Dheeraj Raju

Exploring factors associated with pressure ulcers: A data mining approach

BACKGROUND Pressure ulcers are associated with a nearly three-fold increase in in-hospital mortality. It is essential to investigate how other factors besides the Braden scale could enhance the prediction of pressure ulcers. Data mining modeling techniques can be beneficial to conduct this type of analysis. Data mining techniques have been applied extensively in health care, but are not widely used in nursing research. PURPOSE To remedy this methodological gap, this paper will review, explain, and compare several data mining models to examine patient level factors associated with pressure ulcers based on a four year study from military hospitals in the United States. METHODS The variables included in the analysis are easily accessible demographic information and medical measurements. Logistic regression, decision trees, random forests, and multivariate adaptive regression splines were compared based on their performance and interpretability. RESULTS The random forests model had the highest accuracy (C-statistic) with the following variables, in order of importance, ranked highest in predicting pressure ulcers: days in the hospital, serum albumin, age, blood urea nitrogen, and total Braden score. CONCLUSION Data mining, particularly, random forests are useful in predictive modeling. It is important for hospitals and health care systems to use their own data over time for pressure ulcer risk prediction, to develop risk models based upon more than the total Braden score, and specific to their patient population.

Associations between advanced cancer patients' survival and family caregiver presence and burden

We conducted a randomized controlled trial (RCT) of an early palliative care intervention (ENABLE: Educate, Nurture, Advise, Before Life Ends) for persons with advanced cancer and their family caregivers. Not all patient participants had a caregiver coparticipant; hence, we explored whether there were relationships between patient survival, having an enrolled caregiver, and caregiver outcomes prior to death. One hundred and twenty‐three patient‐caregiver dyads and 84 patients without a caregiver coparticipant participated in the ENABLE early versus delayed (12 weeks later) RCT. We collected caregiver quality‐of‐life (QOL), depression, and burden (objective, stress, and demand) measures every 6 weeks for 24 weeks and every 3 months thereafter until the patients death or study completion. We conducted survival analyses using log‐rank and Cox proportional hazards models. Patients with a caregiver coparticipant had significantly shorter survival (Wald = 4.31, HR = 1.52, CI: 1.02–2.25, P = 0.04). After including caregiver status, marital status (married/unmarried), their interaction, and relevant covariates, caregiver status (Wald = 6.25, HR = 2.62, CI: 1.23–5.59, P = 0.01), being married (Wald = 8.79, HR = 2.92, CI: 1.44–5.91, P = 0.003), and their interaction (Wald = 5.18, HR = 0.35, CI: 0.14–0.87, P = 0.02) were significant predictors of lower patient survival. Lower survival in patients with a caregiver was significantly related to higher caregiver demand burden (Wald = 4.87, CI: 1.01–1.20, P = 0.03) but not caregiver QOL, depression, and objective and stress burden. Advanced cancer patients with caregivers enrolled in a clinical trial had lower survival than patients without caregivers; however, this mortality risk was mostly attributable to higher survival by unmarried patients without caregivers. Higher caregiver demand burden was also associated with decreased patient survival.

Exploring Student Characteristics of Retention that Lead to Graduation in Higher Education Using Data Mining Models

The study used earliest available student data from a flagship university in the southeast United States to build data mining models like logistic regression with different variable selection methods, decision trees, and neural networks to explore important student characteristics associated with retention leading to graduation. The decision tree and logistic regression models indicated first semester GPA, earned credit hours after end of first semester, status (full/part time) at the end of semester, and high school GPA as the most important variables. Of the 22,099 students who were full-time, first time freshmen from 1995-2005, 7,293 did not graduate (33%). Out of the 7,293 who did not graduate, 2,845 students (39%) had first semester GPA < 2.25 with less than 12 earned credit hours. Characteristics of student retention leading to graduation can be predicted as early as end first semester instead of waiting until the end of the first year of school.

The Practice Environment Scale of the Nursing Work Index: An updated review and recommendations for use

OBJECTIVES The Practice Environment Scale of the Nursing Work Index (PES-NWI) is an instrument, which measures the nursing practice environment - defined as factors that enhance or attenuate a nurses ability to practice nursing skillfully and deliver high quality care. The purpose of this paper is to provide an updated review of the Practice Environment Scale of the Nursing Work Indexs use to date and provide recommendations that may be helpful to nursing leaders and researchers who plan to use this instrument. DESIGN A narrative review of quantitative studies. DATA SOURCES PubMed, EMBASE, and the Cumulative Index to Nursing & Allied Health Literature were searched to identify relevant literature using the search terms, Practice Environment Scale of the Nursing Work Index and PES-NWI. REVIEW METHODS Studies were included if they were published in English between 2010 and 2016 and focused on the relationship between the Practice Environment Scale of the Nursing Work Index and patient, nurse, or organizational outcomes. Data extraction focused on the reported survey scores and the significance and strength of the reported associations. RESULTS Forty-six articles, from 28 countries, were included in this review. The majority reported significant findings between the nursing practice environment and outcomes. Although some modifications have been made, the instrument has remained primarily unchanged since its development. Most often, the scores regarding staffing and resource adequacy remained the lowest. CONCLUSION The frequency of use of this instrument has remained high. Many researchers advocate for a move beyond the study of the connection between the Practice Environment Scale and nurse, patient, and organizational outcomes. Research should shift toward identifying interventions that improve the environment in which nurses practice and determining if changing the environment results in improved care quality.

Using Item Response Theory Models to Evaluate the Practice Environment Scale

Background and Purpose: The purpose of this article is to introduce different types of item response theory models and to demonstrate their usefulness by evaluating the Practice Environment Scale. Methods: Item response theory models such as constrained and unconstrained graded response model, partial credit model, Rasch model, and one-parameter logistic model are demonstrated. The Akaike information criterion (AIC) and Bayesian information criterion (BIC) indices are used as model selection criterion. Results: The unconstrained graded response and partial credit models indicated the best fit for the data. Almost all items in the instrument performed well. Conclusions: Although most of the items strongly measure the construct, there are a few items that could be eliminated without substantially altering the instrument. The analysis revealed that the instrument may function differently when administered to different unit types.

Adaptation of the Practice Environment Scale for military nurses: A psychometric analysis

AIM The aim of this study was to confirm the psychometric properties of Practice Environment Scale of the Nursing Work Index in a military population. This study also demonstrates association rule analysis, a contemporary exploratory technique. BACKGROUND One of the instruments most commonly used to evaluate the nursing practice environment is the Practice Environment Scale of the Nursing Work Index. Although the instrument has been widely used, the reliability, validity and individual item function are not commonly evaluated. Gaps exist with regard to confirmatory evaluation of the subscale factors, individual item analysis and evaluation in the outpatient setting and with non-registered nursing staff. DESIGN This was a secondary data analysis of existing survey data. METHODS Multiple psychometric methods were used for this analysis using survey data collected in 2014. First, descriptive analyses were conducted, including exploration using association rules. Next, internal consistency was tested and confirmatory factor analysis was performed to test the factor structure. The specified factor structure did not hold; therefore, exploratory factor analysis was performed. Finally, item analysis was executed using item response theory. The differential item functioning technique allowed the comparison of responses by care setting and nurse type. RESULTS The results of this study indicate that responses differ between groups and that several individual items could be removed without altering the psychometric properties of the instrument. CONCLUSION The instrument functions moderately well in a military population; however, researchers may want to consider nurse type and care setting during analysis to identify any meaningful variation in responses.

Plasma ADAMTS13 activity and von Willebrand factor antigen and activity in patients with subarachnoid haemorrhage

Increased von Willebrand factor (VWF) and reduced ADAMTS13 activity are associated with arterial thrombosis. This may also be the culprit mechanism implicated in delayed cerebral ischaemia after aneurysmal subarachnoid haemorrhage (SAH). It was our objective to determine plasma VWF and ADAMTS13 in patients with SAH and healthy subjects; and to explore the levels of those markers and outcome after SAH. Forty consecutive patients were enrolled between September 2007 and April 2014 in a pilot study. Plasma samples were collected from SAH patients on post-bleed day (PBD) 0, 1, 3, 5, 7 and 10 and healthy controls. VWF antigen (VWFAg) and VWF activity (VWFAc) were determined by enzyme-linked immunoassay and collagen binding assay, respectively. ADAMTS13 activity was determined by the cleavage of a fluorescent substrate. Univariate descriptive statistics and cluster analyses were performed based on outcomes in the group with SAH only. Mean age of SAH patients was 52.4 years (26-84 years) and 30 (75 %) were women. 12/40 (30 %) had a high Hunt and Hess grade (IV-V) and 25 (62.5 %) were treated with coil embolisation. Plasma VWFAg and VWFAc were significantly higher in SAH patients than those in healthy subjects on each PBD (p<0.0001). Concurrently, plasma ADAMTS13 activity in SAH patients was significantly lower than that in healthy subjects (p<0.0001). Among those with SAH, cluster analysis demonstrated that patients with higher VWFAg and VWFAc and/or lower ADAMTS13 activity might be at risk of increased mortality. In conclusion, the relative deficiency of plasma ADAMTS13 activity in SAH patients may associate with worse outcome.

Comparing Data Mining Models in Academic Analytics

The goal of this research study was to compare data mining techniques in predicting student graduation. The data included demographics, high school, ACT profile, and college indicators from 1995-2005 for first-time, full-time freshman students with a six year graduation timeline for a flagship university in the south east United States. The results indicated no difference in misclassification rates between logistic regression, decision tree, neural network, and random forest models. The results from the study suggest that institutional researchers should build and compare different data mining models and choose the best one based on its advantages. The results can be used to predict students at risk and help these students graduate.

“Black cloud” vs. “white cloud” physicians – Myth or reality in apheresis medicine?

Many practitioners believe in the phenomenon of either being labeled a “black cloud” or “white cloud” while on‐call. A “white‐cloud” physician is one who usually gets fewer cases. A “black‐cloud” is one who often has more cases. It is unclear if the designation is only superstitious or if there is some merit. Our aim is to objectively assess this phenomenon in apheresis medicine at our center.

Peri-muscular adipose tissue may play a unique role in determining insulin sensitivity/resistance in women with polycystic ovary syndrome

STUDY QUESTION Do the determinants of insulin sensitivity/resistance differ in women with and without polycystic ovary syndrome (PCOS)? SUMMARY ANSWER Peri-muscular thigh adipose tissue is uniquely associated with insulin sensitivity/resistance in women with PCOS, whereas adiponectin and thigh subcutaneous adipose are the main correlates of insulin sensitivity/resistance in women without PCOS. WHAT IS KNOWN ALREADY In subject populations without PCOS, insulin sensitivity/resistance is determined by body fat distribution and circulating concentrations of hormones and pro-inflammatory mediators. Specifically, visceral (intra-abdominal) adipose tissue mass is adversely associated with insulin sensitivity, whereas thigh subcutaneous adipose appears protective against metabolic disease. Adiponectin is an insulin-sensitizing hormone produced by healthy subcutaneous adipose that may mediate the protective effect of thigh subcutaneous adipose. Testosterone, which is elevated in PCOS, may have an adverse effect on insulin sensitivity/resistance. STUDY DESIGN, SIZE, DURATION Cross-sectional study of 30 women with PCOS and 38 women without PCOS; data were collected between 2007 and 2011. PARTICIPANTS/MATERIALS, SETTING, METHODS Participants were group-matched for obesity, as reflected in BMI (Mean ± SD; PCOS: 31.8 ± 6.0 kg/m2; without PCOS: 31.5 ± 5.0 kg/m2). The whole-body insulin sensitivity index (WBISI) was assessed using a mixed-meal tolerance test; Homeostasis Model Assessment-Insulin resistance (HOMA-IR) was determined from fasting insulin and glucose values. Adipose tissue distribution was determined by computed tomography (CT) scan. Partial correlation analysis, adjusting for total fat mass, was used to identify correlates of WBISI and HOMA-IR within each group of women from measures of body composition, body fat distribution, reproductive-endocrine hormones and adipokines/cytokines. Stepwise multiple linear regression analysis was used to identify the variables that best predicted WBISI and HOMA-IR. MAIN RESULTS AND THE ROLE OF CHANCE Among women with PCOS, both WBISI and HOMA-IR were best predicted by peri-muscular adipose tissue cross-sectional area. Among women without PCOS, both WBISI and HOMA-IR were best predicted by adiponectin and thigh subcutaneous adipose tissue. LIMITATIONS, REASONS FOR CAUTION Small sample size, group matching for BMI and age, and the use of surrogate measures of insulin sensitivity/resistance. WIDER IMPLICATIONS OF THE FINDINGS Because insulin resistance is the root cause of obesity and comorbidities in PCOS, determining its cause could lead to potential therapies. Present results suggest that peri-muscular adipose tissue may play a unique role in determining insulin sensitivity/resistance in women with PCOS. Interventions such as restriction of dietary carbohydrates that have been shown to selectively reduce fatty infiltration of skeletal muscle may decrease the risk for type 2 diabetes in women with PCOS. STUDY FUNDING/COMPETING INTEREST(S) The study was supported by National Institutes of Health grants R01HD054960, R01DK67538, P30DK56336, P60DK079626, M014RR00032 and UL1RR025777. The authors have no conflicts of interest. TRIAL REGISTRATION NUMBER NCT00726908.