Reducing the Effect of Critical Illness by Continuing to Think beyond the Intensive Care Unit

For many patients, critical illness begins before the intensive care unit (ICU) (1, 2). Recognizing that prompt identification and treatment might avert patient deaths, clinicians, scientists, and policymakers have directed considerable resources toward the development and implementation of treatment bundles, early warning systems, quality measures, and other interventions targeting critical illness syndromes in their golden hours (3–7). Although these advances have contributed to improved patient outcomes, both short-term mortality and long-term morbidity remain high.

In this issue of AnnalsATS, Weissman and colleagues (pp. 81–88) ask whether some episodes of critical illness might be stemmed even earlier, preventing ICU admission or hospitalization altogether (8). To answer this question, the authors evaluate 10 years’ worth of inpatient claims drawn from patients with fee-for-service Medicare or a large private payer administering both Medicare Advantage and private insurance plans. In a laudable effort, the cohort included a majority of all hospitalizations among adults over the age of 65 years and a sizable proportion of those among younger adults as well. Among members of this cohort admitted to the ICU, the authors categorized admissions as potentially preventable if their primary reason for admission was an ambulatory care sensitive condition (ACSC) or a life-limiting malignancy (LLM).

ACSCs are defined by the Agency for Healthcare Research and Quality as selected inpatient diagnoses “for which good outpatient care can potentially prevent” hospitalization and limit more serious disease (9). These include acute exacerbations and other complications of chronic diseases (including asthma, chronic obstructive pulmonary disease, diabetes mellitus, and congestive heart failure) and infections that may be amenable to early outpatient antibiotics or vaccination (including bacterial pneumonia and urinary tract infection). LLMs, meanwhile, include malignancies associated with high 1-year mortality and few options for curative treatment (10). In prior work, both ACSCs and LLMs have been used as measures of ambulatory care quality and as benchmarks for evaluating both practice- and system-level interventions.

In total, the authors identified nearly 100 million hospitalizations spanning 10 years, including 16 million ICU admissions. Among these, nearly one in six met ACSC or LLM criteria for being potentially preventable. Notably, these were different from discretionary admissions; the extent to which ICU admissions were driven by bed supply did not differ between these potentially preventable and other ICU admissions, suggesting that they were no more likely to be discretionary than other causes of ICU admission.

A few limitations are worth noting. For one, the estimates of potentially preventable ICU admissions contained in this paper are likely conservative. Definitions of both ACSCs and LLMs are reliant on primary diagnosis codes and were originally designed to categorize hospitalizations more broadly. ICU patients with primary diagnoses of sepsis, respiratory failure, or shock, even as a consequence of an ACSC or LLM, would not qualify. Further, the lists of conditions that qualify leave out many hospitalizations (e.g., those related to substance use disorder) that might be considered ambulatory care sensitive. Finally, the degree to which admissions for ACSCs or LLMs are completely preventable is uncertain: studies evaluating the effects of population-level interventions aimed at improving ambulatory care access and quality have yielded inconsistent results, with even highly successful interventions yielding small absolute decrements in the rates of preventable hospitalizations (11).

Still, these findings are notable for several reasons. First, they suggest that many episodes of critical illness might be averted through improvements to ambulatory care coverage, access, and quality. An implication of this is that the up-front costs associated with these improvements might be at least partially offset by reductions in these high-severity hospitalizations. Future work should focus on identifying who is at highest risk for preventable critical illness and how best to deliver clinical resources to these high-risk patients. At a population level, the state-level variation identified by the authors suggests an opportunity to study how policies and other factors may be effective in reducing rates of preventable critical illness. Such approaches are supported by prior work demonstrating that the Affordable Care Act’s Medicaid expansion, which was variably adopted across states, may have been effective in reducing respiratory failure (12). There is likely more to be learned mechanistically by examining such regional and state-level variation.

Second, such preventable ICU admissions might be useful for benchmarking health care systems or evaluating health policy where other measures of utilization have fallen short. Because of their high acuity, ICU admissions may be less susceptible to nonclinical factors (e.g., patient and clinician discretion, financial pressures, etc.) than other types of utilization, such as ED visits or hospitalizations (13). This may make them more reliable population-level measures of health, particularly when evaluating complex policy interventions (14).

Beyond pre-ICU care, this study should also lead us to consider that the same coverage, access, or quality constraints likely associated with preventable ICU admissions may affect patients after ICU discharge (15). Many rehospitalizations after conditions such as sepsis and respiratory failure are also caused by potentially preventable diagnoses (16, 17). In addition to preventing many ICU admissions, ambulatory care has great potential to improve clinical trajectories among those surviving critical illness.

Finally, this work suggests we continue to expand traditional temporal and geographic boundaries when considering efforts to combat critical illness (18). The emergency department and hospital ward have been important loci for early interventions aimed at many critical illness syndromes. Considering that the seeds of many such illnesses begin well before a hospitalization grants additional opportunities to meaningfully reduce the effect of critical illness.