Determining Risk Factors to Develop a Predictive Model of Incontinence-associated Dermatitis Among Critically Ill Patients with Fecal Incontinence: A Prospective, Quantitative Study

Critically ill patients with fecal incontinence are at high risk of developing incontinence-associated dermatitis (IAD). Scientific prediction and prevention of IAD are essential. PURPOSE: The purpose of this study was to determine the risk factors for IAD among critically ill patients with fecal incontinence. Based on this information, a predictive risk assessment model was developed to provide research evidence for IAD prevention. METHODS: A prospective study was conducted from October 2016 to December 2017. Convenience sampling was used to recruit patients with fecal incontinence treated in intensive care units (ICUs) at Tianjin Medical University General Hospital in China. Trained nurses collected demographic data (age, gender, and ICU type), data related to fecal incontinence (Perineal Assessment Tool [PAT] scores, bowel movement frequency, and stool traits per the Bristol Stool Scale), and clinical data (length of ICU hospitalization, body temperature, diabetes history, hypertension history, consciousness, nutrition support, oxygen supply, number of antibiotic species, sedative use, and albumin levels) from participants and their medical records. The PAT was used to assess patient risk of developing IAD, and the Bristol Stool Scale was used to assess patient stool traits. Names were coded anonymously, and data were entered from paper-and-pencil questionnaires into a software program for statistical analysis. Univariate analysis and multivariate logistic regression were performed to identify risk factors for IAD. A predictive risk factor model was established using a receiver operating characteristic curve. RESULTS: Among 266 critically ill patients with fecal incontinence (182 male, 84 female; mean age 64.18 ± 17.10), IAD incidence was 65.4%. The use of sedative drugs, coma status, higher PAT score, more frequent bowel movements, and loose stool were found to be independent risk factors for IAD (P <.05). The subsequent risk factor predictive model had a sensitivity and specificity of 99.4% and 96.7%, respectively, and the agreement rate was 98.1%. CONCLUSION: The identified risk factors and subsequent predictive model may contribute to timely identification and quantitative risk assessment of IAD among critically ill patients. Additional quantitative research could provide a scientific basis for the development of specific preventive interventions.

Maintaining skin integrity is a major challenge in caring for critically ill patients. Incontinence-associated dermatitis (IAD), a skin inflammation caused by prolonged exposure to urine and/or feces, is characterized by a rash or blisters on the skin surface, serous exudation, erosion, and secondary infection.¹

A literature review² has shown fecal incontinence (ie, involuntary discharge of solid or liquid feces) is a proven risk factor in IAD development because feces contains proteases and lipases and impairs the skin’s protective function. Jia et al³ found the incidence of IAD among Chinese hospitalized incontinence patients with fecal incontinence (11.6%) was higher than those with urinary continence (4.9%). Bliss et al,⁴ Bayón Garcia  et al,⁵ and Drive⁶ reported the incidence of fecal incontinence dermatitis was 20% among hospitalized patients; the incidence of IAD among intensive care unit (ICU) patients ranged from 50% to 86.7%. A comprehensive literature review⁷ found the high incidence of IAD among ICU patients can lead to adverse consequences such as lower quality of life, prolonged hospital stay, and increased medical costs and nursing staff workload. Therefore, special attention should be paid to critically ill patients with IAD. 

Risk factors for IAD development should be identified as early as possible to provide effective and timely prevention. The conceptual framework presented by Brown⁸ suggested 3 major risk factors for IAD: tissue tolerance, perineal environment, and patient mobility (including the ability to go to the toilet and toilet awareness). In his review of the pathophysiology of IAD development, Gray⁹ classified the risk factors for IAD into 6 categories: chronic exposure to wet conditions, incontinence, restriction of device use, alkaline pH, pathogen infection, and friction. According to Brown⁸ and Gray,⁹ other identified risk factors include skin condition, aging, pain, fever, and decreased activity.¹⁰ 

Considering the potential size of the study population, research on risk factors for IAD among critically ill patients with fecal incontinence is scant. Wang et al¹¹ explored the incidence and risk factors of IAD among 109 Chinese ICU patients and found lower Braden Scale score, a lower serum albumin level, and double incontinence to be risk factors for IAD. Cross-sectional research by Van Damme et al¹² identified 7 independent risk factors for IAD development: liquid stool, diabetes, age, smoking, nonuse of diapers, fever, and low oxygen saturation. According to Zhang et al¹³ and Churpek et al,¹⁴ prospective study that includes a predictive risk factor model can examine the relationship between risk factors and calculate the risk for IAD. 

A cross-sectional survey analysis¹⁵ found that although the incidence of IAD among critically ill patients is high, Chinese clinicians usually pay more attention to basic disease treatment and ignore IAD risk factor assessment in this population. In addition, in an investigation by Zhang and Wang,¹⁶ knowledge of IAD prevention among ICU nurses in China has been found to be lacking. 

Because thorough inquiry could establish a scientific clinical foundation for IAD prevention among critically ill Chinese patients, the aim of this study was to prospectively explore the risk factors for IAD among critically ill patients with fecal incontinence and to establish a predictive risk factor model for appropriate, timely, and specific preventive interventions.

Study design. A prospective, quantitative study was performed to analyze characteristics of patients with IAD and fecal incontinence treated in the ICU. 

Sample and setting. This study was conducted at the 3000-bed Tianjin Medical University General Hospital (Tianjin, China) from October 2016 to December 2017. A total of 266 patients was recruited from 4 ICUs (general, neurological [NICU], rehabilitation, and gerontology) using convenience sampling methods. Because the patients in these units often have complex conditions with associated immobility and incontinence, they are at high risk of developing IAD according to epidemiological data.^17-20

Inclusion criteria stipulated participants must be at least 18 years old, without IAD or a pressure injury at admission, have fecal incontinence or urinary incontinence and using an indwelling catheter, and able to provide informed consent (from patient and relative). Patients hospitalized <7 days, experiencing urine leakage despite an indwelling catheter, unable to change position, or who refused study participation were excluded. 

Ethical consideration. The study purpose and method were explained to the patients, and their informed consent was obtained. When it was difficult to obtain informed consent from the patient, it was obtained from the patient’s legal guardian. The study was approved by the Ethics Committees of the General Hospital of Tianjin Medical University (IRB2018-YX-081).

Research variables. Demographic data (age, gender, and ICU type) were collected from the patients’ medical records on admission. Data on length of hospitalization in the ICU were collected after discharge. Data related to fecal incontinence (Perineal Assessment Tool [PAT] scores, bowel movement frequency, and stool traits per the Bristol Stool Scale) were obtained within 24 hours after the first episode of fecal incontinence. Clinical data (body temperature, diabetes history, hypertension history, consciousness, nutritional support, oxygen supply, number of antibiotic species, sedative use, and albumin levels) were collected at the same time as the fecal assessments. Specifically, axillary body temperature was obtained and recorded at 8:00 am for patients with fecal incontinence. Fever was defined as temperature ≥37˚C (normal range is 36˚C to 37˚C²¹). Information on diabetes history, hypertension, and sedative use was collected from clinician records. According to Yang et al,²² antibiotic use can lead to diarrhea, so the number of antibiotic species (based on the treatment plan and the pharmacopoeia) was included in the risk factor assessment. Data on consciousness, nutritional support, and oxygen supply were collected from medical records. Hypertension was defined as follows: first-grade hypertension was noted as systolic blood pressure of 140 mm Hg to 159 mm Hg and/or diastolic blood pressure of 91 mm Hg to 99 mm Hg; secondary hypertension was noted as systolic blood pressure of 160 mm Hg to 179 mm Hg and/or diastolic blood pressure of 100 mm Hg to 109 mm Hg; and tertiary hypertension was defined as systolic blood pressure 180 mm Hg and/or diastolic blood pressure ≥110 mm Hg.²³ Patient albumin level <35g/L 24 hours after the first episode of fecal incontinence was defined as “less than normal” (lower than the normal range of 35 g/L to 50 g/L).²⁴

Care and data collection procedures. A paper-and-pencil questionnaire was used to collect sociodemographic and clinical data. Patient names were coded anonymously. Data collection was completed by 5 nurses; a qualified wound and ostomy therapist was in charge of supervision and 4 experienced ICU nurses from each ICU were responsible for collecting data. At the beginning of the study, the 5 nurse investigators simultaneously were trained on variable measurement and how to use the assessment tools. The 4 investigators in each unit conducted a daily evaluation of all patients to help ensure accurate data acquisition. If an incident of fecal incontinence occurred, the study investigators recorded the research data within 24 hours.²⁵ The qualified wound and ostomy therapist was responsible for supervising the evaluation to ensure data quality.

Patients with fecal incontinence were provided standard incontinence care from trained nurses that included cleaning the area contaminated by feces using warm water and changing the diaper and/or contaminated bed units, along with regular cleaning of the perineal area. The nurses on duty routinely performed skin assessment and care every 2 hours for all patients.²⁶ 

Instruments.

PAT. The PAT is a predictive tool for assessing the risk of developing IAD in patients with incontinence.²⁷ The PAT evaluates risk according to 4 factors and each factor is assessed using 1 to 3 points: the type and intensity of the irritant (1 = formed feces and/or urine, 2 = soft feces and/or urine, 3 = liquid feces and/or urine), the specific amount of time the skin is exposed to irritant (1/2/3: sheets/diapers replaced at least every 8/4/2 hours), the specific condition of genital skin (1 = clean and complete, 2 = erythema/dermatitis, 3 = exfoliation and erosion of skin), and related factors leading to diarrhea (1/2/3: zero to 1/2/3 or more related factors). The total score ranges from 4 to 12. Higher scores represent a greater risk of developing IAD. The PAT score can be divided into low risk (total score <7) and high risk levels (total score ≥7). 

Xie et al²⁸ reported the Cronbach-α value and interrater reliability of the PAT were 0.512 and 0.882, respectively, among Chinese incontinent patients.

Bristol Stool Scale. The Bristol Stool classification, developed by Lewis and Heaton,²⁹ divides stool into 7 types. The first and the second type indicate constipation. The third and the fourth type indicate ideal shape (soft, sausage-shaped, and easy to defecate). The fifth (soft, semisolid, small, uneven edges), sixth (fluffy pieces with ragged edges or mushy), and seventh types (watery/no solid pieces/entirely liquid) indicate the possibility of diarrhea.²⁹ A reliability and validity survey by Blake et al³⁰ concluded this scale showed substantial validity and reliability.

The investigators evaluated the stool traits using Bristol scale reference pictures. The stool trait was identified as the most common recurring stool type for each patient. The frequency of patients’ bowel movements was observed over 24 hours after the first episode of fecal incontinence. 

Data analysis. All data were entered into SPSS, version 20.0 (IBM Corporation, Armonk, NY) software. Descriptive statistics were used to describe frequency, percentage, mean, and standard deviation. Univariate analysis, including the chi-squared test and 2 independent sample t tests, were performed to detect differences among characteristics. Researchers compared the difference between patients with and without IAD among the following factors: age, gender, body temperature, diabetes history, ICU hospitalization, ICU type, coma, sedative use, hypertension history, nutritional support, oxygen supply, number of antibiotic species, PAT score, albumin level, frequency of bowel movements, and stool traits. Binary classification logistic regression was used to analyze the independent risk factors for IAD if a significant difference was noted in univariate analysis. Logistic regression was applied as needed based on model assumptions, and odds ratios were used to quantify the magnitude and direction of any significant associations. 

The significant risk factors (P <.05) from bivariate analysis then were included in a process of stepwise selection to determine the group of strictly significant variables³¹; forward selection was utilized to determine a predictive model. The sensitivity, specificity, and accuracy of the model were analyzed, and the prediction effect of the model was evaluated using the receiver operating characteristic (ROC) curve. All analyses were 2-sided, with a significance level of .05. 

A total of 266 critically ill patients with fecal incontinence (182 male, 84 female; mean age for all participants 64.18 ± 17.10 years) were included in this study, with none lost to follow-up. The incidence of IAD was 65.4%. The mean time from presentation of fecal incontinence to the development of IAD was 3.95 ± 3.47 days. The mean length of hospitalization in the ICU was 22.63 ± 17.10 (range 7–94) days. Stool traits assessed ranged from the third to the seventh type. The investigators classified the third and fourth type as normal because of their similar (ideal) shape.

Risk factors. When comparing the incidence of IAD among patients of different genders, body temperature, and diabetes history, no statistically significant differences were noted (P >.05). However, the incidence of IAD was significantly higher in older patients in the NICU and general ICU; patients with longer ICU hospitalization, serious hypertension history, and enteral and parenteral nutritional support; patients in a coma; patients who were provided sedative drugs, invasive oxygen supply, and more species of antibiotics; and patients having lower albumin levels, higher PAT scores, more frequent bowel movements, and loose stool (P <.05). The results are listed in Table 1 Part 1 and Table 1 Part 2.

Multivariate logistic regression analysis. The factors of statistical significance in univariate analysis were analyzed by binary logistic regression analysis. Coma status, use of sedative drugs, higher PAT score, more frequent bowel movements, and loose stools were independent risk factors of IAD (P <.05) (see Table 2 and Table 3).

Model establishment and predictive analysis. A logistic regression predictive model was established according to the results of multivariate logistic regression analysis. Prob = 1/1+e^-z,  P defined value is 0.5, P ≥.5 means the occurrence of IAD. The predictive model is listed as follows: Z=6.012+1.649×X10+1.463×X8+1.235×X11+ 1.295×X12+2.871×X13+6.835×X14+3.500× X15+3.908 × X16+3.933 × X17. According to the model, sensitivity was found to be 99.4%, specificity was 96.7%, and the agreement rate was 98.1%. The area under the ROC was 0.872; the model has an ideal predictive value (see Table 4 and Figure).

Fecal incontinence and IAD among critically ill patients. The incidence of IAD was 65.4% in this study, compared with rates ranging from 21.35% to 50% in previous research.^6,25 For example, Driver⁶ evaluated 131 patients with fecal incontinence in a critical care unit and reported a 50% prevalence of IAD. Fecal incontinence is also common in the acute care setting. Junkin and Selekof²⁵ surveyed 608 patients in the emergency ward and found fecal incontinence was more common than urinary incontinence overall and associated with a higher rate of skin injury. A cross-sectional study on IAD by Campbell et al³² involving long-term care institutions reported 42% of patients with incontinence had IAD. This previous research and the current study indicated a high incidence of IAD.

Analysis of risk factors among critically ill patients with fecal incontinence. Univariate analysis found that ICU type, length of ICU hospitalization, age, coma, sedative use, nutritional support, oxygen supply, hypertension, number of antibiotic species, PAT score, albumin level, bowel movement frequency, and stool traits were of statistical significance in the occurrence of IAD , which was in line with the results of Junkin and Selekof.²⁵ Kottner et al³³ found high body mass index is a key risk factor for IAD development in patients with diabetes. However, diabetes was not related to the occurrence of IAD in the current study. Because of the small sample sizes for previously published studies, selection bias may have led to different results. According to Xie et al,²⁸ using more species of antibiotics was related to higher risk of developing IAD, which also was in line with the current study. 

The logistic regression predictive model of this study showed coma, sedative drug use, higher PAT score, more frequent bowel movements, and loose stool were independent risk factors for IAD. Level of consciousness correlated with the risk of developing IAD in this study; the risk of developing IAD was 4.3 times higher among comatose patients than among non-coma patients. Bliss et al⁴ demonstrated that patients with poor self-movement ability, low levels of consciousness, and diminished sensory function are more prone to develop IAD. Patients taking sedative medications were 5 times more likely to develop IAD than those who did not. Due to the use of sedative drugs, patients may have complications such as paralysis and a decrease in reactivity to stimuli and movement, which can increase the risk of developing IAD. Comatose critically ill patients and/or patients provided sedative drugs should be observed carefully to reduce the occurrence of injuries.

Patients with PAT ≥7 had a 3.4 times higher risk of developing IAD compared with those with PAT <7 points. Xie et al²⁸ supported these findings and showed persons with PAT scores >7 were more likely to develop IAD. 

The investigative study by Bliss et al³⁴ supported current study findings with regard to bowel movement frequency. The current study also showed that patients with fifth type stool traits (soft, semisolid, small, uneven edges) had a 39-fold increased risk of developing IAD when compared with patients with normal traits, a 49-fold increased risk of developing IAD in patients with the sixth type stool traits (fluffy pieces with ragged edges or mushy), and a 51-fold increased risk of developing IAD in patients with the seventh type stool traits (watery/no solid pieces/entirely liquid). The more watery the stool, the higher the risk of developing IAD; watery feces contain more bile salts and pancreatic lipase, have a larger contact area with skin, and result in greater skin damage.² Tian et al¹⁹ found stool traits had an impact on occurrence of IAD; the specific effects of different stool traits on risk of developing IAD were not compared. 

The greater the frequency of bowel movements, the more watery the stool tended to be, which suggested nurses need to remove the irritant in a timely manner to prevent the occurrence of IAD. In their investigative study, Xu et al³⁵ reported bowel movement frequency and watery stool were risk factors for IAD, although these authors did not analyze the skin damage impact of different stool traits.

This was a cross-sectional study and single-centered. Future research needs to include multicenter studies with larger sample sizes. Due to the use of the same cleaning methods, this study did not explore the impact of different postincontinence episode cleaning methods/products on the occurrence of IAD, which could have affected outcomes. In addition, not all of the variables contributing to IAD were included in the study, such as skin pH and environmental moisture, owing to instrumental limitation, which may affect the clinical applicability of the predictive model. A predictive risk factor model with more comprehensive risk factors is needed.

The results of this prospective study among 266 ICU patients show the use of sedative drugs, coma status, higher PAT score, greater bowel movement frequency, and loose stool to be independent risk factors for IAD among critically ill patients with fecal incontinence. Based on the results, a predictive risk factor model was developed to be used as a quantitative IAD risk assessment tool among critically ill patients. At the same time, the research results provide a scientific basis for quantitative prediction and the development of specific preventive interventions for IAD. 

The authors are grateful to patients and nurses who participated in this study.

Ms. Wei is a wound therapist and Director of the nursing department, Tianjin Medical University General Hospital Airport Hospital, Dongli District, Tianjin, China. Ms. Bao is a registered nurse, intensive care unit, Tianjin Medical University Cancer Institute and Hospital, Hexi District, Tianjin, China. Ms. Chai is a registered nurse, gastroenterology department; Ms. Zheng and Ms. Xuare registered nurses, intensive care unit, Tianjin Medical University General Hospital. 

This study was supported by graduate research fund from Tianjin Medical University, Tianjin, China.  

Please address correspondence to: Li Wei, Tianjin Medical University General Hospital Airport Hospital, No. 85, Dongliu Road, Airport Economic Zone, Dongli District, Tianjin, China 300308; email: weili066100@126.com.

Fecal Incontinence in Acutely and Critically Ill Patients: Options in Management

Improved Quality of Life After Ostomy Surgery