Simulating complex patient populations with hierarchical learning effects to support methods development for post-market surveillance.

BACKGROUND: Validating new algorithms, such as methods to disentangle intrinsic treatment risk from risk associated with experiential learning of novel treatments, often requires knowing the ground truth for data characteristics under investigation. Since the ground truth is inaccessible in real world data, simulation studies using synthetic datasets that mimic complex clinical environments are essential. We describe and evaluate a generalizable framework for injecting hierarchical learning effects within a robust data generation process that incorporates the magnitude of intrinsic risk and accounts for known critical elements in clinical data relationships. METHODS: We present a multi-step data generating process with customizable options and flexible modules to support a variety of simulation requirements. Synthetic patients with nonlinear and correlated features are assigned to provider and institution case series. The probability of treatment and outcome assignment are associated with patient features based on user definitions. Risk due to experiential learning by providers and/or institutions when novel treatments are introduced is injected at various speeds and magnitudes. To further reflect real-world complexity, users can request missing values and omitted variables. We illustrate an implementation of our method in a case study using MIMIC-III data for reference patient feature distributions. RESULTS: Realized data characteristics in the simulated data reflected specified values. Apparent deviations in treatment effects and feature distributions, though not statistically significant, were most common in small datasets (n < 3000) and attributable to random noise and variability in estimating realized values in small samples. When learning effects were specified, synthetic datasets exhibited changes in the probability of an adverse outcomes as cases accrued for the treatment group impacted by learning and stable probabilities as cases accrued for the treatment group not affected by learning. CONCLUSIONS: Our framework extends clinical data simulation techniques beyond generation of patient features to incorporate hierarchical learning effects. This enables the complex simulation studies required to develop and rigorously test algorithms developed to disentangle treatment safety signals from the effects of experiential learning. By supporting such efforts, this work can help identify training opportunities, avoid unwarranted restriction of access to medical advances, and hasten treatment improvements.

Medication Discrepancies in Older Veterans Receiving Home Healthcare.

In a prospective cohort study of Veterans and community health nurses, we enrolled hospitalized older Veterans referred to home care for skilled nursing and/or physical or occupational therapy for posthospitalization care. We assessed preadmission activities of daily living and instrumental activities of daily living, health literacy, numeracy, and cognition. Postdischarge phone calls identified medication errors and medication reconciliation efforts by home healthcare clinicians. Veterans Administration-based community health nurses completed surveys about content and timing of postdischarge interactions with home healthcare clinicians. We determined the types and frequency of medication errors among older Veterans receiving home healthcare, patient-provider communication patterns in this setting, and patient characteristics affecting medication error rates. Most Veterans (24/30, 80%) had at least one discordant medication, and only one noted that errors were identified and resolved. Veterans were asked about medications in the home healthcare setting, but far fewer were questioned about medication-taking details, adherence, and as-needed or nonoral medications. Higher numeracy was associated with fewer errors. Veterans Administration community health nurses reported contact by home healthcare clinicians in 41% of cases (7/17). Given the high rate of medication errors discovered, future work should focus on implementing best practices for medication review in this setting, as well as documenting barriers/facilitators of patient-provider communication.

A transition care coordinator model reduces hospital readmissions and costs.

BACKGROUND: The optimal structure and intensity of interventions to reduce hospital readmission remains uncertain, due in part to lack of head-to-head comparison. To address this gap, we evaluated two forms of an evidence-based, multi-component transitional care intervention. METHODS: A quasi-experimental evaluation design compared outcomes of Transition Care Coordinator (TCC) Care to Usual Care, while controlling for sociodemographic characteristics, comorbidities, readmission risk, and administrative factors. The study was conducted between January 1, 2013 and April 30, 2015 as a quality improvement initiative. Eligible adults (N = 7038) hospitalized with pneumonia, congestive heart failure, or chronic obstructive pulmonary disease were identified for program evaluation via an electronic health record algorithm. Nurse TCCs provided either a full intervention (delivered in-hospital and by post-discharge phone call) or a partial intervention (phone call only). RESULTS: A total of 762 hospitalizations with TCC Care (460 full intervention and 302 partial intervention) and 6276 with Usual Care was examined. In multivariable models, hospitalizations with TCC Care had significantly lower odds of readmission at 30 days (OR = 0.512, 95% CI 0.392 to 0.668) and 90 days (OR = 0.591, 95% CI 0.483 to 0.723). Adjusted costs were significantly lower at 30 days (difference = $3969, 95% CI $5099 to $2691) and 90 days (difference = $5684, 95% CI $7602 to $3627). The effect was similar whether patients received the full or partial intervention. CONCLUSION: An evidence-based multi-component intervention delivered by nurse TCCs reduced 30- and 90-day readmissions and associated health care costs. Lower intensity interventions delivered by telephone after discharge may have similar effectiveness to in-hospital programs.

The Cost of ICU Delirium and Coma in the Intensive Care Unit Patient.

RATIONALE: Intensive care unit (ICU) delirium is highly prevalent and a potentially avoidable hospital complication. The current cost of ICU delirium is unknown. OBJECTIVES: To specify the association between the daily occurrence of delirium in the ICU with costs of ICU care accounting for time-varying illness severity and death. RESEARCH DESIGN: We performed a prospective cohort study within medical and surgical ICUs in a large academic medical center. SUBJECTS: We analyzed critically ill patients (N=479) with respiratory failure and/or shock. MEASURES: Covariates included baseline factors (age, insurance, cognitive impairment, comorbidities, Acute Physiology and Chronic Health Evaluation II Score) and time-varying factors (sequential organ failure assessment score, mechanical ventilation, and severe sepsis). The primary analysis used a novel 3-stage regression method: first, estimation of the cumulative cost of delirium over 30 ICU days and then costs separated into those attributable to increased resource utilization among survivors and those that were avoided on the account of delirium's association with early mortality in the ICU. RESULTS: The patient-level 30-day cumulative cost of ICU delirium attributable to increased resource utilization was $17,838 (95% confidence interval, $11,132-$23,497). A combination of professional, dialysis, and bed costs accounted for the largest percentage of the incremental costs associated with ICU delirium. The 30-day cumulative incremental costs of ICU delirium that were avoided due to delirium-associated early mortality was $4654 (95% confidence interval, $2056-7869). CONCLUSIONS: Delirium is associated with substantial costs after accounting for time-varying illness severity and could be 20% higher (∼$22,500) if not for its association with early ICU mortality.

Risk of Hypoglycemia Following Hospital Discharge in Patients With Diabetes and Acute Kidney Injury.

OBJECTIVE: Hypoglycemia is common in patients with diabetes. The risk of hypoglycemia after acute kidney injury (AKI) is not well defined. The purpose of this study was to compare the risk for postdischarge hypoglycemia among hospitalized patients with diabetes who do and do not experience AKI. RESEARCH DESIGN AND METHODS: We performed a propensity-matched analysis of patients with diabetes, with and without AKI, using a retrospective national cohort of veterans hospitalized between 2004 and 2012. AKI was defined as a 0.3 mg/dL or 50% increase in serum creatinine from baseline to peak serum creatinine during hospitalization. Hypoglycemia was defined as hospital admission or an emergency department visit for hypoglycemia or as an outpatient blood glucose <60 mg/dL. Time to incident hypoglycemia within 90 days postdischarge was examined using Cox proportional hazards models. Prespecified subgroup analyses by renal recovery, baseline chronic kidney disease, preadmission drug regimen, and HbA1c were performed. RESULTS: We identified 65,151 propensity score-matched pairs with and without AKI. The incidence of hypoglycemia was 29.6 (95% CI 28.9-30.4) and 23.5 (95% CI 22.9-24.2) per 100 person-years for patients with and without AKI, respectively. After adjustment, AKI was associated with a 27% increased risk of hypoglycemia (hazard ratio [HR] 1.27 [95% CI 1.22-1.33]). For patients with full recovery, the HR was 1.18 (95% CI 1.12-1.25); for partial recovery, the HR was 1.30 (95% CI 1.23-1.37); and for no recovery, the HR was 1.48 (95% CI 1.36-1.60) compared with patients without AKI. Across all antidiabetes drug regimens, patients with AKI experienced hypoglycemia more frequently than patients without AKI, though the incidence of hypoglycemia was highest among insulin users, followed by glyburide and glipizide users, respectively. CONCLUSIONS: AKI is a risk factor for hypoglycemia in the postdischarge period. Studies to identify risk-reduction strategies in this population are warranted.

Acute kidney injury is a risk factor for subsequent proteinuria.

Acute kidney injury (AKI) is associated with subsequent chronic kidney disease (CKD), but the mechanism is unclear. To clarify this, we examined the association of AKI and new-onset or worsening proteinuria during the 12 months following hospitalization in a national retrospective cohort of United States Veterans hospitalized between 2004-2012. Patients with and without AKI were matched using baseline demographics, comorbidities, proteinuria, estimated glomerular filtration rate, blood pressure, angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker (ACEI/ARB) use, and inpatient exposures linked to AKI. The distribution of proteinuria over one year post-discharge in the matched cohort was compared using inverse probability sampling weights. Subgroup analyses were based on diabetes, pre-admission ACEI/ARB use, and AKI severity. Among the 90,614 matched AKI and non-AKI pairs, the median estimated glomerular filtration rate was 62 mL/min/1.73m2. The prevalence of diabetes and hypertension were 48% and 78%, respectively. The odds of having one plus or greater dipstick proteinuria was significantly higher during each month of follow-up in patients with AKI than in patients without AKI (odds ratio range 1.20-1.39). Odds were higher in patients with Stage II or III AKI (odds ratios 1.32-1.81) than in Stage I AKI (odds ratios 1.18-1.32), using non-AKI as the reference group. Results were consistent regardless of diabetes status or baseline ACEI/ARB use. Thus, AKI is a risk factor for incident or worsening proteinuria, suggesting a possible mechanism linking AKI and future CKD. The type of proteinuria, physiology, and clinical significance warrant further study as a potentially modifiable risk factor in the pathway from AKI to CKD.

Acute Kidney Injury and Risk of Incident Heart Failure Among US Veterans.

BACKGROUND: Acute kidney injury (AKI) is common and associated with poor outcomes. Heart failure is a leading cause of cardiovascular disease among patients with chronic kidney disease. The relationship between AKI and heart failure remains unknown and may identify a novel mechanistic link between kidney and cardiovascular disease. STUDY DESIGN: Observational study. SETTING & PARTICIPANTS: We studied a national cohort of 300,868 hospitalized US veterans (2004-2011) without a history of heart failure. PREDICTOR: AKI was the predictor and was defined as a 0.3-mg/dL or 50% increase in serum creatinine concentration from baseline to the peak hospital value. Patients with and without AKI were matched (1:1) on 28 in- and outpatient covariates using optimal Mahalanobis distance matching. OUTCOMES: Incident heart failure was defined as 1 or more hospitalization or 2 or more outpatient visits with a diagnosis of heart failure within 2 years through 2013. RESULTS: There were 150,434 matched pairs in the study. Patients with and without AKI during the index hospitalization were well matched, with a median preadmission estimated glomerular filtration rate of 69mL/min/1.73m2. The overall incidence rate of heart failure was 27.8 (95% CI, 19.3-39.9) per 1,000 person-years. The incidence rate was higher in those with compared with those without AKI: 30.8 (95% CI, 21.8-43.5) and 24.9 (95% CI, 16.9-36.5) per 1,000 person-years, respectively. In multivariable models, AKI was associated with 23% increased risk for incident heart failure (HR, 1.23; 95% CI, 1.19-1.27). LIMITATIONS: Study population was primarily men, reflecting patients seen at Veterans Affairs hospitals. CONCLUSIONS: AKI is an independent risk factor for incident heart failure. Future studies to identify underlying mechanisms and modifiable risk factors are needed.

Low Health Literacy Is Associated with Increased Transitional Care Needs in Hospitalized Patients.

OBJECTIVE: To examine the association of health literacy with the number and type of transitional care needs (TCN) among patients being discharged to home. DESIGN, SETTING, PARTICIPANTS: A cross-sectional analysis of patients admitted to an academic medical center. MEASUREMENTS: Nurses administered the Brief Health Literacy Screen and documented TCNs along 10 domains: caregiver support, transportation, healthcare utilization, high-risk medical comorbidities, medication management, medical devices, functional status, mental health comorbidities, communication, and financial resources. RESULTS: Among the 384 patients analyzed, 113 (29%) had inadequate health literacy. Patients with inadequate health literacy had needs in more TCN domains (mean = 5.29 vs 4.36; P < 0 .001). In unadjusted analysis, patients with inadequate health literacy were significantly more likely to have TCNs in 7 out of the 10 domains. In multivariate analyses, inadequate health literacy remained significantly associated with inadequate caregiver support (odds ratio [OR], 2.61; 95% confidence interval [CI], 1.37-4.99) and transportation barriers (OR, 1.69; 95% CI, 1.04-2.76). CONCLUSIONS: Among hospitalized patients, inadequate health literacy is prevalent and independently associated with other needs that place patients at a higher risk of adverse outcomes, such as hospital readmission. Screening for inadequate health literacy and associated needs may enable hospitals to address these barriers and improve postdischarge outcomes.

An analysis from the Quality Outcomes Database, Part 2. Predictive model for return to work after elective surgery for lumbar degenerative disease.

OBJECTIVE Current costs associated with spine care are unsustainable. Productivity loss and time away from work for patients who were once gainfully employed contributes greatly to the financial burden experienced by individuals and, more broadly, society. Therefore, it is vital to identify the factors associated with return to work (RTW) after lumbar spine surgery. In this analysis, the authors used data from a national prospective outcomes registry to create a predictive model of patients' ability to RTW after undergoing lumbar spine surgery for degenerative spine disease. METHODS Data from 4694 patients who underwent elective spine surgery for degenerative lumbar disease, who had been employed preoperatively, and who had completed a 3-month follow-up evaluation, were entered into a prospective, multicenter registry. Patient-reported outcomes-Oswestry Disability Index (ODI), numeric rating scale (NRS) for back pain (BP) and leg pain (LP), and EQ-5D scores-were recorded at baseline and at 3 months postoperatively. The time to RTW was defined as the period between operation and date of returning to work. A multivariable Cox proportional hazards regression model, including an array of preoperative factors, was fitted for RTW. The model performance was measured using the concordance index (c-index). RESULTS Eighty-two percent of patients (n = 3855) returned to work within 3 months postoperatively. The risk-adjusted predictors of a lower likelihood of RTW were being preoperatively employed but not working at the time of presentation, manual labor as an occupation, worker's compensation, liability insurance for disability, higher preoperative ODI score, higher preoperative NRS-BP score, and demographic factors such as female sex, African American race, history of diabetes, and higher American Society of Anesthesiologists score. The likelihood of a RTW within 3 months was higher in patients with higher education level than in those with less than high school-level education. The c-index of the model's performance was 0.71. CONCLUSIONS This study presents a novel predictive model for the probability of returning to work after lumbar spine surgery. Spine care providers can use this model to educate patients and encourage them in shared decision-making regarding the RTW outcome. This evidence-based decision support will result in better communication between patients and clinicians and improve postoperative recovery expectations, which will ultimately increase the likelihood of a positive RTW trajectory.

An analysis from the Quality Outcomes Database, Part 1. Disability, quality of life, and pain outcomes following lumbar spine surgery: predicting likely individual patient outcomes for shared decision-making.

OBJECTIVE Quality and outcomes registry platforms lie at the center of many emerging evidence-driven reform models. Specifically, clinical registry data are progressively informing health care decision-making. In this analysis, the authors used data from a national prospective outcomes registry (the Quality Outcomes Database) to develop a predictive model for 12-month postoperative pain, disability, and quality of life (QOL) in patients undergoing elective lumbar spine surgery. METHODS Included in this analysis were 7618 patients who had completed 12 months of follow-up. The authors prospectively assessed baseline and 12-month patient-reported outcomes (PROs) via telephone interviews. The PROs assessed were those ascertained using the Oswestry Disability Index (ODI), EQ-5D, and numeric rating scale (NRS) for back pain (BP) and leg pain (LP). Variables analyzed for the predictive model included age, gender, body mass index, race, education level, history of prior surgery, smoking status, comorbid conditions, American Society of Anesthesiologists (ASA) score, symptom duration, indication for surgery, number of levels surgically treated, history of fusion surgery, surgical approach, receipt of workers' compensation, liability insurance, insurance status, and ambulatory ability. To create a predictive model, each 12-month PRO was treated as an ordinal dependent variable and a separate proportional-odds ordinal logistic regression model was fitted for each PRO. RESULTS There was a significant improvement in all PROs (p < 0.0001) at 12 months following lumbar spine surgery. The most important predictors of overall disability, QOL, and pain outcomes following lumbar spine surgery were employment status, baseline NRS-BP scores, psychological distress, baseline ODI scores, level of education, workers' compensation status, symptom duration, race, baseline NRS-LP scores, ASA score, age, predominant symptom, smoking status, and insurance status. The prediction discrimination of the 4 separate novel predictive models was good, with a c-index of 0.69 for ODI, 0.69 for EQ-5D, 0.67 for NRS-BP, and 0.64 for NRS-LP (i.e., good concordance between predicted outcomes and observed outcomes). CONCLUSIONS This study found that preoperative patient-specific factors derived from a prospective national outcomes registry significantly influence PRO measures of treatment effectiveness at 12 months after lumbar surgery. Novel predictive models constructed with these data hold the potential to improve surgical effectiveness and the overall value of spine surgery by optimizing patient selection and identifying important modifiable factors before a surgery even takes place. Furthermore, these models can advance patient-focused care when used as shared decision-making tools during preoperative patient counseling.

Role of Health Insurance Status in Interfacility Transfers of Patients With ST-Elevation Myocardial Infarction.

Lack of health insurance is associated with interfacility transfer from emergency departments for several nonemergent conditions, but its association with transfers for ST-elevation myocardial infarction (STEMI), which requires timely definitive care for optimal outcomes, is unknown. Our objective was to determine whether insurance status is a predictor of interfacility transfer for emergency department visits with STEMI. We analyzed data from the 2006 to 2011 Nationwide Emergency Department Sample examining all emergency department visits for patients age 18 years and older with a diagnosis of STEMI and a disposition of interfacility transfer or hospitalization at the same institution. For emergency department visits with STEMI, our multivariate logistic regression model included emergency department disposition status (interfacility transfer vs hospitalization at the same institution) as the primary outcome, and insurance status (none vs any [including Medicare, Medicaid, and private insurance]) as the primary exposure. We found that among 1,377,827 emergency department STEMI visits, including 249,294 (18.1%) transfers, patients without health insurance (adjusted odds ratio 1.6, 95% CI 1.5 to 1.7) were more likely to be transferred than those with insurance. Lack of health insurance status was also an independent risk factor for transfer compared with each subcategory of health insurance, including Medicare, Medicaid, and private insurance. In conclusion, among patients presenting to United States emergency departments with STEMI, lack of insurance was an independent predictor of interfacility transfer. In conclusion, because interfacility transfer is associated with longer delays to definitive STEMI therapy than treatment at the same facility, lack of health insurance may lead to important health disparities among patients with STEMI.

Inadequacy of 3-month Oswestry Disability Index outcome for assessing individual longer-term patient experience after lumbar spine surgery.

OBJECTIVE Prospective longitudinal outcomes registries are at the center of evidence-driven health care reform. Obtaining real-world outcomes data at 12 months can be costly and challenging. In the present study, the authors analyzed whether 3-month outcome measurements sufficiently represent 12-month outcomes for patients with degenerative lumbar disease undergoing surgery. METHODS Data from 3073 patients undergoing elective spine surgery for degenerative lumbar disease were entered into a prospective multicenter registry (N(2)QOD). Baseline, 3-month, and 12-month follow-up Oswestry Disability Index (ODI) scores were recorded. The absolute differences between actual 12- and 3-month ODI scores was evaluated. Additionally, the authors analyzed the absolute difference between actual 12-month ODI scores and a model-predicted 12-month ODI score (the model used patients' baseline characteristics and actual 3-month scores). The minimal clinically important difference (MCID) for ODI of 12.8 points and the substantial clinical benefit (SCB) for ODI of 18.8 points were used based on the previously published values. The concordance rate of achieving MCID and SCB for ODI at 3-and 12-months was computed. RESULTS The 3-month ODI scores differed from 12-month scores by an absolute difference of 11.9 ± 10.8, and predictive modeling estimations of 12-month ODI scores differed from actual 12-month scores by a mean (± SD) of 10.7 ± 9.0 points (p = 0.001). Sixty-four percent of patients (n = 1982) achieved an MCID for ODI at 3 months in comparison with 67% of patients (n = 2088) by 12 months; 51% (n = 1731) and 61% (n = 1860) of patients achieved SCB for ODI at 3 months and 12 months, respectively. Almost 20% of patients had ODI scores that varied at least 20 points (the point span of an ODI functional category) between actual 3- and 12-month values. In the aggregate analysis of achieving MCID, 77% of patients were concordant and 23% were discordant in achieving or not achieving MCID at 3 and 12 months. The discordance rates of achieving or not achieving MCID for ODI were in the range of 19% to 27% for all diagnoses and treatments (decompression with and without fusion). The positive and negative predictive value of 3-months ODI to predict 12-month ODI was 86% and 60% for MCID and 82% and 67% for SCB. CONCLUSIONS Based on their findings, the authors conclude the following: 1) Predictive methods for functional outcome based on early patient experience (i.e., baseline and/or 3-month data) should be used to help evaluate the effectiveness of procedures in patient populations, rather than serving as a proxy for long-term individual patient experience. 2) Prospective longitudinal registries need to span at least 12 months to determine the effectiveness of spine care at the individual patient and practitioner level.

Using the WHODAS 2.0 to Assess Functioning Among Veterans Seeking Compensation for Posttraumatic Stress Disorder.

OBJECTIVE: One of the major changes in DSM-5 was removal of the Global Assessment of Functioning (GAF). To determine whether the World Health Organization Disability Assessment Schedule 2.0 (WHODAS 2.0) is a suitable replacement for the GAF, this study compared how well the WHODAS 2.0 and the GAF measured functional impairment and other phenomena related to posttraumatic stress disorder (PTSD) among veterans applying for financial compensation (service connection) for PTSD. METHODS: Clinicians evaluating veteran claimants administered the Clinician Administered PTSD Scale (CAPS) and the WHODAS 2.0 to 177 veterans during their evaluations. Veterans also completed the Inventory of Psychosocial Functioning (IPF), a self-report measure of functional impairment, and received a GAF rating from the examiner. Actual benefit determinations and ratings were obtained. RESULTS: Confirmatory factor analyses demonstrated that the WHODAS 2.0 and the IPF were stronger indicators of a latent variable reflecting functioning compared with the GAF. In receiver operating characteristic curve analyses, the WHODAS 2.0, IPF, and GAF all displayed similar ability to identify veterans with PTSD-related impairment assessed by the CAPS. Compared with the GAF, the WHODAS 2.0 and IPF were less strongly related to PTSD symptom severity and disability ratings by the U.S. Department of Veterans Affairs, but these variables are typically influenced by GAF scores. CONCLUSIONS: The WHODAS 2.0 and IPF are acceptable replacements for the GAF and can be used to assess functional impairment among veterans seeking compensation for PTSD.

National Veterans Health Administration inpatient risk stratification models for hospital-acquired acute kidney injury.

OBJECTIVE: Hospital-acquired acute kidney injury (HA-AKI) is a potentially preventable cause of morbidity and mortality. Identifying high-risk patients prior to the onset of kidney injury is a key step towards AKI prevention. MATERIALS AND METHODS: A national retrospective cohort of 1,620,898 patient hospitalizations from 116 Veterans Affairs hospitals was assembled from electronic health record (EHR) data collected from 2003 to 2012. HA-AKI was defined at stage 1+, stage 2+, and dialysis. EHR-based predictors were identified through logistic regression, least absolute shrinkage and selection operator (lasso) regression, and random forests, and pair-wise comparisons between each were made. Calibration and discrimination metrics were calculated using 50 bootstrap iterations. In the final models, we report odds ratios, 95% confidence intervals, and importance rankings for predictor variables to evaluate their significance. RESULTS: The area under the receiver operating characteristic curve (AUC) for the different model outcomes ranged from 0.746 to 0.758 in stage 1+, 0.714 to 0.720 in stage 2+, and 0.823 to 0.825 in dialysis. Logistic regression had the best AUC in stage 1+ and dialysis. Random forests had the best AUC in stage 2+ but the least favorable calibration plots. Multiple risk factors were significant in our models, including some nonsteroidal anti-inflammatory drugs, blood pressure medications, antibiotics, and intravenous fluids given during the first 48 h of admission. CONCLUSIONS: This study demonstrated that, although all the models tested had good discrimination, performance characteristics varied between methods, and the random forests models did not calibrate as well as the lasso or logistic regression models. In addition, novel modifiable risk factors were explored and found to be significant.

A Multimodal Intervention Improves Postanesthesia Care Unit Handovers.

BACKGROUND: Failures of communication are a major contributor to perioperative adverse events. Transitions of care may be particularly vulnerable. We sought to improve postoperative handovers. METHODS: We introduced a multimodal intervention in an adult and a pediatric postanesthesia care unit (PACU) to improve postoperative handovers between anesthesia providers (APs) and PACU registered nurses (RNs). The intervention included a standardized electronic handover report form, a didactic webinar, mandatory simulation training focused on improving interprofessional communication, and post-training performance feedback. Trained, blinded nurse observers scored PACU handovers during 17 months using a structured tool consisting of 8 subscales and a global score (1-5 scale). Multivariate logistic regression assessed the effect of the intervention on the proportion of observed handovers receiving a global effectiveness rating of ≥3. RESULTS: Four hundred fifty-two clinicians received the simulation-based training, and 981 handovers were observed and rated. In the adult PACU, the estimated percentages of acceptable handovers (global ratings ≥3) among AP-RN pairs, where neither received simulation-based training (untrained dyads), was 3% (95% confidence interval, 1%-11%) at day 0, 10% (5%-19%) at training initiation (day 40), and 57% (33%-78%) at 1-year post-training initiation (day 405). For AP-RN pairs where at least one received the simulation-based training (trained dyads), these percentages were estimated to be 18% (11%-28%) and 68% (57%-76%) on days 40 and 405, respectively. The percentage of acceptable handovers was significantly greater on day 405 than it was on day 40 for both untrained (P < 0.001) and trained dyads (P < 0.001). Similar patterns were observed in the pediatric PACU. Three years later, the unadjusted estimate of the probability of an acceptable handover was 87% (72%-95%) in the adult PACU and 56% (40%-72%) in the pediatric PACU. CONCLUSIONS: A multimodal intervention substantially improved interprofessional PACU handovers, including those by clinicians who had not undergone formal simulation training. An effect appeared to be present >3 years later.

Timeliness of interfacility transfer for ED patients with ST-elevation myocardial infarction.

OBJECTIVES: Most US hospitals lack primary percutaneous coronary intervention (PCI) capabilities to treat patients with ST-elevation myocardial infarction (STEMI) necessitating transfer to PCI-capable centers. Transferred patients rarely meet the 120-minute benchmark for timely reperfusion, and referring emergency departments (EDs) are a major source of preventable delays. We sought to use more granular data at transferring EDs to describe the variability in length of stay at referring EDs. METHODS: We retrospectively analyzed a secondary data set used for quality improvement for patients with STEMI transferred to a single PCI center between 2008 and 2012. We conducted a descriptive analysis of the total time spent at each referring ED (door-in-door-out [DIDO] interval), periods that comprised DIDO (door to electrocardiogram [EKG], EKG-to-PCI activation, and PCI activation to exit), and the relationship of each period with overall time to reperfusion (medical contact-to-balloon [MCTB] interval). RESULTS: We identified 41 EDs that transferred 620 patients between 2008 and 2012. Median MCTB was 135 minutes (interquartile range [IQR] 114,172). Median overall ED DIDO was 74 minutes (IQR 56,103) and was composed of door to EKG, 5 minutes (IQR 2,11); EKG-to-PCI activation, 18 minutes (IQR 7,37); and PCI activation to exit, 44 minutes (IQR 34,56). Door-in door-out accounted for the largest proportion (60%) of overall MCTB and had the largest variability (coefficient of variability, 1.37) of these intervals. CONCLUSIONS: In this cohort of transferring EDs, we found high variability and substantial delays after EKG performance for patients with STEMI. Factors influencing ED decision making and transportation coordination after PCI activation are a potential target for intervention to improve the timeliness of reperfusion in patients with STEMI.

Laboratory test surveillance following acute kidney injury.

BACKGROUND: Patients with hospitalized acute kidney injury (AKI) are at increased risk for accelerated loss of kidney function, morbidity, and mortality. We sought to inform efforts at improving post-AKI outcomes by describing the receipt of renal-specific laboratory test surveillance among a large high-risk cohort. METHODS: We acquired clinical data from the Electronic health record (EHR) of 5 Veterans Affairs (VA) hospitals to identify patients hospitalized with AKI from January 1st, 2002 to December 31st, 2009, and followed these patients for 1 year or until death, enrollment in palliative care, or improvement in renal function to estimated GFR (eGFR) ≥ 60 L/min/1.73 m(2). Using demographic data, administrative codes, and laboratory test data, we evaluated the receipt and timing of outpatient testing for serum concentrations of creatinine and any as well as quantitative proteinuria recommended for CKD risk stratification. Additionally, we reported the rate of phosphorus and parathyroid hormone (PTH) monitoring recommended for chronic kidney disease (CKD) patients. RESULTS: A total of 10,955 patients admitted with AKI were discharged with an eGFR<60 mL/min/1.73 m2. During outpatient follow-up at 90 and 365 days, respectively, creatinine was measured on 69% and 85% of patients, quantitative proteinuria was measured on 6% and 12% of patients, PTH or phosphorus was measured on 10% and 15% of patients. CONCLUSIONS: Measurement of creatinine was common among all patients following AKI. However, patients with AKI were infrequently monitored with assessments of quantitative proteinuria or mineral metabolism disorder, even for patients with baseline kidney disease.

Predicting lung cancer prior to surgical resection in patients with lung nodules.

BACKGROUND: Existing predictive models for lung cancer focus on improving screening or referral for biopsy in general medical populations. A predictive model calibrated for use during preoperative evaluation of suspicious lung lesions is needed to reduce unnecessary operations for a benign disease. A clinical prediction model (Thoracic Research Evaluation And Treatment [TREAT]) is proposed for this purpose. METHODS: We developed and internally validated a clinical prediction model for lung cancer in a prospective cohort evaluated at our institution. Best statistical practices were used to construct, evaluate, and validate the logistic regression model in the presence of missing covariate data using bootstrap and optimism corrected techniques. The TREAT model was externally validated in a retrospectively collected Veteran Affairs population. The discrimination and calibration of the model was estimated and compared with the Mayo Clinic model in both the populations. RESULTS: The TREAT model was developed in 492 patients from Vanderbilt whose lung cancer prevalence was 72% and validated among 226 Veteran Affairs patients with a lung cancer prevalence of 93%. In the development cohort, the area under the receiver operating curve (AUC) and Brier score were 0.87 (95% confidence interval [CI], 0.83-0.92) and 0.12, respectively compared with the AUC 0.89 (95% CI, 0.79-0.98) and Brier score 0.13 in the validation dataset. The TREAT model had significantly higher accuracy (p < 0.001) and better calibration than the Mayo Clinic model (AUC = 0.80; 95% CI, 75-85; Brier score = 0.17). CONCLUSION: The validated TREAT model had better diagnostic accuracy than the Mayo Clinic model in preoperative assessment of suspicious lung lesions in a population being evaluated for lung resection.

Characteristics associated with postdischarge medication errors.

OBJECTIVE: To examine the association of patient- and medication-related factors with postdischarge medication errors. PATIENTS AND METHODS: The Vanderbilt Inpatient Cohort Study includes adults hospitalized with acute coronary syndromes and/or acute decompensated heart failure. We measured health literacy, subjective numeracy, marital status, cognition, social support, educational attainment, income, depression, global health status, and medication adherence in patients enrolled from October 1, 2011, through August 31, 2012. We used binomial logistic regression to determine predictors of discordance between the discharge medication list and the patient-reported list during postdischarge medication review. RESULTS: Among 471 patients (mean age, 59 years), the mean total number of medications reported was 12, and 79 patients (16.8%) had inadequate or marginal health literacy. A total of 242 patients (51.4%) were taking 1 or more discordant medication (ie, appeared on either the discharge list or patient-reported list but not both), 129 (27.4%) failed to report a medication on their discharge list, and 168 (35.7%) reported a medication not on their discharge list. In addition, 279 participants (59.2%) had a misunderstanding in indication, dose, or frequency in a cardiac medication. In multivariable analyses, higher subjective numeracy (odds ratio [OR], 0.81; 95% CI, 0.67-0.98) was associated with lower odds of having discordant medications. For cardiac medications, participants with higher health literacy (OR, 0.84; 95% CI, 0.74-0.95), with higher subjective numeracy (OR, 0.77; 95% CI, 0.63-0.95), and who were female (OR, 0.60; 95% CI, 0.46-0.78) had lower odds of misunderstandings in indication, dose, or frequency. CONCLUSION: Medication errors are present in approximately half of patients after hospital discharge and are more common among patients with lower numeracy or health literacy.

Assisted annotation of medical free text using RapTAT.

OBJECTIVE: To determine whether assisted annotation using interactive training can reduce the time required to annotate a clinical document corpus without introducing bias. MATERIALS AND METHODS: A tool, RapTAT, was designed to assist annotation by iteratively pre-annotating probable phrases of interest within a document, presenting the annotations to a reviewer for correction, and then using the corrected annotations for further machine learning-based training before pre-annotating subsequent documents. Annotators reviewed 404 clinical notes either manually or using RapTAT assistance for concepts related to quality of care during heart failure treatment. Notes were divided into 20 batches of 19-21 documents for iterative annotation and training. RESULTS: The number of correct RapTAT pre-annotations increased significantly and annotation time per batch decreased by ~50% over the course of annotation. Annotation rate increased from batch to batch for assisted but not manual reviewers. Pre-annotation F-measure increased from 0.5 to 0.6 to >0.80 (relative to both assisted reviewer and reference annotations) over the first three batches and more slowly thereafter. Overall inter-annotator agreement was significantly higher between RapTAT-assisted reviewers (0.89) than between manual reviewers (0.85). DISCUSSION: The tool reduced workload by decreasing the number of annotations needing to be added and helping reviewers to annotate at an increased rate. Agreement between the pre-annotations and reference standard, and agreement between the pre-annotations and assisted annotations, were similar throughout the annotation process, which suggests that pre-annotation did not introduce bias. CONCLUSIONS: Pre-annotations generated by a tool capable of interactive training can reduce the time required to create an annotated document corpus by up to 50%.