-Author name in bold denotes the presenting author
-Asterisk * with author name denotes a Non-ASH member
Clinically Relevant Abstract denotes an abstract that is clinically relevant.

PhD Trainee denotes that this is a recommended PHD Trainee Session.

Ticketed Session denotes that this is a ticketed session.

3898 Exploring Differences in Hospital Readmissions in Patients with Sickle Cell Disease By Examining Patterns with mHealth-Acquired Pain and Physiologic Data

Program: Oral and Poster Abstracts
Session: 114. Sickle Cell Disease, Sickle Cell Trait, and Other Hemoglobinopathies, Excluding Thalassemias: Clinical and Epidemiological: Poster III
Hematology Disease Topics & Pathways:
Research, Sickle Cell Disease, Adult, Clinical Research, Hemoglobinopathies, Diseases, Technology and Procedures, Study Population, Human, Machine learning
Monday, December 9, 2024, 6:00 PM-8:00 PM

Shannon Ford, PhD1,2*, Jhana Parikh, BS, MS2,3*, Abhinav Gundala, MSc2*, Caroline Vuong, MD4*, Arvind Subramaniam, MSc2,5*, Elizabeth Hensley, BS2*, Olivia Fernandez6* and Nirmish Shah, MD6

1School of Nursing, University of North Carolina at Wilmington, Wilmington, NC
2Nanbar Health LLC, Winterville, NC
3Campbell School of Osteopathic Medicine, Campbell University, Lillington, NC
4Department of Pediatric Hematology, Amsterdam University Medical Center, Amsterdam, Netherlands
5ECU Brody School of Medicine, East Carolina University, Greenville
6Department of Medicine, Division of Hematology - Duke Sickle Cell Comprehensive Care Unit, Duke University Medical Center, Durham, NC

Introduction

Vaso-occlusive crises (VOCs) are the most common reason for hospital admission for patients with sickle cell disease (SCD). Due to high readmission rates (30–40%) for pain, it is important to understand patterns of physiologic data and self-reported pain that may lead to re-admission. Early recognition of negative patterns can be critical for improving patient outcomes and decreasing healthcare utilization. This study explored the differences in symptom and physiologic data between patients re-admitted and those who were not.

Methods

Patients with SCD aged 18 years and above were eligible for inclusion and approached at a single institution in the Duke University Medical Center. Following informed consent, patients were provided a mobile app (Nanbar) and an Apple Watch. Patients were instructed to continuously wear the Apple Watch and to report pain and other symptoms at least once daily within the Nanbar app for the duration of six months. Pain and physiologic data were collected using the Apple Watch, and demographic and vital signs were acquired from electronic medical records. Network analysis was performed to understand the relationship between re-admission and both physiological data and symptoms. Partial correlation coefficients (PCC) with a (p<.05) threshold were compared for variables of readmitted and not-readmitted networks. Variables included: 1) VOCs for the previous year (VOC); 2) vital signs (systolic blood pressure (SBP), diastolic blood pressure (DBP), pain score, pulse (HR), respiratory rate (RR), blood peripheral oxygen saturation (SP02), and temperature (temp).

Results

The study population included 12 individuals (6 females, 6 males) aged 18–43. Nine individuals had genotype HbSS, two had HbSC, and one had HbSβhalassemia. Ten individuals were currently being treated with hydroxyurea. During the study, a total of 92 days of data were reported by 12 individuals. Median days of symptom reporting (25, range 5-122 days), Apple Watch usage (36, range 6-103 days). Five patients represent the sixteen readmissions within 30 days (median = 2, range 2-8).

For readmitted patients, associations between the partial correlation coefficient (PCC) of the variables from positive to negative PCCs were: VOC with step count (0.93); pain with Sp02 (0.77); VOC and DBP (-0.69); and step count with temp (-0.73). For non-readmitted, associations between PCC of the variables from positive to negative were: pain and VOC (0.46); pain and DBP (0.4); pain and temp (0.26); and VOC and DBP (-0.25). Bootstrapping (n=1000) verified the stability of network edges and strength centrality (absolute sum of all edges).

In the network for readmitted patients, the most influential variables were SBP, DBP, and respiratory rate identified by strength centrality. Expected Influence (EI), which is like strength centrality but takes into account negative associations among variables, found SBP and DBP as the most influential. In the network without readmissions, the most influential variables were DBP, SpO2, and step count identified by strength centrality and pain, SpO2, and step count identified by EI. These findings indicate that the variables with the greatest impact on all other variables, notably including pain and VOC incidence, are different for patients who are readmitted vs those who are not. In particular, the objective clinical variables such as vital signs seem to be more influential for patients who are readmitted, compared to subjective and/or non-clinical variables, including pain and step count for patients who are not.

Conclusions

Using network analysis applied to physiologic and pain data collected from patients with SCD, we identified patterns that are related to hospital readmission. For readmitted patients, vital signs including respiratory rate and blood pressure were the most influential variables, suggesting that changes in vital signs may be a quantifiable risk for re-admission. In those who were not readmitted, the strongest correlation was between VOC and step count, suggesting the potential importance of movement in reducing readmissions. Overall, network analyses offer a unique way to help understand patterns that may result in readmission for patients with SCD; further research is needed to validate these findings in a larger cohort and explore the utility of this technique in supporting early diagnosis and intervention.

Disclosures: Ford: Nanbar Health: Current equity holder in private company. Parikh: Nanbar Health: Current Employment. Gundala: Nanbar Health: Current Employment. Subramaniam: Nanbar: Current Employment. Hensley: Nanbar Health: Current Employment. Shah: Global Blood Therapeutics/ Pfizer: Consultancy, Research Funding, Speakers Bureau; Alexion Pharmaceuticals: Speakers Bureau; Agios Pharmaceuticals: Consultancy; Akirabio: Consultancy, Research Funding; Bluebird bio: Consultancy; Vertex: Consultancy; Forma: Consultancy.

*signifies non-member of ASH