PICOT question

Are Evidence-based Practices Associated with Effective Prevention of Hospital-acquired Pressure Ulcers in U.S. Academic Medical Centers?

William V. Padula, PhD, MS, MSc, Assistant Professor, Department of Health Policy & Management, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD

Robert D. Gibbons, PhD, Professor, Department of Public Health Sciences, University of Chicago, Chicago, IL

Robert J. Valuck, PhD, RPh, Professor, Department of Clinical Pharmacy, University of Colorado, Aurora, CO

Mary Beth F. Makic, PhD, RN, Associate Professor, College of Nursing, University of Colorado, Aurora, CO

Manish K. Mishra, MD, MPH, Assistant Professor, Department of Community and Family Medicine, Geisel School of Medicine at Dartmouth, Hanover, NH

Peter J. Pronovost, MD, PhD, FCCM, and Professor, Departments of Anesthesiology, Critical Care and Surgery, Johns Hopkins School of Medicine, Baltimore, MD

David O. Meltzer, MD, PhD Professor, Department of Medicine, University of Chicago, Chicago, IL

William V. Padula: [email protected]; Robert D. Gibbons: [email protected]; Robert J. Valuck: [email protected]; Mary Beth F. Makic: [email protected]; Manish K. Mishra: [email protected]; Peter J. Pronovost: [email protected]; David O. Meltzer: [email protected]

Abstract

Background—In 2008, the Centers for Medicare and Medicaid Services (CMS) established nonpayment policies resulting from costliness of hospital-acquired pressure ulcers (HAPUs) to

hospitals. This prompted hospitals to adopt quality improvement (QI) interventions that increase

use of evidence-based practices (EBPs) for HAPU prevention.

Objective—To evaluate the longitudinal impact of CMS policy and QI adoption on HAPU rates.

Methods—We characterized longitudinal adoption of 25 QI interventions that support EBPs through hospital leadership, staff, information technology, and performance and improvement.

Quarterly counts of HAPU incidence and inpatient characteristics were collected from 55 UHC

CORRESPONDENCE: Dr. William Padula, Department of Health Policy & Management, 624 N. Broadway Ave., Baltimore, MD 21205; [email protected].

HHS Public Access Author manuscript Med Care. Author manuscript; available in PMC 2017 May 01.

Published in final edited form as: Med Care. 2016 May ; 54(5): 512–518. doi:10.1097/MLR.0000000000000516.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

hospitals between 2007–2012. Mixed-effects regression models tested the longitudinal association

of CMS policy, HAPU coding and QI on HAPU rates. The models assumed level-2 random-

intercepts and random effects for CMS policy and EBP implementation to account for between-

hospital variability in HAPU incidence.

Results—Controlling for all 25 QI interventions, specific updates to EBPs for HAPU prevention had a significant, though modest reduction on HAPU rates (−1.86 cases/quarter; p=0.002) and the

effect of CMS nonpayment policy on HAPU prevention was much greater (−11.32 cases/quarter;

p<0.001).

Conclusions—HAPU rates were significantly lower following changes in CMS reimbursement. Reductions are associated with hospital-wide implementation of EBPs for HAPU prevention.

Given that administrative data were used, it remains unknown whether these improvements were

due to changes in coding or improved quality of care.

Keywords

Pressure Ulcers; Medicare; Quality Improvement; Preventive Care; Evidence-based Practice; Policy

INTRODUCTION

Hospital-acquired pressure ulcers (HAPUs) are important to patients, policymakers, and

health systems in the U.S and globally. In April of 2007, the Centers for Medicare and

Medicaid Services (CMS) announced changes to the Inpatient Prospective Payment System

that would withhold payment for a number of hospital-acquired conditions not present-on-

admission (POA), including HAPUs.(1, 2) This nonpayment policy was not enacted until

October, 2008, giving hospitals 18 months to make adjustments to HAPU prevention

protocols. In January, 2009, CMS established new coding requirements of hospital-acquired

conditions that included use of a POA status indicator, further differentiating HAPUs as an

important measure of hospital quality.(3, 4) And in October, 2014, CMS added a new 1%

reimbursement penalty for hospitals performing in the lowest-quartile with respect to

hospital-acquired condition rates.(5)

HAPUs are the most costly of all hospital-acquired conditions, ranging from $500 to over

$130,000 per patient, which translates into $11 billion in direct medical costs and is one of

the highest rates of settlement costs among U.S. hospitals and long-term care facilities.(6–8)

HAPU rates, measured through administrative data, have declined from a spike of 7% in

2004 to 4.5% as of 2012, although these rates have not declined steadily.(9, 10) Decreases in

these rates are likely a function of changes in coding POA status as well as changes in

quality of care.(11)

Improved adherence to evidence-based practice (EBP) guidelines for HAPU prevention

could have impacted quality of care.(12) The EBP guidelines endorsed by the Agency for

Healthcare Research & Quality and the National Pressure Ulcer Advisory Panel consist of

several steps: (1) skin check and risk assessment with a validated instrument such as the

Braden Scale; (2) repositioning; (3) managing moisture and incontinence; (4) nutrition; (5)

new beds and other support surfaces; and (6) actions to reduce friction and shear.(13, 14)

Padula et al. Page 2

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

These guidelines were first published in 2003, and updated again in 2005 and 2010.(15–17)

Yet for a variety of reasons, many patients do not consistently receive these practices.(18)

According to Nelson et al., quality improvement (QI) interventions can improve adherence

to implementation of EBP guidelines through four domains of hospital culture: leadership;

staff; information technology; and performance and improvement.(19) A complete QI

strategy may contain interventions from multiple domains that are bundled together to

culturally integrate and improve consistent implementation of EBPs.(20)

Given recent updates to the CMS reimbursement policy and coding procedures, there is an

interest in knowing if changes in routine practice, linked to QI interventions, lead to

appreciable decreases in HAPU rates. The study followed an observational cohort of

hospitalized patients using hospital-level administrative data to measure HAPU rates over

time, and analyze associations to QI adoption and CMS policies. It was hypothesized that

declining HAPU rates were associated with increased adoption of QI interventions after

CMS policies were modified.

METHODS

Study Design

This is a hospital-level, retrospective observational cohort study of U.S. academic medical

centers that participated in data sharing through UHC (www.uhc.edu, Chicago, IL) and

responded naturally to updates in CMS policy between October, 2007 and June, 2012. The

UHC represented a cooperative of over 200 academic medical centers that centrally housed

administrative data on hospitalized patients. These de-identified, quarterly hospital-level data

were available for download through queries in the UHC Clinical Database and Resource

Manager.

After the Colorado Multiple Institutional Review Board approved this study as exempt

human subject research, hospital data on HAPU outcomes and inpatient characteristics were

acquired from UHC. These data were managed longitudinally by hospital-quarter. Counts of

HAPUs were obtained from UHC data according to Agency for Healthcare Research and

Quality Patient Safety Indicator #3 (PSI-03, v. 3.1, ©2007).(21) This indicator defined

HAPUs as stage III or IV pressure ulcers (ICD-9 707.23 and 707.24) not POA after five days

length-of-stay.

Patient characteristics were gathered from UHC, which provides hospital-level aggregate

data by each quarter, including: age (counts by category: 18–30; 31–50; 51–64; ≥65); sex

(counts of male and female); length-of-stay (mean); in-hospital mortality (count); intensive

care unit admission (count); case-mix index (hospital-level case-mix per quarter); and

medical or surgical status (counts of each). Patient-level information were not available.

Hospitals were also classified according number of beds, as well as periods of time when

hospitals received Magnet recognition from the American Nurses Credentialing Center as a

way to control for hospitals with excellence in nursing quality.(22)

Padula et al. Page 3

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Data Sources

UHC data were merged with data on the hospital-level adoption patterns of QI interventions

according to the four domains of hospital culture by Nelson et al. and a framework of 25 QI

interventions designated for HAPU prevention.(19, 20) In a study previously described by

Padula et al., Certified Wound, Ostomy and Continence Nurses from a representative sample

of 55 UHC hospitals were surveyed to provide longitudinal information about their own

adoption patterns of each of the 25 QI interventions for HAPU prevention to the nearest

quarter between 2007–2012.(23) These hospitals indicated whether EBPs for HAPU

prevention were standardized in nursing, as well as which of 25 QI interventions across four

domains – leadership, staff, information technology, and performance and improvement –

were used as part of a HAPU prevention bundle. As hospitals indicated which QI

interventions were associated with HAPU prevention, they reviewed documentation to

reference start and end dates of interventions to the nearest quarter.

Not all of the 55 hospitals were available for observation during all quarters since certain

hospitals in the study did not participate in UHC data pooling until after 2007. Each year,

more hospitals joined UHC: 48 hospitals in 2007; 51 in 2008; 54 in 2009–2010; and 55 by

2011. As data on QI adoption were merged with HAPU outcomes, hospitals only

contributed data to the analysis when they became UHC members.

Analysis

We used two-level mixed-effects Poisson regression models to regress HAPU rates over time

by QI intervention and changes in CMS policy.(24) We organized these data as a series of

quarterly, hospital-level measurements of HAPU counts and other patient outcomes by

hospital. Since UHC did not provide patient-level data, individual means were not included

in the model. Changes in counts of HAPUs were studied over time, nested within hospitals,

using offsets to control for variability in the number of patient hospitalizations between

hospitals according to inclusion criteria in PSI-03.

The mixed-effects models were developed and tested in two iterations using SuperMix

(Scientific Software International, Skokie, IL, ©2005–2014). The first model iteration began

by studying the level-1 fixed-effects of each of the 25 categorical QI interventions on HAPU

rates. Each QI intervention was coded as a dichotomous variable for each hospital, where 0

indicated a quarter when a specific QI intervention was not adopted and 1 indicated when it

was in use. This model also controlled each of the following covariates available from UHC

in addition to those already described: time (continuous by quarter); age; sex; length-of-stay

(continuous by day); in-hospital mortality rate; intensive care unit admission; case-mix

index; medical or surgical status; beds (count); Magnet Recognition (dichotomous); and a

standardized EBP protocol for HAPU prevention (dichotomous).

The model also controlled for two binary policy interruptions which varied from 0 to 1 when

the policy was enacted: (1) CMS nonpayment policy in starting in October, 2008; and (2)

inclusion of the POA status indicator in January, 2009. Included in the model design were a

random-intercept and level-two random-effects for both of these aforementioned policy

interruptions. We used random-effects since hospitals began at different points in the process

Padula et al. Page 4

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

towards improving HAPU prevention efforts and HAPU rates varied between hospitals.

Additionally, we assumed that CMS policies would have variable impact on progress at

different hospitals.

A complete model that included all covariates was tested first in Equation 1,

Equation

1

where i was the hospital and j was incremental quarter between the 4th quarter of 2007 and 2nd quarter of 2012. The expected value of Yij, HAPU counts was dependent upon the mean intercept (β0) and subject-specific intercept (ui0), as well as the fixed-effect of time (β1), each of 25 QI interventions (β2), and the mean (β3) and subject-specific impacts of policy

interruptions (ui1). Zij represented the individual-specific error term which varied over time. Between-subject changes in counts of HAPU rates were offset by the total number of

hospitalized patients.

The second model iteration was developed ad-hoc to study the effect size of only statistically

significant QI interventions on HAPU rates from the first model with time-interactions,

while also controlling for significant covariates and CMS nonpayment policy (Equation 2).

A random-intercept and level-two random effects were tested for statistically significant QI

interventions as well as significant policy interruptions. This model iteration also tested for

interactions between the effects of time and policy interruptions on the effect sizes of QI at

reducing HAPUs.

Equation

2

For each of the two model iterations, the best form of each model was determined according

to a likelihood-ratio test based on seven mean-variance adaptive Gauss-Hermite quadrature

points. The simplest model with improved log-likelihood and QI interventions that had

statistically significant reductions on HAPU counts was then selected. Independent and

exchangeable autocorrelation structures were also compared. Furthermore, we used a

Lowess smoother to evaluate the descriptive changes in HAPU rates over time.

RESULTS

The U.S. academic medical centers in this UHC sample recorded 5,208 HAPU cases from a

large cohort of 1,590,022 hospitalized patients between 2007–2012 (Table 1). Most of these

cases had a prolonged length-of-stay and were elderly, male, included intensive care unit

admissions or surgical services, and higher case-mix index.

A plot of HAPU rates indicated that coding of PSI-03 was relatively high in 2007 through

early-2008 in this hospital cohort. According to a Lowess smooth line plot, the rate of

HAPUs decreased between hospitals over time, almost to zero by 2012 (Figure 1). These

reductions suggest that much of the variance in HAPU rates was associated with changes in

policy and coding POA status. Further evaluations also quantified the effect size of QI

interventions on this rate reduction.

Padula et al. Page 5

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Longitudinal Data Analysis

In general, a large increase in the number of QI interventions adopted by each hospital –

from about 7 per hospital in 2007 to over 15 in 2012 – coincided with CMS policy changes

and could be linked to an association with decreased HAPU rates (Figure 2a). A mixed-

effects model of HAPU rates by hospital-quarter showed significant reductions with

advancing time, whether viewing HAPU rates from a pre- or post-CMS policy standpoint

(Table 2). This model also controlled for age, CMS policy and case-mix index. Other

covariates (e.g. sex and death) were omitted following backwards stepwise regression to

improve log-likelihood and eliminate insignificant covariates. For instance, length-of-stay

was omitted due to insignificance (p≥0.05), possibly due to the issue that the study sample

controlled for 5 days length-of-stay or greater according to PSI-03. The model viewing

observations between 2007–2012 identified CMS policy change as the single greatest factor

of changes in HAPU rates. Interestingly, POA coding changes noted in 2009 did not result in

significant changes in HAPU incidence, perhaps because HAPUs had stabilized as a result

of accurately identifying all baseline cases that were POA in conjunction with CMS

nonpayment policy enactment in October, 2008.

Of all QI interventions tested in the first model, only updates to the existing pressure ulcer

prevention protocol (“Prevention Protocol”) showed statistically significant reductions in

HAPU rates. In fact, in 2007 there were 24 hospitals using this QI intervention, and the

amount increased to 43 hospitals by 2012 (Figure 2b). Thus, it would be tested in a second

model. “All-Staff Meetings,” or frequent town hall style meetings to discuss updates in

prevention guidelines were actually associated with increases in HAPU rates.

The second mixed-effects model only controlled for the QI intervention “Prevention

Protocol” in addition to time, case-mix index, CMS policy and patient age. Time-interaction

terms were omitted since these covariates were not significant. This model employed

random-effects for the intercept, QI intervention and CMS policy, and applied an

exchangeable autocorrelation structure (X2 = 10.7, p<0.05). This model calculated an increased effect size of changes in updates to the prevention protocol while controlling for

other covariates, and significantly improved log-likelihood (Table 3). That is, hospitals

adopting the QI intervention for “Prevention Protocol” saw a 27% reduction, or 1.86 fewer

HAPU cases per quarter. CMS policy was associated with a greater reduction of 11.32

HAPUs cases – more than a 100% decrease – suggesting that this policy intervention was

followed by the most dramatic decreases in PSI-03 flags.

DISCUSSION

We performed a series of multivariable longitudinal data analyses about adoption patterns of

QI interventions to support implementation of EBPs for HAPU prevention. Several factors

were identified in reducing HAPU rates in academic medical centers from 2007 through

2012. This decrease may have had much to do with mounting pressure of CMS nonpayment

policy that was enacted in October, 2008, which was the largest observed factor in HAPU

rate reductions. It may also have had to do with overlapping changes to coding pressure

ulcers POA in 2009, such that more pressure ulcers were discovered in the first 4 days of

hospitalization. Following CMS nonpayment policy, the rate of HAPUs fell on average by

Padula et al. Page 6

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

11 cases per quarter. The only significant QI intervention associated with HAPU reductions

was “Prevention Protocol.”

This QI intervention suggested that hospitals investing time for hospital leadership (e.g.

nursing quality educators) and skin team leaders to train frontline providers (e.g. RNs, PTs,

etc.) about updates to EBPs for HAPU prevention saw improved outcomes. Hospitals that

adopted “Prevention Protocol” witnessed an average reduction of 1.18 HAPU cases per

quarter. This reduction in HAPUs could result in significant financial gain for an institution,

which could incur direct costs of $130,000 for a single stage III/IV HAPU case, as well as

potentially much more in legal settlements.(7) In addition, updates to a prevention protocol

that translate commonly scripted guidelines into direct, more interpretable information for

each hospital’s providers and culture could be cost-effective.(25)

Based on a timeline of publications by the Wound, Ostomy and Continence Nursing Society,

EBPs that were first published in 2003 had addenda published in 2005 and 2010, and could

have impacted longitudinal adoption of QI interventions for hospitals in this study.(15–17)

Updates to pressure ulcer staging were also noted in 2007, which could have helped

differentiate a low-stage pressure ulcer from a full-thickness wound not POA that would

qualify for PSI-03.(26) At this rate, detailed EBPs for HAPU prevention changed rapidly as

new technologies and techniques were discovered. It was also recommended that hospitals

which intended to become Magnet centers of excellence for HAPU prevention had to

develop a QI program and appoint formal leaders among nursing staff to learn about EBP

updates and disseminate this information to other hospital providers.(27)

There are a number of limitations to this study. First, the period when QI interventions began

taking effect after adoption was not certain. We assumed immediate effect following the first

quarter of initiation, but there could have been lag between initiation and effectiveness. A

sensitivity analysis addressed the inherent issues of recall bias and lag effect by measuring

effect sizes for altered start- and end-dates by one quarter in either direction, but this

additional analysis returned no significant changes in results. Second, patient-level data were

not available from UHC, so we assumed that within each hospital, QI interventions were

initiated consistently between patients.

Third, response bias potentially limited survey responses to hospitals observing positive

effects following QI adoption, compared to hospitals with net-zero or negative outcomes.

While many additional UHC hospitals contacted were initially willing to participate, these

contacts indicated that they did not have the time to complete the survey or they did not have

permission from their administrations to divulge information about QI strategies despite the

survey’s anonymity. This issue would suggest that actually many more UHC hospitals were

investigating QI strategies for HAPU prevention.

Fourth, the study lacked a control group that did not face the effects of CMS nonpayment

policy or coding changes. The model controlled for time, thereby minimizing the between-

hospital effects of these policies and assigning each hospital as its own control. Additionally,

two hospitals sampled never adopted new QI interventions between 2007–2012, which

offered a helpful comparator within the statistical models.

Padula et al. Page 7

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Fifth, there may have been issues concerning reliability of UHC data due to transcription

errors or oversight in coding at the level of data entry. The use of administrative data makes

reliability of coding PSI-03 questionable. This flag depends on billing coders to accurately

extract all data that qualify a patient for a Stage III/IV HAPU not POA, including the

inclusion and exclusion criteria of PSI-03. Meddings et al. noted that these types of data

sources combined with PSI-03 do not accurately portray the rate of HAPUs, especially

across changes in policy.(28) Since UHC data were most accessible for this study design, we

utilized a method that we believed best controlled for time-dependent changes in coding

policy as published in CMS proceedings. However, despite our attempt to control for it in

regression, we cannot evaluate the extent that coding independently has on HAPU rate

reductions likely due to the collinearlity of changes in POA status with CMS nonpayment

policy. We acknowledge that it likely has a large effect on changes in HAPU rates

independent of nonpayment policy and QI interventions.

Future research should examine the independent effect of coding on reductions in PSI-03, or

for that matter, other patient quality and safety indicators. It would be interesting to

determine if hospitals that effectively updated their prevention protocols also were better at

coding HAPUs properly, among other things. This provides a strong incentive to test

interactions between updates to prevention protocols and accurately coding pressure ulcers

as POA.

In conclusion, HAPU rates were significantly lower following changes in CMS policy over

reimbursements. Reductions were associated with hospital-wide implementation of and

updates to EBPs for HAPU prevention. Given that administrative data were used, it remains

unknown whether these improvements were due to changes in coding or improved quality of

care.

Acknowledgments

William Padula was the recipient of an unrestricted grant from the Agency for Healthcare Research and Quality (1 F32 HS023710-01).

References

1. Stone PW, Glied SA, McNair PD, Matthes N, et al. CMS Changes in Reimbursement for HAIs. Med Care. 2010 May.48:433–439. [PubMed: 20351584]

2. Wald HL, Kramer AM. Nonpayment for Harms Resulting From Medical Care: Catheter-Associated Urinary Tract Infections. JAMA. 2007 Dec 19.298:2782–2784. [PubMed: 18165672]

3. Lindenauer PK, Lagu T, Shieh MS, Pekow PS, Rothberg MB. Association of Diagnostic Coding with Trends in Hospitalizations and Mortality of Patients with Pneumonia, 2003–2009. JAMA. 2012; 307:1405–1413. [PubMed: 22474204]

4. van Rijswijk L, Lyder C. Pressure ulcers: were they there on admission? Am J Nurs. 2008; 108:27– 28. [PubMed: 18946259]

5. CMS Media Relations. Department of Health & Human Services. CMS to Improve Quality of Care during Hospital Inpatient Stays. Washington, DC: Centers for Medicare and Medicaid Services; 2014 Aug 4.

6. Beckrich K, Aronovitch SA. Hospital-Acquired Pressure Ulcers: A Comparison of Costs in Medical vs. Surgical Patients. Nurs Econ. 1999 Sep-Oct;17:263–271. [PubMed: 10711175]

Padula et al. Page 8

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

7. Brem H, Maggi J, Nierman D, et al. High cost of stage IV pressure ulcers. Am J Surg. 2010; 200:473–477. [PubMed: 20887840]

8. Lyder C. Pressure ulcer prevention and management. JAMA. 2003; 289:223–226. [PubMed: 12517234]

9. Lyder CH, Wang Y, Metersky M, Curry M, Kliman R, Verzier NR, Hunt DR. Hospital-acquired pressure ulcers: results from the national Medicare Patient Safety Monitoring System study. J Am Geriatr Soc. 2012 Sep.60:1603–1608. [PubMed: 22985136]

10. Whittington K, Briones R. National Prevalence and Incidence Study: 6-year sequential acute care data. Adv Skin Wound Care. 2004; 17:490–494. [PubMed: 15632743]

11. Meddings JA. Using Administrative Discharge Diagnoses to Track Hospital-acquired Pressure Ulcer Incidence - Limitations, Links, and Leaps. Jt Comm J Qual Patient Saf. 2015; 41:243–245. [PubMed: 25990889]

12. Makic, MB. Evidence-based practice for the prevention of pressure ulcers. College of Nursing. Denver, CO: University of Colorado; 2007.

13. Haesler, E., editor. National Pressure Ulcer Advisory Panel, European Pressure Ulcer Advisory Panel, Pan Pacific Pressure Injury Alliance. Prevention and Treatment of Pressure Ulcers: Clinical Practice Guideline. Perth, Australia: Cambridge Media; 2014.

14. Agency for Healthcare Research and Quality (AHRQ). AHRQ toolkit helps to prevent hospital- acquired pressure ulcers: Research Activities. Rockville, MD: AHRQ; 2011 Jul.

15. Ratliff, CR.; Bryant, DE. WOCN clinical practice guideline. Glenview, IL: Wound, Ostomy, and Continence Nurses Society (WOCN); 2003. Guideline for prevention and management of pressure ulcers; p. 52

16. Ratliff CR, Tomaselli N. WOCN update on evidence-based guideline for pressure ulcers. J Wound Ostomy Continence Nurs. 2010; 37:459–460. [PubMed: 20838310]

17. Ratliff CR. WOCN’s evidence-based pressure ulcer guideline. Adv Skin Wound Care. 2005; 18:204–208. [PubMed: 15920372]

18. Padula WV, Wald HM, Makic HM. Pressure ulcer risk assessment and prevention. Ann Intern Med. 2013; 159:718. [PubMed: 24247681]

19. Nelson, EC.; Batalden, PB.; Godfrey, MM. Quality by Design. San Francisco, CA: Jossey-Bass; 2007.

20. Padula WV, Mishra MK, Makic MB, et al. A framework of quality improvement interventions to implement evidence-based practices for pressure ulcer prevention. Adv Skin Wound Care. 2014; 27:280–284. [PubMed: 24836619]

21. Farguhar, M. AHRQ Quality Indicators. In: Hughes, RG., editor. Patient Safety and Quality: An Evidence-based Handbook for Nurses. Rockville, MD: Agency for Healthcare Research and Quality; 2008 Apr.

22. Clark ML. The Magnet Recognition Program and evidence-based practice. J Perianesth Nurs. 2006 Jun.21:186–189. [PubMed: 16769526]

23. Padula WV, Mishra MK, Makic MB, et al. Increased Adoption of Quality Improvement Interventions to Implement Evidence-Based Practices for Pressure Ulcer Prevention in U.S. Academic Medical Centers. Worldviews Evid Based Nurs. 2015 Oct 13. Epub ahed of print.

24. Hedeker, D.; Gibbons, RD. Longitudinal Data Analysis. Hoboken, NJ: Jon Wiley & Sons, Inc; 2006.

25. Padula WV, Mishra MK, Makic MB, Sullivan PW. Improving the Quality of Pressure Ulcer Care with Prevention: a cost-effectiveness analysis. Med Care. 2011 Apr.49:385–392. [PubMed: 21368685]

26. Black J, Baharestani M, Cuddigan J, et al. National Pressure Ulcer Advisory Panel’s updated pressure ulcer staging system. Urologic nursing. 2007; 27:144–150. 156. [PubMed: 17494455]

27. Creehan S, Cuddigan J, Gonzales D, Nix D, Padula WV, Pittman J, Pontieri-Lewish V, Walden C, Wells B, Wheeler R. Developing Centers of Pressure Ulcer Prevention Excellence: A Framework for Sustainability. J Wound Ostomy Continence Nurs. 2015 In Press.

28. Meddings JA, Reichert H, Hofer T, McMahon LF. Hospital Report Cards for Hospital-Acquired Pressure Ulcers: How Good Are the Grades? Ann Intern Med. 2013 Oct.159:505–513. [PubMed: 24126644]

Padula et al. Page 9

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Figure 1. Smoothed locally weighted regression of aggregate quarterly weights and scatterplot of

hospital-level HAPU rates relative to all hospitalized patients according to Agency for

Healthcare Research and Quality PSI-03 inclusion criteria, while noting CMS nonpayment

policy during October, 2008.

Gray dots indicate quarter when CMS nonpayment policy was established in October, 2008.

Padula et al. Page 10

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Figure 2. (a) Mean number of QI interventions per hospital between all observed hospitals per quarter

relative to changes in CMS reimbursement policy between 3rd quarter of 2007 and 2nd

quarter of 2012; (b) Number of hospitals using the Leadership QI intervention, Prevention

Protocol, per quarter relative to changes in CMS reimbursement policy between 3rd quarter

of 2007 and 2nd quarter of 2012.

Padula et al. Page 11

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Padula et al. Page 12

Table 1

Patient characteristics among a sample of 55 UHC hospitals, 2007–2012.

Characteristic N %

UHC Hospitals 55

Hospital-quarters 1119

Nursing Magnet Hospitals 28 51%

Total Inpatient Admissions 1,590,022

Age

18–30 168,728 11%

31–50 385,282 24%

51–64 462,510 29%

≥65 574,167 36%

Female 776,492 49%

Medical 871,497 55%

Surgical 681,496 43%

Total HAPU Cases 5,208

Age

18–30 268 5%

31–50 850 16%

51–64 1,597 31%

≥65 2,495 48%

Female 2,217 43%

Medical 1,979 38%

Surgical 3,208 62%

HAPU indicates hospital-acquired pressure ulcer.

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Padula et al. Page 13

Table 2

Results of a mixed-effects Poisson regression of hospital-level HAPU rates, controlling for longitudinal

adoption patterns of all QI interventions as well as changes in CMS nonpayment policy.

Variable Estimate Standard Error P

Intercept 1.5250 0.1445 <0.001

Time −0.0323 0.0067 <0.001

Program Mission 0.1100 0.0929 0.2362

Prevention Awareness 0.0194 0.0926 0.8340

Leadership Initiatives −0.1536 0.1039 0.1391

Admin Support −0.0007 0.0981 0.9943

Prevention Protocol −0.1808 0.0845 0.0325

Benchmarking 0.0426 0.0932 0.6474

Wound Team 0.0395 0.0725 0.5855

Performance Measures 0.0065 0.0818 0.9369

Team Huddles 0.0651 0.1050 0.5353

All-Staff Meetings 0.2541 0.1159 0.0283

Wound/QI Team 0.0011 0.0780 0.9887

Prevention Education 0.0002 0.0920 0.9987

Staff Training 0.1219 0.0916 0.1834

Data Tracking −0.0590 0.0796 0.4587

EHR Risk Assess −0.1175 0.1238 0.3423

Electronic Alarm 0.0366 0.0974 0.7070

EHR Implementation −0.0650 0.1126 0.5640

Braden Scale 0.0624 0.0823 0.4484

Visual Tools −0.0833 0.0994 0.4020

Beds 0.0869 0.0778 0.2644

HAPU Staging −0.0378 0.0960 0.3409

Skin Care −0.0820 0.0861 0.1224

Incontinence 0.1339 0.0867 0.8296

Repositioning −0.0331 0.1539 0.3976

Nutrition −0.1267 0.1498 0.9188

Age 31–50 0.0513 0.0079 <0.001

Age 51–64 0.0571 0.0049 <0.001

Age >65 0.0420 0.0030 <0.001

Case-mix Index 0.0467 0.0044 <0.001

CMS Policy (October, 2008) −1.1355 0.1507 <0.001

Present-on-Admission Indicator (January, 2009) −0.0219 0.1399 0.8758

Log-likelihood = −1855.86

Bold indicates statistically significant at 95% confidence

indicates statistically significant QI interventions for reducing HAPUs

Time is measured quarterly from 3rd quarter of 2007 to 2nd quarter of 2012

Med Care. Author manuscript; available in PMC 2017 May 01.

A u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t A

u th

o r M

a n u scrip

t

Padula et al. Page 14

Table 3

Results of a mixed-effects Poisson regression of hospital-level HAPU rates, controlling for longitudinal

adoption patterns of updates to the “Prevention Protocol” and CMS nonpayment policy.

Variable Beta-estimate Standard Error P

Intercept −5.408 0.088 <0.001

Time −0.045 0.006 <0.001

Prevention Protocol −0.2701 0.0889 0.002

Age 31–50 0.047 0.009 <0.001

Age 51–64 0.053 0.005 <0.001

Age >65 0.038 0.003 <0.001

Case-mix Index 0.044 0.004 <0.001

CMS Policy (October, 2008) −1.054 0.101 <0.001

Variance(Intercept) 0.062 0.148 <0.001

Variance(Prevention Protocol) 0.244 0.062 <0.001

Variance(CMS Policy) 0.025 0.061 <0.001

Log-likelihood = −1859.14

Bold indicates statistically significant at 95% confidence

indicates statistically significant QI interventions for reducing HAPUs

Time is measured quarterly from 3rd quarter of 2007 to 2nd quarter of 2012

Med Care. Author manuscript; available in PMC 2017 May 01.