A Machine Learning Methodology for Identification and Triage of Heart Failure Exacerbations.

Inadequate at-home management and self-awareness of heart failure (HF) exacerbations are known to be leading causes of the greater than 1 million estimated HF-related hospitalizations in the USA alone. Most current at-home HF management protocols include paper guidelines or exploratory health applications that lack rigor and validation at the level of the individual patient. We report on a novel triage methodology that uses machine learning predictions for real-time detection and assessment of exacerbations. Medical specialist opinions on statistically and clinically comprehensive, simulated patient cases were used to train and validate prediction algorithms. Model performance was assessed by comparison to physician panel consensus in a representative, out-of-sample validation set of 100 vignettes. Algorithm prediction accuracy and safety indicators surpassed all individual specialists in identifying consensus opinion on existence/severity of exacerbations and appropriate treatment response. The algorithms also scored the highest sensitivity, specificity, and PPV when assessing the need for emergency care. Here we develop a machine-learning approach for providing real-time decision support to adults diagnosed with congestive heart failure. The algorithm achieves higher exacerbation and triage classification performance than any individual physician when compared to physician consensus opinion.

Hypercapnia in Advanced Chronic Obstructive Pulmonary Disease: A Secondary Analysis of the National Emphysema Treatment Trial.

RATIONALE: Hypercapnia develops in one third of patients with advanced chronic obstructive pulmonary disease (COPD) and is associated with increased morbidity and mortality. Multiple factors in COPD are thought to contribute to the development of hypercapnia including increased carbon dioxide (CO2) production, increased dead space ventilation, and the complex interactions of deranged respiratory system mechanics, inspiratory muscle overload and the ventilatory control center in the brainstem. However, these factors have not previously been systematically analyzed in a large, well-characterized population of severe COPD patients. METHODS: This is a secondary analysis of the clinical, physiologic and imaging data from the National Emphysema Treatment Trial (NETT). All patients with complete baseline data for the key predictor variables were included. An inclusive list of 32 potential predictor variables were selected a priori based on consensus of the investigators and literature review. Stepwise variable selection yielded 10 statistically significant associations in multivariate regression. RESULTS: A total of 1419 patients with severe COPD were included in the analysis; mean age 66.4 years (standard deviation 6.3), 38% females, and 422 (29.7%) had baseline hypercapnia. Key variables associated with hypercapnia were low resting partial pressure of oxygen in blood, low minute ventilation (Ve), high volume of exhaled carbon dioxide, low forced expiratory volume in 1 second, high residual volume, lower % emphysema on chest computed tomography, use of oxygen, low ventilatory reserve (high Ve/maximal voluntary ventilation), and not being at high altitude. Low diffusing capacity for carbon monoxide showed a positive association with hypercapnia in univariate analysis but a negative correlation in multivariate analysis. Measures of dyspnea and quality of life did not associate with degree of hypercapnia in multivariable analysis. CONCLUSION: Hypercapnia in a well-characterized cohort with severe COPD and emphysema is chiefly related to poor lung mechanics, high CO2 production, and a reduced ventilatory capability. Hypercapnia is less impacted by gas exchange abnormalities or the presence of emphysema.

Chronic hypersensitivity pneumonitis in the southeastern United States: an assessment of how clinicians reached the diagnosis.

BACKGROUND: Chronic hypersensitivity pneumonitis (cHP) is a disease caused by exposure to inhaled environmental antigens. Diagnosis of cHP is influenced by the awareness of the disease prevalence, which varies significantly in different regions, and how clinicians utilize relevant clinical information. We conducted a retrospective study to evaluate how clinicians in the Southeast United States, where the climate is humid favoring mold growth, diagnosed cHP using items identified in the international modified Delphi survey of experts, i.e., environmental exposure, CT imaging and lung pathology, METHODS: We searched Duke University Medical Center database for patients over the age of 18 with a diagnosis of cHP (ICD-9 code: 495) between Jan. 1, 2008 to Dec. 31, 2013 using a query tool, Duke Enterprise Data Unified Content Explorer (DEDUCE). RESULTS: Five hundred patients were identified and 261 patients had cHP confirmed in clinic notes by a pulmonologist or an allergist. About half of the patients lived in the Research Triangle area where our medical center is located, giving an estimated prevalence rate of 6.5 per 100,000 persons. An exposure source was mentioned in 69.3% of the patient. The most common exposure sources were environmental molds (43.1%) and birds (26.0%). We used Venn diagram to evaluate how the patients met the three most common cHP diagnostic criteria: evidence of environmental exposures (history or precipitin) (E), chest CT imaging (C) and pathology from lung biopsies (P). Eighteen patients (6.9%) met none of three criteria. Of the remaining 243 patients, 135 patients (55.6%) had one (E 35.0%, C 3.3%, P 17.3%), 81 patients (33.3%) had two (E + C 12.3%, E + P 17.3%, C + P 4.9%), and 27 patients (11.1%) had all three criteria (E + C + P). Overall, 49.4% of patients had pathology from lung biopsy compared to 31.6% with CT scan. CONCLUSIONS: Environmental mold was the most common exposure for cHP in the Southeast United States. Lung pathology was available in more than half of cHP cases in our tertiary care center, perhaps reflecting the complexity of referrals. Differences in exposure sources and referral patterns should be considered in devising future diagnostic pathways or guidelines for cHP.

A machine learning approach to triaging patients with chronic obstructive pulmonary disease.

COPD patients are burdened with a daily risk of acute exacerbation and loss of control, which could be mitigated by effective, on-demand decision support tools. In this study, we present a machine learning-based strategy for early detection of exacerbations and subsequent triage. Our application uses physician opinion in a statistically and clinically comprehensive set of patient cases to train a supervised prediction algorithm. The accuracy of the model is assessed against a panel of physicians each triaging identical cases in a representative patient validation set. Our results show that algorithm accuracy and safety indicators surpass all individual pulmonologists in both identifying exacerbations and predicting the consensus triage in a 101 case validation set. The algorithm is also the top performer in sensitivity, specificity, and ppv when predicting a patient's need for emergency care.

Integration of Electronic Patient-Reported Outcomes Into Routine Cancer Care: An Analysis of Factors Affecting Data Completeness.

PURPOSE: Routinely collected patient-reported outcomes (PROs) could provide invaluable data to a patient-centered learning health system but are often highly missing in clinical trials. We analyzed our experience with PROs to understand patterns of missing data using electronic collection as part of routine clinical care. METHODS: This is an analysis of a prospectively collected observational database of electronic PROs captured as part of routine clinical care in four different outpatient oncology clinics at an academic referral center. RESULTS: More than 24,000 clinical encounters from 7,655 unique patients are included. Data were collected via an electronic tablet-based survey instrument (Patient Care Monitor, version 2.0), at the time of clinical care, as part of routine care processes. Missing instruments (ie, no items completed) were submitted for 6.8% of clinical encounters, and 15.8% of encounters had missing items. Nearly 90% of all encounters involved < 10% missing items. In multivariable analyses, younger age, private health insurance, being seen in the breast oncology clinic, less time spent on the instrument, and longitudinal care were significantly associated with less missingness. CONCLUSION: Embedding collection of electronic PRO data into routine clinical care yielded low rates of missing data in this real-world, prospectively collected database. In contrast to clinical trial experience, missingness improve with longitudinal care. This approach may be a solution to minimizing missingness of PROs in research or clinical care settings in support of learning health care systems.

Improving ICU-Based Palliative Care Delivery: A Multicenter, Multidisciplinary Survey of Critical Care Clinician Attitudes and Beliefs.

OBJECTIVE: Addressing the quality gap in ICU-based palliative care is limited by uncertainty about acceptable models of collaborative specialist and generalist care. Therefore, we characterized the attitudes of physicians and nurses about palliative care delivery in an ICU environment. DESIGN: Mixed-methods study. SETTING: Medical and surgical ICUs at three large academic hospitals. PARTICIPANTS: Three hundred three nurses, intensivists, and advanced practice providers. MEASUREMENTS AND MAIN RESULTS: Clinicians completed written surveys that assessed attitudes about specialist palliative care presence and integration into the ICU setting, as well as acceptability of 23 published palliative care prompts (triggers) for specialist consultation. Most (n = 225; 75%) reported that palliative care consultation was underutilized. Prompting consideration of eligibility for specialist consultation by electronic health record searches for triggers was most preferred (n = 123; 41%); only 17 of them (6%) felt current processes were adequate. The most acceptable specialist triggers were metastatic malignancy, unrealistic goals of care, end of life decision making, and persistent organ failure. Advanced age, length of stay, and duration of life support were the least acceptable. Screening led by either specialists or ICU teams was equally preferred. Central themes derived from qualitative analysis of 65 written responses to open-ended items included concerns about the roles of physicians and nurses, implementation, and impact on ICU team-family relationships. CONCLUSIONS: Integration of palliative care specialists in the ICU is broadly acceptable and desired. However, the most commonly used current triggers for prompting specialist consultation were among the least well accepted, while more favorable triggers are difficult to abstract from electronic health record systems. There is also disagreement about the role of ICU nurses in palliative care delivery. These findings provide important guidance to the development of collaborative care models for the ICU setting.

A practical measurement of thoracic sarcopenia: correlation with clinical parameters and outcomes in advanced lung cancer.

Thoracic sarcopenia can feasibly be measured from routine CT scans but does not correlate to patient-centred outcomes http://ow.ly/102UkQ.

Performance of Consultative Palliative Care Model in Achieving Quality Metrics in the ICU.

CONTEXT: Quality metrics for intensive care unit (ICU)-based palliative care have been proposed, but it is unknown how consultative palliative care can contribute to performance on these measures. OBJECTIVES: Assess adherence to proposed quality metrics of ICU-based palliative care by palliative care specialists. METHODS: Surrogates for 9/14 patient-level quality metrics were assessed in all patients who received an initial palliative care specialist consult while in an ICU from 10/26/2012 to 1/16/2015 in the Global Palliative Care Quality Alliance, a nationwide palliative care quality registry. RESULTS: Two hundred fifty-four patients received an initial palliative care consultation in an ICU setting. Mean (SD) age was 67.5 (17.3) years, 52% were female. The most common reasons for consultation were symptom management (33%) and end-of-life transition (24%). Adherence to ICU quality metrics for palliative care was variable: clinicians documented presence or absence of advance directives in 36% of encounters, assessed pain in 52.0%, dyspnea in 50.8%, spiritual support in 62%, and reported an intervention for pain in 100% of patients with documented moderate to severe intensity pain. CONCLUSION: Palliative care consultations in an ICU setting are characterized by variable adherence to candidate ICU palliative care quality metrics. Although symptom management was the most common reason for palliative care consultation, consultants infrequently documented symptom assessments. Palliative care consultants performed better in offering spiritual support and managing documented symptoms. These results highlight specific competencies of consultative palliative care that should be complimented by ICU teams to ensure high-quality comprehensive care for the critically ill.

Usability Testing of an Electronic Patient-Reported Outcome System for Survivors of Critical Illness.

BACKGROUND: Web-based electronic patient-reported outcomes (ePRO) measures are increasingly used to facilitate patient-centered health assessments. However, it is unknown if ePRO completion is feasible for recently ill intensive care unit (ICU) survivors and their families. OBJECTIVE: To develop and evaluate the usability of a novel ePRO system (ePRO to Support People and Enhance Recovery [ePROSPER]) among ICU survivors and their families within an ongoing clinical trial. METHODS: Paper-based PROs were iteratively adapted to electronic forms (ePROs). Then, the usability of ePROSPER was assessed among 60 patients, their family members, and PRO and programming experts via questionnaires (eg, Systems Usability Scale), "think aloud" open-ended feedback, task completion times, and error rates. RESULTS: Input from patients and their families was used to incorporate user-experience modifications into ePROSPER. This feedback also led to inclusion of automated reminders for questionnaire completion and real-time alerts for staff triggered by high symptom levels. Median usability scores increased over testing cycles from 40 to 73 to 95, nearing the maximum score and showing excellent usability. All users completed ePROSPER within 20 minutes; 87% preferred it to a written version. ePROSPER was then implemented in a clinical trial without data errors. CONCLUSIONS: Automated ePRO systems can be successfully integrated in a post-ICU clinical trial setting. The value of integrating such systems in direct clinical care should be assessed in future studies.

End-of-life care in oxygen-dependent ILD compared with lung cancer: a national population-based study.

RATIONALE: Advanced fibrosing interstitial lung disease (ILD) is often progressive and associated with a high burden of symptoms and poor prognosis. Little is known about the symptom prevalence and access to palliative care services at end of life (EOL). OBJECTIVES: Compare prevalence of symptoms and palliative treatments between patients dying with oxygen-dependent ILD and patients dying of lung cancer. METHODS: Nationwide registry-based cohort study of patients with oxygen-dependent ILD and patients with lung cancer who died between 1 January 2011 and 14 October 2013. Prevalence of symptoms and treatments during the last seven days of life were compared using data in Swedish Registry of Palliative Care. MEASUREMENTS AND MAIN RESULTS: 285 patients with ILD and 10 822 with lung cancer were included. In ILD, death was more likely to be 'unexpected' (15% vs 4%), less likely to occur in a palliative care setting (17% vs 40%) and EOL discussions with the patients (41% vs 59%) were less common than in lung cancer. Patients with ILD suffered more from breathlessness (75% vs 42%) while patients with lung cancer had more pain (51% vs 73%) (p<0.005 for all comparisons). Patients with ILD had more unrelieved breathlessness, pain and anxiety. The survival time from initiation of oxygen therapy in ILD was a median 8.4 months (IQR 3.4-19.2 months). CONCLUSIONS: Patients with ILD receive poorer access to specialist EOL care services and experience more breathlessness than patients with lung cancer. This study highlights the need of better EOL care in oxygen-dependent ILD.

Development and Refinement of a Learning Health Systems Training Program.

CONTEXT: In the emerging Learning Health System (LHS), the application and generation of medical knowledge are a natural outgrowth of patient care. Achieving this ideal requires a physician workforce adept in information systems, quality improvement methods, and systems-based practice to be able to use existing data to inform future care. These skills are not currently taught in medical school or graduate medical education. CASE DESCRIPTION: We initiated a first-ever Learning Health Systems Training Program (LHSTP) for resident physicians. The curriculum builds analytical, informatics and systems engineering skills through an active-learning project utilizing health system data that culminates in a final presentation to health system leadership. FINDINGS: LHSTP has been in place for two years, with 14 participants from multiple medical disciplines. Challenges included scheduling, mentoring, data standardization, and iterative optimization of the curriculum for real-time instruction. Satisfaction surveys and feedback were solicited mid-year in year 2. Most respondents were satisfied with the program, and several participants wished to continue in the program in various capacities after their official completion. MAJOR THEMES: We adapted our curriculum to successes and challenges encountered in the first two years. Modifications include a revised approach to teaching statistics, smaller cohorts, and more intensive mentorship. We continue to explore ways for our graduates to remain involved in the LHSTP and to disseminate this program to other institutions. CONCLUSION: The LHSTP is a novel curriculum that trains physicians to lead towards the LHS. Successful methods have included diverse multidisciplinary educators, just in time instruction, tailored content, and mentored projects with local health system impact.

Symptom Burden of Chronic Lung Disease Compared with Lung Cancer at Time of Referral for Palliative Care Consultation.

RATIONALE: A growing evidence base supports provision of palliative care services alongside life-prolonging care. Whereas palliative care processes have been implemented widely in the care of patients with lung cancer, the same is not true for patients with chronic, progressive lung disease. OBJECTIVES: To compare the symptom burden of chronic lung disease with that of lung cancer at the time of initial palliative care consultation. METHODS: Data were abstracted from the Carolinas Palliative Care Consortium's Quality Data Collection Tool, an electronic database used by seven academic and community palliative care practices in multiple states for quality improvement purposes. We analyzed data derived from first palliative care encounters collected during a 2-year period, including the primary diagnosis of chronic lung disease or lung cancer, unresolved symptoms, setting of initial palliative care encounter, Palliative Performance Scale status, and on that basis we estimated prognosis for survival. We compared key clinical variables between chronic lung disease and lung cancer using Kruskal-Wallis and χ(2) tests. MEASUREMENTS AND MAIN RESULTS: We identified 152 patients with lung cancer and 86 patients with chronic lung disease. Of the total sample, 53% were women and 87% were white. Patients with chronic lung disease were more likely than those with lung cancer to have the initial palliative care encounter occur in the intensive care unit (17% vs. 6%; P = 0.005) and less likely as an outpatient (20% vs. 56%; P < 0.0001). Patients with chronic lung disease were also less likely to have a high Palliative Performance Scale status (14% vs. 30%; P = 0.009) but more likely to have an estimated prognosis for survival longer than 6 months (51% vs. 28%; P = 0.002). The most prevalent symptoms were dyspnea (55% vs. 42%) and pain (40% vs. 52%), neither of which differed between groups (P = 0.08). CONCLUSIONS: Patients with chronic lung disease have symptom burdens similar to those of patients with lung cancer at the time of first palliative care encounter. Given the population burden of chronic lung disease and limitations in the palliative care workforce, attention should be focused on ensuring that pulmonologists are prepared to assess and manage the common palliative care needs of patients with chronic lung disease.

Practical Dyspnea Assessment: Relationship Between the 0-10 Numerical Rating Scale and the Four-Level Categorical Verbal Descriptor Scale of Dyspnea Intensity.

CONTEXT: Measurement of dyspnea is important for clinical care and research. OBJECTIVES: To characterize the relationship between the 0-10 Numerical Rating Scale (NRS) and four-level categorical Verbal Descriptor Scale (VDS) for dyspnea assessment. METHODS: This was a substudy of a double-blind randomized controlled trial comparing palliative oxygen to room air for relief of refractory breathlessness in patients with life-limiting illness. Dyspnea was assessed with both a 0-10 NRS and a four-level categorical VDS over the one-week trial. NRS and VDS responses were analyzed in cross section and longitudinally. Relationships between NRS and VDS responses were portrayed using descriptive statistics and visual representations. RESULTS: Two hundred twenty-six participants contributed responses. At baseline, mild and moderate levels of breathlessness were reported by 41.9% and 44.6% of participants, respectively. NRS scores demonstrated increasing mean and median levels for increasing VDS intensity, from a mean (SD) of 0.6 (±1.04) for VDS none category to 8.2 (1.4) for VDS severe category. The Spearman correlation coefficient was strong at 0.78 (P < 0.0001). Based on the distribution of NRS scores within VDS categories, we calculated test characteristics of two different cutpoint models. Both models yielded 75% correct translations from NRS to VDS; however, Model A was more sensitive for moderate or greater dyspnea, with fewer misses downcoded. CONCLUSION: There is strong correlation between VDS and NRS measures for dyspnea. Proposed practical cutpoints for the relationship between the dyspnea VDS and NRS are 0 for none, 1-4 for mild, 5-8 for moderate, and 9-10 for severe.

Development and usability testing of a Web-based decision aid for families of patients receiving prolonged mechanical ventilation.

BACKGROUND: Web-based decision aids are increasingly important in medical research and clinical care. However, few have been studied in an intensive care unit setting. The objectives of this study were to develop a Web-based decision aid for family members of patients receiving prolonged mechanical ventilation and to evaluate its usability and acceptability. METHODS: Using an iterative process involving 48 critical illness survivors, family surrogate decision makers, and intensivists, we developed a Web-based decision aid addressing goals of care preferences for surrogate decision makers of patients with prolonged mechanical ventilation that could be either administered by study staff or completed independently by family members (Development Phase). After piloting the decision aid among 13 surrogate decision makers and seven intensivists, we assessed the decision aid's usability in the Evaluation Phase among a cohort of 30 surrogate decision makers using the Systems Usability Scale (SUS). Acceptability was assessed using measures of satisfaction and preference for electronic Collaborative Decision Support (eCODES) versus the original printed decision aid. RESULTS: The final decision aid, termed 'electronic Collaborative Decision Support', provides a framework for shared decision making, elicits relevant values and preferences, incorporates clinical data to personalize prognostic estimates generated from the ProVent prediction model, generates a printable document summarizing the user's interaction with the decision aid, and can digitally archive each user session. Usability was excellent (mean SUS, 80 ± 10) overall, but lower among those 56 years and older (73 ± 7) versus those who were younger (84 ± 9); p = 0.03. A total of 93% of users reported a preference for electronic versus printed versions. CONCLUSIONS: The Web-based decision aid for ICU surrogate decision makers can facilitate highly individualized information sharing with excellent usability and acceptability. Decision aids that employ an electronic format such as eCODES represent a strategy that could enhance patient-clinician collaboration and decision making quality in intensive care.

Setting the vision: applied patient-reported outcomes and smart, connected digital healthcare systems to improve patient-centered outcomes prediction in critical illness.

PURPOSE OF REVIEW: Prediction models in critical illness are generally limited to short-term mortality and uncommonly include patient-centered outcomes. Current outcome prediction tools are also insensitive to individual context or evolution in healthcare practice, potentially limiting their value over time. Improved prognostication of patient-centered outcomes in critical illness could enhance decision-making quality in the ICU. RECENT FINDINGS: Patient-reported outcomes have emerged as precise methodological measures of patient-centered variables and have been successfully employed using diverse platforms and technologies, enhancing the value of research in critical illness survivorship and in direct patient care. The learning health system is an emerging ideal characterized by integration of multiple data sources into a smart and interconnected health information technology infrastructure with the goal of rapidly optimizing patient care. We propose a vision of a smart, interconnected learning health system with integrated electronic patient-reported outcomes to optimize patient-centered care, including critical care outcome prediction. SUMMARY: A learning health system infrastructure integrating electronic patient-reported outcomes may aid in the management of critical illness-associated conditions and yield tools to improve prognostication of patient-centered outcomes in critical illness.

Long-term persistence of quality improvements for an intensive care unit communication initiative using the VALUE strategy.

PURPOSE: Communication in the intensive care unit (ICU) is an important component of quality ICU care. In this report, we evaluate the long-term effects of a quality improvement (QI) initiative, based on the VALUE communication strategy, designed to improve communication with family members of critically ill patients. MATERIALS AND METHODS: We implemented a multifaceted intervention to improve communication in the ICU and measured processes of care. Quality improvement components included posted VALUE placards, templated progress note inclusive of communication documentation, and a daily rounding checklist prompt. We evaluated care for all patients cared for by the intensivists during three separate 3 week periods, pre, post, and 3 years following the initial intervention. RESULTS: Care delivery was assessed in 38 patients and their families in the pre-intervention sample, 27 in the post-intervention period, and 41 in follow-up. Process measures of communication showed improvement across the evaluation periods, for example, daily updates increased from pre 62% to post 76% to current 84% of opportunities. CONCLUSIONS: Our evaluation of this quality improvement project suggests persistence and continued improvements in the delivery of measured aspects of ICU family communication. Maintenance with point-of-care-tools may account for some of the persistence and continued improvements.

An opportunistic infection associated with ruxolitinib, a novel janus kinase 1,2 inhibitor.

We report a case of Cryptococcus neoformans pneumonia in a patient taking ruxolitinib, a janus kinase 1,2 inhibitor approved for the treatment of myelofibrosis. We hypothesize that ruxolitinib contributed to this infection through its effects on cell-mediated immunity. Clinicians should be aware of the potential for intracellular or opportunistic infections associated with this novel drug class.