Predictive modeling with electronic health record (EHR) data is anticipated to drive personalized medicine and improve healthcare quality. Constructing predictive statistical models typically requires extraction of curated predictor variables from normalized EHR data, a labor-intensive process that discards the vast majority of information in each patient’s record. We propose a representation of patients’ entire raw EHR records based on the Fast Healthcare Interoperability Resources (FHIR) format. We demonstrate that deep learning methods using this representation are capable of accurately predicting multiple medical events from multiple centers without site-specific data harmonization. We validated our approach using de-identified EHR data from two US academic medical centers with 216,221 adult patients hospitalized for at least 24 h. In the sequential format we propose, this volume of EHR data unrolled into a total of 46,864,534,945 data points, including clinical notes. Deep learning models achieved high accuracy for tasks such as predicting: in-hospital mortality (area under the receiver operator curve [AUROC] across sites 0.93–0.94), 30-day unplanned readmission (AUROC 0.75–0.76), prolonged length of stay (AUROC 0.85–0.86), and all of a patient’s final discharge diagnoses (frequency-weighted AUROC 0.90). These models outperformed traditional, clinically-used predictive models in all cases. We believe that this approach can be used to create accurate and scalable predictions for a variety of clinical scenarios. In a case study of a particular prediction, we demonstrate that neural networks can be used to identify relevant information from the patient’s chart.

/pdf/scalable-and-accurate-deep-learning-with-electronic-health-eu4vc3mdb6.pdf

Scalable and accurate deep learning with electronic health records

Predictive modeling with electronic health record (EHR) data is anticipated to drive personalized medicine and improve healthcare quality. Constructing predictive statistical models typically requires extraction of curated predictor variables from normalized EHR data, a labor-intensive process that discards the vast majority of information in each patient's record. We propose a representation of patients' entire, raw EHR records based on the Fast Healthcare Interoperability Resources (FHIR) format. We demonstrate that deep learning methods using this representation are capable of accurately predicting multiple medical events from multiple centers without site-specific data harmonization. We validated our approach using de-identified EHR data from two U.S. academic medical centers with 216,221 adult patients hospitalized for at least 24 hours. In the sequential format we propose, this volume of EHR data unrolled into a total of 46,864,534,945 data points, including clinical notes. Deep learning models achieved high accuracy for tasks such as predicting in-hospital mortality (AUROC across sites 0.93-0.94), 30-day unplanned readmission (AUROC 0.75-0.76), prolonged length of stay (AUROC 0.85-0.86), and all of a patient's final discharge diagnoses (frequency-weighted AUROC 0.90). These models outperformed state-of-the-art traditional predictive models in all cases. We also present a case-study of a neural-network attribution system, which illustrates how clinicians can gain some transparency into the predictions. We believe that this approach can be used to create accurate and scalable predictions for a variety of clinical scenarios, complete with explanations that directly highlight evidence in the patient's chart.

/pdf/scalable-and-accurate-deep-learning-for-electronic-health-55ghjalidu.pdf

Scalable and accurate deep learning for electronic health records

In older patients, acute medical illness that requires hospitalization is a sentinel event that often precipitates disability. This results in the subsequent inability to live independently and complete basic activities of daily living (ADLs). This hospitalization-associated disability occurs in approximately one-third of patients older than 70 years of age and may be triggered even when the illness that necessitated the hospitalization is successfully treated. In this article, we describe risk factors and risk stratification tools that identify older adults at highest risk of hospitalization-associated disability. We describe hospital processes that may promote hospitalization-associated disability and models of care that have been developed to prevent it. Since recognition of functional status problems is an essential prerequisite to preventing and managing disability, we also describe a pragmatic approach toward functional status assessment in the hospital focused on evaluation of ADLs, mobility, and cognition. Based on studies of acute geriatric units, we describe interventions hospitals and clinicians can consider to prevent hospitalization-associated disability in patients. Finally, we describe approaches clinicians can implement to improve the quality of life of older adults who develop hospitalization-associated disability and that of their caregivers.

Hospitalization-associated Disability: "She Was Probably Able to Ambulate, but I'm Not Sure"

BACKGROUND
Depression, anxiety and somatization are the most common mental disorders in primary care as well as medical specialty populations; each is present in at least 5-10% of patients and frequently comorbid with one another. An efficient means for measuring and monitoring all three conditions would be desirable.


METHODS
Evidence regarding the psychometric and pragmatic characteristics of the Patient Health Questionnaire (PHQ)-9 depression, generalized anxiety disorder (GAD)-7 anxiety and PHQ-15 somatic symptom scales are synthesized from two sources: (1) four multisite cross-sectional studies (three conducted in primary care and one in obstetric-gynecology practices) comprising 9740 patients, and (2) key studies from the literature that have studied these scales.


RESULTS
The PHQ-9 and its abbreviated eight-item (PHQ-8) and two-item (PHQ-2) versions have good sensitivity and specificity for detecting depressive disorders. Likewise, the GAD-7 and its abbreviated two-item (GAD-2) version have good operating characteristics for detecting generalized anxiety, panic, social anxiety and post-traumatic stress disorder. The optimal cutpoint is > or = 10 on the parent scales (PHQ-9 and GAD-7) and > or = 3 on the ultra-brief versions (PHQ-2 and GAD-2). The PHQ-15 is equal or superior to other brief measures for assessing somatic symptoms and screening for somatoform disorders. Cutpoints of 5, 10 and 15 represent mild, moderate and severe symptom levels on all three scales. Sensitivity to change is well-established for the PHQ-9 and emerging albeit not yet definitive for the GAD-7 and PHQ-15.


CONCLUSIONS
The PHQ-9, GAD-7 and PHQ-15 are brief well-validated measures for detecting and monitoring depression, anxiety and somatization.

Iconography : The Patient Health Questionnaire Somatic, Anxiety, and Depressive Symptom Scales: a systematic review

Obesity increases the risks of diabetes, hypertension, and cardiovascular diseases, ultimately contributing to mortality. Korean Society for the Study of Obesity (KSSO) was established to improve the management of obesity through research and education; to that end, the Committee of Clinical Practice Guidelines of KSSO reviews systemic evidence using expert panels to develop clinical guidelines. The clinical practice guidelines for obesity were revised in 2018 using National Health Insurance Service Health checkup data from 2006 to 2015. Following these guidelines, we added a category, class III obesity, which includes individuals with body mass index (BMI) ≥35 kg/m2. Agreeing with the International Federation for the Surgery of Obesity and Metabolic Disorders, Asian Pacific Chapter consensus, we determined that bariatric surgery is indicated for Korean patients with BMI ≥35 kg/m2 and for Korean patients with BMI ≥30 kg/m2 who have comorbidities. The new guidelines focus on guiding clinicians and patients to manage obesity more effectively. Our recommendations and treatment algorithms can serve as a guide for the evaluation, prevention, and management of overweight and obesity.

2018 Korean Society for the Study of Obesity Guideline for the Management of Obesity in Korea.

OBJECTIVES: To examine the association between mobility levels of older hospitalized adults and functional outcomes.

DESIGN: Prospective cohort study.

SETTING: A 900-bed teaching hospital in Israel.

PARTICIPANTS: Five hundred twenty-five older (≥70) acute medical patients hospitalized for a nondisabling condition.

MEASUREMENTS: In-hospital mobility was assessed using a previously validated scale. The main outcomes were decline from premorbid baseline functional status at discharge (activities of daily living (ADLs)) and at 1-month follow-up (ADLs and instrumental ADLs (IADLs)). Hospital mobility levels and functional outcomes were assessed according to prehospitalization functional trajectories. Logistic regressions were modeled for each outcome, controlling for functional status, morbidity, and demographic characteristics.

RESULTS: Forty-six percent of participants had declined in ADLs at discharge and 49% at follow-up; 57% had declined in IADLs at follow-up. Mobility during hospitalization was twice as high in participants with no preadmission functional decline. Low versus high in-hospital mobility was associated with worse basic functional status at discharge (adjusted odds ratio (AOR)=18.03, 95% confidence interval (CI)=7.68–42.28) and at follow-up (AOR=4.72, 95% CI=1.98–11.28) and worse IADLs at follow-up (AOR=2.00, 95% CI=1.05–3.78). The association with poorer discharge functional outcomes was present in participants with preadmission functional decline (AOR for low vs high mobility=15.26, 95% CI=4.80–48.42) and in those who were functionally stable (AOR for low vs high mobility=10.12, 95% CI=2.28–44.92).

CONCLUSION: In-hospital mobility is an important modifiable factor related to functional decline in older adults in immediate and short-term (1-month follow-up) functional outcomes.

Low Mobility During Hospitalization and Functional Decline in Older Adults

Objectives
To investigate the combined contribution of processes of hospitalization and preadmission individual risk factors in explaining functional decline at discharge and at 1-month follow-up in older adults with nondisabling conditions.

Design
Prospective cohort study.

Setting
Internal medicine wards in two Israeli medical centers.

Participants
Six hundred eighty-four individuals aged 70 and older admitted for a nondisabling problem.

Measurements
Functional decline was measured according to change in modified Barthel Index from premorbid to discharge and from premorbid to 1 month after discharge. In-hospital mobility, continence care, sleep medication consumption, satisfaction with hospital environment, and nutrition intake were assessed using previously tested self-report instruments.

Results
Two hundred eighty-two participants (41.2%) reported functional decline at discharge and 317 (46.3%) at 1 month after discharge. Path analysis indicated that in-hospital mobility (standardized maximum likelihood estimate (SMLE) = −0.48, P < .001), continence care (SMLE = −0.12, P < .001), and length of stay (LOS) (SMLE = 0.06, P < .001) were directly related to functional decline at discharge and, together with personal risk factors, explained 64% of variance. In-hospital mobility, continence care, and LOS were indirectly related to functional decline at 1 month after discharge through functional decline at discharge (SMLE = 0.45, P < .001). Nutrition consumption (SMLE = −0.07, P < .001) was significantly related to functional decline at 1 month after discharge, explaining, together with other risk factors, 32% of variance.

Conclusion
In-hospital low mobility, suboptimal continence care, and poor nutrition account for immediate and 1-month posthospitalization functional decline. These are potentially modifiable hospitalization risk factors for which practice and policy should be targeted in efforts to curb the posthospitalization functional decline trajectory.

https://www.smgg.es/images/articulos/hospitalizaciondecline.pdf

Hospital-associated functional decline: the role of hospitalization processes beyond individual risk factors.

COMPLAINTS OF SLEEP DISTURBANCE ARE COMMON IN OLDER ADULTS, WITH PREVALENCE RATES OF OVER 50% IN THE COMMUNITY, AND 70% IN THE assisted living setting.1,2 Some of the consequences of poor sleep quality in the elderly include cognitive decline,3,4 increased risk of falls,5 daytime fatigue,6 and reduced physical and mental health and health-related quality of life status.1,7

Factors that are associated with poor sleep in the elderly include physical and mental comorbidity, polypharmacy, functional status, primary sleep disorders, and changes in circadian sleep-wake patterns.8 Circadian rhythm changes which underlie common sleep complaints in older adults include an advanced sleep phase, reduced sleep consolidation and duration and early morning awakenings.9,10

Although light exposure has been demonstrated as the most powerful zeitgeber (time-cue) contributing to human entrainment to the 24-h day,11 there is evidence suggesting that social and activity rhythms also contribute to circadian entrainment in humans.12 To capture and to quantify such social and behavioral rhythms in daily life behaviors, Monk et al13 developed and validated the Social Rhythm Metric (SRM). The SRM is a 2-week diary in which subjects prospectively record the timing of a list of various daily routine events, such getting out of bed and making first daily contact with another person. Based on these reports, an average regularity score is computed, which represents personal level of daily routinization, with increasing routinization indicating a more stable lifestyle rhythm. Using this tool, they discovered that lifestyle rhythm stabilizes with age.14 Furthermore, increased lifestyle regularity was associated with fewer sleep problems15 and with a tendency towards morning types.16

Social and activity rhythms may be viewed as part of the daily life routine. Routine is a concept pertaining to strategically designed behavioral patterns used to organize and coordinate activities along the axes of time, duration, social and physical contexts, and order.17 Young18 suggested 4 functional advantages of routines and habits, including skill and performance enhancement and energy and resource conservation.

Maintenance of routine may be particularly adaptive in older individuals. Thus, routine has been associated with lower disruptive behavior and higher functional status in Alzheimer patients.19 In various studies with community dwelling older adults, researchers have found that daily routine facilitates functional status and well-being.20,21 Clark et al.22,23 showed improvement in functional, mental and physical health status, life satisfaction, social functioning, body pain, and emotional problems following a routine activity enhancement intervention among older adults.

As changes in the circadian system are considered a hallmark of aging, and are implicated as underlying factors of reduced sleep quality in the elderly, it is plausible to assume that routine lifestyle rhythms may serve as a protective mechanism contributing to the maintenance of high quality of sleep. However, in a comparison of sleep quality and daily social rhythms measured by the SRM in young and older adults, healthy 70-year-old individuals exhibited poor sleep compared to younger counterparts in their twenties, despite significantly greater lifestyle regularity in the older adults.24 It is yet to be determined whether differences in lifestyle regularity predict sleep quality within the elderly population.

Routine lifestyle rhythms may be characterized by stability in the timing, the frequency and the duration of performed activities, such as the daily timing, the number of times per day and the duration of time spent eating meals, watching television, or reading a book. Moreover, other than daily regularity, it is possible to identify weekly patterns of regularity, for example, shopping, cleaning, exercise, and social engagements. Previously, Monk et al15 have assessed the relationships between the timing of daily activities and measures of sleep quality. To date, the role of routine, as a concept related to a wide range of activities measured by their frequency and duration, has not been investigated in relation to sleep quality.

In the current prospective study, we investigated the relationship between routine and sleep quality in the elderly, using multiple indicators of routine, and controlling for risk factors for poor sleep quality. We hypothesize, that in older adults residing in a retirement community, daily routine is associated with good sleep quality, beyond functional status, comorbidities, and age.

Contribution of routine to sleep quality in community elderly

Title. A concept analysis of routine: relevance to nursing

Aim.  This paper reports a concept analysis identifying the attributes, antecedents and consequences of the concept of routine and examining the implications and applications of this concept in the field of nursing.

Background.  Routine may be a pivotal concept in understanding functional adaptation and wellbeing. Nurses in institutional settings work according to scheduled routines, patient care is largely orchestrated in routines set by organizations and regulations, and persons receiving care have their own life routines determining identity, capacities and frame of reference. However, to date, nursing has paid little attention to the relevance of routine and the role it may play in patient care.

Method.  A concept analysis was conducted using Rodgers’ guidelines. The literature search was based on the following databases: PsycInfo, CINAHL, MedLine, Social Services, and Social Work abstracts. To be included in the analysis, papers had to relate directly and essentially to the concept of routine. Seventy-four papers published from 1977 to 2005 were included in the final stage of the analysis. The analysis included target populations, disciplinary perspectives, type of manuscript, themes and definitions, theoretical models, antecedents and consequences, as well as related terms.

Results.  Routine is a concept pertaining to strategically designed behavioural patterns (conscious and subconscious) and is used to organize and coordinate activities along different axes of time, duration, social and physical contexts, sequence and order. It emerges from the literature as a strategy that serves adaptation, in general, especially in the face of change and stressful situations. The conceptual structure, relations with other concepts, antecedents and consequences are described.

Conclusion.  The concept of routine is ill-defined and seldom used in the field of nursing, despite the promise it may hold for a better understanding of a wide range of health-related issues. This concept analysis offers an integrative view of routine and suggests directions for future research and practice.

A concept analysis of routine: relevance to nursing

Among patients whose refill records suggested they had 7 days’ supply or less on hand, patients in the anticipatory outreach group were more likely than controls to refill a medication within 7 days (26.9% vs 26.3%; OR, 1.03, 95% CI, 1.001.06) (Table 2). However, this effect was not consistent across medication types and was most notably observed among patients taking antiseizure medications (29.7% vs 26.1%; OR, 1.20; 95% CI, 1.02-1.41).

https://jamanetwork.com/journals/jamainternalmedicine/articlepdf/2588809/jamainternal_Agmon_2016_ld_160066.pdf

Anna Zisberg

Papers

Low Mobility During Hospitalization and Functional Decline in Older Adults

Hospital-associated functional decline: the role of hospitalization processes beyond individual risk factors.

Contribution of routine to sleep quality in community elderly

A concept analysis of routine: relevance to nursing

Association Between 900 Steps a Day and Functional Decline in Older Hospitalized Patients.