Slings and Their Effect on Skin Pressure

Preventing the development of pressure wounds in patients and preventing staff injuries and a safe work environment has been an ongoing effort at Christiana Care Health System (CCHS). Skin damage on hospitalized patients can be devastating. Not only does it increase patient's hospital stay, but it also causes pain and affects patients quality of life and subsequently costs. One practice in prevention and treatment of wounds is the use of pressure reducing, relieving or pressure redistribution surfaces (mattresses) and repositioning devices (electric lifts). Manufacturers of mattresses emphasize the need to limit layers between surface (mattress) and the patient.

Safe work environments focus on programs that help to ensure the prevention of work related injuries. Injury prevention programs have focused on equipment to eliminate manual lifting in an effort to decrease staff injuries from patient transfers and repositioning. Part of this practice is to leave the sling under the patient in order to reduce the amount of times staff are manually repositioning or turning the patient. With an increasing patient population that is either obese or extremely obese (bariatric patients) positioning and lifting equipment need to manage large weight limits. In response, manufacturers have constructed slings with increased fabric strength to handle heavier patients. There is concern that the newer sling fabrics may interfere with the low airloss and pressure redistribution purpose of mattresses. Information from this study may help in making slings with materials that can support higher weights while not causing any added skin problems.

A convenience sample of 240 adult patients meeting the study criteria will be recruited from the inpatient units at Wilmington Hospitaland Christiana Hospital for this comparison study. After obtaining informed consent, subjects will be randomized by a member of the study team to one of six groups, yielding 40 subjects per group. Groups will be assigned as follows

• Supine and with sling
• 30 degrees HOB (Head of bed)elevation without leg elevation and with sling
• 30 degrees HOB elevation with 10 degrees leg elevation and with sling
• Supine and without sling
• 30 degrees HOB elevation without leg elevation and without sling
• 30 degrees HOB elevation with 10 degrees leg elevation and without sling

Study Methodology

Staff nurses will identify patients who need use of a sling for transfers and repositioning and contact the research team.

A member of the research team will determine patient eligibility, obtain informed consent, and collect demographic information such as

• age
• height
• weight
• BMI, (Body Mass Index)
• gender
• race
• Braden score, a list of risk factors in determining potential skin breakdown
• diagnosis
• activity level

Once consented, the patients sacral skin temperature and skin pH will be measured using a thermometer and nitrazine paper and a control measurement of both temperature and pH will be taken on anterior thigh.

The X sensor or pressure mapping system, will be placed directly under the patient for patients randomized to the no sling groups. For patients randomized to the sling groups, the sling will be placed between the X sensor and the patient. The use of incontinence pads, etc. if needed, will be limited to one and will be noted on data collection tool.

Pressure mapping, temperature and pH will be recorded at one minute baseline and at 20 minutes. Pressure mapping over the sacral area will be recorded and will include the highest, lowest and average measurements.

Data Analysis

The proposed study will have three dependent variables:

• pressure
• temperature
• pH.

All three dependent variables will be on the ratio scales of measure. The goal of the study is to test the null hypothesis that the three dependent variables are equal among six conditions. Subjects will be randomly assigned to the six conditions. Each of the dependent variables will be collected twice. Therefore, the study will incorporate one between subjects factor, condition and one within subjects factor, time period. The between subjects factor will have six levels and the within subjects factor will have two levels.

The traditional approach to analyzing data from the proposed research design is a univariate, repeated measures analysis of variance (ANOVA). Increasingly, random effects or RE models are being employed with data from repeated measures designs (Hedeker \& Gibbons, 1997 and Singer and Willett, 2003). RE models offer two important advantages over ANOVA, (Raudenbush and Bryk, 2002). First, unequal samples sizes such as occurs with missing data do not present the difficulties they impose for ANOVA. Second, it is easy to specify complex models that include multiple random effects. The proposed research model is straightforward and there are only two time periods where data will be collected. In such instances, a third alternative is available: a multivariate repeated measures ANOVA (Tabachnick \& Fidell, 2006). Like RE models, the multivariate approach is robust to violations of the sphericity assumption (Stevens, 2002). Unlike RE models, the analysis is straight forward and results are interpreted according to traditional ANOVA conventions, a time effect, a group effect, and a group by time interaction, which makes them understandable to a wider audience (Maxwell and Delaney 2004). Consequently, data for the proposed study will be analyzed using the multivariate repeated measures ANOVA.

A priori power was estimated for the three, multivariate repeated measures ANOVA for each dependent variable in the study. The proposed analyses will use a two tailed alpha levels set to .05. Overall power was set to .80, meaning the study will have an 80 percent probability of finding a significant difference if such differences exist in the population. Equal sample sizes are assumed. The power analysis concentrated on the multivariate group by time interaction because it usually is the least powerful element in a repeated measures ANOVA (Field, 2005 and Wickens and Keppel, 2004). A medium effect size is anticipated Cohens 1988 f = .25. In addition, a drop out rate of 20 percent is expected. Results show an overall sample size of 240 will be required, 40 subjects per group. This sample size was selected as the smallest sample that would be important to detect.