Highlights
• A 20-year retrospective analysis of 205,393 mechanically ventilated ARDS admissions reveals mortality declined during the ICD-9 era (OR 0.96/year) but paradoxically increased during the ICD-10 period (OR 1.05/year).
• The ICD-10 cohort demonstrated greater comorbidity burden and higher prevalence of pulmonary ARDS etiology, potentially explaining the mortality reversal.
• Hospitalization costs and length of stay improved during ICD-9 but plateaued in the ICD-10 era, reflecting increasing complexity of surviving patients.
• These findings underscore the critical importance of accounting for coding transitions and patient case-mix when interpreting longitudinal ARDS outcomes.
Background: The Enduring Challenge of ARDS
Acute respiratory distress syndrome remains one of the most formidable challenges in critical care medicine. Characterized by severe hypoxemia, bilateral pulmonary infiltrates, and non-cardiogenic pulmonary edema, ARDS affects approximately 200,000 to 300,000 patients annually in the United States alone. Despite decades of research, mortality rates have historically hovered between 30% and 50%, with survivors often experiencing prolonged Intensive Care Unit (ICU) stays and significant long-term functional impairment.
The advent of lung-protective ventilation strategies, prone positioning, and conservative fluid management has transformed ARDS care over the past two decades. However, large-scale longitudinal analyses capturing these practice changes have been scarce, particularly studies examining outcomes spanning pre-COVID eras with sufficient statistical power to detect secular trends.
The study by Padappayil and colleagues addresses this critical gap by leveraging the National Inpatient Sample—the largest publicly available all-payer inpatient healthcare database in the United States—to examine how in-hospital mortality among mechanically ventilated ARDS (MV-ARDS) patients evolved between 2000 and 2019.
Study Design and Methods
The investigators conducted a retrospective cohort study utilizing the National Inpatient Sample from January 2000 through December 2019. The study population comprised adult patients (aged 18 years or older) admitted non-electively with a diagnosis of ARDS who concurrently received invasive mechanical ventilation.
Notably, the researchers stratified the cohort by ICD coding era to account for the fundamental change in diagnostic coding that occurred in October 2015. The ICD-9 era encompassed admissions from 2000 through the third quarter of 2015 (n = 146,888), while the ICD-10 era included admissions from the fourth quarter of 2015 through 2019 (n = 58,505). This stratification acknowledges that the transition from ICD-9 to ICD-10 coding systems introduced substantial changes in how ARDS diagnoses are captured and classified.
The primary outcome was in-hospital mortality. Secondary outcomes included length of stay (LOS) and total hospitalization charges, with charges adjusted to 2019 dollars using the Consumer Price Index for hospital services to ensure temporal comparability. Adjusted trend analyses were performed separately for each coding era using logistic regression (for mortality) and linear regression (for LOS and charges).
The study accounted for potential confounders including patient demographics, comorbidity burden, and hospital characteristics, though the specific adjustment variables were not detailed in the available abstract.
Key Findings
Mortality Trends: A Tale of Two Eras
The most striking finding was the divergent mortality trajectory between coding eras. During the ICD-9 period (2000–Q3 2015), in-hospital mortality demonstrated a statistically significant annual decline, with an odds ratio of 0.96 per year (95% CI 0.95–0.97, p<0.001). This translates to approximately a 4% reduction in the odds of death per year, representing meaningful progress in ARDS survival over 15 years.
However, this encouraging trend reversed dramatically after ICD-10 implementation. During the ICD-10 era (Q4 2015–2019), mortality increased significantly at a rate of 1.05 per year (95% CI 1.01–1.08, p=0.004). This 5% annual increase in the odds of in-hospital death marks a sobering counterpoint to earlier gains.
Length of Stay and Hospitalization Costs
Paralleling the mortality trends, resource utilization patterns also shifted between eras. During the ICD-9 period, both length of stay and total hospitalization charges demonstrated decreases, reflecting improvements in care efficiency and possibly the declining severity of the patient population surviving acute illness. However, during the ICD-10 period, these metrics showed no significant improvement, and the abstract suggests this may reflect the rising costs associated with managing increasingly complex ARDS patients.
Patient Characteristics and Case-Mix
The investigators identified important differences between the two coding era cohorts that may explain the mortality reversal. The ICD-10 cohort exhibited a greater comorbidity burden compared to the ICD-9 group, suggesting that advances in medical care allowed sicker patients with more underlying conditions to survive acute events and subsequently develop ARDS. Additionally, patients in the ICD-10 era were more likely to have a pulmonary etiology for their ARDS, which may carry different prognostic implications than extrapulmonary causes.
Expert Commentary: Interpreting the Paradox
These findings present a compelling case for the complexity inherent in interpreting longitudinal outcomes data. The apparent mortality reversal coinciding with ICD-10 adoption raises several important considerations.
First, the ICD-10 coding system introduced substantially greater granularity in diagnostic documentation, with thousands of new codes compared to the relatively limited ICD-9 vocabulary. This enhanced specificity may have led to more accurate capture of ARDS diagnoses, potentially including milder cases that would have been missed or coded differently under ICD-9. Alternatively, the new coding framework may have improved detection of more severe presentations that were previously undercoded.
Second, the temporal coincidence with broader changes in critical care practice warrants consideration. The study period includes the dissemination of evidence-based ARDS interventions, including low tidal volume ventilation, prone positioning for severe hypoxemia, and neuromuscular blockade in early severe ARDS. The mortality decline during ICD-9 may partially reflect adoption of these proven strategies, while the subsequent increase may indicate either their limitations in the face of increasingly complex patients or the emergence of new challenges in patient management.
Third, the greater comorbidity burden in the ICD-10 cohort suggests that the population developing ARDS has evolved substantially over two decades. Patients who would have succumbed to their underlying conditions in earlier years may now survive long enough to experience respiratory failure, bringing greater baseline frailty and complexity to the ARDS population.
The study’s strengths include its large sample size, spanning two decades of clinical practice, and its innovative approach to addressing coding system transitions. Limitations include the inherent constraints of administrative data, which may not capture important clinical nuances such as ARDS severity (as defined by Berlin criteria), ventilator settings, or ICU-specific processes of care. Additionally, the National Inpatient Sample captures only in-hospital outcomes, leaving long-term mortality and functional outcomes unaddressed.
Conclusion and Clinical Implications
This 20-year analysis of mechanically ventilated ARDS patients reveals that progress in outcomes is neither linear nor irreversible. The initial mortality decline during the ICD-9 era likely reflects genuine improvements in ARDS care, including evidence-based ventilation strategies and enhanced supportive care. The subsequent mortality increase during the ICD-10 period, however, challenges simplistic interpretations and demands careful consideration of multiple factors.
Healthcare providers and researchers must recognize that coding transitions introduce artifacts that can profoundly influence apparent outcomes. Future analyses of ARDS trends must incorporate strategies to adjust for coding system changes, or acknowledge these limitations when interpreting findings. The greater comorbidity burden and shifting etiology patterns in contemporary ARDS patients suggest that clinicians are increasingly managing a more complex patient population with unique challenges.
From a research perspective, these findings underscore the need for granular clinical registries that capture disease severity, treatment details, and long-term outcomes—data elements that administrative databases cannot reliably provide. From a clinical standpoint, the evolving case-mix of ARDS patients has implications for prognostic counseling, resource allocation, and the design of future interventional trials.
The paradox of declining then increasing mortality over two decades ultimately reflects the dynamic nature of critical care medicine—a field continuously shaped by advances in technology, changes in patient populations, and the complex interplay between what we measure and what matters to patients.
Reference
Padappayil P, Shah D, Jackson T, Helander ME, Kaul V, Ghosh AJ. Secular Trends in Acute Respiratory Distress Syndrome: A 20-Year Analysis of the National Inpatient Sample. Chest. 2026-03-26. PMID: 41903837.