INTRODUCTION
Pressure injuries (PIs) are common complications in healthcare settings.1 The health consequences of PIs include decreased independence with activities of daily living and quality of life; increased pain, likelihood of infection, and mortality; and longer hospital stays.2-7 In post-acute care (PAC) settings, PIs are associated with readmission to acute care hospitals.8 Pressure injuries are frequently avoidable with appropriate medical care9 and are one of the eight conditions that the CMS highlights as preventable.10 The development of new or worsened PIs is therefore an important indicator of the quality of care provided by a PAC facility.
In 2014, the US Congress passed the Improving Medicare Post-Acute Care Transformation (IMPACT) Act of 2014 requiring the collection of standardized patient assessment data elements and public reporting of quality measures across PAC settings; skin integrity and changes in skin integrity are one of those quality measure domains.11 The skin integrity quality measure Percent of Residents or Patients with Pressure Ulcers That Are New or Worsened (NQF 0678) is calculated using standardized patient assessment data and quantifies new or worsened stages 2 to 4 PIs as an indicator of care quality, enabling better understanding of the incidence of new or worsened PIs in each PAC setting.
However, the patient populations in each PAC setting, and even between facilities of similar classification, are different.12-15 For example, research across three institutional PAC settings (long-term care hospitals [LTCHs], skilled nursing facilities [SNFs], and inpatient rehabilitation facilities [IRFs]) indicates that LTCH patients have longer stays and a greater number of comorbidities and chronic illnesses in comparison with patients in SNFs, which treat populations who require skilled nursing care or rehabilitation following an injury, illness, or surgery and patients in IRFs, which treat patients who benefit from intensive rehabilitation.16-19 To account for patient differences and obtain accurate quality measurements, it is necessary to adjust for PI risk factors present at the time of admission that are beyond the control of the facility. This protects facilities that treat a higher proportion of high-risk patients from being mislabeled as providing lower quality relative to their peers when they have a higher observed percentage of patients with new or worsened PIs.
The quality measure Percent of Residents or Patients with Pressure Ulcers That Are New or Worsened (Short Stay) (NQF #0678) was publicly reported by CMS for the assessment of PIs in SNFs, IRFs, and LTCHs between 2012 and 2018,20 and this measure was incorporated in the skin integrity domain of the IMPACT Act (note that this measure was replaced in 2018 with the measure Changes in Skin Integrity Post-Acute Care: Pressure Ulcer/Injury). The measure was calculated using discharge data on the number of stage 2 through stage 4 PIs that were new or worsened since admission to the PAC setting. The calculated quality measure score reported the percentage of patients or residents with at least one new or worsened PI. The measure adjusted for functional limitations based on bed mobility abilities, low body mass index (BMI), bowel incontinence, and a diagnosis of peripheral vascular disease (PVD)/peripheral artery disease (PAD) or diabetes. Limitation in functional mobility is a well-documented risk factor for PIs because it negatively impacts multiple organ systems including the integumentary system.21,22 Low BMI increases risk for PI development through reduced thickness of the skin and underlying tissues.23-25 Bowel incontinence increases risk of developing PIs through the presence of moisture that affects skin integrity.26,27 Disorders such as PVD/PAD and diabetes increase risk for PIs through reduced sensory perception and perfusion.23
Although the quality measure was implemented in 2012 across the three institutional PAC settings, assessment items used in the calculation of the measure have been updated. Because of advances in PI research, experts have called for updates to the risk-adjustment models across PAC settings.28-37 Previous studies have identified several other factors as predictors for the development of PIs. Four such factors are advanced age, bladder incontinence, dual bladder and bowel incontinence, and high BMI.38-44 Increased risk for PI among those who are of advanced age can be attributed to changes in the integrity of skin tissue,45 with risk increasing greatly after the age of 85 years.46 Similar to bowel incontinence, urinary incontinence contributes to the development of PIs through the presence of moisture on the skin's surface.36 Those with both urinary and bowel incontinence are at higher risk for PIs because fecal matter enzymes convert the acid in urine into ammonia. This causes the skin to become more permeable due to higher pH levels and puts individuals with dual incontinence at higher risk than those with either singular type of incontinence.36,47 In addition, obesity is related to the development of PIs through reduction of mobility.48 Moderate or severe obesity (BMI >=35 kg/m2) is associated with 18.9% higher odds of PIs, and mild obesity (BMI 30-34.9 kg/m2) is associated with 4.5% higher odds for PIs.39
The purpose of this study is to examine the association between risk factors and the development of new or worsened stages 2 to 4 PIs across IRFs, LTCHs, and SNFs. Study results can inform future updates to the risk adjusters for cross-setting and setting-specific PI quality measures. Therefore, the researchers investigated the following research questions:
1. To what extent are the clinical risk factors of functional limitation (bed mobility), bowel incontinence, diabetes/PVD/PAD, and low BMI associated with new or worsened stages 2 to 4 PIs across the SNF, IRF, and LTCH populations?
2. To what extent do high BMI, urinary incontinence, dual urinary and bowel incontinence, and advanced age increase risk for new or worsened stages 2 to 4 PI development in SNF, IRF, and LTCH populations?
METHODS
Data Sources
Researchers obtained data for this study from the following sources: Minimum Data Set (MDS) 3.0 data for SNFs, the LTCH Continuity Assessment Record and Evaluation Data Set (LTCH CARE Data Set) for LTCHs, and the Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI) data for IRFs. These sources are national, setting-specific standardized item sets. All data describing stages 2 to 4 PIs and risk factors are defined operationally by items in the version of the setting-specific assessment tools in use at the time (MDS 3.0 version 1.14.1, LTCH CARE Data Set version 3.0, and IRF-PAI version 1.4). The populations in each setting vary based on CMS data collection policy: all Medicare fee-for-service patients are included for the SNF setting, all patients (regardless of payer) are included for the LTCH setting, and all Medicare (fee-for-service and Medicare Advantage) patients are included for the IRF setting. The data were obtained from CMS under a Data Use Agreement, and the analyses were conducted as part of a larger project that was determined to be exempt by RTI International Institutional Review Board.
Data used in these analyses were the most recent available at the time of the study. Data for SNFs and LTCHs consisted of patient stays with discharge dates spanning October 1, 2016 to December 31, 2016. Data for IRFs included patient stays with discharge dates from October 1, 2016 to March 31, 2017.
Consistent with the quality measure calculation,49 the authors included only patients with PAC stays with defined admission and discharge dates and included each stay occurring within the data measurement period. Also consistent with the measure specifications, stays that met the following criteria were excluded: (1) no usable data regarding stages 2 to 4 PIs and/or (2) no data on risk-factor-related items. After exclusions, this yielded a study population that included 394,372 SNF resident stays, 40,696 LTCH patient stays, and 244,891 IRF patient stays.
Risk Factors
The authors examined risk categories for the four admission items used as risk-adjustors for PIs in the Percent of Residents with Pressure Ulcers that are New or Worsened (short stay) measure: functional limitation (bed mobility), bowel incontinence, diagnosis of diabetes or PVD/PAD, and low BMI. They also examined additional risk adjustors: high BMI, urinary incontinence, dual bowel and urinary incontinence, and advanced age. These risk factors reflect the patient's status at the time of admission. Risk categories for all factors are displayed in Table 1. The items used in risk adjustment in this measure are aligned (in terms of the specifications for each response category) across the three PAC settings. Of the items representing potential additional risk factors, all but one (urinary incontinence) are standardized across settings. In the IRF-PAI and LTCH CARE data sets, urinary incontinence is measured on a 7-point scale; on the MDS, urinary incontinence is measured on a 5-point scale.
Statistical Analysis
The dependent variable for all analyses is new or worsened PIs, which is defined as the incidence of one or more new or worsened PIs of stages 2, 3, and/or 4, acquired after PAC admission. Stage 1 PIs and injuries that are unstageable because of suspected deep tissue injury, nonremovable dressings or devices, and/or the presence of slough or eschar were not included in the dependent variable definition to be consistent with the measure specifications. The authors first calculated the incidence of at least one new or worsened stages 2 to 4 PI for each PAC population. This was calculated by taking the number of stays in which one or more stages 2 to 4 PI(s) developed or worsened from the time of admission divided by the total number of included stays. All rates were converted to percentage format.
The authors then calculated frequencies and percentages of new or worsened PIs for each clinical characteristic for the eight risk factors, stratified by PAC population, along with 95% classical CI estimates about those proportions. They also calculated relative risks (RRs) along with 95% classical CI estimates; here, RR is defined as the ratio of the incidence of new or worsened PIs for one clinical categorization of risk divided by the incidence for the category associated with the lowest risk (the reference category). For example, the RR for dependent in bed mobility in SNF is calculated using independent in bed mobility in SNF as the reference category; an RR of 2.0 indicates that the risk for new or worsened PIs associated with a category is twice as great as that for the reference category.
RESULTS
Table 2 presents descriptive summaries of the prevalence of selected patient characteristics at admission, PI incidence, proportion of stays with PI present, and the RR of PI incidence for each of the eight risk factors stratified by PAC setting, along with 95% frequentist CIs. The unadjusted PI incidence, which refers to new or worsened stages 2 to 4 PIs since admission, was 1.23% for SNFs, 3.07% for LTCHs, and 1.56% for IRFs (Table 2).
Clinical Characteristics across PAC Settings
The admission clinical characteristics of patients varied across the three PAC settings. The prevalence of PVD/PAD and/or diabetes at admission was similar across SNFs, LTCHs, and IRFs: less than half of the patients had PVD/PAD and/or diabetes. Disparate characteristics among settings included functional limitation (bed mobility), bowel incontinence, BMI, urinary incontinence, and age (Table 2).
Regarding bed mobility, patients in SNFs and IRFs were more similar to each other at admission than to LTCH patients. The majority of patients in SNFs/IRFs needed supervision or touching assistance, partial/moderate assistance from the helper, or substantial/maximal assistance; in contrast, LTCHs displayed greater proportions of patients who were dependent on the helper for almost all assistance or for whom bed mobility was not attempted due to medical condition or safety concerns.
In comparison with IRFs, SNFs had higher proportions of patients who were occasionally or frequently incontinent. In contrast, many patients in LTCHs were always bowel incontinent or not rated due to ostomy or having no bowel movement reported.
With respect to BMI, SNFs and LTCHs had higher proportions of patients with low BMI compared with IRFs. In addition, LTCHs had higher proportions of patients with BMI greater than 30 or 35 kg/m2 at admission.
The frequency of urinary incontinence at admission varied across the three PAC settings. Among SNF residents, one-fifth were incontinent less than daily. More than half of the patients in IRFs were urinary continent. Although more than a third of patients in LTCHs were always continent, this item was not applicable to another third of LTCH patients who used a catheter.
In addition, more than half of patients in LTCHs (55.53%) were both bowel incontinent and urinary incontinent at admission. Comparatively, SNFs and IRFs had lower proportions of patients with both bowel and urinary incontinence (36.18% and 48.63%, respectively).
Skilled nursing facilities had proportionally more patients of advanced age compared with LTCHs and IRFs.
Risk for New or Worsened PI Incidence across PAC Settings
Limited bed mobility, bowel incontinence, low BMI, diabetes and/or PVD/PAD, advanced age, urinary incontinence, and dual urinary and bowel incontinence were all associated with greater incidence and elevated risk of new or worsened PIs in each setting (Table 2). In contrast, high BMI was associated with lower risk for new or worsened PIs across all settings.
When comparing RR for each of the eight risk factors in terms of most to least disparate among settings, functional limitation varied the most in RR across the three settings. The greatest risks relative to reference conditions were all observed for bed mobility items: SNF, not attempted due to medical condition or safety concerns, RR = 8.74 (95% CI, 7.01-10.89); IRF, dependent, RR = 9.41 (95% CI, 7.31-12.12); LTCH, not applicable: resident did not perform this activity prior to the current illness, exacerbation, or injury, RR = 11.08 (95% CI, 7.27-16.89). The variance of the RRs was 3.03. The high BMI category (defined as BMI >=30 kg/m2) showed the least variation across settings: high BMI was associated with RRs for SNF, IRF, and LTCH of 0.77 (95% CI, 0.72-0.82), 0.76 (95% CI, 0.71-0.82), and 0.72 (95% CI, 0.64-0.81), respectively (Table 2). The variance of those RRs was 0.002.
DISCUSSION
To the authors' knowledge, this is the first study that uses national data to assess the risk factors associated with PI incidence in three institutional PAC settings. The findings indicate that the clinical risk factors currently utilized for risk adjustment (low BMI, bowel incontinence, functional limitation, and diabetes and/or PVD/PAD) for the quality measure Percent of Residents with Pressure Ulcers that are New or Worsened are indeed risk factors of new or worsened PI incidence across SNF, IRF, and LTCH populations, supporting the validity of the risk-adjustment model for that measure. The findings provide empirical evidence to help inform current policy efforts to design quality measures across PAC settings to make them comparable. In addition, advanced age, urinary incontinence, and dual bowel/urinary incontinence were also positively associated with PI incidence in all three PAC settings.
The prevalence of each of the clinical categorizations of risk varied among settings, highlighting the dissimilarities of the patient populations. Despite the clinical differences among populations, each of the tested risk factors was associated with PI incidence across settings (all positively correlated with the exception of high BMI), providing support for measure alignment for the PI quality measure. Urinary incontinence, advanced age, and dual urinary and bowel incontinence could be considered for inclusion in the development of risk-adjustment models for new skin integrity quality measures, such as the Changes in Skin Integrity Post-Acute Care: Pressure Ulcer/Injury measure49 implemented across the three PAC settings as of October 1, 2018. Adding these risk factors would bolster efforts to evaluate the quality of care that is being provided to patients in PAC settings nationally.
Differences in the incidence of new of worsened PIs by PAC population may also reflect differences in care delivery. For example, the use of specialized equipment (eg, specialized beds, surfaces), differences in staffing, and differences in length of stay may vary; these topics should be addressed in future research.
Although there were similar risk factors, the strength of the association for each clinical categorization varied somewhat by setting, prompting further discussion. The factors that varied most in terms of RR for PI development appeared to be less characteristic of the patient's long-term health status and more likely to be symptoms related to other causes that could dually drive risk for new or worsened PIs. For example, RRs varied most widely among populations by functional limitation (bed mobility). Although mobility limitations may be associated with greater risk for new or worsened PIs, individuals who are immobile due to critical illness and multiple comorbidities may be more at risk than those who are immobile following a joint replacement but are otherwise healthy. Similarly, wider variation across settings for urinary incontinence and bowel incontinence may be due to a wide range of underlying conditions that led to the incontinence.
Prior studies comparing patient populations across SNFs, LTCHs, and IRFs have shown that the most common diagnoses in LTCHs are pulmonary edema and respiratory failure, respiratory system diagnosis with ventilator support >=96 hours, septicemia without ventilator support, osteomyelitis with major complication or comorbidity, and skin injuries with major complication or comorbidity.50 In contrast, the most common conditions in IRFs are stroke, other neurologic conditions, fracture of the lower extremity, debility, and brain injury,17 and the most common conditions in SNF are muscle weakness, pneumonia, urinary tract infection, aftercare following joint replacement, and other specified aftercare.51 The clinical data used in the current analyses do not capture the medical conditions of patients, which may itself increase risk. Future research is needed to isolate the effect of each of the identified PI risk factors to aid clinicians in screening for risk.
Limitations
First, because this study did not include home health agencies-the fourth, non-facility-based PAC setting-the findings are not generalizable to all PAC settings. Follow-up studies that investigate cross-setting risk factors for PIs should include home health agencies. Second, the research was limited to data available on the item sets, but the authors recognize that there are other clinical case-mix characteristics that may influence the development of PIs (eg, underlying medical conditions). Future testing should consider including other risk factors, such as primary PAC condition, mental health status, nutrition status indicators, comorbid conditions, differences in prevention measures across settings, and mechanical ventilator use. Third, comparisons were limited by the fact that the available patient stay data varied based on payer and that populations across the three PAC settings differed overall. Future research could examine possible differences across the settings for patients with the same PAC diagnosis (eg, stroke), as well as possible differences between Medicare beneficiaries and those utilizing other types of insurance. Finally, the study data were collected prior to the declaration of the COVID-19 pandemic in 2020, and PAC populations may have changed in recent years.
CONCLUSIONS
This study demonstrated that functional limitation (bed mobility), bowel incontinence, urinary incontinence, dual urinary and bowel incontinence, advanced age, presence of diabetes and/or PVD/PAD, and low BMI are all risk factors for new and worsened PIs in SNF, LTCH, and IRF patients. This study also showed the prevalence and RR for each clinical categorization of the aforementioned risk factors for new and worsened PIs. Further research could improve PI risk screening.
PRACTICE PEARLS
* The incidence of new or worsened PIs varies among inpatient PAC settings-LTCHs, IRFs, and SNFs.
* The associations between risk factors for and incidence of new or worsened PIs are consistent across PAC settings.
* Seven risk factors identified in this study predicted increased incidence of new or worsened PIs in all three inpatient PAC settings: limited bed mobility, bowel incontinence, low BMI, diabetes/PVD/PAD, advanced age, urinary incontinence, and dual urinary and bowel incontinence.
REFERENCES