C55-36 Seeing Post-hospital Morbidity In Acute Respiratory Distress Syndrome Using Real-world Data

Rationale

Survivors of critical illness can suffer significant morbidity, characterized as Post-ICU Syndrome (PICS). Diagnosis involves a battery of time-and cost-intensive in-person testing. We evaluated the potential to identify morbidity in patients with acute respiratory distress syndrome (ARDS) using real-world data (RWD), and whether this could be mapped to PICS domains. This would enable many more patients to be ‘seen’ and inform the development of targeted therapies.

Methods

We conducted a retrospective cohort study using Optum Claims data. We identified adults with a first ICU admission for ARDS from Jan 2016-October 2022 and alive ≥30 days after hospital discharge. Incident morbidity was identified using comorbidities (ICD-10 codes), procedures (Current Procedural Terminology codes), and medications (prescription codes) within 30-365 days post discharge, which were not seen in the year before ICU admission. Codes were collapsed into validated categories and mapped to possible PICS domains. One-year cumulative incidences were calculated accounting for death and loss to follow-up. In an exploratory analysis, incident morbidity was compared to that seen among survivors in a propensity-score-matched cohort of non-ICU pneumonia admissions, hypothesizing that PICS-related morbidity would be more common in post-ICU ARDS survivors.

Results

28,090 ARDS patients were included. The most common new comorbidities were respiratory problems (>50%), metabolic disorders such as hypertension (>40%) and lipid metabolism disorders (>35%), as well as malaise and fatigue disorders, aplastic anemia, and heart failure (>30% in each). Common incident procedures included those related to follow-up care and radiographic imaging of the chest and heart. Common medications included antibacterials, antihypertensives and other cardiac drugs. PICS domain-mapped codes were seen in many survivors, ranging from >15% experiencing cognition disorders to > 50% with pain-related disorders (Figure). Overall incident morbidity and PICS-related morbidity (Figure) were also both high among matched non-ICU pneumonia survivors.

Conclusion

High post-hospital incident morbidity can be seen using RWD in both ARDS ICU survivors and non-ICU pneumonia survivors. Much of this morbidity is related to PICS domains, and metabolic syndrome disorders, which were more commonly seen in the ARDS survivors suggesting an increased risk after critical illness and the ICU experience. Such RWD analyses could facilitate the identification of post-hospital burden and enable predictive modeling to understand pre- and in-hospital risk factors. Collaborations with patients and clinicians are needed to validate our assumptions related to PICS domain mapping, tailor SeeMe to stakeholder priorities, and validate claims-based RWD inferences with measured incident morbidity.

This abstract is funded by: Bayer AG