ADVERTISEMENT
Comparison of Air-Fluidized Therapy with Other Support Surfaces Used to Treat Pressure Ulcers in Nursing Home Residents—Part 3.
Duration on support surfaces. Residents placed on Group 1 surfaces remained on the surface for an average of 30.1 days, as compared to a mean duration of 29.4 days on Group 2 and 43.8 days on Group 3 surfaces (P <0.0001). Duration on a support surface at the episode level was the same for all three support groups because each episode ranged from 7 to 10 days; mean durations were comparable between episodes on Group 1 (7.3 days), Group 2 (7.4 days), and Group 3 (7.3 days) support surfaces.
Severity of illness. Using the CSI®, residents on Group 2 (mean = 108.5) or Group 3 surfaces (mean = 108.1) were significantly sicker than those on Group 1 surfaces (mean = 82.0) (P <0.0001) (see Table 3). The nutritional component of the CSI® at the resident level indicated a significant difference in the nutritional status between residents on Group 1, 2, or 3 support surfaces (P = 0.0013). The Duncan Multiple Range Test indicated that nutritional impairment for residents on Group 3 surfaces was significantly greater than that of residents on Group 1 surfaces (see Table 3).
Similar to the resident level, assessment at the episode level indicated that residents were significantly sicker during the episodes when they were placed on Group 2 (mean = 107.1) or Group 3 (mean = 109.5) surfaces as compared to episodes on Group 1 surfaces (mean = 82.0) (P <0.0001) (see Table 4). Additionally, the nutritional component of the CSI indicated that residents were most nutritionally impaired during episodes on Group 3 surfaces (P <0.0001); Duncan’s Multiple Range Test showed significantly more impairment on Group 3 as compared to Group 2 surfaces, with significantly more impairment on Group 2 as compared with Group 1 surfaces (see Table 3).
Correspondingly, the subset of the episode data, with comparable baseline sizes, showed that residents were more ill during episodes obtained on Group 2 (mean CSI = 105.4) and Group 3 (mean CSI = 111.3) as compared with Group 1 (mean CSI = 85.8) surfaces (P <0.0001). In contrast, significant differences were not found in the nutritional index (see Table 3).
Hospitalizations and emergency room visits. Significant differences in the number of hospitalizations and emergency room visits were found between residents on Group 1 (47 out of 461; 10.2%), Group 2 (23 out of 121; 19.0%), and Group 3 surfaces (6 out of 82; 7.3%) (P = 0.0116). The Chi-square test was used to compare hospitalizations and emergency room visits between the groups. Residents on Group 2 surfaces had significantly more hospitalizations and emergency room visits than did residents on Group 1 (P = 0.008) or on Group 3 (P = 0.0195) surfaces, while no significant difference was found between residents on Group 1 or Group 3 surfaces (P = 0.4184) (see Table 4). Regression analyses, controlling for severity of illness, confirmed fewer hospitalizations with Group 3 as compared with Group 2 surfaces (P = 0.0251). Additionally, an increased severity of illness correlated positively with an increased rate of hospitalizations and emergency room visits (P = 0.0012).
Healing rate.
Resident level. Healing rates, determined at the resident level, were greatest for residents on air-fluidized therapy (Group 3) (mean = 5.2 cm2/week) as compared to those on Group 1 (mean = 1.5 cm2/week) or Group 2 (mean = 1.8 cm2/week) surfaces (ANOVA: P = 0.0071; Duncan’s Multiple Range Test: Group 3 significantly different from Groups 1 and 2). Differences in healing rates on Group 2 versus Group 3 surfaces were significantly different when assessed individually by pressure ulcer stage. Stage I/II pressure ulcers healed more rapidly on Group 3 (8.8 cm2/week) as compared with a Group 1 (mean = 1.6 cm2/week) or Group 2 (mean = 2.4 cm2/week) surfaces (ANOVA: P = 0.0229; Duncan’s Multiple Range Test: Group 3 significantly different from Groups 1 and 2). Similarly, Stage III/IV pressure ulcers healed more rapidly on Group 3 (4.1 cm2/week) as compared with Group 1 (mean = 1.1 cm2/week) or Group 2 (mean = 1.4 cm2/week) surfaces (ANOVA: P = 0.0259; Duncan’s Multiple Range Test: Group 3 significantly different from Groups 1 and 2) (see Table 5).
Episode level. Episode level analyses also demonstrated a higher healing rate for ulcers placed on Group 3 (mean = 2.1 cm2/week) as compared to Group 1 (mean = 0.7 cm2/week) or Group 2 (mean = 0.9 cm2/week) surfaces (ANOVA: P = 0.002; Duncan’s Multiple Range Test: Group 3 significantly different from Groups 1 and 2; paired analysis: Group 2 versus Group 3: P = 0.0735). Statistically significant differences were found for the healing rate of Stage III/IV pressure ulcers on Group 3 (mean = 3.1 cm2/week) as compared to Group 1 (mean = 0.6 cm2/week) or Group 2 (mean = 0.7 cm2/week) surfaces (ANOVA: P = 0.0009; Duncan’s Multiple Range Test: Group 3 significantly different from Groups 1 and 2; paired analysis for Group 2 versus Group 3: P = 0.002).
Analysis of the subset of Stage III/IV pressure ulcers, performed to provide a comparable baseline ulcer size (20 cm2 to 75 cm2) between ulcers on Group 2 and Group 3 surfaces, showed significantly greater healing rates for ulcers on Group 1 (mean = +2.5 cm2/week) and Group 3 (mean = +2.3cm2/week) as compared with those on Group 2 (mean = -2.1 cm2/week) surfaces (ANOVA: P = 0.0399; Duncan’s Multiple Range Test: Group 3 significantly different from Groups 1 and 2; paired testing showed P = 0.1192) (see Table 5).
Hierarchical analyses indicated a lack of bias, attributed to multiple ulcers and/or support surfaces related to the same resident, for all three episode level analyses, including 1) all stages and sizes (P = 0.0026); 2) all stages and sizes controlling statistically for starting area (P = 0.0346); and 3) Stage III/IV ulcers with comparable baseline sizes of 20 cm2 to 75 cm2 (P = 0.0174).
Additional factors affecting healing rates. Regression analyses were performed to assess the potential effects of resident and ulcer variables on healing rates. The following variables were included: age; female gender; patient factors contributing to pressure ulcer formation and healing/comorbid conditions (atherosclerotic vascular disease, cancer, congestive heart failure, coma, deep vein thrombosis, dementia, diabetes, initial ulcer size, multiple sclerosis, Parkinson’s disease, peripheral vascular disease, stroke, history of radiation, limited mobility, paraplegia, quadriplegia, restlessness, severity of illness, and weight); and wound burden factors such as number of pressure ulcers (one, three to five, or >5), initial ulcer size, and pressure ulcer stage. The Braden scale was excluded from the analyses to increase the power of the analysis because the Braden score was missing in 264 out of 664 (39.8%) of the residents.
Resident level. The initial resident level regression analyses (R2 = 0.14) showed a positive correlation with healing for initial pressure ulcer size (P <0.0001; F = 104.6). Because the marked effects of initial pressure ulcer size masked other factors with a potential impact on the healing rate, regression analyses also were performed excluding initial size. Results (R2 = 0.05) showed a positive correlation with healing for Stage I pressure ulcers (P = 0.0008; F = 11.3), Group 3 support surfaces (P = 0.0042; F = 8.3), and three to five pressure ulcers/resident (P = 0.0152; F =5 .9). Quadriplegia showed a negative correlation with healing (P = 0.0428; F = 4.1).
Episode level. Regression data at the episode level (R2 = 0.14) demonstrated a positive correlation with healing for an increased initial ulcer size (P < 0.0001; F = 633.9), female gender (P = 0.0086; F = 6.9), and Stage I ulcers (P = 0.0085; F = 6.9). Negative correlations with healing included Stage IV ulcers (P <0.0001; F = 21.6) and quadriplegia (P = 0.0343; F = 4.4). Omitting initial size from the regression analyses (R2 = 0.01) because the strong correlation masked other significant factors resulted in a positive correlation with healing for Group 3 support surfaces (P = 0.0034; F = 8.6). A negative correlation with healing was found for Stage II ulcers (P = 0.0200; F = 5.4), limited mobility (P = 0.0233; F = 5.2), and quadriplegia (P = 0.045; F = 4.0).
Regression analyses for the subset of Stage III/IV ulcers with a comparable baseline size (R2 = 0.06), including the initial pressure ulcer size, did not show a correlation between initial ulcer size and healing rates. Factors with a positive correlation with healing included female gender (P = 0.0021; F = 9.6), an interaction between Group 1 surfaces and Stage 3 pressure ulcers (P= 0.0187; F = 5.6), and an interaction between Group 3 surfaces and severe illness (P = 0.0486; F = 3.9). Impaired healing was associated with the interaction between Group 2 surfaces and Stage IV pressure ulcers (P = 0.0131; F = 6.2).
Discussion
Published pressure ulcer treatment studies are generally limited in sample size and underpowered41; therefore, most practice guidelines are based on expert opinion rather than empirical evidence.5 Furthermore, current investment levels in pressure ulcer research are minute, especially when compared to the cost of treatment.59,60 To help meet the need for additional information, this study provides empirical data to help assess the effectiveness of pressure relief from support surfaces to heal pressure ulcers through redistribution of the pressure that contributes to the formation and maintenance of these wounds.52,60 In contrast to previous studies, this report includes a greater number of residents and provides a broad range of clinical data in a typical treatment setting.52
Rationale for resident and episode level analyses. This study used two primary analyses to assess healing rates for pressure ulcers of subjects placed on Group 3 support surfaces, or air-fluidized therapy, as compared to healing rates for residents placed on Group 1 or Group 2 surfaces. First, the resident analysis assessed data from a single ulcer for each resident. Although investigators have used various approaches to assess healing rates when subjects have multiple pressure ulcers, such as using the sum of all sizes of wounds61 or selecting the largest and/or most severe wound by stage, this study used the largest pressure ulcer at the time of the initial assessment. This approach avoids bias introduced by repetition of the effects of a resident’s health status variables that may not have been identified in the medical record.
Episode level analyses provided additional information by including data from all pressure ulcers affecting the 49.1% of the residents with more than one ulcer. Of those residents with multiple ulcers, about half had two pressure ulcers, while the other half had between three and 14 ulcers. The episode approach also included data for all support surfaces, including data for those residents who changed support surfaces during the treatment period. Although residents remained on Group 3 surfaces longer than on the other two surfaces, the episode approach provides uniform treatment duration of 7 to 10 days. Finally, the increased amount of episode data allowed researchers to calculate healing rates using a subset of data limited to Stage III/IV ulcers with a comparable baseline size between Group 2 and Group 3 surfaces. Hierarchical analyses indicated that bias was not introduced because of multiple ulcers, or multiple episodes of assessment, from the same resident for overall episode analyses including all sizes and stages, analyses of all sizes of Stage III/IV ulcers, or Stage III/IV ulcers with comparable baseline areas between 20 cm2 and 75 cm2.
Interestingly, both approaches demonstrated improved healing of pressure ulcers on patients place on Group 3 as compared with Group 1 or Group 2 surfaces.
Effects of initial pressure ulcer size. Initial pressure ulcer size is a predictor of pressure treatment outcome.41 However, no consensus exists for the most reliable and consistent way to analyze pressure ulcer healing rates. The literature reports a variety of approaches, including time to complete healing, rate of change in the area or volume, and the percentage change from baseline.62 Calculating the change in area tends to exaggerate the healing of larger wounds, while accounting for baseline area by calculating the percent reduction exaggerates the healing of smaller wounds.41 This problem is best solved with a prospective study that matches wound sizes between groups at the onset of the study. Alternatively, this study analyzed pressure ulcer outcomes by grouping Stage III/IV ulcers with comparable mean initial baseline size for episodes on Group 2 and Group 3 support surfaces. Using this approach, the healing rate remained significantly greater for residents on Group 3 support surfaces, as compared with those on Group 2 surfaces. This difference remained even though the illness severity for residents on both Group 2 and Group 3 support surfaces was similar. Regression analyses indicated that the initial pressure ulcer size did not correlate with the healing rate, while placement on a Group 2 surface correlated with impaired healing.
Baseline differences between groups.
Wound burden. Overall, the wound burden was significantly greater for residents on Group 3 surfaces (mean = 3.5 pressure ulcers/resident). Additionally, pressure ulcers were most severe for residents on Group 3 (more Stage IV and larger pressure ulcers), as compared with Group 1 or Group 2 surfaces (see Tables 1 and 2). This finding is consistent with the expectation that the most severely affected residents are placed on the more sophisticated specialty beds. Despite an increased wound burden, healing rates were significantly higher for residents on Group 3 as compared with Group 2 surfaces. Although improved healing rates, at the resident level, were still significantly greater for Stage I/II ulcers on Group 3, as compared with Group 1 and Group 2 surfaces, the number of residents with Stage I/II ulcers on Group 3 surfaces was low (N = 18) as compared to the more robust number of residents with Stage III/IV ulcers at baseline.
Baseline differences between subsets of Group 1 and Group 2 support surfaces. Study data did not discern any significant differences in the healing rates for residents placed on subsets of Group 1 (alternating pressure or foam or water or gel surfaces) or Group 2 surfaces (powered air floatation or low-air-loss, powered pressure-reducing, powered air overlay, or non-powered advanced pressure surfaces) using both resident level and episode level analyses. The data from the literature that compares the effectiveness of different Group 1 or Group 2 support surfaces on the healing of pressure ulcers is limited. However, results of a recent meta-analysis confirmed that “The relative merits of alternating and constant low pressure devices and of the different alternating pressure devices for pressure ulcer prevention are unclear.”63
Severity of illness. This study confirms that specialized beds in Group 2 and Group 3 (air-fluidized therapy) are generally reserved for the most seriously ill patients. Even though residents on Group 2 or Group 3 surfaces were significantly sicker than those on Group 1 support surfaces, residents on air-fluidized therapy exhibited higher healing rates than those on other surfaces. The improved healing with air-fluidized therapy was confirmed by regression analyses of both the resident and episode data, including data controlled for initial pressure ulcer size. These findings are consistent with the improved healing with air-fluidized therapy noted in randomized trials, as well as in case reports, described in the literature review section.
Nutritional status. Adequate nutritional status plays a crucial role in healing pressure ulcers because the inflammatory process within the pressure ulcer induces a catabolic state that requires additional calories. This nutritional impairment is exacerbated by the loss of serum protein in exudate drainage, especially with large wounds.60 The nutritional index scores were significantly higher (more impaired nutrition) for residents on Group 2 and Group 3 versus Group 1 surfaces (see Table 3). Distinctions between these groups became more marked at the episode level, possibly because of the increased power from multiple units of measurement. Statistically significant differences were shown at the episode level between all three groups of support surfaces; the most severe impairment was found with Group 3 surfaces, with the best level of nutrition on Group 1 surfaces (see Table 3).
This impaired nutritional status mirrors the increased illness severity and would be expected to predict impaired healing of residents on Group 2 or Group 3 surfaces. However, pressure ulcer healing rates of residents on Group 3 surfaces were higher than those on Group 2 and Group 1 surfaces. The impaired nutritional index in the overall episode data reflects that of resident analyses. However, in the subset of episode data with a comparable baseline size, nutritional status, while numerically more impaired on Group 2 and Group 3 surfaces as compared with Group 1 surfaces, did not show statistically significant differences.
Demographic factors. Residents on Group 3 surfaces were significantly younger than those on Group 1 or Group 2 surfaces (see Table 1). Although increased age is thought to adversely affect healing, age did not appear to significantly affect the healing rate in the regression analyses. Additionally, baseline differences in gender existed between the groups; more female residents were on Group 1 or Group 2 surfaces, while slightly more men were on Group 3 surfaces. Regression analyses at the resident level did not identify gender as a factor that affected the healing rate. Analyses at the episode level identified female gender as a factor with positive correlation with healing. Interestingly, only the Group 3 surface residents had a greater percentage of men. Therefore, based on the regression analysis, the influence of gender would be consistent with an impaired healing rate for residents on Group 3 surfaces. In contrast, the healing rate was significantly better for residents on Group 3 as compared to Group 1 or Group 2 surfaces. In addition, while women are generally thought to be at greater risk than men for the development and maintenance of pressure ulcers, improved healing of venous ulcers has been reported in women.64
Regression analyses. Regression analyses provide some evidence of factors related to pressure ulcer healing, although the R2 values indicate a relatively small impact of factors included in the analyses. While results varied to some extent between the data sets examined, initial size and Stage I ulcers showed a strong positive correlation with healing; whereas, quadriplegia showed a consistent negative correlation with healing — findings consistent with clinical experience. In addition, resident level analysis showed a positive correlation between healing and use of Group 3 surfaces and with multiple pressure ulcers/resident. Although improved healing on a Group 3 surface is consistent with the literature, the correlation between multiple pressure ulcers and improved healing seems counterintuitive. However, this finding may be related to the strong correlation between multiple ulcers/resident and severity of illness associated with greater use of Group 3 surfaces.
Additional findings from regression analyses of episode data included a positive correlation between healing and female gender (described above) and a strong negative correlation with Stage IV pressure ulcers. The strong negative correlation between a Stage IV ulcer and healing is consistent with the literature. Additionally, the improved healing on Group 3 surfaces occurred despite a significantly greater number of Stage IV ulcers on Group 3 as compared with Group 2 or Group 1 surfaces. Analyses excluding initial area also identified impaired healing for Stage II ulcers and for residents with limited mobility. Impaired healing with limited mobility is, again, consistent with clinical experience. The impaired healing with Stage II ulcers may reflect the greater proportion of Stage II ulcers on Group 1 as compared with those on Group 2 or Group 3 surfaces (see Table 2).
Regression analyses of the subset of episode data used for the comparable baseline analysis also showed a positive correlation with female gender, despite a greater number of episodes assessed from male residents placed on Group 1 and Group 3 surfaces. Again, this finding may reflect a correlation between improved healing and the use of Group 3 as compared with Group 2 surfaces in this subset. Additionally, a correlation with improved healing was found for the interaction between Group 1 surfaces and Stage III ulcers, possibly reflecting the decreased healing rate on Group 2 surfaces and the relatively low number of Stage III ulcers on Group 3 surfaces. Finally, a positive correlation for increased healing was found for interaction between Group 3 surfaces and a greater illness severity. This finding is consistent with improved healing for residents placed on Group 3 surfaces despite their high illness severity. On the other hand, a negative correlation with healing was found for an interaction between Group 2 surfaces and Stage IV ulcers. This interaction correlates with the negative healing rate for ulcers on Group 2 surfaces within this subset of episode data.
Hospitalizations and emergency room visits. Fewer hospitalizations and emergency room visits occurred for residents placed on air-fluidized therapy, or Group 3, as compared with residents placed on Group 2 surfaces. This decreased rate occurred even though residents on both Group 2 and Group 3 support surfaces had similar levels of illness severity and nutritional impairment. Additionally, residents on Group 3 surfaces had the largest ulcers, with a significantly greater number of Stage IV pressure ulcers (see Table 1 and Table 2). Furthermore, regression analyses identified severity of illness as a factor associated with increased hospitalizations or emergency room visits. The increased rate of hospitalizations for residents on Group 2 as compared to Group 1 surfaces is consistent with the greater illness in residents on Group 2 as compared to those on Group 1 surfaces (P <0.0001). In contrast, a lower rate of hospitalizations or emergency room visits occurred for residents on Group 3 surfaces — a rate similar to that of residents on Group 1 surfaces, despite the significantly greater illness severity and nutritional impairment found in residents on Group 3 as compared to Group 1 surfaces. Possibly, physiological improvements, in addition to pressure relief, not only improved the pressure ulcer healing rate, but also contributed to fewer hospitalizations and emergency room for residents on Group 3 surfaces.
Limitations
Most of the study limitations result from the use of retrospective data — ie, the original intent of resident care in the long-term care facilities was not to evaluate the relationship between support surfaces and the healing rate of pressure ulcers. For example, several parameters that affect healing were not included in the regression analyses, either because of missing data or because the timing of data collection could not be accurately correlated with pressure ulcer size measurements used to determine healing rates. These parameters include debridement, the level of continence, pressure ulcer infection, and the effect of different dressings. Debridement was not included because the recorded time of debridement usually differed from the time of pressure ulcer size measurements. The level of continence was not included because of difficulties in determining the actual state of continence from the records. For example, residents with external catheters were deemed continent, although these catheters frequently leak. Pressure ulcer infection was documented too infrequently to provide meaningful data. Finally, dressings were not included because of the difficulty in correlating the timing of dressing placement with pressure ulcer measurements and because of the complexity of sorting out the scores of dressings used by type and function. Furthermore, the terms used to describe dressings are confusing, while functional differences between dressings have not been clinically validated.65
In addition, a prospective study would have used exclusion criteria to control for variables such as baseline size, severity of illness, and nutritional status. Instead, this study controlled for variability using statistical analyses to demonstrate the effectiveness of Group 3 support surfaces to heal pressure ulcers.
Another potential concern is the accuracy of pressure ulcer staging by clinicians.66 However, experts acknowledge that clinician ability to stage pressure ulcers improved for ulcers at Stage II and greater.41 Because the data from this study indicate improved healing on Group 3 surfaces for pressure ulcers at, or greater than, Stage III, the conclusions relating to the more severe pressure ulcers are most likely to be accurately recorded. For this reason, analyses to adjust for the initial size between support surface groups were limited to Stage III/IV pressure ulcers.
The importance of providing optimal nutrition to support pressure ulcer treatment is well recognized.67 While a nutritional assessment was available in the existing database, an optimal study would correlate changes in the pre-albumin levels with both pressure ulcer size measurements and with the support surface used. Pre-albumin is preferable to albumin because the 2-day half-life of pre-albumin reflects rapid changes in nutritional status when compared with the 20-day half-life of albumin.68,69 Additionally, Church and Hill70 found that pre-albumin most accurately correlated with nitrogen balance. Although some albumin levels were available, most of the values were not obtained at the same time as pressure ulcer measurements on a designated support surface.
Conclusion
Despite limitations in this retrospective study, a statistically significantly greater healing rate, particularly for Stage III/IV pressure ulcers, was found for residents of long-term care facilities placed on Group 3 as compared to Group 1 or Group 2 support surfaces. This increased healing rate was found using both an analysis of individual residents and an analysis of episode data. Furthermore, improved healing on Group 3 as compared with Group 2 support surfaces was found using episode data selected to provide a comparable baseline ulcer size between Group 2 and Group 3 surfaces. These findings are consistent with the literature to date and occurred despite the greater severity of illness, greater wound burden, and more impaired nutrition found in residents placed on Group 3 or Group 2 support surfaces as compared to those placed on Group 1 surfaces. The impaired healing on Group 2 surfaces is probably related to the significant severity of illness in those patients (equivalent to those on a Group 3 surface, but significantly more impaired than those on a Group 1 surface). Additionally, significantly fewer residents on Group 3 as compared to Group 2 support surfaces had hospitalizations and emergency room visits, although residents on both types of support surfaces demonstrated a comparable marked severity of illness and high degree of nutritional impairment.
Future research should supplement these findings with a prospective trial to control for many variables including initial pressure ulcer size, use of dressings, effect of debridement, pressure ulcer infection, and level of continence. Confirmation of a reduced rate of hospitalization and emergency room visits would be particularly useful in a controlled setting. Additionally, a controlled assessment of the relationship between support surfaces, the rate of pressure ulcer healing, and both the illness severity and nutritional status as measured by pre-albumin levels, may help determine additional physiological benefits of air-fluidized therapy.
For clinicians, it is important to note that in clinical practice, very sick patients are often placed on Group 3 surfaces. It also was found that, despite their level of illness, residents, especially those with Stage III/IV pressure ulcers, heal better on air-fluidized therapy as compared with Group 2 support surfaces. Additionally, patients placed on Group 3 surfaces (air-fluidized therapy) may require fewer hospitalizations than patients placed on Group 2 support surfaces.
