Laura J. Myers

Estimating and Reporting on the Quality of Inpatient Stroke Care by Veterans Health Administration Medical Centers

Background— Reporting of quality indicators (QIs) in Veterans Health Administration Medical Centers is complicated by estimation error caused by small numbers of eligible patients per facility. We applied multilevel modeling and empirical Bayes (EB) estimation in addressing this issue in performance reporting of stroke care quality in the Medical Centers. Methods and Results— We studied a retrospective cohort of 3812 veterans admitted to 106 Medical Centers with ischemic stroke during fiscal year 2007. The median number of study patients per facility was 34 (range, 12–105). Inpatient stroke care quality was measured with 13 evidence-based QIs. Eligible patients could either pass or fail each indicator. Multilevel modeling of a patients pass/fail on individual QIs was used to produce facility-level EB-estimated QI pass rates and confidence intervals. The EB estimation reduced interfacility variation in QI rates. Small facilities and those with exceptionally high or low rates were most affected. We recommended 8 of the 13 QIs for performance reporting: dysphagia screening, National Institutes of Health Stroke Scale documentation, early ambulation, fall risk assessment, pressure ulcer risk assessment, Functional Independence Measure documentation, lipid management, and deep vein thrombosis prophylaxis. These QIs displayed sufficient variation across facilities, had room for improvement, and identified sites with performance that was significantly above or below the population average. The remaining 5 QIs were not recommended because of too few eligible patients or high pass rates with little variation. Conclusions— Considerations of statistical uncertainty should inform the choice of QIs and their application to performance reporting.

Lower Endoscopy Reduces Colorectal Cancer Incidence in Older Individuals

BACKGROUND & AIMS In older individuals, there are unclear effects of lower endoscopy on incidence of colorectal cancer (CRC) and of colonoscopy on site of CRC. We investigated whether sigmoidoscopy or colonoscopy is associated with a decreased incidence of CRC in older individuals, and whether the effect of colonoscopy differs by anatomic location. METHODS We performed a case-control study using linked US Veterans Affairs and Medicare data. Cases were veterans aged 75 years or older diagnosed with CRC in fiscal year 2007. Cases were matched for age and sex to 3 individuals without a CRC diagnosis (controls). We determined the number of cases and controls who received colonoscopies or sigmoidoscopies from fiscal year 1997 to a date 6 months before the diagnosis of CRC (for cases) or to a corresponding index date (for controls). The probability of exposure was modeled using generalized linear mixed equations, adjusted for potential confounders. For the analysis of CRC risk in different anatomic locations, the proximal colon was defined as proximal to the splenic flexure. RESULTS We identified 623 cases and 1869 controls (mean age, 81 y; 98.7% male, 86.2% Caucasian). Among cases, 243 (39.0%) underwent any lower endoscopy (177 colonoscopies). Among controls, 978 (52.3%) underwent any lower endoscopy (758 colonoscopies). Cases were significantly less likely than controls to have undergone lower endoscopy within the preceding 10 years (adjusted odds ratio [aOR], 0.58; 95% confidence interval [CI], 0.48-0.69). This effect was significant for colonoscopy (aOR, 0.57; 95% CI, 0.47-0.70), but not sigmoidoscopy. Similar results were observed when a 5-year exposure window was applied. Colonoscopy was associated with a reduced risk of distal CRC (aOR, 0.45; 95% CI, 0.32-0.62) and proximal CRC (aOR, 0.65; 95% CI, 0.46-0.92). CONCLUSIONS In a study of the US Veterans Affairs and Medicare databases, lower endoscopy in the preceding 10 years was associated with a significant reduction in CRC incidence among older veterans. Colonoscopy was associated with significant reductions in distal and proximal CRC.

Do-not-resuscitate orders, quality of care, and outcomes in veterans with acute ischemic stroke.

Objective: There is concern that do-not-resuscitate (DNR) orders may lead to stroke patients receiving less aggressive treatment and poorer care. Our objectives were to assess the relationship between DNR orders and quality of stroke care among veterans. Methods: A cohort of 3,965 acute ischemic stroke patients admitted to 131 Veterans Health Administration (VHA) facilities in fiscal year 2007 underwent chart abstraction. DNR codes were identified through electronic orders or by documentation of “no code,” “no cardiopulmonary resuscitation,” or “no resuscitation.” Quality of care was measured using 14 inpatient ischemic stroke quality indicators. The association between DNR orders and quality indicators was examined using multivariable logistic regression. Results: Among 3,965 ischemic stroke patients, 535 (13.5%) had DNR code status, 71% of whom had orders first documented within 1 day of admission. Overall, 4.9% of patients died in-hospital or were discharged to hospice; these outcomes were substantially higher in patients with DNR orders (29.7%), particularly if they were not documented until ≥2 days after admission (47.1%). Patients with DNR orders were significantly older, had more comorbidities, and had greater stroke severity. Following adjustment there were few significant associations between DNR status and the 14 quality indicators, with the exception of lower odds of early ambulation (odds ratio = 0.58, 95% confidence interval = 0.41–0.81) in DNR patients. Conclusions: DNR orders were associated with limited differences in the select quality indicators investigated, which suggests that DNR orders did not impact quality of care. However, whether DNR orders influence treatment decisions that more directly affect survival remains to be determined.

Effect of Clinical and Social Risk Factors on Hospital Profiling for Stroke Readmission: A Cohort Study

Context Whether adjustment for medical and social risk factors available from administrative and electronic health records improves models to predict hospital readmission after stroke is not known. Contribution The addition of detailed clinical information and social risk factors to a Centers for Medicare & Medicaid Services model that includes only age, sex, and comorbid conditions did not substantially alter the evaluation of 30-day readmission performance of Veterans Health Administration hospitals caring for patients with stroke. Implication More comprehensive models to evaluate hospital readmission after stroke might not improve upon simpler models. The Editors Stroke is the fourth leading cause of death and a leading cause of disability among U.S. adults and the second most common cause of hospitalization in elderly persons (1). In the Veterans Health Administration (VA), more than 6000 veterans are hospitalized annually for acute ischemic stroke in a VA medical center (2). Within the VA, stroke is common and costly; understanding factors predictive of readmission is important to reducing 30-day readmission rates and improving outcomes. In some mortality models, differences in disease severity explain much of the variation in mortality rates among patients. However, most published 30-day readmission models, even those with detailed clinical data, explain little of the variation in readmission rates among patients (3). Recent research has shown that the predictive ability of 30-day readmission models for patients with congestive heart failure improves with the addition of clinically detailed information, such as severity of disease indices and social risk factors that represent the degree of chaos and social risk in a patients life (4). Identifying the social and clinical factors associated with 30-day stroke readmission would allow targeted interventions, such as supported discharge transitions, care coordination, home visits, physical therapy, earlier appointments, and greater education efforts, to be implemented to prevent readmission and potentially intervene on behalf of patients who are at the highest risk for readmission. A recent systematic review of stroke readmission showed limited literature on 30-day stroke readmission. An improved understanding of factors predictive of this outcome is warranted (5). In addition, the Centers for Medicare & Medicaid Services (CMS) has selected 30-day stroke readmission as a measure of hospital quality. It will also be used in the VA and reported on Hospital Compare; however, the proposed CMS stroke readmission model, like all other prediction models developed by the CMS for Hospital Compare, includes only age, sex, and comorbid conditions. The CMS has taken the position that including race in the model holds hospitals that care for minority populations to a different standard (6). Similarly, other variables of social risk (for example, low income and homelessness) are not included. However, current readmission models might penalize hospitals caring for disadvantaged populations with more needs. In response to this concern, the National Quality Forum invited public comment on whether sociodemographic variables should be included in hospital profiling and has published a draft report suggesting that adjustment for these factors may be warranted (7). Nevertheless, given the lack of availability of such variables in Medicare data, information is limited on how they may affect hospital profiling. Because the VA shares outcome data with the Hospital Compare program for other conditions, understanding the effects of including the best clinical and social risk factors available on hospital-level comparisons in the VA is informative to policymakers. In this article, we examine the effect of including these factors on hospital profiling based on 30-day readmission rates. Methods Overview We used data from the 2007 VA Office of Quality and Performance Stroke Special Project to construct 3 patient-level models that examined predictors of 30-day readmission. First, we developed a 30-day readmission model using methods outlined by the CMS (model 1) (8). Second, we compared this model with one including some measures of social risk (for example, homelessness and substance abuse) from VA administrative data (model 2). Finally, we compared model 1 with a model that included social risk and clinical factors using data from medical record review (such as stroke severity and Acute Physiology and Chronic Health Evaluation [APACHE] and Morse Fall Scale scores) (model 3). We then ranked hospitals by their 30-day risk-standardized readmission rates (RSRRs) for each model and examined facility rankings among the 3 models to determine whether a more comprehensive model (models 2 or 3) classified hospital performance differently from the CMS-based readmission model (model 1). Data Source and Sample A sample of 5000 veterans admitted to a VA hospital in fiscal year 2007 (9) with a primary discharge diagnosis of ischemic stroke was identified from VA administrative data by using a modified high-specificity algorithm of International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), codes (10). Abstractors from the West Virginia Medical Institute who were specially trained for this study collected data through retrospective chart review of medical records. A total of 1013 patients were excluded because they had carotid endarterectomy during the stroke hospitalization, the stroke occurred after admission, or they were admitted only for poststroke rehabilitation. Patients were further excluded if they were transferred and ultimately discharged from a non-VA hospital (because we were interested only in VA hospitals), were discharged against medical advice (because providers did not have the opportunity to deliver full care and prepare the patient for discharge), died during the index hospitalization (because they were not eligible for readmission), or were enrolled in a Medicare HMO in the month before or after the index stroke admission (because readmission information may not be available). The final analytic sample included 3436 patients from 114 hospitals. Dependent Variable Our main dependent variable was unplanned, all-cause, 30-day readmission rates. We adhered to the principles outlined in CMS technical documents to define hospital readmissions (6, 8, 11). Specifically, readmissions were defined as a subsequent inpatient admission to any acute care facility in the VA or a CMS facility within 30 days of discharge from the index stroke hospitalization. We used the recently updated CMS Planned Readmission Algorithm, version 3.0, for stroke to identify planned readmissions (8), which determines hospitalizations that may be planned for follow-up stroke care (for example, carotid endarterectomy) and other commonly planned hospitalizations (such as elective cholecystectomy). Per the updated CMS Planned Readmission Algorithm, if the first readmission after discharge is planned, any subsequent unplanned readmissions are not counted as an outcome because the unplanned readmission may be related to care received during the planned readmission. Thus, patients with a planned readmission followed by an unplanned readmission within 30 days of discharge were considered not to have had a readmission. Readmissions were assessed from several data sources, including VA medical SAS inpatient data (SAS Institute), the VA Non-VA Medical Care (formerly Fee Basis) inpatient files (which include care at non-VA facilities that is paid for by the VA), and CMS Medical Provider Analysis and Review files for hospitalizations in the Medicare program. Independent Variables Independent variables included demographic, clinical, and social risk factors associated with 30-day readmission. These domains were built on previous work that identified domains of health associated with 30-day readmission (4, 12). Demographic Characteristics Demographic variables included age, sex, and race data available in VA national data sets for each veteran. Data on race were 98% complete. Clinical Characteristics For clinical characteristics, we included data on disease severity, functional status, and utilization. Disease Severity. The 25 clinical covariates outlined in the CMS technical documents for creating the 30-day stroke readmission measure for the CMS-based model (model 1) were identified in VA and CMS administrative data in the 12 months before the stroke admission and were included as separate dichotomous variables in the CMS risk-standardized readmission models. For the other 2 models, we included additional clinical measures based on data from the medical record review. Trained abstractors retrospectively constructed the National Institutes of Health Stroke Scale, a measure of stroke severity, from physician notes within 24 hours of admission (13). Disease severity was represented by a modified APACHE score (14) calculated using admission data for each veteran. Data on hypoxia, dysphagia, and code status were collected during medical record abstraction. We calculated a Charlson Comorbidity Index score (15) based on data on each patients medical history. Finally, as another measure of health status, we created a dichotomous variable that indicated whether a veterans copayment was waived for VA health care services because of a military serviceconnected disability. Functional Status. We used 3 measures to categorize functional status. First, the Morse Fall Scale is a validated measure documented in the medical record for each veteran during hospitalization (1618). The scale ranges from 0 to 150, and a score greater than 50 indicates a high risk for falls. Second, chart abstractors classified patients as ambulatory before the stroke, nonambulatory, or of unknown status. Third, using administrative data, we categorized patients on the basis of whether they were receiving VA home-based primary care o

Does the Inclusion of Stroke Severity in a 30-Day Mortality Model Change Standardized Mortality Rates at Veterans Affairs Hospitals?

Background— The Centers for Medicare and Medicaid Services is considering developing a 30-day ischemic stroke hospital-level mortality model using administrative data. We examined whether inclusion of the National Institutes of Health Stroke Scale (NIHSS), a measure of stroke severity not included in administrative data, would alter 30-day mortality rates in the Veterans Health Administration. Methods and Results— A total of 2562 veterans admitted with ischemic stroke to 64 Veterans Health Administration Hospitals in the fiscal year 2007 were included. First, we examined the distribution of unadjusted mortality rates across the Veterans Health Administration. Second, we estimated 30-day all-cause, risk standardized mortality rates (RSMRs) for each hospital by adjusting for age, sex, and comorbid conditions using hierarchical models with and without the inclusion of the NIHSS. Finally, we examined whether adjustment for the NIHSS significantly changed RSMRs for each hospital compared with other hospitals. The median unadjusted mortality rate was 3.6%. The RSMR interquartile range without the NIHSS ranged from 5.1% to 5.6%. Adjustment with the NIHSS did not change the RSMR interquartile range (5.1%–5.6%). Among veterans ≥65 years, the RSMR interquartile range without the NIHSS ranged from 9.2% to 10.3%. With adjustment for the NIHSS, the RSMR interquartile range changed from 9.4% to 10.0%. The plot of 30-day RSMRs estimated with and without the inclusion of the NIHSS in the model demonstrated overlapping 95% confidence intervals across all hospitals, with no hospital significantly below or above the mean-unadjusted 30-day mortality rate. The 30-day mortality measure did not discriminate well among hospitals. Conclusions— The impact of the NIHSS on RSMRs was limited. The small number of stroke admissions and the narrow range of 30-day stroke mortality rates at the facility level in the Veterans Health Administration cast doubt on the value of using 30-day RSMRs as a means of identifying outlier hospitals based on their stroke care quality.

Implementing Geriatric Resources for Assessment and Care of Elders Team Care in a Veterans Affairs Medical Center: Lessons Learned and Effects Observed.

In a randomized clinical trial, Geriatric Resources for Assessment and Care of Elders (GRACE), a model of care that works in collaboration with primary care providers (PCPs) and patient‐centered medical homes to provide home‐based geriatric care management focusing on geriatric syndromes and psychosocial problems commonly found in older adults, improved care quality and reduced acute care use for high‐risk, low‐income older adults. To assess the effect of GRACE at a Veterans Affairs (VA) Medical Center (VAMC), veterans aged 65 and older from Marion County, Indiana, with PCPs from four of five VAMC clinics who were not on hospice or dialysis were enrolled in GRACE after discharge home from an acute hospitalization. After an initial home‐based transition visit to GRACE enrollees, the GRACE team returned to conduct a geriatric assessment. Guided by 12 protocols and input from an interdisciplinary panel and the PCP, the GRACE team developed and implemented a veteran‐centric care plan. Hospitalized veterans from the fifth clinic, who otherwise met enrollment criteria, served as a usual‐care comparison group. Demographic, comorbidity, and usage data were drawn from VA databases. The GRACE and comparison groups were similar in age, sex, and burden of comorbidity, although predicted risk of 1‐year mortality in GRACE veterans was higher. Even so, GRACE enrollment was associated with 7.1% fewer emergency department visits, 14.8% fewer 30‐day readmissions, 37.9% fewer hospital admissions, and 28.5% fewer total bed days of care, saving the VAMC an estimated

Factors associated with program utilization of radiation therapy treatment for VHA and medicare dually enrolled patients.

200,000 per year after program costs during the study for the 179 veterans enrolled in GRACE. Having engaged, enthusiastic VA leadership and GRACE staff; aligning closely with the medical home; and accommodating patient acuity were among the important lessons learned during implementation.

Quality of care for veterans with transient ischemic attack and minor stroke

We examine how distance to a Veterans Health Administration (VHA) facility, patient hometown classification (e.g., small rural town), and service-connected disability are associated with veterans’ utilization of radiation therapy services across the VHA and Medicare. In 2008, 45,914 dually-enrolled veteran patients received radiation therapy. Over 3-quarters (35,513) of the patients received radiation therapy from the Medicare program. Younger age, male gender, shorter distance to a VHA facility, and VHA priority or disability status increased the odds of utilizing the VHA. However, veterans residing in urban areas were less likely to utilize the VHA. Urban dwelling patients’ utilization of Medicare instead of the VHA suggests a complex decision that incorporates geographic access to VHA services, financial implications of veteran priority status, and the potential availability of multiple sources of radiation therapy in competitive urban markets.

National Implementation of Acute Stroke Care Centers in the Veterans Health Administration (VHA): Formative Evaluation of the Field Response

Importance The timely delivery of guideline-concordant care may reduce the risk of recurrent vascular events for patients with transient ischemic attack (TIA) and minor stroke. Although many health care organizations measure stroke care quality, few evaluate performance for patients with TIA or minor stroke, and most include only a limited subset of guideline-recommended processes. Objective To assess the quality of guideline-recommended TIA and minor stroke care across the Veterans Health Administration (VHA) system nationwide. Design, Setting, and Participants This cohort study included 8201 patients with TIA or minor stroke cared for in any VHA emergency department (ED) or inpatient setting during federal fiscal year 2014 (October 1, 2013, through September 31, 2014). Patients with length of stay longer than 6 days, ventilator use, feeding tube use, coma, intensive care unit stay, inpatient rehabilitation stay before discharge, or receipt of thrombolysis were excluded. Outlier facilities for each process of care were identified by constructing 95% CIs around the facility pass rate and national pass rate sites when the 95% CIs did not overlap. Data analysis occurred from January 16, 2016, through June 30, 2017. Main Outcomes and Measures Ten elements of care were assessed using validated electronic quality measures. Results In the 8201 patients included in the study (mean [SD] age, 68.8 [11.4] years; 7877 [96.0%] male; 4856 [59.2%] white), performance varied across elements of care: brain imaging by day 2 (6720/7563 [88.9%]; 95% CI, 88.2%-89.6%), antithrombotic use by day 2 (6265/7477 [83.8%]; 95% CI, 83.0%-84.6%), hemoglobin A1c measurement by discharge or within the preceding 120 days (2859/3464 [82.5%]; 95% CI, 81.2%-83.8%), anticoagulation for atrial fibrillation by day 7 after discharge (1003/1222 [82.1%]; 95% CI, 80.0%-84.2%), deep vein thrombosis prophylaxis by day 2 (3253/4346 [74.9%]; 95% CI, 73.6%-76.2%), hypertension control by day 90 after discharge (4292/5979 [71.8%]; 95% CI, 70.7%-72.9%), neurology consultation by day 1 (5521/7823 [70.6%]; 95% CI, 69.6%-71.6%), electrocardiography by day 2 or within 1 day prior (5073/7570 [67.0%]; 95% CI, 65.9%-68.1%), carotid artery imaging by day 2 or within 6 months prior (4923/7685 [64.1%]; 95% CI, 63.0%-65.2%), and moderate- to high-potency statin prescription by day 7 after discharge (3329/7054 [47.2%]; 95% CI, 46.0%-48.4%). Performance varied substantially across facilities (eg, neurology consultation had a facility outlier rate of 53.0%). Performance was higher for admitted patients than for patients cared for only in EDs with the greatest disparity for carotid artery imaging (4478/5927 [75.6%] vs 445/1758 [25.3%]; P < .001). Conclusions and Relevance This national study of VHA system quality of care for patients with TIA or minor stroke identified opportunities to improve care quality, particularly for patients who were discharged from the ED. Health care systems should engage in ongoing TIA care performance assessment to complement existing stroke performance measurement.

Short-Term Medical Costs of a VHA Health Information Exchange: A CHEERS-Compliant Article.

BackgroundIn 2011, the Veterans Health Administration (VHA) released the Acute Ischemic Stroke (AIS) Directive, which mandated reorganization of acute stroke care, including self-designation of stroke centers as Primary (P), Limited Hours (LH), or Supporting (S).ObjectivesIn partnership with the VHA Offices of Emergency Medicine and Specialty Care Services, the VA Stroke QUERI conducted a formative evaluation in a national sample of three levels of stroke centers in order to understand barriers and facilitators.Design and ApproachThe evaluation consisted of a mixed-methods assessment that included a qualitative assessment of data from semi-structured interviews with key informants and a quantitative assessment of stroke quality-of-care data reporting practices by facility characteristics.ParticipantsThe final sample included 38 facilities (84 % participation rate): nine P, 24 LH, and five S facilities. In total, we interviewed 107 clinicians and 16 regional Veterans Integrated Service Network (VISN) leaders.ResultsAcross all three levels of stroke centers, stroke teams identified the specific need for systematic nurse training to triage and initiate stroke protocols. The most frequently reported barriers centered around quality-of-care data collection. A low number of eligible veterans arriving at the VAMC in a timely manner was another major impediment. The LH and S facilities reported some unique barriers: access to radiology and neurology services; EMS diverting stroke patients to nearby stroke centers, maintaining staff competency, and a lack of stroke clinical champions. Solutions that were applied included developing stroke order sets and templates to provide systematic decision support, implementing a stroke code in the facility for a coordinated response to stroke, and staff resource allocation and training. Data reporting by facility evaluation demonstrated that categorizing site volume did indicate a lower likelihood of reporting among VAMCs with 25–49 acute stroke admissions per year.ConclusionsThe AIS Directive brought focused attention to reorganizing stroke care across a wide range of facility types. Larger VA facilities tended to follow established practices for organizing stroke care, but the unique addition of the LH designation presented some challenges. S facilities tended to report a lack of a coordinated stroke team and champion to drive process changes.