Preventing Pressure Ulcers in Long-term Care:  A Cost-effectiveness Analysis FREE

Pressure ulcers are localized injury to the skin and underlying tissue, usually over bony prominences.¹ They are mostly a preventable complication of medical care² but remain common in many care settings.³ In particular, long-term care residents are at high risk of developing pressure ulcers because of immobility,⁴ poor nutritional status,⁵ impaired mental status,⁶ and incontinence.^4,6 Pressure ulcers are associated with adverse effects on health,⁷ social well-being,⁸ quality of life,⁹ and high treatment costs—approximately US$3.3 billion per year in 2008.¹⁰

Preventing the development of pressure ulcers is an important aspect of improving long-term care quality.¹¹ However, the cost implications of different quality improvement strategies are not well described.¹² In 2008, the Ontario Health Technology Advisory Committee, a panel that makes recommendations to the Ontario Ministry of Health and Long-Term Care regarding the uptake and diffusion of health technologies,¹³ commissioned an evidence synthesis and a cost-effectiveness analysis of pressure ulcer prevention.^14- 15 Based on our previous work and recent systematic reviews of pressure ulcer prevention,^{12,14,16- 18} the present study evaluates the cost-effectiveness of evidence-based strategies to improve current prevention practice in long-term care facilities.

We conducted the cost-effectiveness analysis according to guidelines by the Canadian Agency for Drugs and Technologies in Health¹⁹ and the Panel on Cost Effectiveness in Health and Medicine.²⁰ Health outcomes included lifetime risk of pressure ulcers and quality-adjusted life-years (QALYs). All costs, expressed in 2009 Canadian dollars (Can$1 = US$0.84),²¹ were calculated from the perspective of a single health care payer. Future health outcomes and costs were discounted at 3%.

Our base case analysis considered long-term care residents, with a mean age of 83 years, a life expectancy of 3.8 years, and a 3-month risk of a stage 2 to 4 pressure ulcer of 2.6% (eTable 1).

We considered the following evidence-informed strategies to improve current prevention practice.

Table 1 describes the preventive interventions in details. Because the strategies address different impairments, they are not mutually exclusive. We therefore compared each strategy with current practice. We did not consider strategies that combined these interventions because their synergistic effect is uncertain.¹² In practice, however, these interventions are key elements of a comprehensive skin care program.^30- 32

The methodological quality of the supporting evidence is generally suboptimal (Table 1), according to current systematic reviews.^{12,14,16- 18} The multiple RCTs supporting strategies 1 and 2 were mostly conducted with inadequate blinding of outcome assessment. Nevertheless, the results appear to be clear-cut with low risk of error in accepting the significant prevention effect.³³ Strategy 3 or 4 is supported by a small RCT reporting a significantly large prevention effect.^22- 23 There could be a moderate to high risk of error in accepting the reported effect.^33- 34

We used a validated Markov model with a 1-week cycle to simulate pressure ulcer–related health events whose likelihoods depend on the prevention effectiveness.¹⁵ Every week, a simulated resident is at risk of developing a pressure ulcer. According to the original staging system from the National Pressure Ulcer Advisory Panel, simulated pressure ulcers are classified as stage 1 (persistent redness of skin), stage 2 (loss of partial thickness of skin, appearing as an abrasion, blister, or shallow crater), stage 3 (loss of full thickness of skin, presented as a deep crater), and stage 4 (loss of full thickness of skin, exposing muscle or bone).³⁵ We originally considered deep-tissue injury¹⁵ but excluded it because of its low prevalence in nursing homes (eg, 0.1%).³⁶

The eFigure displays the stage-specific Markov states, with additional stratification by the following prognostic indicators for wound healing: age groups (ie, 80-84 years, 85-89 years, ≥90 years), risk status (ie, low or high risk),³⁷ wound status (chronic or healable, defined as progressing toward skin closure within 6 months for stage 2-4),³⁸ local infection (stage 2-4 only),³⁹ systemic infection (stage 3-4 only),⁴⁰ and care setting (long-term care facilities or hospitals).⁴¹ Residents with systemic infection were assumed to be hospitalized, and in rare instances, could die from the infection.⁴² At any time, simulated residents could be hospitalized or die because of other causes. The model was constructed in TreeAge Pro 2009 Suite (TreeAge Software, Williamstown, Massachusetts).

Table 2 displays key inputs derived from a population-based cohort, a telephone survey of current prevention practice, linked administrative databases (eAppendix), and literature reviews.

We included all residents from 89 long-term care facilities in Ontario, Canada (overall, 613 facilities), that participated in the first implementation phase of the Resident Assessment Instrument–Minimum Data Set (RAI-MDS, version 2.0) from May 2004 to November 2007. The RAI-MDS is mandated for minimum data collection to inform care monitoring and planning.⁵⁴ Included residents must have had a full RAI-MDS assessment (eg, physical functioning, cognition, nutritional status, skin health, and incontinence) and at least 1 reassessment (eg, every 3 months or when health status changed).⁵⁵ The cohort consisted of 18 325 residents (overall, approximately 92 000 residents). The 3-month incidence of stage 2 to 4 pressure ulcers in our cohort was 2.6%, similar to that reported in US long-term care facilities (eTable 1).^43- 44

We surveyed the directors of care from a random sample of 34 facilities (26 facilities responded) regarding their facility size; staffing ratios and salaries; types, numbers, and costs of mattresses; repositioning programs; nutritional supplements; and incontinence care (Table 2).¹⁵ Still common in the surveyed facilities are standard mattresses (55% of long-term care beds) and incontinence care with soap and water (eg, 50% of care episodes).

We identified high-risk residents using a validated risk-adjustment scale developed by Berlowitz et al^43- 44 (2001) (Table 2). The scale uses 17 RAI-MDS variables (including a stage 1 pressure ulcer) to predict the risk of stage 2 or higher pressure ulcers. High-risk residents (63%) were defined as those who had a 3-month predicted pressure ulcer incidence of greater than 1%.

Few studies report comprehensive data on pressure ulcer prognosis.⁵⁶ We therefore estimated stage-specific transition rates in the eFigure using RAI-MDS data from successive skin assessments (eAppendix and eTable 2).¹⁵ Among high-risk residents, the estimated incidence of developing a stage 1 pressure ulcer was 0.68% per week (Table 2).

We used the Minimum Data Set–Health Status Index,^46,57 a validated algorithm for predicting Health Utilities Index scores from selected RAI-MDS items (Table 2).⁵⁸ The mean health utility score was 0.36 (range, 0-1) and was low because 44% of the residents were bedfast and 65% were incontinent (eTable 1). Residents with stage 2 to 4 pressure ulcers had lower mean scores. After we adjusted for comorbidities, the associated mean (SE) decrement was 0.022 (0.004).⁹ We combined utility and longevity to calculate QALYs.⁵⁹

Long-term care costs cover nursing and personal care, food, basic accommodation, and other costs (Table 2).⁶⁰ The base cost for nursing and personal care was adjusted using the RAI-MDS case-mix index to reflect individual care.⁴⁷ The mean weekly cost attributable to additional care was $106 for stage 2 and $128 for stage 3 or 4 pressure ulcers. For strategy 1, the mean weekly cost was $1.26 per resident, derived by amortizing the cost of upgrading to pressure redistribution foam mattresses over their lifespan (Table 2). The weekly cost was $39.55 for strategy 2 and $6.51 for strategy 3, including both supply and staff time costs. For strategy 4, the weekly supply cost was $2.80, but incontinence care could be faster with foam cleansing than with soap and water, resulting in a mean staff time saving of $3.50 per week, although the variation in staff time cost was large.

We projected the absolute reduction in lifetime risk of stage 2 to 4 pressure ulcers between each quality-improvement strategy and current practice. We derived the corresponding number needed to treat by taking the reciprocal of the absolute risk reduction.⁶¹ We calculated the incremental cost and incremental QALY of each strategy, and if applicable, derived the incremental cost per QALY gained.⁵⁹

Our base case analysis considered the risk of death due to systemic infections that are related to pressure ulcers. In a large study of medical injuries during hospitalization, Zhan and Miller⁴⁵ estimate that pressure ulcers are associated with approximately 7% excess risk of all-cause mortality (Table 2). We conducted a series of sensitivity analyses: (1) with the excess mortality risk estimate, (2) from a long-term care perspective, and (3) for strategies 2 to 4, with supply costs only. For the latter, staff time required to conduct preventive interventions that are recommended by practice guidelines was considered part of current practice.^31- 32,62 To account for the joint input uncertainty, we conducted a probabilistic sensitivity analysis by sampling 10 000 random values from the input distributions specified in Table 1 and Table 2 and eTable 2.⁶³

Table 3 summarizes results of the base case analysis, stratified by quality of the clinical evidence. Strategies 1 and 2 are supported by relatively high-quality evidence, whereas strategies 3 and 4 are supported by low-quality evidence. The projected lifetime risk of pressure ulcers was 8.3% with current practice, and ranged from 6.1% to 8.0% with quality improvement strategies. Over a lifetime, the number needed to treat to prevent 1 pressure ulcer was 45 for strategy 1, 63 for strategy 4, 158 for strategy 3, and 333 for strategy 2.

In Table 3, the expected health gain was mainly due to reduced morbidity rather than mortality and was very small: 0.31quality-adjusted life-days gained (discounted) for strategy 1, 0.20 for strategy 2, 0.11 for strategy 3, and 0.05 for strategy 4. With current practice, the mean lifetime cost (including accommodation, food, and support services for activities of daily living) was approximately $149 000 per resident (discounted). Strategies 1 and 4 minimally improved QALYs and reduced the mean lifetime cost by $115 and $179, respectively. Strategy 2 increased the mean lifetime cost by $731 (0.5% of $149 000), corresponding to an incremental cost per QALY gained of approximately $7.8 million. The corresponding figures for strategy 3 was $24 and approximately $78 000, respectively.

Table 4 displays the sensitivity analysis results. Taking the excess mortality risk attributable to pressure ulcers into account, improved prevention increased life expectancy as well as mean lifetime costs. For example, strategy 1 increased life expectancy by 1.61 quality-adjusted life-days (discounted, or 4.59 life-days, undiscounted) and increased the mean lifetime cost by $389, corresponding to a cost per QALY gained of approximately $88 000. From the long-term care perspective, strategies 1 and 4 still reduced lifetime costs. According to the results from Table 4, excluding staff time cost improved the cost-effectiveness of strategy 2, turned strategy 3 into a cost-saving option, and decreased the mean cost saving of strategy 4.

The Figure displays results of the probabilistic sensitivity analysis, in the form of cost-effectiveness acceptability curves. Each curve shows the probability that each quality improvement strategy is more cost-effective than current practice for different willingness-to-pay values. If decision makers are willing to pay up to $50 000 (or $100 000) per additional QALY, the probability that quality improvement is cost-effective is 94% (96%) for strategy 4, 82% (87%) for strategy 1, 43% (53%) for strategy 3, and 1% (8%) for strategy 2.

According to current systematic reviews, pressure redistribution foam mattresses (including cubed foams,^24- 25 visco-elastic foam,²⁶ and high-density foams),^25- 27 significantly reduce pressure ulcer incidence.^12,14,16 We showed the associated quality improvement strategy is likely to improve health and save costs. Other cost-effectiveness analyses report qualitatively similar results; the overall cost increase due to improving prevention was minimal compared with the cost avoidance associated with the reduction or delay of pressure ulcers.^65- 67

By improving tissue tolerance to pressure and oxygen shortage,⁶⁸ nutritional supports are an important measure to reduce pressure ulcer risk.⁶⁹ According to current systematic reviews,^12,14,17 oral nutritional supplements (eg, daily drinks of 237 mL, 2 kcal/mL) are associated with a significant but relatively small prevention effect.²⁸ We showed that the associated strategy is not cost-effective, but there may be wider health benefits (eg, supporting other strategies, reversing unintentional weight loss, and reducing infections) that we could not fully capture in our analysis.⁷⁰

Dry sacral skin seems to be a significant and independent risk factor for pressure ulcer development.⁵ Compared with a matched greasy placebo, a hyperoxygenated fatty acid regimen for dry skin was reported to significantly reduce pressure ulcer incidence.²² This trial result, however, could be at risk of a false-positive error because of a lack of an intent-to-treat analysis.³⁴ Citing results from this trial, a recent international guideline recommended skin emollients to hydrate dry skin to reduce risk of skin damage.³² Given the available clinical evidence, we showed that the associated strategy may be cost-effective. But firmer conclusion is not possible without additional clinical data.

No-rinse cleansing is the recommended approach for incontinence care.^62,64 According to our results, the use of soap and water, however, is still common. With increasing frequency, washing with soap and water has a significant disrupting effect on the skin's barrier function.⁷¹ Instead of soap and water, incontinence care with a foam cleanser (in combination with an emollient, a water-repellent barrier and a water-repellent deodorant) was reported to significantly reduce pressure ulcer incidence.^18,23 The finding of this small trial, however, could be influenced by publication bias, the phenomenon that small studies with significant results are likely to be published.⁷² Given the existing clinical evidence, we showed that the associated strategy could improve health, save staff time, and reduce costs. Additional clinical data, however, are needed to confirm our finding.

Although the focus of our study is on improving prevention in long-term care residents, many of the supporting trials were conducted with acute care patients.^24- 28 A few trials were conducted as long as 17 years ago^24- 25; generalizability to current practice may be limited by the fact that the types of available devices may have changed. Because of the low prevalence of deep-tissue injuries in long-term care,³⁶ we did not explicitly consider their consequences in our analysis. Also, we did not evaluate any repositioning strategies because the evidence in this area is evolving. Repositioning will be the focus of a separate analysis.

Partially based on data reported herein, the Ontario Health Technology Assessment Committee recommended pressure-redistribution foam mattresses for all long-term care facilities in Ontario and the evaluation of repositioning schedules on the new support surfaces.⁷³ Subsequently, the Ontario Ministry of Health and Long-Term Care contributed funding to an on-going trial evaluating repositioning schedules that had been funded originally by the National Institutes of Health.

The clinical and economic evidence supports pressure redistribution mattresses for all long-term care residents. Improved prevention with perineal foam cleansers and dry skin emollients appears to be cost-effective, but firm conclusions are limited by the available clinical evidence.