Predictive models for pressure ulcers from intensive care unit electronic health records using Bayesian networks.

BACKGROUND: We develop predictive models enabling clinicians to better understand and explore patient clinical data along with risk factors for pressure ulcers in intensive care unit patients from electronic health record data. Identifying accurate risk factors of pressure ulcers is essential to determining appropriate prevention strategies; in this work we examine medication, diagnosis, and traditional Braden pressure ulcer assessment scale measurements as patient features. In order to predict pressure ulcer incidence and better understand the structure of related risk factors, we construct Bayesian networks from patient features. Bayesian network nodes (features) and edges (conditional dependencies) are simplified with statistical network techniques. Upon reviewing a network visualization of our model, our clinician collaborators were able to identify strong relationships between risk factors widely recognized as associated with pressure ulcers. METHODS: We present a three-stage framework for predictive analysis of patient clinical data: 1) Developing electronic health record feature extraction functions with assistance of clinicians, 2) simplifying features, and 3) building Bayesian network predictive models. We evaluate all combinations of Bayesian network models from different search algorithms, scoring functions, prior structure initializations, and sets of features. RESULTS: From the EHRs of 7,717 ICU patients, we construct Bayesian network predictive models from 86 medication, diagnosis, and Braden scale features. Our model not only identifies known and suspected high PU risk factors, but also substantially increases sensitivity of the prediction - nearly three times higher comparing to logistical regression models - without sacrificing the overall accuracy. We visualize a representative model with which our clinician collaborators identify strong relationships between risk factors widely recognized as associated with pressure ulcers. CONCLUSIONS: Given the strong adverse effect of pressure ulcers on patients and the high cost for treating pressure ulcers, our Bayesian network based model provides a novel framework for significantly improving the sensitivity of the prediction model. Thus, when the model is deployed in a clinical setting, the caregivers can suitably respond to conditions likely associated with pressure ulcer incidence.

Outcomes of an Oral Care Protocol in Postmechanically Ventilated Patients.

BACKGROUND: Oral care is standard practice to prevent hospital-associated infections while patients are intubated and in the intensive care unit. Following extubation and transfer, infections remain an important risk for patients, but less attention is paid to oral care. Few studies have assessed the impact of oral care in recently extubated acutely ill patients. AIMS: To develop an evidence-based oral care protocol for hospitalized patients and determine the impact of this protocol on health outcomes in recently extubated patients. METHODS: In this randomized controlled trial, subjects were randomized to usual care or an intervention protocol that included tooth brushing, tongue scraping, flossing, mouth rinsing, and lip care. Major outcome measures were the revised THROAT (R-THROAT; oral cavity assessment) and overall prevalence of methicillin-sensitive Staphylococcus aureus and methicillin-resistant S. aureus on oral cultures. RESULTS: Seventy-four subjects were randomized. As measured by the R-THROAT, oral cavity health improved over time in both groups, but the intervention group demonstrated significantly more improvement than the control group (R-THROAT score improved by 1.97 intervention vs. 0.87 control; p = .04). Two categories, tongue and mouth comfort, demonstrated the most significant improvement. There was no difference in MSSA/MRSA colonization between the groups at the conclusion of the study. Overall, subjects in the intervention group were more satisfied with their protocol than subjects in the usual care group. LINKING EVIDENCE TO ACTION: This study offers an important evaluation of an oral care protocol after extubation. Results demonstrated improvement in the oral cavity assessment with the designed oral care protocol. Patients expressed a preference for the intervention protocol, which included a battery-operated toothbrush, higher-quality toothpaste and mouth rinse, tongue scraper, floss, and lip balm. The implementation of an oral care protocol specifically addressing patients in the immediate postintubation is essential.

Nurses' Attitudes Regarding the Safe Handling of Patients Who Are Morbidly Obese: Instrument Development and Psychometric Analysis.

BACKGROUND AND PURPOSE: Nurses' attitudes play an important role in the consistent practice of safe patient handling behaviors. The purposes of this study were to develop and assess the psychometric properties of a newly developed instrument measuring attitudes of nurses related to the care and safe handling of patients who are obese. METHODS: Phases of instrument development included (a) item generation, (b) content validity assessment, (c) reliability assessment, (d) cognitive interviewing, and (e) construct validity assessment through factor analysis. RESULTS: The final data from the exploratory factor analysis produced a 26-item multidimensional instrument that contains 9 subscales. CONCLUSIONS: Based on the factor analysis, a 26-item instrument can be used to examine nurses' attitudes regarding patients who are morbidly obese and related safe handling practices.

Predictive Modeling for Pressure Ulcers from Intensive Care Unit Electronic Health Records.

Our goal in this study is to find risk factors associated with Pressure Ulcers (PUs) and to develop predictive models of PU incidence. We focus on Intensive Care Unit (ICU) patients since patients admitted to ICU have shown higher incidence of PUs. The most common PU incidence assessment tool is the Braden scale, which sums up six subscale features. In an ICU setting it's known drawbacks include omission of important risk factors, use of subscale features not significantly associated with PU incidence, and yielding too many false positives. To improve on this, we extract medication and diagnosis features from patient EHRs. Studying Braden, medication, and diagnosis features and combinations thereof, we evaluate six types of predictive models and find that diagnosis features significantly improve the models' predictive power. The best models combine Braden and diagnosis. Finally, we report the top diagnosis features which compared to Braden improve AUC by 10%.

Body mass index and pressure ulcers: improved predictability of pressure ulcers in intensive care patients.

BACKGROUND: Obesity contributes to immobility and subsequent pressure on skin surfaces. Knowledge of the relationship between obesity and development of pressure ulcers in intensive care patients will provide better understanding of which patients are at high risk for pressure ulcers and allow more efficient prevention. OBJECTIVES: To examine the incidence of pressure ulcers in patients who differ in body mass index and to determine whether inclusion of body mass index enhanced use of the Braden scale in the prediction of pressure ulcers. METHODS: In this retrospective cohort study, data were collected from the medical records of 4 groups of patients with different body mass index values: underweight, normal weight, obese, and extremely obese. Data included patients' demographics, body weight, score on the Braden scale, and occurrence of pressure ulcers. RESULTS: The incidence of pressure ulcers in the underweight, normal weight, obese, and extremely obese groups was 8.6%, 5.5%, 2.8%, and 9.9%, respectively. When both the score on the Braden scale and the body mass index were predictive of pressure ulcers, extremely obese patients were about 2 times more likely to experience an ulcer than were normal weight patients. In the final model, the area under the curve was 0.71. The baseline area under the curve for the Braden scale was 0.68. CONCLUSIONS: Body mass index and incidence of pressure ulcers were related in intensive care patients. Addition of body mass index did not appreciably improve the accuracy of the Braden scale for predicting pressure ulcers.

Predictive validity of the Braden scale for patients in intensive care units.

BACKGROUND: Patients in intensive care units are at higher risk for development of pressure ulcers than other patients. In order to prevent pressure ulcers from developing in intensive care patients, risk for development of pressure ulcers must be assessed accurately. OBJECTIVES: To evaluate the predictive validity of the Braden scale for assessing risk for development of pressure ulcers in intensive care patients by using 4 years of data from electronic health records. Methods Data from the electronic health records of patients admitted to intensive care units between January 1, 2007, and December 31, 2010, were extracted from the data warehouse of an academic medical center. Predictive validity was measured by using sensitivity, specificity, positive predictive value, and negative predictive value. The receiver operating characteristic curve was generated, and the area under the curve was reported. RESULTS: A total of 7790 intensive care patients were included in the analysis. A cutoff score of 16 on the Braden scale had a sensitivity of 0.954, specificity of 0.207, positive predictive value of 0.114, and negative predictive value of 0.977. The area under the curve was 0.672 (95% CI, 0.663-0.683). The optimal cutoff for intensive care patients, determined from the receiver operating characteristic curve, was 13. CONCLUSIONS: The Braden scale shows insufficient predictive validity and poor accuracy in discriminating intensive care patients at risk of pressure ulcers developing. The Braden scale may not sufficiently reflect characteristics of intensive care patients. Further research is needed to determine which possibly predictive factors are specific to intensive care units in order to increase the usefulness of the Braden scale for predicting pressure ulcers in intensive care patients.

Beyond the bundle--journey of a tertiary care medical intensive care unit to zero central line-associated bloodstream infections.

INTRODUCTION: We set a goal to reduce the incidence rate of catheter-related bloodstream infections to rate of <1 per 1,000 central line days in a two-year period. METHODS: This is an observational cohort study with historical controls in a 25-bed intensive care unit at a tertiary academic hospital. All patients admitted to the unit from January 2008 to December 2011 (31,931 patient days) were included. A multidisciplinary team consisting of hospital epidemiologist/infectious diseases physician, infection preventionist, unit physician and nursing leadership was convened. Interventions included: central line insertion checklist, demonstration of competencies for line maintenance and access, daily line necessity checklist, and quality rounds by nursing leadership, heightened staff accountability, follow-up surveillance by epidemiology with timely unit feedback and case reviews, and identification of noncompliance with evidence-based guidelines. Molecular epidemiologic investigation of a cluster of vancomycin-resistant Enterococcus faecium (VRE) was undertaken resulting in staff education for proper acquisition of blood cultures, environmental decontamination and daily chlorhexidine gluconate (CHG) bathing for patients. RESULTS: Center for Disease Control/National Health Safety Network (CDC/NHSN) definition was used to measure central line-associated bloodstream infection (CLA-BSI) rates during the following time periods: baseline (January 2008 to December 2009), intervention year (IY) 1 (January to December 2010), and IY 2 (January to December 2011). Infection rates were as follows: baseline: 2.65 infections per 1,000 catheter days; IY1: 1.97 per 1,000 catheter days; the incidence rate ratio (IRR) was 0.74 (95% CI=0.37 to 1.65, P=0.398); residual seven CLA-BSIs during IY1 were VRE faecium blood cultures positive from central line alone in the setting of findings explicable by noninfectious conditions. Following staff education, environmental decontamination and CHG bathing (IY2): 0.53 per 1,000 catheter days; the IRR was 0.20 (95% CI=0.06 to 0.65, P=0.008) with 80% reduction compared to the baseline. Over the two-year intervention period, the overall rate decreased by 53% to 1.24 per 1,000 catheter-days (IRR of 0.47 (95% CI=0.25 to 0.88, P=0.019) with zero CLA-BSI for a total of 15 months. CONCLUSIONS: Residual CLA-BSIs, despite strict adherence to central line bundle, may be related to blood culture contamination categorized as CLA-BSIs per CDC/NHSN definition. Efforts to reduce residual CLA-BSIs require a strategic multidisciplinary team approach focused on epidemiologic investigations of practitioner- or unit-specific etiologies.