Tempur-Pedic® Medical Mattresses Heal Wounds: A Retrospective Multi-site Study
Contemporary Topics in Skin, Wound, Ostomy, and Incontinence Care September 2000, Vol. 46, Issue 9
Literature Review
In an effort to confirm these impressions and investigate study design precedence, a literature search was conducted primarily through Medline and augmented with additional periodical and electronic resources. Concerns about funding for support surfaces are not unfounded. The Health Care Financing Administration commissioned a series of studies to investigate support surface efficacy to determine if the additional cost of advanced support surfaces resulted in better outcomes. This analysis of the literature and manufacturers’ claims led to a determination that although support surface use does tend to improve outcomes compared with treatment provided without a support surface, there is no evidence that more expensive support surfaces are capable of producing better outcomes than less expensive alternatives. This determination has led to an effort to review support coverage criteria, especially for the most costly support surface group, but this effect has been complicated by the lack of compelling efficacy data.
The Agency for Health Care Policy and Research (AHCPR) Guidelines,8 published in 1993, criticized the lack of studies proving support surface efficacy. In an effort to see if research conducted after the release of the Guidelines improved this situation, Maklebust reviewed the medical literature for support surface research. She confirmed that little had been done to advance the state of support surface literature in the past 6 years and cited a number of references that reinforced the need for rigorous research, despite the high costs, while acknowledging that funding for these studies is not forthcoming. Ultimately, the literature implied that clinicians must continue to attempt to subjectively match a support surface to the needs of the patient.
These observations aside, there is no paucity of support surface research. The literature contains hundreds of studies investigating the appropriate use of support surfaces along with various measures of performance. Most literature presents, sometimes as comparatives, intermediate measures of performance that include laboratory measurements of interface pressure, mattress cover vapor permeability, mattress cover friction coefficient, shear force, or other physical measurement. Although these factors are important in predicting whether or not a support surface has the potential to work, and may give clues as to effectiveness, they do not support efficacy to an extent necessary to quell cost/benefit concerns.
The following question arises: If intermediate measures are insufficient measures to support efficacy, would clinical outcomes prove a more reliable indicator? The answer may be a firm, maybe. The list of wound management outcomes can be lengthy and complex; whether all pressure ulcers or other ischemic wounds can be healed remains questionable. It is reasonable that support surfaces will not contribute to every good wound healing outcome, nor should a non-healing ulcer be blamed solely on support surface ineffectiveness. Regardless, the most effective clinical indicators for evaluating support surface efficacy may be assessing wound incidence and wound healing rates.
Purpose
Payers and providers are demanding compelling evidence of support surface performance. Meanwhile, funding constraints, expediency, and subject availability have a cumulative impact on a manufacturer’s ability to provide what is promised. With these issues in mind, it was theorized that truly compelling evidence could be attained reliably and efficiently – even retrospectively – using data collected at multiple evaluation sites.
There are many benefits to using this approach. A composite population may be more useful than a single source in establishing trends or norms. Analysis of raw data by a third party not involved in data collection may reduce bias. Data collected in multiple facilities provide a larger data set in a shorter time, overcoming sample size/time limitations and making support surface outcomes studies more timely, informative, and cost effective. The purpose of the study discussed here is to test this theory using preexisting data collected by different researchers investigating the effectiveness of a non-powered therapeutic support surface (Tempur-Med Mattress, Tempur-Medical, Inc., Lexington, Ky). Each data collection was performed to serve the purposes of those conducting the research.
Methods and Procedures
Two primary research goals were defined. The first was to determine if commonly available data points are suitable for measuring outcomes of performance. The second was to verify the value of combining data collected in a number of small, essentially anecdotal reports in a way that results in a larger sample suitable to substantiate conclusions.
The data available for analysis were derived from three independent studies conducted at six institutions (one acute care hospital, one veterans administration hospital [VAMC] , one rehabilitation hospital [Rehab], and three skilled nursing facilities [SNF]). This research was largely conducted in support of internal decision-making processes. The data collected at four of these facilities (rehab and SNFs) were used in support of a regulatory decision-making project (coverage criteria).
The data used in analysis were provided by the original investigators using photocopies of the original data sheets completed at least 1 year before the start of this project. These data sheets were encoded in a manner to preclude knowledge of the subject’s identity. Nothing associated with this study could negatively impact the dignity or quality of care provided to any of the subjects involved. Institutional Review Board (IRB) approval was not necessary.
Each of the data collection sheets was reviewed, and numerous data points common to each of the individual studies were identified. Common data included wound size and stage, duration of support surface use, and specific notations regarding the date a wound was measured or deemed healed by the evaluating clinician. Other, more randomly available data were useful in determining general demographics, diagnosis, and level of acuity. Demographic information illustrated the types of subjects treated using the evaluated mattress, while specific wound data was used to objectively calculate results.
For the sake of clarity, the data and subsequent presentation are segregated into three main areas of study: 1) general demographics; 2) the subset of patients treated prophylactically; and 3) patients with existing wounds.
General demographics. The general demographic statistic represents a composite of all patients for whom the evaluated mattress was ordered and for whom any information was available. The sole purpose of presenting this data was to establish some understanding about the types of indications for which support surfaces are used. The data for many of these subjects were not included in either of the more specific analyses as the result of the filtering techniques used in this study. Subset-specific demographics are presented along with each of the more detailed analyses.
Prophylaxis. The prevention statistic is based on a simple question, “Did an individual without a pressure ulcer develop one while on the mattress?â€? In order to be included in the subset of prophylactic treatment, subjects had to be identified as not having any open wound. To make the statistic more meaningful, risk factors were identified where possible. Evidence of risk included low Braden scores 15 and notations of preexisting pressure ulcers.
Wound healing. Changes in wound size can be observed from the available data and are clinically relevant indicators of healing or deterioration. Data sheets that contained wound staging, physical wound measurements, and duration of treatment sufficient
to support analysis (at least two measurement periods at least 2 weeks apart) were included. Data sheets that were illegible, did not meet the above criteria, or contained inconsistent data were removed.
A subset of data sets (n = 45), roughly 47% of the initial pool of data sets (n = 95), was complete enough to establish wound size at a number of specific points in time and to calculate changes in wound size.
Because the data available is limited to what was provided, the more intricate tools to calculate wound change over time found in the literature could not be employed. A mathematical formula (reduction in size divided by time with elements necessary to preclude division by zero errors) was found that considers incremental wound size reduction over time (see Figure 1).
Figure 1. This mathematical formula considers incremental wound size reduction over time.
Effect of combining data
Combining data from multiple sources contains an element of risk in that difference in any single source may have a disproportionate influence on the whole. Such influences could be the result of different populations, different treatment protocols, or any of a number of undefined variables. To this end, an effect size analysis was desirable. It was determined that the impact could be measured by combining, for each of the source studies, the total sum of final wound size minus the total sum of initial wound size (difference [DF]) across site and initial stage for all wounds included in the final analysis. For each stage and site for which data existed, sample size and confidence interval for the mean DF could be determined. If the DF confidence level contained common values within the range, and was not significant (P > 0.05), the data could reliably be combined. Two-way ANOVA was used to measure the impact of combining the DF for site and stage.