Search

ImportanceDiscussions about goals of care are important for high-quality palliative care yet are often lacking for hospitalized older patients with serious illness.ObjectiveTo evaluate a communication-priming intervention to promote goals-of-care discussions between clinicians and hospitalized older patients with serious illness.Design, Setting, and ParticipantsA pragmatic, randomized clinical trial of a clinician-facing communication-priming intervention vs usual care was conducted at 3 US hospitals within 1 health care system, including a university, county, and community hospital. Eligible hospitalized patients were aged 55 years or older with any of the chronic illnesses used by the Dartmouth Atlas project to study end-of-life care or were aged 80 years or older. Patients with documented goals-of-care discussions or a palliative care consultation between hospital admission and eligibility screening were excluded. Randomization occurred between April 2020 and March 2021 and was stratified by study site and history of dementia.InterventionPhysicians and advance practice clinicians who were treating the patients randomized to the intervention received a 1-page, patient-specific intervention (Jumpstart Guide) to prompt and guide goals-of-care discussions.Main Outcomes and MeasuresThe primary outcome was the proportion of patients with electronic health record–documented goals-of-care discussions within 30 days. There was also an evaluation of whether the effect of the intervention varied by age, sex, history of dementia, minoritized race or ethnicity, or study site.ResultsOf 3918 patients screened, 2512 were enrolled (mean age, 71.7 [SD, 10.8] years and 42% were women) and randomized (1255 to the intervention group and 1257 to the usual care group). The patients were American Indian or Alaska Native (1.8%), Asian (12%), Black (13%), Hispanic (6%), Native Hawaiian or Pacific Islander (0.5%), non-Hispanic (93%), and White (70%). The proportion of patients with electronic health record–documented goals-of-care discussions within 30 days was 34.5% (433 of 1255 patients) in the intervention group vs 30.4% (382 of 1257 patients) in the usual care group (hospital- and dementia-adjusted difference, 4.1% [95% CI, 0.4% to 7.8%]). The analyses of the treatment effect modifiers suggested that the intervention had a larger effect size among patients with minoritized race or ethnicity. Among 803 patients with minoritized race or ethnicity, the hospital- and dementia-adjusted proportion with goals-of-care discussions was 10.2% (95% CI, 4.0% to 16.5%) higher in the intervention group than in the usual care group. Among 1641 non-Hispanic White patients, the adjusted proportion with goals-of-care discussions was 1.6% (95% CI, −3.0% to 6.2%) higher in the intervention group than in the usual care group. There was no evidence of differential treatment effects of the intervention on the primary outcome by age, sex, history of dementia, or study site.Conclusions and RelevanceAmong hospitalized older adults with serious illness, a pragmatic clinician-facing communication-priming intervention significantly improved documentation of goals-of-care discussions in the electronic health record, with a greater effect size in racially or ethnically minoritized patients.Trial RegistrationClinicalTrials.gov Identifier: NCT04281784

CONTEXT: Documented goals-of-care discussions are an important quality metric for patients with serious illness. Natural language processing (NLP) is a promising approach for identifying goals-of-care discussions in the electronic health record (EHR). OBJECTIVES: To compare three NLP modeling approaches for identifying EHR documentation of goals-of-care discussions and generate hypotheses about differences in performance. METHODS: We conducted a mixed-methods study to evaluate performance and misclassification for three NLP featurization approaches modeled with regularized logistic regression: bag-of-words (BOW), rule-based, and a hybrid approach. From a prospective cohort of 150 patients hospitalized with serious illness over 2018–2020, we collected 4,391 inpatient EHR notes; 99 (2.3%) contained documented goals-of-care discussions. We used leave-one-out cross-validation to estimate performance by comparing pooled NLP predictions to human abstractors with receiver-operating-characteristic (ROC) and precision-recall (PR) analyses. We qualitatively examined a purposive sample of 70 NLP-misclassified notes using content analysis to identify linguistic features that allowed us to generate hypotheses underpinning misclassification. RESULTS: All three modeling approaches discriminated between notes with and without goals-of-care discussions (AUC(ROC): BOW, 0.907; rule-based, 0.948; hybrid, 0.965). Precision and recall were only moderate (precision at 70% recall: BOW, 16.2%; rule-based, 50.4%; hybrid, 49.3%; AUC(PR): BOW, 0.505; rule-based, 0.579; hybrid, 0.599). Qualitative analysis revealed patterns underlying performance differences between BOW and rule-based approaches. CONCLUSION: NLP holds promise for identifying EHR-documented goals-of-care discussions. However, the rarity of goals-of-care content in EHR data limits performance. Our findings highlight opportunities to optimize NLP modeling approaches, and support further exploration of different NLP approaches to identify goals-of-care discussions.

Goals-of-care discussions are important for patient-centered care among hospitalized patients with serious illness. However, there are little data on the occurrence, predictors, and timing of these discussions.To examine the occurrence, predictors, and timing of electronic health record (EHR)-documented goals-of-care discussions for hospitalized patients.This retrospective cohort study used natural language processing (NLP) to examine EHR-documented goals-of-care discussions for adults with chronic life-limiting illness or age ≥80 hospitalized 2015-2019. The primary outcome was NLP-identified documentation of a goals-of-care discussion during the index hospitalization. We used multivariable logistic regression to evaluate associations with baseline characteristics.Of 16,262 consecutive, eligible patients without missing data, 5,918 (36.4%) had a documented goals-of-care discussion during hospitalization; approximately 57% of these discussions occurred within 24 hours of admission. In multivariable analysis, documented goals-of-care discussions were more common for women (OR=1.26, 95%CI 1.18-1.36), older patients (OR=1.04 per year, 95%CI 1.03-1.04), and patients with more comorbidities (OR=1.11 per Deyo-Charlson point, 95%CI 1.10-1.13), cancer (OR=1.88, 95%CI 1.72-2.06), dementia (OR=2.60, 95%CI 2.29-2.94), higher acute illness severity (OR=1.12 per National Early Warning Score point, 95%CI 1.11-1.14), or prior advance care planning documents (OR=1.18, 95%CI 1.08-1.30). Documentation of these discussions was less common for racially or ethnically minoritized patients (OR=0.823, 95%CI 0.75-0.90).Among hospitalized patients with serious illness, documented goals-of-care discussions identified by NLP were more common among patients with older age and increased burden of acute or chronic illness, and less common among racially or ethnically minoritized patients. This suggests important disparities in goals-of-care discussions.

ImportanceMany clinical trial outcomes are documented in free-text electronic health records (EHRs), making manual data collection costly and infeasible at scale. Natural language processing (NLP) is a promising approach for measuring such outcomes efficiently, but ignoring NLP-related misclassification may lead to underpowered studies.ObjectiveTo evaluate the performance, feasibility, and power implications of using NLP to measure the primary outcome of EHR-documented goals-of-care discussions in a pragmatic randomized clinical trial of a communication intervention.Design, Setting, and ParticipantsThis diagnostic study compared the performance, feasibility, and power implications of measuring EHR-documented goals-of-care discussions using 3 approaches: (1) deep-learning NLP, (2) NLP-screened human abstraction (manual verification of NLP-positive records), and (3) conventional manual abstraction. The study included hospitalized patients aged 55 years or older with serious illness enrolled between April 23, 2020, and March 26, 2021, in a pragmatic randomized clinical trial of a communication intervention in a multihospital US academic health system.Main Outcomes and MeasuresMain outcomes were natural language processing performance characteristics, human abstractor-hours, and misclassification-adjusted statistical power of methods of measuring clinician-documented goals-of-care discussions. Performance of NLP was evaluated with receiver operating characteristic (ROC) curves and precision-recall (PR) analyses and examined the effects of misclassification on power using mathematical substitution and Monte Carlo simulation.ResultsA total of 2512 trial participants (mean [SD] age, 71.7 [10.8] years; 1456 [58%] female) amassed 44 324 clinical notes during 30-day follow-up. In a validation sample of 159 participants, deep-learning NLP trained on a separate training data set identified patients with documented goals-of-care discussions with moderate accuracy (maximal F1 score, 0.82; area under the ROC curve, 0.924; area under the PR curve, 0.879). Manual abstraction of the outcome from the trial data set would require an estimated 2000 abstractor-hours and would power the trial to detect a risk difference of 5.4% (assuming 33.5% control-arm prevalence, 80% power, and 2-sided α = .05). Measuring the outcome by NLP alone would power the trial to detect a risk difference of 7.6%. Measuring the outcome by NLP-screened human abstraction would require 34.3 abstractor-hours to achieve estimated sensitivity of 92.6% and would power the trial to detect a risk difference of 5.7%. Monte Carlo simulations corroborated misclassification-adjusted power calculations.Conclusions and RelevanceIn this diagnostic study, deep-learning NLP and NLP-screened human abstraction had favorable characteristics for measuring an EHR outcome at scale. Adjusted power calculations accurately quantified power loss from NLP-related misclassification, suggesting that incorporation of this approach into the design of studies using NLP would be beneficial.

Natural Language Processing (NLP) is a key enabling technology for re-use of information in free-text clinical notes. However, a barrier to deployment is the availability of labeled corpora for supervised machine learning, which are expensive to acquire as they must be annotated by experienced clinicians. Where corpora are available, they may be opportunistically collected and thus vulnerable to bias. Here we evaluate an approach for accounting for dataset bias in the context of identifying specific patient-provider interactions. In this context, bias is the result of a phenomenon being over or under-represented in a particular type of clinical note as a result of the way a dataset was curated. Using a clinical dataset which represents a great deal of variation in terms of author and setting, we control for confounding variables using a backdoor adjustment approach [1, 2], which to our knowledge has not been previously applied the clinical domain. This approach improves precision by up to 5% and the adjusted models’ scores for false positives are generally lower, resulting in a more generalizable model with the potential to enhance the downstream utility of models trained using opportunistically collected clinical corpora.

Context Goals-of-care discussions are an important quality metric in palliative care. However, goals-of-care discussions are often documented as free text in diverse locations. It is difficult to identify these discussions in the electronic health record (EHR) efficiently. Objectives To develop, train, and test an automated approach to identifying goals-of-care discussions in the EHR, using natural language processing (NLP) and machine learning (ML). Methods From the electronic health records of an academic health system, we collected a purposive sample of 3183 EHR notes (1435 inpatient notes and 1748 outpatient notes) from 1426 patients with serious illness over 2008–2016, and manually reviewed each note for documentation of goals-of-care discussions. Separately, we developed a program to identify notes containing documentation of goals-of-care discussions using NLP and supervised ML. We estimated the performance characteristics of the NLP/ML program across 100 pairs of randomly partitioned training and test sets. We repeated these methods for inpatient-only and outpatient-only subsets. Results Of 3183 notes, 689 contained documentation of goals-of-care discussions. The mean sensitivity of the NLP/ML program was 82.3% (SD 3.2%), and the mean specificity was 97.4% (SD 0.7%). NLP/ML results had a median positive likelihood ratio of 32.2 (IQR 27.5–39.2) and a median negative likelihood ratio of 0.18 (IQR 0.16–0.20). Performance was better in inpatient-only samples than outpatient-only samples. Conclusion Using NLP and ML techniques, we developed a novel approach to identifying goals-of-care discussions in the EHR. NLP and ML represent a potential approach toward measuring goals-of-care discussions as a research outcome and quality metric.

Background: Multiple chronic conditions (MCCs) are associated with increased intensity of end-of-life (EOL) care, but their effect is not well explored in patients with cancer. Objective: We examined EOL health care intensity and advance care planning (ACP) documentation to better understand the association between MCCs and these outcomes. Design: Retrospective cohort study. Setting/Subjects: Patients aged 18+ years at UW Medicine who died during 2010-2017 with poor prognosis cancer, with or without chronic liver disease, chronic pulmonary disease, coronary artery disease, dementia, diabetes with end-stage organ damage, end-stage renal disease, heart failure, or peripheral vascular disease. Measurements: ACP documentation 30+ days before death, in-hospital death, and inpatient or intensive care unit (ICU) admission in the last 30 days. We performed logistic regression for outcomes. Results: Of 15,092 patients with cancer, 10,596 (70%) had 1+ MCCs (range 1-8). Patients with cancer and heart failure had highest odds of hospitalization (odds ratio [OR] 1.67, 95% confidence interval [CI] 1.46-1.91), ICU admission (OR 2.06, 95% CI 1.76-2.41), or in-hospital death (OR 1.62, 95% CI 1.43-1.84) versus patients with cancer and other conditions. Patients with ACP 30+ days before death had lower odds of in-hospital death (OR 0.65, 95% CI 0.60-0.71), hospitalization (OR 0.67, 95% CI 0.61-0.74), or ICU admission (OR 0.71, 95% CI 0.64-0.80). Conclusions: Patients with ACP 30+ days before death had lower odds of high-intensity EOL care. Further research needs to explore how to best use ACP to ensure patients receive care aligned with patient and family goals for care.

Most people prefer to die at home, yet most do not. Understanding factors associated with terminal hospitalization may inform interventions to improve care.Among patients with chronic illness receiving care in a multihospital healthcare system, we identified the following: (1) predictors of death in any hospital; (2) predictors of death in a hospital outside the system; and (3) trends from 2010 to 2015.Retrospective cohort using death certificates and electronic health records. Settings/Subjects: Decedents with one of nine chronic illnesses.Among 20,486 decedents, those most likely to die in a hospital were younger (odds ratio [OR] 0.977, confidence interval [CI] 0.974-0.980), with more comorbidities (OR 1.188, CI 1.079-1.308), or more outpatient providers (OR 1.031, CI 1.015-1.047); those with cancer or dementia, or more outpatient visits were less likely to die in hospital. Among hospital deaths, patients more likely to die in an outside hospital had lower education (OR 0.952, CI 0.923-0.981), cancer (OR 1.388, CI 1.198-1.608), diabetes (OR 1.507, CI 1.262-1.799), fewer comorbidities (OR 0.745, CI 0.644-0.862), or fewer hospitalizations within the system during the prior year (OR 0.900, CI 0.864-0.938). Deaths in hospital did not change from 2010 to 2015, but the proportion of hospital deaths outside the system increased (p 0.022).Patients dying in the hospital who are more likely to die in an outside hospital, and therefore at greater risk for inaccessibility of advance care planning, were more likely to be less well-educated and have cancer or diabetes, fewer comorbidities, and fewer hospitalizations. These findings may help target interventions to improve end-of-life care.

Context Recent analyses of Medicare data show decreases over time in intensity of end-of-life care. Few studies exist regarding trends in intensity of end-of-life care for those under 65 years of age. Objectives To examine recent temporal trends in place of death, and both hospital and intensive care unit (ICU) utilization, for age-stratified decedents with chronic, life-limiting diagnoses ( Methods Retrospective cohort using death certificates and electronic health records for 22,068 patients with chronic illnesses who died between 2010 and 2015. We examined utilization overall and stratified by age using multiple regression. Results The proportion of deaths at home did not change, but hospital admissions in the last 30 days of life decreased significantly from 2010 to 2015 (hospital b = −0.026; CI = −0.041, −0.012). ICU admissions in the last 30 days also declined over time for the full sample and for patients aged 65 years or older (overall b = −0.023; CI = −0.039, −0.007), but was not significant for younger decedents. Length of stay (LOS) did not decrease for those using the hospital or ICU. Conclusion From 2010 to 2015, we observed a decrease in hospital admissions for all age groups and in ICU admissions for those over 65 years. As there were no changes in the proportion of patients with chronic illness who died at home nor in hospital or ICU LOS in the last 30 days, hospital and ICU admissions in the last 30 days may be a more responsive quality metric than site of death or LOS for palliative care interventions.

Introduction There is relatively sparse literature on the use of administrative datasets for research in patients with adult congenital heart disease (ACHD). The goal of this analysis is to examine the accuracy of administrative data for identifying patients with ACHD who died. Methods A list of the International Classification of Diseases codes representing ACHD of moderate- or great-complexity was created. A search for these codes in the electronic health record of adults who received care in 2010-2016 was performed, and used state death records to identify patients who died during this period. Manual record review was completed to evaluate performance of this search strategy. Identified patients were also compared with a list of patients with moderate- or great-complexity ACHD known to have died. Results About 134 patients were identified, of which 72 had moderate- or great-complexity ACHD confirmed by manual review, yielding a positive predictive value of 0.54 (95% CI 0.45, 0.62). Twenty six patients had a mild ACHD diagnosis. Thirty six patients had no identified ACHD on record review. Misidentifications were attributed to coding error for 19 patients (53%), and to acquired ventricular septal defects for 11 patients (31%). Diagnostic codes incorrect more than 50% of the time were those for congenitally corrected transposition, endocardial cushion defect, and hypoplastic left heart syndrome. Only 1 of 21 patients known to have died was not identified by the search, yielding a sensitivity of 0.95 (0.76, 0.99). Conclusion Use of administrative data to identify patients with ACHD of moderate or great complexity who have died had good sensitivity but suboptimal positive predictive value. Strategies to improve accuracy are needed. Administrative data is not ideal for identification of patients in this group, and manual record review is necessary to confirm these diagnoses.

Over the past 15 years, we have made a series of innovations and systematic improvements to the lab courses offered in the Department of Chemical and Biological Engineering (CHBE) at The University of British Columbia (UBC). Prior to 2003, CHBE teaching laboratories used a more traditional laboratory course model where students performed "cookbook" experiments and individually wrote formal lab reports. Redevelopment began with our second-year course and improvements from the second-year course were progressively added to senior years. The integrated lab sequence now culminates in a capstone problem-based learning laboratory experience. We will present our optimized laboratory sequence model that is currently used in the 2nd, 3rd and 4th year CHBE program at UBC.

CONTEXT. Advance care planning (ACP) is difficult in the setting of a life-threatening trauma but may be equally important in this context, especially with increasing numbers of trauma victims being elderly or having multimorbidity. OBJECTIVES. Identify predictors of absent ACP documentation in the electronic health records of patients with underlying chronic illness who died of traumatic injury. METHODS. We used death records and electronic health records to identify decedents with chronic life-limiting illness who died of traumatic injury between 2010 and 2015 and to evaluate factors associated with documentation of living wills, durable powers of attorney, or physician orders for life-sustaining treatment. RESULTS. Only 22% of decedents had ACP documentation at time of injury. Among those without preinjury ACP documentation, 4% completed ACP documentation after injury. In multipredictor analyses, patients were less likely to have ACP documentation at the time of injury if they were younger (P < 0.001), had fewer chronic illnesses (P = 0.002), and had fewer nonsurgical hospitalizations (P = 0.042) in the year before injury. Among patients without ACP documentation before injury, those with fewer postinjury nonsurgical hospitalizations were less likely to complete ACP documentation after injury (P = 0.019). CONCLUSIONS. Our findings suggest that patient characteristics play an important role in the completion of ACP among patients with chronic life-limiting illness and who died from sudden severe injury. Interventions to improve ACP completion by patients with serious chronic conditions have the potential for increasing goal-concordant care in the event of traumatic injury.

Objective: To evaluate the association between the number of chronic conditions and hospital utilization at the end of life. Background: An understanding of the association of multimorbidi...

Importance Patients with chronic illness frequently use Physician Orders for Life-Sustaining Treatment (POLST) to document treatment limitations. Objectives To evaluate the association between POLST order for medical interventions and intensive care unit (ICU) admission for patients hospitalized near the end of life. Design, Setting, and Participants Retrospective cohort study of patients with POLSTs and with chronic illness who died between January 1, 2010, and December 31, 2017, and were hospitalized 6 months or less before death in a 2-hospital academic health care system. Exposures POLST order for medical interventions (“comfort measures only” vs “limited additional interventions” vs “full treatment”), age, race/ethnicity, education, days from POLST completion to admission, histories of cancer or dementia, and admission for traumatic injury. Main Outcomes and Measures The primary outcome was the association between POLST order and ICU admission during the last hospitalization of life; the secondary outcome was receipt of a composite of 4 life-sustaining treatments: mechanical ventilation, vasopressors, dialysis, and cardiopulmonary resuscitation. For evaluating factors associated with POLST-discordant care, the outcome was ICU admission contrary to POLST order for medical interventions during the last hospitalization of life. Results Among 1818 decedents (mean age, 70.8 [SD, 14.7] years; 41% women), 401 (22%) had POLST orders for comfort measures only, 761 (42%) had orders for limited additional interventions, and 656 (36%) had orders for full treatment. ICU admissions occurred in 31% (95% CI, 26%-35%) of patients with comfort-only orders, 46% (95% CI, 42%-49%) with limited-interventions orders, and 62% (95% CI, 58%-66%) with full-treatment orders. One or more life-sustaining treatments were delivered to 14% (95% CI, 11%-17%) of patients with comfort-only orders and to 20% (95% CI, 17%-23%) of patients with limited-interventions orders. Compared with patients with full-treatment POLSTs, those with comfort-only and limited-interventions orders were significantly less likely to receive ICU admission (comfort only: 123/401 [31%] vs 406/656 [62%], aRR, 0.53 [95% CI, 0.45-0.62]; limited interventions: 349/761 [46%] vs 406/656 [62%], aRR, 0.79 [95% CI, 0.71-0.87]). Across patients with comfort-only and limited-interventions POLSTs, 38% (95% CI, 35%-40%) received POLST-discordant care. Patients with cancer were significantly less likely to receive POLST-discordant care than those without cancer (comfort only: 41/181 [23%] vs 80/220 [36%], aRR, 0.60 [95% CI, 0.43-0.85]; limited interventions: 100/321 [31%] vs 215/440 [49%], aRR, 0.63 [95% CI, 0.51-0.78]). Patients with dementia and comfort-only orders were significantly less likely to receive POLST-discordant care than those without dementia (23/111 [21%] vs 98/290 [34%], aRR, 0.44 [95% CI, 0.29-0.67]). Patients admitted for traumatic injury were significantly more likely to receive POLST-discordant care (comfort only: 29/64 [45%] vs 92/337 [27%], aRR, 1.52 [95% CI, 1.08-2.14]; limited interventions: 51/91 [56%] vs 264/670 [39%], aRR, 1.36 [95% CI, 1.09-1.68]). In patients with limited-interventions orders, older age was significantly associated with less POLST-discordant care (aRR, 0.93 per 10 years [95% CI, 0.88-1.00]). Conclusions and Relevance Among patients with POLSTs and with chronic life-limiting illness who were hospitalized within 6 months of death, treatment-limiting POLSTs were significantly associated with lower rates of ICU admission compared with full-treatment POLSTs. However, 38% of patients with treatment-limiting POLSTs received intensive care that was potentially discordant with their POLST.

Background: Although racial/ethnic minorities receive more intense, nonbeneficial healthcare at the end of life, the role of race/ethnicity independent of other social determinants of health is not well understood. Objectives: Examine the association between race/ethnicity, other key social determinants of health, and healthcare intensity in the last 30 days of life for those with chronic, life-limiting illness. Subjects: We identified 22,068 decedents with chronic illness cared for at a single healthcare system in Washington State who died between 2010 and 2015 and linked electronic health records to death certificate data. Design: Binomial regression models were used to test associations of healthcare intensity with race/ethnicity, insurance status, education, and median income by zip code. Path analyses tested direct and indirect effects of race/ethnicity with insurance, education, and median income by zip code used as mediators. Measurements: We examined three measures of healthcare intensity: (1) intensive care unit admission, (2) use of mechanical ventilation, and (3) receipt of cardiopulmonary resuscitation. Results: Minority race/ethnicity, lower income and educational attainment, and Medicaid and military insurance were associated with higher intensity care. Socioeconomic disadvantage accounted for some of the higher intensity in racial/ethnic minorities, but most of the effects were direct effects of race/ethnicity. Conclusions: The effects of minority race/ethnicity on healthcare intensity at the end of life are only partly mediated by other social determinants of health. Future interventions should address the factors driving both direct and indirect effects of race/ethnicity on healthcare intensity.

OBJECTIVE Overall health care resource utilization by adults with congenital heart disease has increased dramatically in the past two decades, yet little is known about utilization patterns at the end of life. The objective of this study is to better understand the patterns and influences on end-of-life care intensity for adults with congenital heart disease. METHODS We identified a sample of adults with congenital heart disease (n = 65), cancer (n = 10 784), or heart failure (n = 3809) who died between January 2010 and December 2015, cared for in one multi-hospital health care system. We used multivariate analysis to evaluate markers of resource utilization, location of death, and documentation of advance care planning among patients with congenital heart disease versus those with cancer and those with heart failure. RESULTS Approximately 40% of adults with congenital heart disease experienced inpatient and intensive care unit (ICU) hospitalizations in the last 30 days of life; 64% died in the hospital. Compared to patients with cancer, patients with adult congenital heart disease (ACHD) were more likely to have inpatient (adjusted risk ratio 1.57; 95% CI 1.12-2.18) and ICU admissions in the last 30 days of life (adjusted risk ratio 2.56; 95% CI 1.83-3.61), more likely to die in the hospital (adjusted risk ratio 1.75; 95% CI 1.43-2.13), and more likely to have documentation of advance care planning (adjusted risk ratio 1.46; 95% CI 1.09-1.96). Compared to patients with heart failure (HF), patients with ACHD were less likely to have an ICU admission in the last 30 days of life (adjusted risk ratio 0.73; 95% CI 0.54-0.99). CONCLUSIONS Adults with congenital heart disease have significant hospital resource utilization near the end of life compared to patients with cancer, notable for more hospitalizations and a higher likelihood of death in the hospital. This population represents an important opportunity for the application of palliative and supportive care.

As our population ages and the burden of chronic illness rises, there is increasing need to implement quality metrics that measure and benchmark care of the seriously ill, including the delivery of both primary care and specialty palliative care. Such metrics can be used to drive quality improvement, value-based payment, and accountability for population-based outcomes.In this article, we examine use of the electronic health record (EHR) as a tool to assess quality of serious illness care through narrative review and description of a palliative care quality metrics program in a large healthcare system.In the search for feasible, reliable, and valid palliative care quality metrics, the EHR is an attractive option for collecting quality data on large numbers of seriously ill patients. However, important challenges to using EHR data for quality improvement and accountability exist, including understanding the validity, reliability, and completeness of the data, as well as acknowledging the difference between care documented and care delivered. Challenges also include developing achievable metrics that are clearly linked to patient and family outcomes and addressing data interoperability across sites as well as EHR platforms and vendors. This article summarizes the strengths and weakness of the EHR as a data source for accountability of community- and population-based programs for serious illness, describes the implementation of EHR data in the palliative care quality metrics program at the University of Washington, and, based on that experience, discusses opportunities and challenges. Our palliative care metrics program was designed to serve as a resource for other healthcare systems.Although the EHR offers great promise for enhancing quality of care provided for the seriously ill, significant challenges remain to operationalizing this promise on a national scale and using EHR data for population-based quality and accountability.

47 Background: Multiple chronic conditions (MCCs) are associated with increased intensity of end-of-life (EOL) care in many conditions but their effect has not been well explored in patients with cancer. We examined intensity of EOL care and advance care planning (ACP) by patients with cancer and MCCs versus those without MCCs to better understand how MCCs affect EOL healthcare use in this population. Methods: Our sample included patients with cancer in the UW Medicine system who died between 2010-2017. MCCs were defined using the Dartmouth Atlas of Healthcare for the most common categories of severe illness in the last two years of life. These included chronic pulmonary disease, coronary artery disease (CAD), heart failure, severe chronic liver disease, chronic renal disease, dementia, diabetes with end organ damage, or peripheral vascular disease. Patients were categorized as having none or 1+ MCCs. We used a claims-based indicator for the presence of functional limitation. Outcomes included ACP documentation in the electronic health record, death in hospital, and inpatient or ICU admission in the last month of life. We performed logistic regression for all outcomes controlling for confounders defined a priori (age at death, race, sex, marital status, insurance, education, functional limitation). Results: Of 15,092 patients with cancer, 10,596 (70%) had 1+MCCs (range 1-8 MCCs). Common MCCs were pulmonary (25%), CAD (23%), and renal (18%). Those with MCCs were older (median 66 years (range 18-104) vs 63 years (range 18-102)), with more functional limitations (65% vs 43%). Those with 1+ MCC had more ACP documentation (43% vs. 23%) compared to those with no MCCs. Patients with 1+ MCC were more likely to die in hospital (OR 1.86, 95% CI 1.72-2.02) and to have inpatient (OR 2.45, 95% CI 2.20-2.72) or ICU admissions (OR 2.95, 95% CI 2.55-3.42) in their last 30 days versus patients with cancer and no MCCs. Conclusions: Among patients with cancer in a single healthcare system, patients with cancer and MCCs were more likely to have ACP documentation, die in hospital and experience high-intensity hospital-based care at or near EOL. Further research is needed to explore if such high-intensity care is aligned with patient and family goals for care.

Little is known about psychiatric illness and utilization of end-of-life care.We hypothesized that preexisting psychiatric illness would increase hospital utilization at end of life among patients with chronic medical illness due to increased severity of illness and care fragmentation.We reviewed electronic health records to identify decedents with one or more of eight chronic medical conditions based on International Classification of Diseases-9 codes. We used International Classification of Diseases-9 codes and prescription information to identify preexisting psychiatric illness. Regression models compared hospital utilization among patients with and without psychiatric illness. Path analyses examined the effect of severity of illness and care fragmentation.Eleven percent of 16,214 patients with medical illness had preexisting psychiatric illness, which was associated with increased risk of death in nursing homes (P = 0.002) and decreased risk of death in hospitals (P 0.001). In the last 30 days of life, psychiatric illness was associated with reduced inpatient and intensive care unit utilization but increased emergency department utilization. Path analyses confirmed an association between psychiatric illness and increased hospital utilization mediated by severity of illness and care fragmentation, but a stronger direct effect of psychiatric illness decreasing hospitalizations.Our findings differ from the increased hospital utilization for patients with psychiatric illness in circumstances other than end-of-life care. Path analyses confirmed hypothesized associations between psychiatric illness and increased utilization mediated by severity of illness and care fragmentation but identified more powerful direct effects decreasing hospital use. Further investigation should examine whether this effect represents a disparity in access to preferred care.

This silent film from 1928 is a re-imagining of Poe's The Fall of the House of Usher. Includes musical accompaniment.