Although mortality rates due to critical illness have decreased during the past several decades as a result of improvements in clinical care in the ICU, as many as two thirds of survivors experience one or more persistent impairments as part of a post–intensive care syndrome (PICS), which manifests as physical, cognitive, and mental health impairments for years after hospitalization (1, 2). Prolonged immobility in the ICU is a key independent risk factor for the acquisition of muscle and nerve disease during critical illness, referred to as ICU-acquired weakness, which leads to the long-term physical impairments seen in PICS and ultimately disability (3). To reduce the profound insults of critical illness on skeletal muscle and physical function, early mobilization and rehabilitation in the ICU has been studied as a PICS prevention strategy (4–7). Notably, early mobility has been shown to yield a 19% absolute risk reduction in cognitive impairment at 1 year in ICU survivors after mechanical ventilation in a recent single-center trial, in addition to reducing ICU-acquired weakness at 1 year (8). Although studies of early mobility performed ⩾72 hours after initiation of mechanical ventilation have not shown benefit (6, 7), trials initiating early mobility within 72 hours have demonstrated improved outcomes, including better functional status before hospital discharge and reductions in ICU-acquired weakness in survivors (4, 5, 8). However, these trials left several unanswered questions about the optimal timing and “dose” of early mobility.

The Treatment of Mechanically Ventilated Adults with Early Activity and Mobilization (TEAM) randomized trial (9), published in 2022, was an international multicenter randomized controlled trial that randomized 750 mechanically ventilated adults to receive increased early mobilization versus usual care to investigate the optimal dose of early mobility in the ICU. To put this trial into context with earlier trials, patients in both arms were enrolled an average of approximately 80 hours after admission to the hospital, and the intervention group received approximately 21 minutes of early mobility compared to approximately 9 minutes in the control group. In comparison with usual care, the trial found no improvement from the intervention on the primary outcome, which was the number of days alive and out of the hospital at Day 180. Similarly, no benefit to early mobility was seen for secondary outcomes, including health-related quality of life and disability at 180 days. In contrast, there was a signal of an increased number of adverse events in the early mobility treatment group. The TEAM trial was a landmark study in the realm of early mobility trials and generated important additional questions about identifying responders to ICU rehabilitation.

In this issue of the Journal, Serpa Neto and colleagues (pp. 779–787) conducted an analysis of the TEAM trial evaluating whether a baseline diagnosis of diabetes mellitus modified the participants’ response to early mobilization versus usual care (10). Importantly, this is a post hoc analysis by the investigators. The methodology included exploratory logistic, linear, and median regression models to evaluate the associations between diabetes and clinical outcomes. In addition, no corrections were made for multiple comparisons. These data, as indicated by the authors, should be considered hypothesis-generating and in need of more rigorous study to discern cause and effect.

In this analysis, all participants included in the original TEAM trial were subsequently stratified based on the presence or absence of a diagnosis of diabetes, and outcomes were compared between groups. The authors found that patients with diabetes had fewer days alive and out of the hospital by Day 180 than those without diabetes, in addition to a substantially higher 180-day mortality rate (30% vs. 18%; P = 0.044). In Tables 3 and E7 of their study, the authors present 11 outcome variables to assess for an interaction between the effect of early mobilization stratified by the presence of diabetes with and without adjustment for confounders. Of the 11 outcomes presented, one yielded a significant P value for an interaction, which was 180-day mortality (P = 0.001). Importantly, there was not a significant interaction between early mobility and a diagnosis of diabetes mellitus on the primary outcome of days alive and out of the hospital at Day 180 (P = 0.108). There was also no difference between groups in functional outcomes at 180 days. There was no overall difference in adverse events between patients with and without diabetes. Among patients with diabetes, there were increased probabilities of severe hyperglycemia and hypoglycemia, tachy- and bradycardias, and hypotension in the first 5 days after randomization among those assigned early mobility in an exploratory Bayesian logistic model.

The results of this study are thought-provoking, yet they should be interpreted with several caveats in mind. Such findings may very well be explained by the inherent limitations of post hoc subgroup analyses of clinical trials that are not prespecified, namely the diminished benefit of randomization by selecting a smaller sample from the larger cohort, limited power due to smaller sample sizes, and the problem of multiple comparisons (11). Although randomization is the ideal approach to balancing characteristics between entire study groups, chance imbalances are more likely to occur in subgroups that are not prespecified or stratified for at randomization. Such imbalances may confound the associations between the treatment assignment and the outcome in the subgroup of interest, particularly when the power to detect such associations is also reduced by smaller treatment groups. In this study, diabetes was not a stratification variable at randomization, and patients with diabetes were older, had a greater severity of illness as defined by Acute Physiology and Chronic Health Evaluation II score, and higher baseline frailty and disability scores. Additionally, with multiple comparisons that occur in subgroup analyses, the risk of a false-positive finding caused by type I error increases. To properly confirm such findings, as the authors note, adequately powered clinical trials enrolling the subgroup of interest are required to confirm the hypothesis that there is a differential treatment effect with regard to early mobility in the ICU in patients with diabetes.

Is there biological plausibility to the findings of increased mortality with early mobility in diabetic patients? A recent pooled analysis of four earlier ICU rehabilitation trials found that patients with two or more comorbidities who underwent physical rehabilitation interventions achieved significantly higher quality of life scores at 6 months compared with the post-ICU population without multiple comorbidities (12). Why, then, would critically ill patients with diabetes who received increased early mobility experience greater mortality compared with nondiabetic patients? In many patients with critical illness (not just those with diabetes), there is impaired mitochondrial biogenesis, deranged lipid oxidation, and intramuscular inflammation, all of which can push the neuromuscular system toward catabolism (13, 14). One might think that this is precisely the time for more mobility to engage the diseased muscle beds to prevent atrophy and catabolism. However, diabetic individuals have metabolic inflexibility and impaired energy metabolism that may lead to worsening muscle injury and glucose handling when greater physical exertion and increased energy demands are levied in the midst of critical illness. This phenomenon may be reflected in the higher probability of glycemic and physiologic derangements seen in the diabetic patients in the early mobility group. This idea remains untested, but these data from the TEAM trial’s post hoc analysis help build a hypothesis worthy of future testing.

What about the brain? Although there was not a statistically significant difference in delirium and coma, there was a numerically greater incidence of delirium on multiple study days among diabetic patients in the early mobility group. Given that the brain consumes a significant proportion of energy relative to its overall mass, it is particularly susceptible to alterations in energy metabolism. Do diabetic patients who receive early mobility during critical illness experience bioenergetic failure with impaired glucose metabolism at the neuronal level? Might this manifest as more delirium, or is overuse of sedation a primary driver? Again, these hypotheses are likely best tested via a prospective, randomized investigation.

This secondary analysis by Serpa Neto and colleagues advances the field of early mobility in the ICU by asking a vital question about ICU rehabilitation trials: which patients are most likely to benefit or be harmed by increased early mobilization? Future trials should consider the possibility of heterogeneity of treatment effect a priori by prespecifying the subgroups that will be evaluated and hypothesize a directionality of the expected effect modification with an appropriate biologic rationale and review of prior data (15). As a scientific community, we also desperately need greater understanding of biological mechanisms that drive persistent functional impairments such as impaired cellular metabolism and mitochondrial dysfunction. Finally, it must be remembered that early mobility in the ICU cannot improve outcomes as a standalone intervention. For example, the most obvious limitation to early mobility in an ICU patient undergoing ventilation is oversedation. Among 23,000 patients included in three large cohorts studying the ABCDEF bundle (Assess, prevent, and manage pain; Both spontaneous awakening trials and spontaneous breathing trials; Choice of analgesia and sedation; Delirium: assess, prevent, and manage; Early mobility and Exercise; and Family engagement and empowerment)—a six-step, evidence-based bundle of care that synergistically pairs decreasing sedation with nonpharmacological delirium management and family involvement—higher proportional performance of all components of the bundle was an independent predictor of improved outcomes in multiple domains, including survival rate, length of stay, and cost of care (16–18). Such an approach marries the rigor of the scientific method with humanism to create an “awake and walking ICU” by which to improve patient-centered outcomes. Understanding which patient populations are most likely to benefit from or be harmed by such interventions and bundles of care will help clinicians and scientists develop and tailor treatments to reduce the devastating public health problem of PICS and maximize recovery.

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES