Beyond Hospital-Level Aggregated Data: A Methodology to Adapt Clinical Data From the Electronic Health Record for Nursing Unit-Level Research.

BACKGROUND: Studies of nurse staffing frequently use data aggregated at the hospital level that do not provide the appropriate context to inform unit-level decisions, such as nurse staffing. OBJECTIVES: Describe a method to link patient data collected during the provision of routine care and recorded in the electronic health record (EHR) to the nursing units where care occurred in a national dataset. RESEARCH DESIGN: We identified all Veterans Health Administration acute care hospitalizations in the calendar year 2019 nationwide. We linked patient-level EHR and bar code medication administration data to nursing units using a crosswalk. We divided hospitalizations into segments based on the patient's time-stamped location (ward stays). We calculated the number of ward stays and medication administrations linked to a nursing unit and the unit-level and facility-level mean patient risk scores. RESULTS: We extracted data on 1117 nursing units, 3782 EHR patient locations associated with 1,137,391 ward stays, and 67,772 bar code medication administration locations associated with 147,686,996 medication administrations across 125 Veterans Health Administration facilities. We linked 89.46% of ward stays and 93.10% of medication administrations to a nursing unit. The average (standard deviation) unit-level patient severity across all facilities is 4.71 (1.52), versus 4.53 (0.88) at the facility level. CONCLUSIONS: Identification of units is indispensable for using EHR data to understand unit-level phenomena in nursing research and can provide the context-specific information needed by managers making frontline decisions about staffing.

An Alternative Method of Public Reporting of Comparative Hospital Quality and Performance Data for Transparency Initiatives.

BACKGROUND: Hospital performance comparisons for transparency initiatives may be inadequate if peer comparison groups are poorly defined. OBJECTIVE: The objective of this study was to evaluate a new approach identifying hospital peers for comparison. DESIGN/SETTING: We used Mahalanobis distance as a new method of developing peer-specific groupings for hospitals to incorporate both external and internal complexity. We compared the overlap in groups with an existing method used by the Veterans' Health Administration's Office for Productivity, Efficiency, and Staffing (OPES). PARTICIPANTS: One hundred twenty-two acute-care Veterans' Health Administration's Medical Facilities as defined in the OPES fiscal year 2014 report. MEASURES: Using 15 variables in 9 categories developed from expert input, including both hospital internal measures and community-based external measures, we used principal components analysis and calculated Mahalanobis distance between each hospital pair. This method accounts for correlation between variables and allows for variables having different variances. We identified the 50 closest hospitals, then eliminated any potential peer whose score on the first component was >1 SD from the reference hospital. We compared overlap with OPES measures. RESULTS: Of 15 variables, 12 have SDs exceeding 25% of their means. The first 2 components of our analysis explain 24.8% and 18.5% of variation among hospitals. Eight of 9 variables scaling positively on the first component measure internal complexity, aligning with OPES groups. Four of 5 variables scaling positively on the second component but not the first are factors from the policy environment; this component reflects a dimension not considered in OPES groups. CONCLUSION: Individualized peers that incorporate external complexity generate more nuanced comparators to evaluate quality.

Statins and the risk of colorectal carcinoma: a nested case-control study in veterans with diabetes.

OBJECTIVES: Experimental data indicate a possible preventive effect for statins in colorectal cancer (CRC). However, the available epidemiological data are conflicting. METHODS: We conducted a nested case-control study of veterans with diabetes in national databases of the Department of Veterans Affairs (VA) and Medicare-linked files. Cases were defined as incident CRC during January 2001-December 2002, sampled on incidence density. VA pharmacy benefits management (PBM) files were used to identify filled prescriptions for statins. Multivariable conditional logistic regression models were used to estimate odds ratios (ORs) after adjusting for potential confounding variables. Stratified analyses were conducted for potential effect modifiers. RESULTS: A total of 6,080 cases and 24,320 controls were examined. The mean age was 74 years, and the majority of patients were Caucasian (88%) and male (99%). Filled prescriptions of statins were recorded less frequently in cases (49%) than in controls (52%; OR: 0.88; 95% confidence interval (95% CI): 0.83-0.93). This inverse association remained significant after adjusting for inflammatory bowel disease, diabetes severity, cholecystectomy, liver disease, filled prescriptions for sulfonylurea, aspirin or NSAID use, or colorectal evaluation. Simvastatin comprised the majority (87%) of statin-filled prescriptions, and the association with risk of CRC with simvastatin was very similar to that of any statin. No significant associations were observed between the risk of CRC and nonstatin cholesterol (OR: 1.02; 95% CI 0.88-1.18) or triglyceride-lowering medications (OR: 0.96; 95% CI: 0.87-1.05). The significant inverse association was limited to Caucasians, patients without history of polyps, patients aged 65 years and older, and patients with colon cancer (excluding rectum). CONCLUSIONS: The use of statins was associated with a small reduction in the risk of colon cancer in patients with diabetes. However, the causal link is not clear.

Effect of using information from only one system for dually eligible health care users.

OBJECTIVE: We sought to determine whether all diagnoses and total illness burden of patients who use both the VA and Medicare health care systems can be obtained from examination of data from only one of these systems. METHODS: Cohorts included all age-eligible Medicare users who also used the VA health care system in fiscal years 2000-2002 but were not enrolled in a Medicare HMO. Relative risk scores (RRS; a measure of illness burden developed by DxCG, Inc., Boston, MA) were calculated using VA, Medicare, and all diagnoses from both VA and Medicare data sources. The relationship between RRS and reliance on Medicare versus the VA system also was explored. We explored whether differences in VA and Medicare RRS were caused by veterans who mainly used pharmacy services or by an underweighting in the RRS calculation of mental health diagnoses. Finally, we explored the relationship between inpatient utilization and RRS in each system. RESULTS: On average for a given patient who used both VA and Medicare services, more diagnoses were recorded in Medicare ( approximately 13-15) than in the VA system ( approximately 8) for dual users. On average only 2 diagnoses were common to both the VA and Medicare. Medicare data alone accounted for approximately 80% of individuals' total illness burden, and VA data alone lead to RRSs that capture one-third of the total illness burden. The ratio of RRS when calculated using Medicare and VA separately was approximately 2.4. RRS was only weakly to moderately correlated with inpatient utilization in each system. CONCLUSION: Using data from just Medicare or VA data sources when conducting research on dually eligible veterans may seriously underestimate total illness burden of the population and also may lead to an underidentification of individuals in a particular disease class.

Using GIS in government: an overview of the VHA's Healthcare Atlas, FY-2000.

The amount of VA data available for analysis can be overwhelming to individuals who need to translate these data into usable information. The Atlas, using current GIS technology, was funded to provide data in a comprehensive guide. Patients were identified using a disease classification scheme based on Kaiser Permanente methodology and the Clinical Classifications Software (AHRQ). Utilization data were extracted from the Medical SAS Datasets. Cost data were obtained from the HERC. GIS tools were used to create the Atlas. The Atlas overviews the location of VA hospitals; profiles veteran, VA enrollee and patient populations; examines overall utilization; depicts patterns in healthcare use by specific disease cohorts; and examines geographic variations in costs. This product will enhance knowledge of VA's enrolled patient population and their healthcare needs, and provide background information that will improve the formulation of specific research questions to address those needs.

Using GIS to profile health-care costs of VA Quality-Enhancement Research Initiative diseases.

The Health Services Research and Development (HSR&D) Service at the Department of Veterans Affairs (VA) Health Care System launched a Quality Enhancement Research Initiative (QUERI) in 1998. This study estimated health-care costs of nine diseases under the QUERI project and analyzed geographic differences in health-care costs and utilization across 22 VA Integrated Service Networks (VISNs), using a geographic information system (GIS). Patients with these diseases were identified from diagnoses recorded between October 1999 and September 2000. Annual health-care costs for each disease were estimated in four categories: inpatient medical or surgical, other inpatient, outpatient, and outpatient pharmacy. Geographic differences of costs and health-care utilization across the 22 VISNs for chronic heart failure, diabetes, and spinal-cord injury were mapped using a GIS package. Average costs and patterns of health-care utilization varied substantially across the 22 VISNs. The observed differences in health-care utilization across geographic regions raised questions for further investigation.

Proton pump inhibitors are associated with reduced incidence of dysplasia in Barrett's esophagus.

BACKGROUND AND AIMS: Esophageal acid exposure is important in the pathogenesis of Barrett's esophagus (BE), and possibly in the progression of BE to dysplasia and carcinoma. The aim of this study is to compare the development of dysplasia in BE patients treated with or without proton pump inhibitor (PPI) or histamine 2-receptor antagonist (H2RA). METHODS: We analyzed prospectively collected data by a single endoscopist on patients with BE in a VA (Veterans Affairs) setting over a 20-yr time period (1981-2000). A pathologist used standard criteria to diagnose BE/dysplasia. Pharmacy information after 1994 was retrieved from a computerized database, and from research files for the period before that. The receipt and the duration of H2RA and/or PPI use was compared between those with and without dysplasia. The incidence of dysplasia was examined in a Kaplan-Meier survival analysis stratified by PPI treatment status, and the risk of dysplasia was examined in a Cox multiple regression analysis controlling for demographic features, length of BE, and the year of BE diagnosis. RESULTS: We analyzed data for 236 unique veteran patients with a mean age at BE diagnosis of 61.5 yr, 86% Caucasian, and 98% male. During 1,170 patient-yr of follow-up, 56 patients developed dysplasia giving an annual incidence rate of 4.7%. Of those, 14 had high-grade dysplasia. The cumulative incidence of dysplasia was significantly lower among patients who received PPI after BE diagnosis than in those who received no therapy or H2RA; log rank test (p < 0.001). Furthermore, among those on PPIs, a longer duration of use was associated with less frequent occurrence of dysplasia. In multivariate analysis, the use of PPI after BE diagnosis was independently associated with reduced risk of dysplasia, hazards ratio: 0.25 (95% CI 0.13-0.47), p < 0.0001. Longer segments of BE and Caucasian race were other independent risk factors for developing dysplasia. In general, similar findings were observed when only cases with high-grade dysplasia were analyzed. CONCLUSIONS: These results indicate that PPI therapy is associated with a significant reduction in the risk of developing dysplasia in patients with BE. However, more studies are required to confirm this finding.

Is the length of newly diagnosed Barrett's esophagus decreasing? The experience of a VA Health Care System.

BACKGROUND & AIMS: Secular trends in the length of newly diagnosed Barrett's esophagus (BE) are unknown. We have anecdotally noticed less frequent new diagnoses of long segments of BE. METHODS: This is a retrospective analysis of prospectively collected information on a well-characterized large cohort of patients with documented BE that was diagnosed between 1981 and 2000 at Southern Arizona Department of Veterans Affairs Health Care System. We examined temporal changes in the length of BE at the time of diagnosis (frequency and proportions). We conducted correlation analyses, as well as multivariate linear regression analyses, to examine the association between year of diagnosis and BE length while adjusting for temporal differences in age, sex, ethnicity, previous use of antisecretory therapy, and the presence of intestinal metaplasia (IM) of the gastric cardia. RESULTS: There were 340 patients with BE first diagnosed between 1981 and 2000. All cases were defined by the presence of areas of salmon-colored mucosa in the lower end of the tubular esophagus and IM in biopsy specimens obtained from these areas on at least 2 endoscopic examinations. There were no significant changes over time in mean age of patients with BE (61 yr) or proportion of white patients (84%). The mean length of BE at the time of first diagnosis declined progressively over time. In the earliest period (1981-1985), mean BE length was 6 +/- 3.8 cm, whereas mean BE length in 1996-2000 was 3.6 +/- 2.9 cm. This observation was explained not only by more frequent diagnoses of short BE, but also by less frequent diagnoses of long BE (> or =3 cm). There was a strong inverse correlation between BE length at the time of diagnosis and year of diagnosis (Pearson's correlation coefficient, -0.29; P <0.0001). In the multivariate linear regression model, a more recent year of BE diagnosis was an independent predictor of shorter BE length (P <0.0001). Similar results were obtained in analyses restricted to veteran patients or those with BE > or = 3 cm. CONCLUSIONS: There has been a progressive decline in the length of newly diagnosed BE as a result of an increase in short-segment BE, but, curiously, also because of a decline in long-segment BE (> or =3 cm). These changes cannot be explained fully by changes in demographic features of patients, previous therapy, or the increasing emphasis on IM of the gastric cardia. The role of referral bias and/or temporal changes in the definitions cannot be excluded.