OBJECTIVE: To explore Veterans Health Administration clinicians' perspectives on the idea of redesigning electronic consultation (e-consult) delivery in line with a hub-and-spoke (centralized) model. MATERIALS AND METHODS: We conducted a qualitative study in VA New England Healthcare System (VISN 1). Semi-structured phone interviews were conducted with 35 primary care providers and 38 specialty care providers, including 13 clinical leaders, at 6 VISN 1 sites varying in size, specialist availability, and e-consult volume. Interviews included exploration of the hub-and-spoke (centralized) e-consult model as a system redesign option. Qualitative content analysis procedures were applied to identify and describe salient categories. RESULTS: Participants saw several potential benefits to scaling up e-consult delivery from a decentralized model to a hub-and-spoke model, including expanded access to specialist expertise and increased timeliness of e-consult responses. Concerns included differences in resource availability and management styles between sites, anticipated disruption to working relationships, lack of incentives for central e-consultants, dedicated staff's burnout and fatigue, technological challenges, and lack of motivation for change. DISCUSSION: Based on a case study from one of the largest integrated healthcare systems in the United States, our work identifies novel concerns and offers insights for healthcare organizations contemplating a scale-up of their e-consult systems. CONCLUSIONS: Scaling up e-consults in line with the hub-and-spoke model may help pave the way for a centralized and efficient approach to care delivery, but the success of this transformation will depend on healthcare systems' ability to evaluate and address barriers to leveraging economies of scale for e-consults. Informatics Association 2021. This work is written by US Government employees and is in the public domain in the US.

OBJECTIVE: The study sought to investigate whether consistent use of the Veterans Health Administration's My HealtheVet (MHV) online patient portal is associated with improvement in diabetes-related physiological measures among new portal users. MATERIALS AND METHODS: We conducted a retrospective cohort study of new portal users with type 2 diabetes that registered for MHV between 2012 and 2016. We used random-effect linear regression models to examine associations between months of portal use in a year (consistency) and annual means of the physiological measures (hemoglobin A1c [HbA1c], low-density lipoproteins [LDLs], and blood pressure [BP]) in the first 3 years of portal use. RESULTS: For patients with uncontrolled HbA1c, LDL, or BP at baseline, more months of portal use in a year was associated with greater improvement. Compared with 1 month of use, using the portal 12 months in a year was associated with annual declines in HbA1c of -0.41% (95% confidence interval [CI], -0.46% to -0.36%) and in LDL of -6.25 (95% CI, -7.15 to -5.36) mg/dL. Twelve months of portal use was associated with minimal improvements in BP: systolic BP of -1.01 (95% CI, -1.33 to -0.68) mm Hg and diastolic BP of -0.67 (95% CI, -0.85 to -0.49) mm Hg. All associations were smaller or not present for patients in control of these measures at baseline. CONCLUSIONS: We found consistent use of the patient portal among new portal users to be associated with modest improvements in mean HbA1c and LDL for patients at increased risk at baseline. For patients with type 2 diabetes, self-management supported by online patient portals may help control HbA1c, LDL, and BP. Informatics Association 2021. This work is written by US Government employees and is in the public domain in the US.

BACKGROUND: Strong relationships and effective communication between clinicians support care coordination and contribute to care quality. As a new mechanism of clinician communication, electronic consultations (e-consults) may have downstream effects on care provision and coordination. OBJECTIVE: The objective of this study was to understand primary care providers' and specialists' perspectives on how e-consults affect communication and relationships between clinicians. RESEARCH DESIGN: Qualitative study using thematic analysis of semistructured interviews. SUBJECTS: Six of 8 sites in the VISN 1 (Veterans Integrated Service Network) in New England were chosen, based on variation in organization and received e-consult volume. Seventy-three respondents, including 60 clinicians in primary care and 3 high-volume specialties (cardiology, pulmonology, and neurology) and 13 clinical leaders at the site and VISN level, were recruited. MEASURES: Participants' perspectives on the role and impact of e-consults on communication and relationships between clinicians. RESULTS: Clinicians identified 3 types of e-consults' social affordances: (1) e-consults were praised for allowing specialist advice to be more grounded in patient data and well-documented, but concerns about potential legal liability and increased transparency of communication to patients and others were also noted; (2) e-consults were perceived as an imperfect modality for iterative communication, especially for complex conversations requiring shared deliberation; (3) e-consults were understood as a factor influencing clinician relationships, but clinicians disagreed on whether e-consults promote or undermine relationship building. CONCLUSIONS: Clinicians have diverse concerns about the implications of e-consults for communication and relationships. Our findings may inform efforts to expand and improve the use of e-consults in diverse health care settings.

BACKGROUND: Diabetes is a complex, chronic disease that requires patients' effective self-management between clinical visits; this in turn relies on patient self-efficacy. The support of patient autonomy from healthcare providers is associated with better self-management and greater diabetes self-efficacy. Effective provider-patient secure messaging (SM) through patient portals may improve disease self-management and self-efficacy. SM that supports patients' sense of autonomy may mediate this effect by providing patients ready access to their health information and better communication with their clinical teams. OBJECTIVE: We examined the association between healthcare team-initiated SM and diabetes self-management and self-efficacy, and whether this association was mediated by patients' perceptions of autonomy support from their healthcare teams. DESIGN: We surveyed and analyzed content of messages sent to a sample of patients living with diabetes who use the SM feature on the VA's My HealtheVet patient portal. PARTICIPANTS: Four hundred forty-six veterans with type 2 diabetes who were sustained users of SM. MAIN MEASURES: Proactive (healthcare team-initiated) SM (0 or > /= 1 messages); perceived autonomy support; diabetes self-management; diabetes self-efficacy. KEY RESULTS: Patients who received at least one proactive SM from their clinical team were significantly more likely to engage in better diabetes self-management and report a higher sense of diabetes self-efficacy. This relationship was mediated by the patient's perception of autonomy support. The majority of proactive SM discussed scheduling, referrals, or other administrative content. Patients' responses to team-initiated communication promoted patient engagement in diabetes self-management behaviors. CONCLUSIONS: Perceived autonomy support is important for diabetes self-management and self-efficacy. Proactive communication from clinical teams to patients can help to foster a patient's sense of autonomy and encourage better diabetes self-management and self-efficacy.

BACKGROUND: Hypoglycemic events are common and potentially dangerous conditions among patients being treated for diabetes. Automatic detection of such events could improve patient care and is valuable in population studies. Electronic health records (EHRs) are valuable resources for the detection of such events. OBJECTIVE: In this study, we aim to develop a deep-learning-based natural language processing (NLP) system to automatically detect hypoglycemic events from EHR notes. Our model is called the High-Performing System for Automatically Detecting Hypoglycemic Events (HYPE). METHODS: Domain experts reviewed 500 EHR notes of diabetes patients to determine whether each sentence contained a hypoglycemic event or not. We used this annotated corpus to train and evaluate HYPE, the high-performance NLP system for hypoglycemia detection. We built and evaluated both a classical machine learning model (ie, support vector machines [SVMs]) and state-of-the-art neural network models. RESULTS: We found that neural network models outperformed the SVM model. The convolutional neural network (CNN) model yielded the highest performance in a 10-fold cross-validation setting: mean precision=0.96 (SD 0.03), mean recall=0.86 (SD 0.03), and mean F1=0.91 (SD 0.03). CONCLUSIONS: Despite the challenges posed by small and highly imbalanced data, our CNN-based HYPE system still achieved a high performance for hypoglycemia detection. HYPE can be used for EHR-based hypoglycemia surveillance and population studies in diabetes patients.

OBJECTIVE: Electronic consultations (e-consults) are clinician-to-clinician communications that may obviate face-to-face specialist visits. E-consult programs have spread within the US and internationally despite limited data on outcomes. We conducted a systematic review of the recent peer-reviewed literature on the effect of e-consults on access, cost, quality, and patient and clinician experience and identified the gaps in existing research on these outcomes. MATERIALS AND METHODS: We searched 4 databases for empirical studies published between 1/1/2015 and 2/28/2019 that reported on one or more outcomes of interest. Two investigators reviewed titles and abstracts. One investigator abstracted information from each relevant article, and another confirmed the abstraction. We applied the GRADE criteria for the strength of evidence for each outcome. RESULTS: We found only modest empirical evidence for effectiveness of e-consults on important outcomes. Most studies are observational and within a single health care system, and comprehensive assessments are lacking. For those outcomes that have been reported, findings are generally positive, with mixed results for clinician experience. These findings reassure but also raise concern for publication bias. CONCLUSION: Despite stakeholder enthusiasm and encouraging results in the literature to date, more rigorous study designs applied across all outcomes are needed. Policy makers need to know what benefits may be expected in what contexts, so they can define appropriate measures of success and determine how to achieve them. Informatics Association 2019. This work is written by US Government employees and is in the public domain in the US.

BACKGROUND: Improper dosing of medications such as insulin can cause hypoglycemic episodes, which may lead to severe morbidity or even death. Although secure messaging was designed for exchanging nonurgent messages, patients sometimes report hypoglycemia events through secure messaging. Detecting these patient-reported adverse events may help alert clinical teams and enable early corrective actions to improve patient safety. OBJECTIVE: We aimed to develop a natural language processing system, called HypoDetect (Hypoglycemia Detector), to automatically identify hypoglycemia incidents reported in patients' secure messages. METHODS: An expert in public health annotated 3000 secure message threads between patients with diabetes and US Department of Veterans Affairs clinical teams as containing patient-reported hypoglycemia incidents or not. A physician independently annotated 100 threads randomly selected from this dataset to determine interannotator agreement. We used this dataset to develop and evaluate HypoDetect. HypoDetect incorporates 3 machine learning algorithms widely used for text classification: linear support vector machines, random forest, and logistic regression. We explored different learning features, including new knowledge-driven features. Because only 114 (3.80%) messages were annotated as positive, we investigated cost-sensitive learning and oversampling methods to mitigate the challenge of imbalanced data. RESULTS: The interannotator agreement was Cohen kappa=.976. Using cross-validation, logistic regression with cost-sensitive learning achieved the best performance (area under the receiver operating characteristic curve=0.954, sensitivity=0.693, specificity 0.974, F1 score=0.590). Cost-sensitive learning and the ensembled synthetic minority oversampling technique improved the sensitivity of the baseline systems substantially (by 0.123 to 0.728 absolute gains). Our results show that a variety of features contributed to the best performance of HypoDetect. CONCLUSIONS: Despite the challenge of data imbalance, HypoDetect achieved promising results for the task of detecting hypoglycemia incidents from secure messages. The system has a great potential to facilitate early detection and treatment of hypoglycemia.

BACKGROUND: Many health care systems now allow patients to access their electronic health record (EHR) notes online through patient portals. Medical jargon in EHR notes can confuse patients, which may interfere with potential benefits of patient access to EHR notes. OBJECTIVE: The aim of this study was to develop and evaluate the usability and content quality of NoteAid, a Web-based natural language processing system that links medical terms in EHR notes to lay definitions, that is, definitions easily understood by lay people. METHODS: NoteAid incorporates two core components: CoDeMed, a lexical resource of lay definitions for medical terms, and MedLink, a computational unit that links medical terms to lay definitions. We developed innovative computational methods, including an adapted distant supervision algorithm to prioritize medical terms important for EHR comprehension to facilitate the effort of building CoDeMed. Ten physician domain experts evaluated the user interface and content quality of NoteAid. The evaluation protocol included a cognitive walkthrough session and a postsession questionnaire. Physician feedback sessions were audio-recorded. We used standard content analysis methods to analyze qualitative data from these sessions. RESULTS: Physician feedback was mixed. Positive feedback on NoteAid included (1) Easy to use, (2) Good visual display, (3) Satisfactory system speed, and (4) Adequate lay definitions. Opportunities for improvement arising from evaluation sessions and feedback included (1) improving the display of definitions for partially matched terms, (2) including more medical terms in CoDeMed, (3) improving the handling of terms whose definitions vary depending on different contexts, and (4) standardizing the scope of definitions for medicines. On the basis of these results, we have improved NoteAid's user interface and a number of definitions, and added 4502 more definitions in CoDeMed. CONCLUSIONS: Physician evaluation yielded useful feedback for content validation and refinement of this innovative tool that has the potential to improve patient EHR comprehension and experience using patient portals. Future ongoing work will develop algorithms to handle ambiguous medical terms and test and evaluate NoteAid with patients.