Development of the interRAI Pressure Ulcer Risk Scale (PURS) for use in long-term care and home care settings

Pressure ulcers are an important and potentially preventable problem in LTC homes [1]. Estimates of the incidence of pressure ulcers vary among facilities from 6.5% [2] to 23.9% [3], despite the availability of validated tools [4] for evaluating an individual's risk of developing a pressure ulcer and clinical guidelines for prevention and care [5]. The key risk factors reported in the literature are impaired mobility, under-nutrition and low body mass index (BMI), poor physical health including factors affecting oxygenation and perfusion, advanced age, body temperature, friction and shear, skin moisture, pain, drugs, some kinds of medical devices, and impaired cognition and sensory perception [6–11]. Prevention is important for protecting or improving quality of life but is often costly and time consuming. Therefore, it must be carefully targeted based on risk factors for the outcome of interest [12]. As illustrated by the development and implementation of quality measures [13] for determining the quality of long-term care, pressure ulcer prevention has become a major focus of quality improvement initiatives across the US and Canada [1, 14].

In 2007 the Ontario Ministry of Health and Long-Term Care sponsored the implementation of a Pressure Ulcer Awareness and Prevention Program (PUAP Program) by the Canadian Association of Wound Care (CAWC) in Ontario LTC homes. The basis of the PUAP Program is risk assessment and intervention using the Braden Scale[4]. Within the program LTC staff complete the Braden Scale at admission, then every week for four weeks and quarterly thereafter [5], or more frequently if a change in health status occurs. Resources and plan of care are based on a resident's risk level and individual sub-scale scores. Completing both the mandated MDS [15] and the Braden Scale presented LTC home staff with considerable assessment and paperwork burden, often perceived as duplicated effort. This perception may place the PUAP at potential risk of facility non-adherence to protocols, and residents at risk of pressure ulcer development.

Few studies have been published that examine the Braden Scale and the MDS together. A chart review of 8 US homes in which MDS and Braden Scale data were available compared the Braden Scale and the MDS Resident Assessment Protocol (RAP), concluding that there was good potential for MDS information to take the place of the Braden Scale [16]. An examination of MDS items and Braden Scale domains indicated that some Braden Scale domains are not well-operationalized in the MDS (nutritional status and friction/shear), and some that are available (sensory perception items) are not used in the pressure ulcer RAP [17]. Braden Scale data were collected along with MDS data in the National Pressure Ulcer Long-Term Care Study, with MDS items used primarily as descriptor variables [18].

The purpose of this study was to examine the potential of the MDS assessment, mandated for use in Ontario LTC homes, to inform the risk for pressure ulcers. Specifically, the study sought to simulate Braden Scale domains with MDS items, using clinical expert input, and use this as a starting point to develop a bedside MDS-based scale to identify LTC home residents at various levels of risk for developing pressure ulcers, in order to facilitate targeting of risk factors for prevention.

Using the best candidate single items available in each of the six domains to give weights equivalent to the Braden Scale scoring, a six-item cross-walk was constructed using the following MDS items: cognitive skills for daily decision making, bowel incontinence, walk in room, transfer, leaves food uneaten, and bed mobility. While suitable for informing a Braden cross-walk, it resulted in cell sizes that were too small for validation (i.e., prediction of a new pressure ulcer) in the sample of only 257.

Shifting to the larger baseline/reassessment dataset (n = 12,896), the Braden Scale cross-walk total score was tested for its predictive ability, with pressure ulcer incidence ranging from about 9% down to about 1% across its scoring range from 6 (high risk) to 23 (low risk), respectively, and a c-statistic in the logistic regression model of 0.676. The receiver operator characteristic (ROC) curve is provided as Figure 1. The six component items were also tested in multivariable logistic regression models to examine if they all contributed in the prediction of a new pressure ulcer at reassessment. Of the six items, cognitive skills for decision making and transfer were not significantly associated with the dependent variable when adjusting for the other four. For reasons of compatibility with the newer generation interRAI instruments, weight loss was substituted for leaves food uneaten; while the two items were only weakly correlated with one another, weight loss had a much stronger association with a new pressure ulcer, and it was consistent with the spirit of the Braden Scale nutrition domain.

Of the seven items, history of resolved pressure ulcer was the strongest predictor of new pressure ulcers at follow-up in the multivariable model, with an odds ratio of 2.5. Other items had odds ratios between 1.4 and 1.8. An additive score was then constructed with history of resolved pressure ulcer counting for two points if present, and the other six items counting one if present, resulting in a scale of 0 to 8 (higher numbers representing higher risk of a new pressure ulcer at reassessment). Values of 8 were very rare (2 cases only), and were grouped with the 7's. This additive scale achieved a c-statistic of 0.708 in the model predicting new pressure ulcers. Results are shown in Table 4, Figure 2 and Figure 3, with increasing incidence rates and odds ratios by increasing scale value. The new scale is named the interRAI Pressure Ulcer Risk Scale (interRAI PURS).

interRAI PURS Score	Proportion of residents at baseline	Pressure ulcer rate at reassessment	Odds ratio, 95% confidence
0	18.9%	0.8%	Reference
1	22.3%	2.0%	2.53 (1.53, 4.16)
2	17.9%	3.3%	4.21 (2.59, 6.83)
3	24.0%	5.4%	7.11 (4.50, 11.23)
4	10.1%	9.0%	12.28 (7.65, 19.72)
5	4.7%	9.9%	13.62 (8.01, 23.14)
6	1.8%	14.2%	20.53 (11.09, 38.01)
7-8	0.3%	21.1%	33.11 (10.14, 108.13)

The seven independent variables in interRAI PURS were also used in modeling new pressure ulcer using an interactive decision tree tool. Similar associations to the regression results were found, and generally parallel structures were supported among candidate tree models. Predictive performance was slightly weaker than the additive model, with c-statistic values around 0.690, and when considering the additional complexity of scoring an algorithm or tree at the bedside, this approach was not pursued.

Applying the interRAI PURS score to Ontario CCC patients without a pressure ulcer at their initial assessment (n = 13,062), distribution of the scale was skewed more towards the high risk end (only 20% scored 0 or 1, compared to over 41% among LTC residents). This was consistent with an overall higher pressure ulcer incidence rate of 9.7% in CCC compared to 3.9% in LTC. The scale values showed consistently higher pressure ulcer incidence with higher scores (4.4% at scale value 0, to 23.7% at scale value 7). The c-statistic in this sample was 0.607.

In long-stay home care clients, interRAI PURS is skewed more towards lower risk, with nearly three quarters of clients scoring 0 or 1. Risk of pressure ulcer incidence was in fact much lower in this population compared to either LTC or CCC at about 2.6%. The scale was able to clearly and consistently differentiate risk, with incidence rates ranging from 1.6% among those scoring 0, to 13.1% for those scoring 6 or greater. Here, the c-statistic for interRAI PURS was 0.629.

interRAI PURS provides an estimate of graded risk for developing a pressure ulcer among individuals who do not present with a pressure ulcer at the time of assessment. It can be calculated using a completed MDS assessment, presumably as a computer-generated output similar to other MDS-based scales and clinical assessment protocols. It has a simple 7-item additive structure which lends itself to manual bedside scoring during early admission periods prior to completion of the full set of MDS items, or prior to a scheduled MDS reassessment. Since it uses items drawn directly from the MDS, assessor training skills and resources already present in that home or organization can be drawn upon, minimizing any duplication often required to support a separate pressure ulcer clinical screening program. interRAI PURS can be calculated using interRAI instruments in various settings, as well as for the new integrated suite of instruments, including the long-term care facilities (interRAI LTCF), home care (interRAI HC), and palliative care (interRAI PC) in which good psychometric properties have been demonstrated [31].

Of note is the property that the scale distributes roughly one-fifth of the assessed residents into each of the lowest four scale values among LTC residents, with progressively fewer in each of the higher risk scores. This makes it possible to identify increasingly rare individuals at the highest level of risk. Identification of such individuals may add clinical value, as pressure ulcer risk is associated with high need for care due to functional deficits [32].

The ability of interRAI PURS to predict individuals at higher risk of a new pressure ulcer could not be directly compared to established instruments such as the Braden Scale from data available in this study, since the concurrent Braden and MDS observations were too few to establish stable estimates for pressure ulcer incidence. However, by constructing a cross-walk of the Braden Scale using evidence from the small matched dataset, a much larger sample of MDS assessments could be employed. Clinical expert input was used to guide the choice of domain areas and items, and hence the process aimed to strike a balance between evidence available in the data and in the clinical literature. While the Braden cross-walk using MDS items was informative, it was not pursued due to the failure of two of the six Braden domains to predict new pressure ulcers in the MDS data, along with the superior performance of interRAI PURS.

Given that the starting point of this work was the Braden scale, it is notable that of its six sub-scales, many have parallels in interRAI PURS: mobility (walking); nutrition (weight loss); friction/shear (bed mobility). As well, shortness of breath can be related to the Braden Scale activity domain. The sensory perception sub-scale of the Braden Scale does not have an analogous item in interRAI PURS. Also, two items in interRAI PURS do not match to an analogous domain in the Braden: daily pain and history of resolved pressure ulcer, although both are described in the literature as risk factors for pressure ulcer development [33, 34].

Several considerations need to be observed from a clinical perspective with any summary scale such as the interRAI PURS. The quality of contributing items can be dependent on the skill of the assessor, and consistent training and monitoring of MDS assessment skills are important. The administration of interRAI PURS as close to the admission as possible, along with reassessment conditioned on baseline risk and clinical change (not merely when the next MDS quarterly assessment is due) are also central to an effective pressure ulcer program. RAPs provide clinicians with a structured, problem and prevention-oriented approach to care planning for potential or actual problems, such as pressure ulcers, and may be used along with summary scales such as interRAI PURS to guide care planning and targeting of pressure ulcer program activities. Longitudinal change in interRAI PURS score should also be tracked to signal changes in care plans. interRAI PURS, like any automated scale, should be used to support but not replace clinical judgment.

interRAI PURS may be used clinically both as an automated output from a computer-entered MDS assessment, as well as from a manually completed bedside form. Next steps will need to evaluate how interRAI PURS performs as a clinical tool, considering both usability and clinical outcomes at the bedside. In Ontario, a collaborative pressure ulcer awareness program in LTC homes is underway that will be evaluating both interRAI PURS and the process of its implementation. While the interRAI PURS has been shown to identify individuals more likely to develop a pressure ulcer in the future, it cannot be assumed that its use as a clinical indicator to guide practice will reduce the incidence of pressure ulcers. In fact, there is a lack of high quality evidence of clinical effectiveness for any risk scale of this kind [27]. Randomized controlled trial quality studies would be required to show that assessing pressure ulcer risk is more than a burden to care staff. However, with risk assessment firmly established in clinical practice, the interRAI PURS can reduce duplication in this assessment burden.

Of note in this study is the low reported prevalence of pressure ulcers in the long-term care sample, at 8.4%. While this falls near the lower boundary of reported rates, it may underreport some residents with pressure ulcers, in particular Stage 1 ulcers. This is not entirely surprising as detection of Stage 1 types can be problematic [35]. In addition to basic issues of detection and recording, the 3-month interval of the MDS data will tend to under-report new pressure ulcers, as some will occur and subsequently heal before the new assessment look-back window, and risk factors may also change during this time. This work uses secondary data not explicitly collected for this purpose, and while data quality has generally been found to be good in jurisdictions like Ontario where MDS implementation was planned and supported [19], there is of course some degree of imprecision in measurement of both pressure ulcer and risk factors. However, these errors tend to attenuate associations in the data, supporting the robustness of results found. There is likely a bias in the data towards under-reporting the least important wounds (pressure ulcers quickly healed, minor stage 1 pressure ulcers); whether these types of individuals have distinct risk factors is unknown, but it is likely that measured risk factors operate in a similar fashion to those experiencing more severe pressure ulcers.

This study provides a new application for MDS data to support interventions aimed at targeting care for those at risk of developing pressure ulcers. The reduction of assessment burden achieved by harmonizing the risk assessment with the MDS and the ability to derive this scale in multiple service settings make interRAI PURS an attractive scale for system-wide approaches to pressure ulcer reduction strategies.