Faculty Bibliography

Objectives: Discharges against medical advice (AMA) occur when patients leave the hospital prior to a physician-recommended endpoint. It is unknown whether AMA discharges are associated with higher healthcare costs within 30 days of discharge. We examine healthcare costs following a hospital discharge in a commercially insured population.
Method(s): This retrospective cohort study examined individuals aged 18 to 64 with a hospitalization during 2007-2015 from a 10% random sample of enrollees in the IQVIA^TM Adjudicated Health Plan Claims Data. We included individuals with insurance coverage 6 months before and 30 days after their first hospitalization. Individuals with AMA and non-AMA discharges were matched on baseline covariates. Generalized linear models and cost ratios (CR) were used to quantify the association between AMA discharges and 30-day costs. We report CRs overall and by points of service (inpatient, emergency department (ED), physician office, non-physician outpatient encounter (NPOE) and prescription drug fill).
Result(s): Of the 467,746 individuals in the unmatched sample, 2,164 (0.46%) were discharged AMA. Mean (median) costs were 20% (5%) higher in the AMA group compared to the non-AMA group. In the matched sample and relative to those discharged routinely, individuals with an AMA discharge incurred 1.20 times (95% CI: 1.08, 1.34) higher costs. Similarly, individuals with an AMA discharge incurred higher inpatient (CR: 1.71, 95% CI: 1.45, 2.01) and ED (CR: 2.10, 95% CI: 1.84, 2.39) costs within 30 days post-discharge. Conversely, individuals with an AMA discharge incurred lower NPOE (CR: 0.84, 95% CI: 0.74, 0.95) and prescription drug fill (CR: 0.81; 95% CI: 0.73, 0.91) costs. There were no differences in physician office visit costs across the two groups.
Conclusion(s): An AMA discharge is associated with higher 30-day costs compared to those discharged routinely, particularly for acute care services. Future work should determine whether these findings extend to publicly-insured individuals.
Copyright

BACKGROUND:Childhood anxiety prevents optimal diabetes management yet may be underrecognized by guardians. OBJECTIVE:We aimed to investigate associations among anxiety, diabetes treatment adherence, and diabetes symptom control through child and guardian report. METHODS:Cross-sectional pilot study surveying a convenience sample of children (ages 2-21) in a pediatric endocrinology clinic. Behavior Assessment System for Children, Second Edition 2, Self-Care Inventory Report, and Pediatric Quality of Life measured anxiety, diabetes treatment adherence, and diabetes symptom control. Analyses were performed with Spearman correlations. RESULTS: = -0.38, P = 0.02]). Child- and guardian-reported anxiety were positively correlated (rho = 0.426, P = 0.017)-particularly for children aged >12 (rho = 0.686, P = 0.003)-although not significantly for children ≤ 12 (rho = 0.201, P = 0.473). CONCLUSION:Anxiety in children with type 1 diabetes varies with the domain of diabetes management (treatment adherence vs. symptom control) and reporting source (child vs. guardian). Children aged ≤12 exhibited a stronger relationship between higher anxiety and worse diabetes management with worse treatment adherence and symptom control in the presence of higher anxiety. Guardians of younger children were less effective at recognizing symptoms. Challenges identifying anxiety and its detrimental effects on diabetes management suggest routine screening of anxiety in pediatric endocrinology clinics is especially salient.

OBJECTIVE:Inpatient discharge education is often suboptimal. Measures of parents' perceived comprehension of discharge instructions are included in national metrics given linkage to morbidity; few studies compare parents' perceived and actual comprehension. We (1) compared parent perceived and actual comprehension of discharge instructions and (2) assessed associations between plan complexity and parent health literacy with overestimation of comprehension (perceive comprehension but lack actual comprehension). METHODS:Prospective cohort study of English/Spanish-speaking parents (n=192) of inpatients ≤12 years old and discharged on ≥1 daily medication from an urban public hospital. We used McNemar's tests to compare parent perceived (agree/strongly agree on 5-point Likert scale) and actual comprehension (concordance of parent report with medical record) of instructions (domains: medications, appointments, return precautions, and restrictions). Generalized estimating equations were performed to assess associations between low parent health literacy (Newest Vital Sign score ≤3) and plan complexity with overestimation of comprehension. RESULTS:Medication side effects were the domain with lowest perceived comprehension (80%), while >95% of parents perceived comprehension for other domains. Actual comprehension varied by domain (41-87%) and was lower than perceived comprehension. Most (84%) parents overestimated comprehension in ≥1 domain. Plan complexity (adjusted odds ratio [aOR] 3.6 [95% CI 2.9-4.7]) and low health literacy (aOR 1.9 [1.3-2.6]) were associated with overestimation of comprehension. CONCLUSIONS:Parental perceived comprehension of discharge instructions overestimated actual comprehension in most domains. Plan complexity and low health literacy were associated with overestimation of comprehension. Future interventions should incorporate assessment of actual comprehension and standardization of discharge instructions.

BACKGROUND:While mass-casualty incidents (MCIs) may have competing absolute definitions, a universally ac-cepted criterion is one that strains locally available resources. In the fall of 2017, a MCI occurred in New York and Bellevue Hospital received multiple injured patients within minutes; lessons learned included the need for a formal-ized, efficient patient and injury tracking system. Our objective was to create an organized MCI clinical tracking form for civilian trauma centers. METHODS:After the MCI, the notes of the surgeon responsible for directing patient triage were analyzed. A suc-cinct, organized template was created that allows MCI directors to track demographics, injuries, interventions, and other important information for multiple patients in a real-time fashion. This tool was piloted during a subsequent MCI. RESULTS:In late 2018, the hospital received six patients following another MCI. They arrived within a 4-minute window, with 5 patients being critically injured. Two emergent surgeries and angioembolizations were performed. The tool was used by the MCI director to prioritize and expedite care. All physicians agreed that the tool assisted in orga-nizing diagnostic and therapeutic triage. CONCLUSIONS:During MCIs, a streamlined patient tracking template assists with information recall and communica-tion between providers and may allow for expedited care.

BACKGROUND:Optical coherence tomography (OCT) studies have demonstrated differences between people with schizophrenia and controls. Many questions remain including the agreement between scanners. The current study seeks to determine inter-device agreement of OCT data in schizophrenia compared to controls and to explore the relations between OCT and visual function measures. METHODS:Participants in this pilot study were 12 individuals with schizophrenia spectrum disorders and 12 age- and sex-matched controls. Spectralis and Cirrus OCT machines were used to obtain retinal nerve fiber layer (RNFL) thickness and macular volume. Cirrus was used to obtain ganglion cell layer + inner plexiform layer (GCL + IPL) thickness. Visual function was assessed with low-contrast visual acuity and the King-Devick test of rapid number naming. RESULTS:There was excellent relative agreement in OCT measurements between the two machines, but poor absolute agreement, for both patients and controls. On both machines, people with schizophrenia showed decreased macular volume but no difference in RNFL thickness compared to controls. No between-group difference in GCL + IPL thickness was found on Cirrus. Controls showed significant associations between King-Devick performance and RNFL thickness and macular volume, and between low-contrast visual acuity and GCL + IPL thickness. Patients did not show significant associations between OCT measurements and visual function. CONCLUSIONS:Good relative agreement suggests that the offset between machines remains constant and should not affect comparisons between groups. Decreased macular volume in individuals with schizophrenia on both machines supports findings of prior studies and provides further evidence that similar results may be found irrespective of OCT device.

Background and Objectives: To rapidly develop, implement, and evaluate emergency department (ED) clinical protocols for initiation of buprenorphine (E

Background: Primary care clinics often struggle to choose the approach to alcohol and drug screening that is best suited to their resources, workflows, and patient populations. We are conducting a multi-site study to inform the implementation and feasibility of electronic health record (EHR)-integrated screening.
Method(s): In two urban academic health systems, researchers worked with stakeholders from 6 clinics to define and implement their optimal screening approach. All clinics used single-item screening questions for alcohol/drugs followed by AUDIT-C/DAST-10. Clinics chose between: (a) screening at routine vs. annual visits; and (b) staff-administered vs computer self-administered screening. Results were recorded in the EHR, and data was extracted quarterly to describe implementation outcomes including screening rate and detected prevalence of unhealthy (moderate-high risk) use among those screened. Findings are from the first 3 to 12 months post-implementation at each clinic.
Result(s): Across sites, of 84 311 patients with primary care visits, 58 492 (69%) were screened. In the four clinics with mature (9-12 months) implementation, screening rates ranged from 42% to 95%. Rates were lower (10%-22%) in the two clinics that recently launched. Screening at routine encounters, in comparison to annual visits, achieved higher screening rates for alcohol (90%-95% vs 42%-62%) and drugs (90%-94% vs 38%-60%). Staff-administered screening, in comparison to patient self-administered screening, had lower rates of detection of unhealthy alcohol use (2% vs 15-37%). Detection of unhealthy drug use was low, ranging from 0.3% to 1.5%.
Conclusion(s): EHR-integrated screening was feasible to implement in at least four of the six clinics; 1-year results (available Fall 2019) will determine feasibility at all sites. Self-administered screening at routine primary care visits achieved the highest rates of screening and detection of unhealthy alcohol use. Although limited by differences among clinics and their patient populations, this study provides insight into outcomes that may be expected with commonly used screening strategies in primary care.
Summary: This multi-site study conducted in the NIDA Clinical Trials Network seeks to inform the implementation and feasibility of EHR-integrated screening for substance use in primary care. This study will provide insight into outcomes that may be expected with commonly used screening strategies in primary care and may assist in fine-tuning the most appropriate approach to alcohol and drug screening best suited for primary care clinics, based on their individual resources, workflows, and patient populations

OBJECTIVE:To explore the impact of an adapted goals-of-care communications skills workshop created for ED physicians from the physicians' perspective. METHODS:Semi-structured, one-on-one audio-recorded interviews lasting 30-60 min were conducted with twelve physicians who had completed the training workshop. Interviews explored the experience of undertaking the workshop, its impact on their clinical practice and their ability to teach new skills to other clinicians using learnt techniques. Descriptive content analysis was performed on interview transcripts. RESULTS:Participants reported positive experiences of the workshop. The analysis identified four main themes and ten subthemes dealing with workshop content and its impact on subsequent ED-based clinical practice. There were: 1) value and future improvements for the course; 2) value of the course to practice; 3) value of the course for teaching residents and 4) barriers to application of learning. Specifically recommended components include the use of mnemonics, mechanisms to introduce difficult conversations and a positive feedback environment. Participant-recommended refinements to the workshop included emphasising urgency, replicating the chaotic ED environment and expanding content to include more ED-focused goals-of-care discussions. CONCLUSION/CONCLUSIONS:A short, focused training workshop directed at improving palliative care communication skills among ED clinicians appears to be welcomed and useful.

PURPOSE:Cohort participants usually have lower mortality rates than nonparticipants, but it is unclear if this survival advantage decreases or increases as cohort studies age. METHODS:We used a 1975 private census of Washington County, Maryland, to compare mortality among cohort participants to nonparticipants for three cohorts, Campaign Against Cancer and Stroke (CLUE I), Campaign Against Cancer and Heart Disease (CLUE II), and Atherosclerosis Risk In Communities (ARIC) initiated in 1974, 1989, and 1986, respectively. We analyzed mortality risk using time-truncated Cox regression models. RESULTS:Participants had lower mortality risk in the first 10 years of follow-up compared with nonparticipants (fully adjusted average hazard ratio [95% confidence intervals] were 0.72 [0.68, 0.77] in CLUE I, 0.69 [0.65, 0.73] in CLUE II, and 0.74 [0.63, 0.86] in ARIC), which persisted over 20 years of follow-up (0.81 [0.78, 0.84] in CLUE I, 0.87 [0.84, 0.91] in CLUE II, and 0.90 [0.83, 0.97] in ARIC). This lower average hazard for mortality among participants compared with nonparticipants attenuated with longer follow-up (0.99 [0.96, 1.01] after 30+ years in CLUE I, 1.02 [0.99, 1.05] after 30 years in CLUE II, and 0.95 [0.89, 1.00] after 30+ years in ARIC). In ARIC, participants who did not attend visits had higher mortality, but those who did attend visits had similar mortality to the community. CONCLUSIONS:Our results suggest the volunteer selection for mortality in long-standing epidemiologic cohort studies often diminishes as the cohort ages.