Precision Phenotyping for Curating Research Cohorts of Patients with Post-Acute Sequelae of COVID-19 (PASC) as a Diagnosis of Exclusion.

Scalable identification of patients with the post-acute sequelae of COVID-19 (PASC) is challenging due to a lack of reproducible precision phenotyping algorithms and the suboptimal accuracy, demographic biases, and underestimation of the PASC diagnosis code (ICD-10 U09.9). In a retrospective case-control study, we developed a precision phenotyping algorithm for identifying research cohorts of PASC patients, defined as a diagnosis of exclusion. We used longitudinal electronic health records (EHR) data from over 295 thousand patients from 14 hospitals and 20 community health centers in Massachusetts. The algorithm employs an attention mechanism to exclude sequelae that prior conditions can explain. We performed independent chart reviews to tune and validate our precision phenotyping algorithm. Our PASC phenotyping algorithm improves precision and prevalence estimation and reduces bias in identifying Long COVID patients compared to the U09.9 diagnosis code. Our algorithm identified a PASC research cohort of over 24 thousand patients (compared to about 6 thousand when using the U09.9 diagnosis code), with a 79.9 percent precision (compared to 77.8 percent from the U09.9 diagnosis code). Our estimated prevalence of PASC was 22.8 percent, which is close to the national estimates for the region. We also provide an in-depth analysis outlining the clinical attributes, encompassing identified lingering effects by organ, comorbidity profiles, and temporal differences in the risk of PASC. The PASC phenotyping method presented in this study boasts superior precision, accurately gauges the prevalence of PASC without underestimating it, and exhibits less bias in pinpointing Long COVID patients. The PASC cohort derived from our algorithm will serve as a springboard for delving into Long COVID's genetic, metabolomic, and clinical intricacies, surmounting the constraints of recent PASC cohort studies, which were hampered by their limited size and available outcome data.

Structure and Funding of Clinical Informatics Fellowships: A National Survey of Program Directors.

BACKGROUND: In 2011, the American Board of Medical Specialties established clinical informatics (CI) as a subspecialty in medicine, jointly administered by the American Board of Pathology and the American Board of Preventive Medicine. Subsequently, many institutions created CI fellowship training programs to meet the growing need for informaticists. Although many programs share similar features, there is considerable variation in program funding and administrative structures. OBJECTIVES: The aim of our study was to characterize CI fellowship program features, including governance structures, funding sources, and expenses. METHODS: We created a cross-sectional online REDCap survey with 44 items requesting information on program administration, fellows, administrative support, funding sources, and expenses. We surveyed program directors of programs accredited by the Accreditation Council for Graduate Medical Education between 2014 and 2021. RESULTS: We invited 54 program directors, of which 41 (76%) completed the survey. The average administrative support received was $27,732/year. Most programs (85.4%) were accredited to have two or more fellows per year. Programs were administratively housed under six departments: Internal Medicine (17; 41.5%), Pediatrics (7; 17.1%), Pathology (6; 14.6%), Family Medicine (6; 14.6%), Emergency Medicine (4; 9.8%), and Anesthesiology (1; 2.4%). Funding sources for CI fellowship program directors included: hospital or health systems (28.3%), clinical departments (28.3%), graduate medical education office (13.2%), biomedical informatics department (9.4%), hospital information technology (9.4%), research and grants (7.5%), and other sources (3.8%) that included philanthropy and external entities. CONCLUSION: CI fellowships have been established in leading academic and community health care systems across the country. Due to their unique training requirements, these programs require significant resources for education, administration, and recruitment. There continues to be considerable heterogeneity in funding models between programs. Our survey findings reinforce the need for reformed federal funding models for informatics practice and training.

Pharmacist-Led Diabetes Control Intervention and Health Outcomes in Hispanic Patients With Diabetes.

Importance: Among patients with type 2 diabetes (T2D), Hispanic individuals are more likely than non-Hispanic White individuals to develop diabetes-related complications. Objective: To examine the association of a pharmacist-led intervention (UCMyRx) with hemoglobin A1c (HbA1c) and systolic blood pressure (SBP) among Hispanic patients with T2D. Design, Setting, and Participants: This quality improvement study used electronic health record data and a difference-in-differences study design to evaluate the association of UCMyRx exposure with changes in HbA1c concentration and SBP among Hispanic patients with T2D, relative to usual care, at University of California, Los Angeles primary care clinics between February and April of 2023. The study population included patients with an International Classification of Diseases, Ninth Revision/International Statistical Classification of Diseases and Related Health Problems, Tenth Revision diagnosis of T2D, self-reporting Hispanic ethnicity, age 18 years or older, with 1 or more visits with a UCMyRx pharmacist (treatment) or 2 or more visits, 2 or more years apart, during the study window (comparison). Additionally, patients had to have the following observations during the study window (March 2, 2013-December 31, 2018): (1) a HbA1c 8% or higher, anywhere between 365 days before and 14 days after the index date (date of the first UCMyRx visit or a randomly generated index date) and a follow-up HbA1c measure within 120 to 365 days after the index date (n = 396) and/or (2) a SBP 140 mm Hg or higher between 365 days before and 14 days after the index date, and a follow-up SBP measure within 120 to 450 days after the index date (n = 795). Exposure: Pharmacists review laboratory results/vital signs, perform medication reconciliation, and develop personally tailored interventions to address adherence barriers and increase guideline-concordant care. Main Outcomes and Measures: Pre- to post-index date changes in HbA1c and SBP. Results: Of the 931 unique patients with T2D analyzed, the mean (SD) age was 64 (14.1) years, and 552 (59.3%) were female. In adjusted analyses, having 1 or more UCMyRx visits was associated with a reduction in HbA1c concentration (ß = -0.46%; 95% CI, -0.84% to -0.07%) but no change in SBP (ß = -1.71 mm Hg; 95% CI, -4.00 to 0.58 mm Hg). Conclusions and Relevance: In this quality improvement study of UCMyRx among Hispanic patients with T2D, a negative association was observed between UCMyRx exposure and HbA1c concentration but not SBP. Pharmacist-led intervention may be a strategy for improving outcomes among Hispanic patients with T2D.

Evaluating the predictive ability of natural language processing in identifying tertiary/quaternary cases in prioritization workflows for interhospital transfer.

Objectives: Tertiary and quaternary (TQ) care refers to complex cases requiring highly specialized health services. Our study aimed to compare the ability of a natural language processing (NLP) model to an existing human workflow in predictively identifying TQ cases for transfer requests to an academic health center. Materials and methods: Data on interhospital transfers were queried from the electronic health record for the 6-month period from July 1, 2020 to December 31, 2020. The NLP model was allowed to generate predictions on the same cases as the human predictive workflow during the study period. These predictions were then retrospectively compared to the true TQ outcomes. Results: There were 1895 transfer cases labeled by both the human predictive workflow and the NLP model, all of which had retrospective confirmation of the true TQ label. The NLP model receiver operating characteristic curve had an area under the curve of 0.91. Using a model probability threshold of ≥0.3 to be considered TQ positive, accuracy was 81.5% for the NLP model versus 80.3% for the human predictions (P = .198) while sensitivity was 83.6% versus 67.7% (P<.001). Discussion: The NLP model was as accurate as the human workflow but significantly more sensitive. This translated to 15.9% more TQ cases identified by the NLP model. Conclusion: Integrating an NLP model into existing workflows as automated decision support could translate to more TQ cases identified at the onset of the transfer process.

Embedding research study recruitment within the patient portal preCheck-in.

OBJECTIVE: Patient portals are increasingly used to recruit patients in research studies, but communication response rates remain low without tactics such as financial incentives or manual outreach. We evaluated a new method of study enrollment by embedding a study information sheet and HIPAA authorization form (HAF) into the patient portal preCheck-in (where patients report basic information like allergies). MATERIALS AND METHODS: Eligible patients who enrolled received an after-visit patient-reported outcomes survey through the patient portal. No additional recruitment/messaging efforts were made. RESULTS: A total of 386 of 843 patients completed preCheck-in, 308 of whom signed the HAF and enrolled in the study (37% enrollment rate). Of 93 patients who were eligible to receive the after-visit survey, 45 completed it (48% completion rate). CONCLUSION: Enrollment and survey completion rates were higher than what is typically seen with recruitment by patient portal messaging, suggesting that preCheck-in recruitment can enhance research study recruitment and warrants further investigation.

A Multicenter Retrospective Cohort Study to Characterize Patients Hospitalized With Multisystem Inflammatory Syndrome in Adults and Coronavirus Disease 2019 in the United States, 2020-2021.

BACKGROUND: The diagnosis of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-associated multisystem inflammatory syndrome in adults (MIS-A) requires distinguishing it from acute coronavirus disease 2019 (COVID-19) and may affect clinical management. METHODS: In this retrospective cohort study, we applied the US Centers for Disease Control and Prevention case definition to identify adults hospitalized with MIS-A at 6 academic medical centers from 1 March 2020 to 31 December 2021. Patients MIS-A were matched by age group, sex, site, and admission date at a 1:2 ratio to patients hospitalized with acute symptomatic COVID-19. Conditional logistic regression was used to compare demographic characteristics, presenting symptoms, laboratory and imaging results, treatments administered, and outcomes between cohorts. RESULTS: Through medical record review of 10 223 patients hospitalized with SARS-CoV-2-associated illness, we identified 53 MIS-A cases. Compared with 106 matched patients with COVID-19, those with MIS-A were more likely to be non-Hispanic black and less likely to be non-Hispanic white. They more likely had laboratory-confirmed COVID-19 ≥14 days before hospitalization, more likely had positive in-hospital SARS-CoV-2 serologic testing, and more often presented with gastrointestinal symptoms and chest pain. They were less likely to have underlying medical conditions and to present with cough and dyspnea. On admission, patients with MIS-A had higher neutrophil-to-lymphocyte ratio and higher levels of C-reactive protein, ferritin, procalcitonin, and D-dimer than patients with COVID-19. They also had longer hospitalization and more likely required intensive care admission, invasive mechanical ventilation, and vasopressors. The mortality rate was 6% in both cohorts. CONCLUSIONS: Compared with patients with acute symptomatic COVID-19, adults with MIS-A more often manifest certain symptoms and laboratory findings early during hospitalization. These features may facilitate diagnosis and management.

Distinct components of alert fatigue in physicians' responses to a noninterruptive clinical decision support alert.

OBJECTIVE: Clinical decision support (CDS) alerts may improve health care quality but "alert fatigue" can reduce provider responsiveness. We analyzed how the introduction of competing alerts affected provider adherence to a single depression screening alert. MATERIALS AND METHODS: We analyzed the audit data from all occurrences of a CDS alert at a large academic health system. For patients who screen positive for depression during ambulatory visits, a noninterruptive alert was presented, offering a number of relevant documentation actions. Alert adherence was defined as the selection of any option offered within the alert. We assessed the effect of competing clinical guidance alerts presented during the same encounter and the total of all CDS alerts that the same provider had seen in the prior 90 days, on the probability of depression screen alert adherence, adjusting for physician and patient characteristics. RESULTS: The depression alert fired during 55 649 office visits involving 418 physicians and 40 474 patients over 41 months. After adjustment, physicians who had seen the most alerts in the prior 90 days were much less likely to respond (adjusted OR highest-lowest quartile, 0.38; 95% CI 0.35-0.42; P < .001). Competing alerts in the same visit further reduced the likelihood of adherence only among physicians in the middle two quartiles of alert exposure in the prior 90 days. CONCLUSIONS: Adherence to a noninterruptive depression alert was strongly associated with the provider's cumulative alert exposure over the past quarter. Health systems should monitor providers' recent alert exposure as a measure of alert fatigue.

International comparisons of laboratory values from the 4CE collaborative to predict COVID-19 mortality.

Given the growing number of prediction algorithms developed to predict COVID-19 mortality, we evaluated the transportability of a mortality prediction algorithm using a multi-national network of healthcare systems. We predicted COVID-19 mortality using baseline commonly measured laboratory values and standard demographic and clinical covariates across healthcare systems, countries, and continents. Specifically, we trained a Cox regression model with nine measured laboratory test values, standard demographics at admission, and comorbidity burden pre-admission. These models were compared at site, country, and continent level. Of the 39,969 hospitalized patients with COVID-19 (68.6% male), 5717 (14.3%) died. In the Cox model, age, albumin, AST, creatine, CRP, and white blood cell count are most predictive of mortality. The baseline covariates are more predictive of mortality during the early days of COVID-19 hospitalization. Models trained at healthcare systems with larger cohort size largely retain good transportability performance when porting to different sites. The combination of routine laboratory test values at admission along with basic demographic features can predict mortality in patients hospitalized with COVID-19. Importantly, this potentially deployable model differs from prior work by demonstrating not only consistent performance but also reliable transportability across healthcare systems in the US and Europe, highlighting the generalizability of this model and the overall approach.

Displaying Cost and Completion Time for Reference Laboratory Test Orders-A Randomized Controlled Trial.

OBJECTIVES: Reduction in unnecessary services is one strategy for increasing the value of health care. Reference laboratory, or send-out, tests are associated with considerable costs. We investigated whether displaying cost and turnaround time (TAT), or time-to-result, for reference laboratory tests at the time of order entry in the electronic health record (EHR) system would impact provider ordering practices. METHODS: Reference laboratory test cost and TAT data were randomized prior to the study and only displayed for the intervention group. A 24-month dataset composed of 12 months each for baseline and study periods was extracted from the clinical data mart. A difference-in-differences (DID) analysis was conducted using a linear mixed-effects model to estimate the association between the intervention and changes in test-ordering patterns. RESULTS: In the inpatient setting, the DIDs of aggregate test-order costs and volume were not different among the control and intervention groups (p = 0.31 and p = 0.26, respectively). In the ambulatory setting, the DIDs of aggregate test-order costs and volume were not different among the control and intervention groups (p = 0.82 and p = 0.51, respectively). For both inpatient and ambulatory settings, no significant difference was observed in the DID of aggregate test-order costs and volumes calculated in respect to stratified relative cost and TAT groups (p > 0.05). CONCLUSION: Lack of alternative tests, test orders placed at a late step in patient management, and orders facilitated by trainees or mid-level providers may have limited the efficacy of the intervention. Our randomized study demonstrated no significant association between the display of cost or TAT display and ordering frequency.

Medical conditions of primary care patients with documented cannabis use and cannabis use disorder in electronic health records: a case control study from an academic health system in a medical marijuana state.

BACKGROUND: Approximately 3.8% of adults worldwide have used cannabis in the past year. Understanding how cannabis use is associated with other health conditions is crucial for healthcare providers seeking to understand the needs of their patients, and for health policymakers. This paper analyzes the relationship between documented cannabis use disorders (CUD), cannabis use (CU) and other health diagnoses among primary care patients during a time when medical use of marijuana was permitted by state law in California, United States of America. METHODS: The study utilized primary care electronic health record (EHR) data from an academic health system, using a case-control design to compare diagnoses among individuals with CUD/CU to those of matched controls, and those of individuals with CUD diagnoses with individuals who had CU otherwise documented. Associations of documented CU and CUD with general medical conditions and health conditions associated with cannabis use (both medical and behavioral) were analyzed using conditional logistic regression. RESULTS: Of 1,047,463 patients with ambulatory encounters from 2013-2017, 729 (0.06%) had CUD diagnoses and 3,731 (0.36%) had CU documented in their EHR. Patients with documented CUD and CU patients had significantly (p < 0.01) higher odds of most medical and behavioral diagnoses analyzed. Compared to matched controls, CUD-documented patients had highest odds of other substance use disorders (OR = 21.44: 95% CI 9.43-48.73), any mental health disorder (OR = 6.99; 95% CI 5.03-9.70) social anxiety disorder (OR = 13.03; 95% CI 2.18-77.94), HIV/AIDS (OR = 7.88: 95% CI 2.58-24.08), post-traumatic stress disorder (OR = 7.74: 95% CI 2.66-22.51); depression (OR = 7.01: 95% CI 4,79-10.27), and bipolar disorder (OR = 6.49: 95% CI 2.90-14.52). Compared to matched controls, CU-documented patients had highest odds of other substance use disorders (OR = 3.64; 95% CI 2.53-5.25) and post-traumatic stress disorder (OR = 3.41; 95% CI 2.53-5.25). CUD-documented patients were significantly more likely than CU-documented patients to have HIV/AIDS (OR = 6.70; 95% CI 2.10-21.39), other substance use disorder (OR = 5.88; 95% CI 2.42-14.22), depression (OR = 2.85; 95% CI 1.90-4.26), and anxiety (OR = 2.19: 95% CI 1.57-3.05) diagnoses. CONCLUSION: The prevalence of CUD and CU notation in EHR data from an academic health system was low, highlighting the need for improved screening in primary care. CUD and CU documentation were associated with increased risk for many health conditions, with the most elevated risk for behavioral health disorders and HIV/AIDS (among CUD-documented, but not CU-documented patients). Given the strong associations of CUD and CU documentation with health problems, it is important for healthcare providers to be prepared to identify CU and CUD, discuss the pros and cons of cannabis use with patients thoughtfully and empathically, and address cannabis-related comorbidities among these patients.

Characteristics of the National Applicant Pool for Clinical Informatics Fellowships (2018-2020).

In a prior survey, we found that applicants for 2017 ACGME-accredited clinical informatics fellowship positions were only 24% female and only 3% were members of underrepresented minorities (URM, consisting of American Indian or Alaska Native, Black or African American, Hispanic, Latino, or Spanish Origin, or Native Hawaiian or Other Pacific Islander). Since 2018, applications for clinical informatics fellowships have been accepted through the AAMC's Electronic Residency Application Service (ERAS). We analyzed national data from ERAS on applicants to clinical informatics fellowship programs for 2018 to 2020 positions. We also obtained applicants' subsequent success in starting clinical informatics fellowship training from the AAMC's GME Track survey. Over these 3 years, we found that the fellowship applicant pool grew from 63 to 74 (17%) and the number of positions offered grew from 34 to 41 (17%). The proportion of women applicants grew to 34% by 2020 and the proportion of underrepresented minorities grew to 12% by 2020. By comparison, medical students 7 years earlier (2013) were 47% female (P=.01) and 16% URM (P>.20), and applicants to many other subspecialties were similar. Applicants' sex and URM membership were not associated with success in starting fellowship training. We conclude that the underrepresentation of women and URM members in clinical informatics fellowships has improved but not resolved. Urgent efforts are needed to increase the both the numbers and the diversity of clinical informatics applicants by promoting the field among medical students and residents, particularly among women and URM members.

Association of Antiosteoporotic Medication Bisphosphonates and Denosumab with Primary Breast Cancer: An Electronic Health Record Cohort Study.

Background: The risks of osteoporosis and breast cancer are increasing in elderly women. Bisphosphonates and denosumab are recommended for treatment of osteoporosis. They have different and overlapping pharmacodynamics and previous studies have shown conflicting results regarding their risk association with breast cancer. We intend to further look into this issue through a comparative study. Methods: Electronic health records of 91,626 women older than 50 years with no previous history of malignancy and no nonbreast cancer during follow-up were retrieved from southern California and retrospectively analyzed using univariate, bivariate, and log-rank tests. Medication use, breast cancer risk, and associated demographic and clinical history were assessed. Results: Over an average of 3.6 years follow-up, the breast cancer relative risks (RRs) counted after 365 days of latency are 1.12 (95% confidence interval [CI]: 0.64-1.97) for denosumab ever users and 0.37 (95% CI: 0.21-0.66) for bisphosphonates ever users, when covariates are comparable. The significant difference is supported by the Log-rank test (p = 0.0004). Excluding statins coprescribers, the breast cancer RR is 1.31 (0.71, 2.43) in denosumab group and 0.26 (0.11, 0.62) in bisphosphonates group. There is a reduced RR in statins ever users (0.47, 95% CI: 0.38-0.58), and the breast cancer risk difference is not significant between concomitant denosumab/statins and bisphosphonates/statins ever users with RR 0.65 (0.16, 2.58) versus 0.55 (0.26, 1.16), p = 0.692. Conclusions: Our data support an association of lower breast cancer risk with bisphosphonates use in elderly women. We did not observe a lower breast cancer risk in denosumab group; however, our data revealed a potential lower breast cancer risk in denosumab users with concurrent statins use and this requires further study.

Privacy-protecting, reliable response data discovery using COVID-19 patient observations.

OBJECTIVE: To utilize, in an individual and institutional privacy-preserving manner, electronic health record (EHR) data from 202 hospitals by analyzing answers to COVID-19-related questions and posting these answers online. MATERIALS AND METHODS: We developed a distributed, federated network of 12 health systems that harmonized their EHRs and submitted aggregate answers to consortia questions posted at https://www.covid19questions.org. Our consortium developed processes and implemented distributed algorithms to produce answers to a variety of questions. We were able to generate counts, descriptive statistics, and build a multivariate, iterative regression model without centralizing individual-level data. RESULTS: Our public website contains answers to various clinical questions, a web form for users to ask questions in natural language, and a list of items that are currently pending responses. The results show, for example, that patients who were taking angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers, within the year before admission, had lower unadjusted in-hospital mortality rates. We also showed that, when adjusted for, age, sex, and ethnicity were not significantly associated with mortality. We demonstrated that it is possible to answer questions about COVID-19 using EHR data from systems that have different policies and must follow various regulations, without moving data out of their health systems. DISCUSSION AND CONCLUSIONS: We present an alternative or a complement to centralized COVID-19 registries of EHR data. We can use multivariate distributed logistic regression on observations recorded in the process of care to generate results without transferring individual-level data outside the health systems.

Comparative Effectiveness of Aspirin Dosing in Cardiovascular Disease.

BACKGROUND: The appropriate dose of aspirin to lower the risk of death, myocardial infarction, and stroke and to minimize major bleeding in patients with established atherosclerotic cardiovascular disease is a subject of controversy. METHODS: Using an open-label, pragmatic design, we randomly assigned patients with established atherosclerotic cardiovascular disease to a strategy of 81 mg or 325 mg of aspirin per day. The primary effectiveness outcome was a composite of death from any cause, hospitalization for myocardial infarction, or hospitalization for stroke, assessed in a time-to-event analysis. The primary safety outcome was hospitalization for major bleeding, also assessed in a time-to-event analysis. RESULTS: A total of 15,076 patients were followed for a median of 26.2 months (interquartile range [IQR], 19.0 to 34.9). Before randomization, 13,537 (96.0% of those with available information on previous aspirin use) were already taking aspirin, and 85.3% of these patients were previously taking 81 mg of daily aspirin. Death, hospitalization for myocardial infarction, or hospitalization for stroke occurred in 590 patients (estimated percentage, 7.28%) in the 81-mg group and 569 patients (estimated percentage, 7.51%) in the 325-mg group (hazard ratio, 1.02; 95% confidence interval [CI], 0.91 to 1.14). Hospitalization for major bleeding occurred in 53 patients (estimated percentage, 0.63%) in the 81-mg group and 44 patients (estimated percentage, 0.60%) in the 325-mg group (hazard ratio, 1.18; 95% CI, 0.79 to 1.77). Patients assigned to 325 mg had a higher incidence of dose switching than those assigned to 81 mg (41.6% vs. 7.1%) and fewer median days of exposure to the assigned dose (434 days [IQR, 139 to 737] vs. 650 days [IQR, 415 to 922]). CONCLUSIONS: In this pragmatic trial involving patients with established cardiovascular disease, there was substantial dose switching to 81 mg of daily aspirin and no significant differences in cardiovascular events or major bleeding between patients assigned to 81 mg and those assigned to 325 mg of aspirin daily. (Funded by the Patient-Centered Outcomes Research Institute; ADAPTABLE ClinicalTrials.gov number, NCT02697916.).

Primary nonadherence to statin medications: Survey of patient perspectives.

Statin medications reduce cardiovascular events, but many patients never start taking their prescribed statin (primary nonadherence). Limited knowledge exists about the attitudes and beliefs of those with primary nonadherence. In this study, patients with primary nonadherence to statin medications (n = 173) completed a self-administered cross-sectional survey that assessed their attitudes and beliefs related to primary nonadherence and to potential motivators for statin use. Patients were recruited in 2019 from two academic health systems and nationwide internet advertisements. Only 49 of 173 (28.3%) patients with primary nonadherence reported having cardiovascular disease (CVD). Ninety-nine patients (57.2%) never filled their prescription, and 74 (42.8%) filled but never took any statin. Over half failed to initially inform their prescriber they might not take the statin. Patients strongly or somewhat agreed that they desired alternate treatment plans such as diet and/or exercise (n = 134; 77.4%) or natural remedies/dietary supplements (n = 125; 72.3%). Ninety-eight (56.6%) stronglyor somewhat worried about the possibility of statin dependence or addiction. Twenty-seven (15.6%) patients noted that they would not take a statin based solely on CVD risk estimates; 50 (28.9%) selected a CVD risk threshold of >20%; and 23 (13.3%) a threshold of >50% as motivating factors to take statins. Patients with primary nonadherence have attitudes about taking statins based on CVD risk that differ from scientific recommendations, may not tell providers about their hesitation to take statins, and likely prefer alternative initial approaches to cholesterol lowering. Early shared decision-making and assessment of patient attitudes about statins could potentially better align initial approaches for CVD risk reduction.

Validation of an internationally derived patient severity phenotype to support COVID-19 analytics from electronic health record data.

OBJECTIVE: The Consortium for Clinical Characterization of COVID-19 by EHR (4CE) is an international collaboration addressing coronavirus disease 2019 (COVID-19) with federated analyses of electronic health record (EHR) data. We sought to develop and validate a computable phenotype for COVID-19 severity. MATERIALS AND METHODS: Twelve 4CE sites participated. First, we developed an EHR-based severity phenotype consisting of 6 code classes, and we validated it on patient hospitalization data from the 12 4CE clinical sites against the outcomes of intensive care unit (ICU) admission and/or death. We also piloted an alternative machine learning approach and compared selected predictors of severity with the 4CE phenotype at 1 site. RESULTS: The full 4CE severity phenotype had pooled sensitivity of 0.73 and specificity 0.83 for the combined outcome of ICU admission and/or death. The sensitivity of individual code categories for acuity had high variability-up to 0.65 across sites. At one pilot site, the expert-derived phenotype had mean area under the curve of 0.903 (95% confidence interval, 0.886-0.921), compared with an area under the curve of 0.956 (95% confidence interval, 0.952-0.959) for the machine learning approach. Billing codes were poor proxies of ICU admission, with as low as 49% precision and recall compared with chart review. DISCUSSION: We developed a severity phenotype using 6 code classes that proved resilient to coding variability across international institutions. In contrast, machine learning approaches may overfit hospital-specific orders. Manual chart review revealed discrepancies even in the gold-standard outcomes, possibly owing to heterogeneous pandemic conditions. CONCLUSIONS: We developed an EHR-based severity phenotype for COVID-19 in hospitalized patients and validated it at 12 international sites.

International Comparisons of Harmonized Laboratory Value Trajectories to Predict Severe COVID-19: Leveraging the 4CE Collaborative Across 342 Hospitals and 6 Countries: A Retrospective Cohort Study.

Objectives: To perform an international comparison of the trajectory of laboratory values among hospitalized patients with COVID-19 who develop severe disease and identify optimal timing of laboratory value collection to predict severity across hospitals and regions. Design: Retrospective cohort study. Setting: The Consortium for Clinical Characterization of COVID-19 by EHR (4CE), an international multi-site data-sharing collaborative of 342 hospitals in the US and in Europe. Participants: Patients hospitalized with COVID-19, admitted before or after PCR-confirmed result for SARS-CoV-2. Primary and secondary outcome measures: Patients were categorized as "ever-severe" or "never-severe" using the validated 4CE severity criteria. Eighteen laboratory tests associated with poor COVID-19-related outcomes were evaluated for predictive accuracy by area under the curve (AUC), compared between the severity categories. Subgroup analysis was performed to validate a subset of laboratory values as predictive of severity against a published algorithm. A subset of laboratory values (CRP, albumin, LDH, neutrophil count, D-dimer, and procalcitonin) was compared between North American and European sites for severity prediction. Results: Of 36,447 patients with COVID-19, 19,953 (43.7%) were categorized as ever-severe. Most patients (78.7%) were 50 years of age or older and male (60.5%). Longitudinal trajectories of CRP, albumin, LDH, neutrophil count, D-dimer, and procalcitonin showed association with disease severity. Significant differences of laboratory values at admission were found between the two groups. With the exception of D-dimer, predictive discrimination of laboratory values did not improve after admission. Sub-group analysis using age, D-dimer, CRP, and lymphocyte count as predictive of severity at admission showed similar discrimination to a published algorithm (AUC=0.88 and 0.91, respectively). Both models deteriorated in predictive accuracy as the disease progressed. On average, no difference in severity prediction was found between North American and European sites. Conclusions: Laboratory test values at admission can be used to predict severity in patients with COVID-19. Prediction models show consistency across international sites highlighting the potential generalizability of these models.

Reducing Emergency Department Visits Among Patients With Diabetes by Embedding Clinical Pharmacists in the Primary Care Teams.

BACKGROUND: Pharmacists are effective at improving control of cardiovascular risk factors, but it less clear whether these improvements translate into less emergency department (ED) use and fewer hospitalizations. The UCMyRx program embed pharmacists in primary care. OBJECTIVE: The objective of this study was to examine if the integration of pharmacists into primary care was associated with lower ED and hospital use for patients with diabetes. DESIGN: This was a quasi-experimental study with a comparator group. SUBJECTS: The analytic sample included patients with diabetes with uncontrolled cardiovascular risk factors (A1C >9%, blood pressure >140/90 mm Hg, low-density lipoprotein-cholesterol >130 mg/dL) who had 1 or more visits in either a UCMyRx (648 patients, 14 practices) or usual care practice (1944 patients, 14 practices). MEASURES: Our outcomes were ED and hospitalization rates as measured before and after the consultations between UCMyRx and usual care. Our predictor variable was the pharmacist consultation. Poisson generalized estimating equations model was used to estimate the adjusted predicted change in utilization before and after the pharmacist consultation. The Average Treatment Effect on the Treated was estimated. RESULTS: In models adjusted, the adjusted mean predicted number of emergency department visits/month during the year before the consultation was 0.09 among UCMyRx patients. During the year after initiating the care with the pharmacists, this rate decreased to an adjusted mean monthly rate of 0.07, with an Average Treatment Effect on the Treated=0.021 (P=0.035), a predicted reduction of 21% in emergency department visits associated with the clinical pharmacist consults. There was a nonsignificant predicted 3.2% reduction in hospitalizations over time for patients in the UCMyRx program. CONCLUSION: Clinical pharmacists are an important addition to clinical care teams in primary care practices and significantly decreased utilization of the ED among patients with poorly controlled diabetes.

The Impact of Resident Holdover Admissions on Length of Hospital Stay and Risk of Transfer to an Intensive Care Unit.

OBJECTIVE: Implementation of residency duty hour standards has led to adoption of different staffing models, such as the "holdover" model, whereby nighttime teams admit patients and transfer their care to daytime teams who provide ongoing care. In contrast, nonholdover teams at our institution are responsible for both admitting patients and providing ongoing care. We sought to determine whether patients admitted by holdover teams experience worse outcomes than those admitted by nonholdover teams. METHODS: This is a retrospective cohort study of patients admitted to the internal medicine hospital service at a quaternary care hospital from July 2013 to June 2015. Primary outcomes included hospital length of stay (LOS) and transfer to an intensive care unit within 72 hours of admission. Secondary outcomes were any transfer to an intensive care unit, in-hospital mortality, discharge to home (versus discharge to postacute care facility), and readmission to the health system within 30 days of discharge. RESULTS: We analyzed 5518 encounters, 64% of which were admitted by a holdover team. Outcomes were similar between study groups, except the LOS, which was 5.5 hours longer for holdover encounters in unadjusted analyses (5.18 versus 4.95 days, P = 0.04) but not significantly different in adjusted analyses. The mean discharge time was 4:00 p.m. for both groups, whereas the mean admission times were 12:00 a.m. and 4:00 p.m. for holdover and nonholdover encounters, respectively. CONCLUSIONS: Holdover encounters at our institution were not associated with worse patient safety outcomes. A small increase in LOS may have been attributable to holdover patients having earlier admission and identical discharge times.

Accuracy of an Electronic Health Record Patient Linkage Module Evaluated between Neighboring Academic Health Care Centers.

BACKGROUND: Patients often seek medical treatment among different health care organizations, which can lead to redundant tests and treatments. One electronic health record (EHR) platform, Epic Systems, uses a patient linkage tool called Care Everywhere (CE), to match patients across institutions. To the extent that such linkages accurately identify shared patients across organizations, they would hold potential for improving care. OBJECTIVE: This study aimed to understand how accurate the CE tool with default settings is to identify identical patients between two neighboring academic health care systems in Southern California, The University of California Los Angeles (UCLA) and Cedars-Sinai Medical Center. METHODS: We studied CE patient linkage queries received at UCLA from Cedars-Sinai between November 1, 2016, and April 30, 2017. We constructed datasets comprised of linkages ("successful" queries), as well as nonlinkages ("unsuccessful" queries) during this time period. To identify false positive linkages, we screened the "successful" linkages for potential errors and then manually reviewed all that screened positive. To identify false-negative linkages, we applied our own patient matching algorithm to the "unsuccessful" queries and then manually reviewed a sample to identify missed patient linkages. RESULTS: During the 6-month study period, Cedars-Sinai attempted to link 181,567 unique patient identities to records at UCLA. CE made 22,923 "successful" linkages and returned 158,644 "unsuccessful" queries among these patients. Manual review of the screened "successful" linkages between the two institutions determined there were no false positives. Manual review of a sample of the "unsuccessful" queries (n = 623), demonstrated an extrapolated false-negative rate of 2.97% (95% confidence interval [CI]: 1.6-4.4%). CONCLUSION: We found that CE provided very reliable patient matching across institutions. The system missed a few linkages, but the false-negative rate was low and there were no false-positive matches over 6 months of use between two nearby institutions.