Citations

Abstract Introduction Delirium occurrence is common and preventive strategies are resource intensive. Screening tools can prioritize patients at risk. Using machine learning, we can capture time and treatment effects that pose a challenge to delirium prediction. We aim to develop a delirium prediction model that can be used as a screening tool. Methods From the eICU Collaborative Research Database (eICU-CRD) and the Medical Information Mart for Intensive Care version III (MIMIC-III) database, patients with one or more Confusion Assessment Method-Intensive Care Unit (CAM-ICU) values and intensive care unit (ICU) length of stay greater than 24 h were included in our study. We validated our model using 21 quantitative clinical parameters and assessed performance across a range of observation and prediction windows, using different thresholds and applied interpretation techniques. We evaluate our models based on stratified repeated cross-validation using 3 algorithms, namely Logistic Regression, Random Forest, and Bidirectional Long Short-Term Memory (BiLSTM). BiLSTM represents an evolution from recurrent neural network-based Long Short-Term Memory, and with a backward input, preserves information from both past and future. Model performance is measured using Area Under Receiver Operating Characteristic, Area Under Precision Recall Curve, Recall, Precision (Positive Predictive Value), and Negative Predictive Value metrics. Results We evaluated our results on 16 546 patients (47% female) and 6294 patients (44% female) from eICU-CRD and MIMIC-III databases, respectively. Performance was best in BiLSTM models where, precision and recall changed from 37.52% (95% confidence interval [CI], 36.00%–39.05%) to 17.45 (95% CI, 15.83%–19.08%) and 86.1% (95% CI, 82.49%–89.71%) to 75.58% (95% CI, 68.33%–82.83%), respectively as prediction window increased from 12 to 96 h. After optimizing for higher recall, precision and recall changed from 26.96% (95% CI, 24.99%–28.94%) to 11.34% (95% CI, 10.71%–11.98%) and 93.73% (95% CI, 93.1%–94.37%) to 92.57% (95% CI, 88.19%–96.95%), respectively. Comparable results were obtained in the MIMIC-III cohort. Conclusions Our model performed comparably to contemporary models using fewer variables. Using techniques like sliding windows, modification of threshold to augment recall and feature ranking for interpretability, we addressed shortcomings of current models.

Abstract Background A patent foramen ovale (PFO) is a persistent embryonic defect in the interatrial septum. Platypnoea-orthodeoxia syndrome is characterized by positional hypoxaemia that is most commonly due to right-to-left shunting through a PFO. Dynamic right-to-left shunting through a PFO can also exacerbate positional hypoxaemia without platypnea-orthodeoxia syndrome. Case summary A 78-year-old woman with hyperthyroidism and paroxysmal atrial fibrillation (AF) presented with positional hypoxaemia exacerbated by supine positioning. Diagnostic testing revealed intermittent right-to-left shunting through a PFO triggered by worsening atrial functional tricuspid regurgitation and elevated right atrial pressures. Diuresis, rate control, and thyroidectomy initially led to resolution of positional hypoxaemia, but recurrent AF episodes triggered right-to-left shunting with recurrent desaturation. Left atrial and cavo-tricuspid isthmus ablation led to restoration of normal sinus rhythm and resolution of positional hypoxaemia without PFO closure. Discussion The clinical presentation of intermittent intracardiac right-to-left shunting can mimic decompensated heart failure with pulmonary oedema. Persistent hypoxaemia out of proportion to the degree of pulmonary oedema and minimally responsive to supplemental O2 should raise suspicion for right-to-left shunt aetiology. Positional arterial blood gases can facilitate the diagnostic evaluation of refractory hypoxaemia in cases of suspected shunting. Diagnostic imaging for PFO detection includes both transthoracic and transesophageal echocardiography with Valsalva manoeuver and agitated saline injection. Closure of a PFO for management of arterial deoxygenation syndromes should not be performed before treating other causes of arterial deoxygenation and optimizing factors that may exacerbate shunting across the PFO.

Background: Automated insulin delivery (AID) systems have proven effective in increasing time-in-range during both clinical trials and real-world use. Further improvements in outcomes for single-hormone (insulin only) AID may be limited by suboptimal insulin delivery settings. Methods: Adults (≥18 years of age) with type 1 diabetes were randomized to either sensor-augmented pump (SAP) (inclusive of predictive low-glucose suspend) or adaptive zone model predictive control AID for 13 weeks, then crossed over to the other arm. Each week, the AID insulin delivery settings were sequentially and automatically updated by an adaptation system running on the study phone. Primary outcome was sensor glucose time-in-range 70–180 mg/dL, with noninferiority in percent time below 54 mg/dL as a hierarchical outcome. Results: Thirty-five participants completed the trial (mean age 39 ± 16 years, HbA1c at enrollment 6.9% ± 1.0%). Mean time-in-range 70–180 mg/dL was 66% with SAP versus 69% with AID (mean adjusted difference +2% [95% confidence interval: −1% to +6%], P = 0.22). Median time <70 mg/dL improved from 3.0% with SAP to 1.6% with AID (−1.5% [−2.4% to −0.5%], P = 0.002). The adaptation system decreased initial basal rates by a median of 4% (−8%, 16%) and increased initial carbohydrate ratios by a median of 45% (32%, 59%) after 13 weeks. Conclusions: Automated adaptation of insulin delivery settings with AID use did not significantly improve time-in-range in this very well-controlled population. Additional study and further refinement of the adaptation system are needed, especially in populations with differing degrees of baseline glycemic control, who may show larger benefits from adaptation.

AbstractThis study assessed the effectiveness of genetic testing in shortening the time to diagnosis of late infantile neuronal ceroid lipofuscinosis type 2 (CLN2) disease. Individuals who received epilepsy gene panel testing through Behind the Seizure®, a sponsored genetic testing program (Cohort A), were compared to children outside of the sponsored testing program during the same period (Cohort B). Two cohorts were analyzed: children aged ≥24 to ≤60 months with unprovoked seizure onset at ≥24 months between December 2016 and January 2020 (Cohort 1) and children aged 0 to ≤60 months at time of testing with unprovoked seizure onset at any age between February 2019 and January 2020 (Cohort 2). The diagnostic yield in Cohort 1A (n = 1814) was 8.4% (n = 153). The TPP1 diagnostic yield within Cohort 1A was 2.9‐fold higher compared to Cohort 1B (1.0%, n = 18/1814 vs. .35%, n = 8/2303; p = .0157). The average time from first symptom to CLN2 disease diagnosis was significantly shorter than previously reported (9.8 vs. 22.7 months, p < .001). These findings indicate that facilitated access to early epilepsy gene panel testing helps to increase diagnostic yield for CLN2 disease and shortens the time to diagnosis, enabling earlier intervention.

Background: It is known that hypokalemia, hypocalcemia and hypomagnesemia cause a prolonged QTc interval and are associated with an increased risk of ventricular arrhythmias. However, there are limited data on optimal serum electrolytes for patients admitted to the coronary care unit (CCU). Objective: Here, we aimed to evaluate the relationship between serum potassium, ionized calcium and magnesium levels and in-hospital mortality in the CCU setting. Methods: This retrospective cohort study included 8,498 consecutive patients admitted to the CCU at an academic tertiary medical center from 2004 through 2013. The mean of serum electrolytes was categorized as <3.5, 3.5-<4.0, 4.0-<4.5 (reference), 4.5-<5.0, and ≥5.0 mEq/L for potassium; <4.4, 4.4-<4.6, 4.6-<4.8, 4.8-<5.0 (reference), and ≥5.0 mg/dL for ionized calcium; and <1.8, 1.8-<2.0, 2.0-<2.2 (reference), 2.2-<2.4, and ≥2.4 mg/dL for magnesium. Multivariate logistic regression analysis was used to determine the association between the mean of serum electrolytes and in-hospital mortality. Results: Of the 8,498 patients, 599 (7%) died during hospitalization. Relationship between mean serum potassium (U-shaped), ionized calcium (inverted J-shape) and magnesium (J-shaped) and in-hospital mortality was present in Figure. Serum potassium levels <3.5, 4.5-<5.0 and ≥5.0 mEq/L were independently associated with an increased risk of in-hospital mortality (adjusted OR 1.7 95% CI 1.1-2.5, OR 1.7 95% CI 1.3-2.2, and OR 3.4 95% CI 2.4-4.8, respectively), compared to the reference group. As for serum ionized calcium, the levels of <4.4, 4.4-<4.6, and 4.6-<4.8 mg/dL independently increased risk of mortality (adjusted OR 3.0 95% CI 2.3-4.0, OR 2.0 95% CI 1.5-2.6, and OR 1.4 95% CI, 1.1-1.8, respectively). As for serum magnesium, an only level above 2.4 mg/dL was associated with an increased mortality risk (OR 1.6 95% CI, 1.1-2.2). Conclusions: In the CCU setting, serum potassium 3.5-<4.5 mEq/L, ionized calcium ≥5.0 mg/dL and magnesium <2.4 mg/dL are associated with the lowest in-hospital mortality. These findings do not support supratherapeutic repletion of potassium and magnesium.

Abstract Background: As the development and use of oral anticancer agents increases, it is critical to understand patient adherence to both standard and investigational agents. The open label, phase 3 multicenter PALLAS trial investigates whether adding 2 years of the CDK4/6 inhibitor palbociclib (P) to adjuvant endocrine therapy (ET) improves invasive disease-free survival (iDFS) over adjuvant ET alone in patients (pts) with HR- positive, HR2-negative, stage II-III breast cancer. Pts were randomly assigned to either Palbociclib (125 mg/day, 3 weeks on, 1 week off, in a 28-day cycle) plus ongoing provider/patient-choice adjuvant ET (P+ET) versus ET alone. We examined patient-reported adherence to ET +/- P during the first two years of study treatment. Methods: Adherence outcomes were measured in English-speaking pts in the U.S., UK, Ireland and Australia, Spanish-speaking pts in Spain and Mexico, and German-speaking pts in Germany and Austria. Adherence measures included drug diaries completed at each cycle, pill counts (for P only) collected at each study visit, and the Morisky Medication Adherence Scale-4 item and the McHorney Adherence Estimator questionnaires completed at cycles 2, 3, 6, 12, 18, and 24 (22 months). Mean adherence for each cycle was defined as the average proportion of prescribed pills taken (via drug diary) across all patients who initiated that cycle. Persistence was defined as the duration of drug initiation to treatment cessation (via drug diary). Generalized estimating equations were used to model the “most adherent” pts on the Morisky (score = 5 vs score &lt;5) and “low risk” for adherence problems on the McHorney (score=0 vs score &gt;0) to compare the average difference between arms over time for ET, adjusting for baseline demographic and clinical variables. Results: 81% (N=4688) of PALLAS pts were included in this analysis. Median persistence to ET was 23.6 months in P + ET (n=2169), 23.7 months in ET alone (2136) and 20.4 months for P (n=2194). The number of pts who initiated each cycle for ET declined over time and was similar between arms; the decline was more marked for P (Table 1). Mean adherence range as measured by drug diary was 98.2-99.3% for ET in P+ET and 98.0 - 99.4% in ET alone; and for P, ranged from 93.4 - 98.8%. The adherence and persistence results were nearly identical whether measured by drug diary or pill count for P. The observed percent “most adherent” for P measured by the Morisky scale ranged from 71.9% - 79.6% and the percent “low risk” for adherence problems measured by the McHorney scale, 64.0% - 73.4%. The percent of pts “most adherent” and “low risk” for adherence problems to ET was higher, on average over time, in the P+ET group compared to ET alone (75% vs 68%, p&lt;0.01; 75% vs 72%, p=.05, respectively). Conclusions: Self-reported mean adherence for both P + ET and ET alone was strikingly high for pts who remained on therapy; persistence was also high with ET during the 2-year treatment period. Current analyses suggest that nonadherence to either P or ET was likely not a major contributor to the iDFS results seen in the overall PALLAS trial. These results illustrate the importance of measuring and monitoring patient adherence to oral study agents. Support: AFT, Pfizer; ClinicalTrials.gov (NCT02513394) and EudraCT (2014-005181-30). https://acknowledgments.alliancefound.org Table 1.Adherence and persistence in PALLAS over the 24 month treatment periodPalbocicilb + ETETPalbociclibETETTreatment cycleMean Adherence (SD)N*Mean Adherence (SD)NMean Adherence (SD)N193.4 (13.8)229098.7 (6.1)230999.0 (5.2)2343294.7 (13.3)218198.8 (6.4)220398.4 (7.1)2206397.4 (10.1)211298.7 (7.3)214298.8 (6.3)2168497.6 (10.2)203598.6 (7.8)211099.0 (5.7)2154597.6 (11.1)197598.3 (7.9)209398.1 (8.3)2148698.1 (8.7)189998.7 (6.6)203798.5 (8.2)2116798.1 (9.3)185599.0 (6.3)201298.6 (7.3)2102898.0 (9.1)181098.2 (8.2)199798.1 (8.2)2092998.4 (7.4)173399.0 (6.4)196298.8 (6.8)20511098.3 (8.6)171499.2 (4.9)194699.0 (5.5)20381198.3 (8.0)169198.5 (7.3)193798.2 (8.2)20311298.7 (6.2)161299.1 (5.3)191098.7 (7.5)19861398.5 (7.7)159698.9 (6.4)189398.8 (7.1)19691497.9 (9.3)157698.4 (7.3)188298.0 (9.1)19611598.5 (6.9)151499.2 (4.6)185499.0 (6.0)19281698.2 (8.7)148699.1 (4.6)183599.1 (5.7)19131798.1 (9.8)147998.3 (7.8)182698.5 (6.4)19031898.7 (6.8)141699.3 (4.1)176899.2 (5.3)18601998.7 (7.3)140899.3 (3.8)175699.4 (3.1)18482098.4 (8.9)139798.9 (5.3)174298.7 (5.8)18382198.5 (7.5)130199.0 (6.1)167698.8 (7.1)17842297.8 (9.7)124698.8 (6.2)164799.2 (5.7)17692398.1 (8.8)118098.2 (8.5)161998.4 (8.0)17552498.8 (7.2)108699.0 (6.3)146898.9 (7.1)16192598.8 (6.7)99699.2 (5.2)143298.8 (7.1)15892697.9 (11.2)84599.1 (5.7)140998.6 (8.8)1578ET=Endocrine therapy * Number of pts who initiated the treatment cycle. Citation Format: Eileen Shinn, David Zahrieh, Angela DeMichele, Nick Zdenkowski, Julie Lemieux, Jun Mao, Vesna Bjelic-Radisic, Michelle Naughton, Georg Pfeiler, Karen Gelmon, Ingrid Mayer, Daniel Egle, Gabriele Zoppoli, Tiffany Traina, Miguel Martin Jiménez, Silvia Antolin Novoa, Tufia Haddad, Arlene Chan, Alistair Edward Ring, Antonio Wolff, Jose JuanPonce Lorenzo, Dhanusha Sabanathan, Hal Burstein, Zbigniew Ireneusz Nowecki, Gunda Pristauz-Telsnigg, Adam Brufsky, Meritxell Bellet-Ezquerra, Theodoros Foukakis, Yelena Novik, Gabor Rubovszky, Karoline Muehlbacher, Otto Metzger, Theodora Goulioti, Ernest Law, Ann Partridge, Lisa Carey, Alex Zoroufy, Dominik Hlauschek, Christian Fesl, Erica Mayer, Michael Gnant. Adherence with adjuvant endocrine therapy with or without Palbociclib in the PALLAS trial [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P4-12-01.