA score as an adjunctive risk scale in addition to the either Braden or the mJ/C. If no PU risk assessment scale is in use, the total SOFA score should be used; clinical PU risk assessment by health care professional should always be performed and documented. Additional mu

Statisticians Hanne Koskela and Riku Kivimäki provided valuable contributions for the analysis of the results. The language of the article was reviewed by Robert Paul, MD, PhD, certified translator.

Dr. Soppi was chairman of the board, Carital Group, Finland, a group of companies manufacturing and providing global marketing of mattresses for the prevention and treatment of pressure ulcers, through 2017. Financial support for the study included state research funding (grant 13693), the Turku University Hospital Foundation (year 2016), and the Foundation for Nurse Education.

Ms. Ahtiala is an authorized wound care nurse, Service Division, Perioperative Services, Intensive Care Medicine and Pain Management, Turku University Hospital, Turku, Finland. Dr. Soppi is a senior consultant in internal medicine, Eira Hospital, Helsinki, Finland. Dr. Saari is an associate professor, Department of Anaesthesiology and Intensive Care, University of Turku and Turku University Hospital.

Please address correspondence to: Maarit Ahtiala, RN, Perioperative Services, Intensive Care Medicine and Pain Management, Turku University Hospital, Hämeenkatu 11, FI-20520 Turku, Finland; email: maarit.ahtiala@tyks.fi.