The Combined Effects of Frailty and Cognitive Impairment on Post-ICU Disability among Older ICU Survivors

To the Editor:

Frailty and cognitive impairment represent distinct vulnerability factors among older adults. With advancing age, the likelihood that these vulnerability factors will co-occur increases. Our objective was to evaluate the combined effects of preexisting frailty and cognitive impairment on the course of disability after an ICU admission among older adults. Some of these results have been previously reported in the form of an abstract (1).

Data were obtained from participants in the Precipitating Events Project, which included 754 adults aged 70+ years from the greater New Haven, Connecticut, region (2). Participants were evaluated monthly for disability in 13 functional activities from 1998 to 2014, with data available for 99.2% of 81,188 interviews. Frailty was assessed every 18 months, using the Fried index (range, 0–5, with higher scores representing greater frailty), based on the presence of weight loss, exhaustion, muscle weakness, slow gait speed, and low physical activity (3, 4). Cognitive status was assessed every 18 months, using the Folstein and colleagues Mini-Mental State Examination (MMSE; range, 0–30) (5). Based on prior results (6), showing that minimal (MMSE; range, 24–27) and moderate (MMSE < 24) cognitive impairment exerted comparable adverse effects on post-ICU disability, relative to intact cognition (MMSE 28–30), we classified participants as cognitively intact (MMSE ≥ 28) or not (MMSE ≤ 27). As previously described (4, 6), the analytic sample included 266 ICU admissions from participants who survived to the first post-ICU monthly interview; participants with a pre-ICU disability count of 13/13 (maximum) were excluded, as they were not eligible for worsening disability. Using the last frailty assessment and the last cognitive assessment before ICU admission, we evaluated the interaction of frailty and cognitive impairment on the disability count over the course of 6 months after an ICU admission (the primary outcome), using a Bayesian negative binomial model. Statistical significance for the corresponding rate ratio was defined as a credible interval exclusive of the value 1. We subsequently evaluated the association of frailty count with the primary outcome in strata defined by cognitive status (7). Covariates included age, sex, race, education, chronic conditions, pre-ICU disability count, ICU length of stay, mechanical ventilation, and shock, as previously described (4). A series of sensitivity analyses were performed to assess for potential bias because of the competing risk for death during follow-up (8). Finally, we performed a supplementary analysis to illustrate the absolute mean counts of disability for 3-level Fried frailty (frailty = 3–5 criteria, prefrailty = 1–2 criteria, and not frail = 0 criteria) within strata defined by cognitive status. All analyses were performed with SAS version 9.4 (9) and its GENMOD procedure, which allows a choice of distributions, including the negative binomial, and a Bayes option, which automatically generates posterior output data and an array of graphical checks.

As shown in Table 1, the mean age was 83.5 years, and the mean disability count in the month before ICU admission was 4.6 (range, 0–13). Nearly two-thirds of the sample was not cognitively intact (n = 169, 63.5%) in the assessment preceding ICU admission. The Fried frailty counts were evenly distributed, except for the maximum score of 5/5, which comprised only 4.1% of this sample of ICU survivors. Over the course of follow-up, there were 32 (18.9%) deaths in the cognitively impaired group and 11 (11.3%) deaths in the cognitively intact group (P = 0.07, Fisher’s exact chi-square).

Table 1.

Descriptive Characteristics of the 266 ICU Admissions Contributed by Older ICU Survivors

Characteristic	Mean ± SD or n (%)
Age, yr	83.5 ± 5.4
Female sex	156 (58.7)
Non-Hispanic white	235 (88.4)
Education	12.0 ± 2.9
Frailty count*† (0–5)	2.3 ± 1.4
0 (not frail)	32 (12.0)
1 (prefrail)	52 (19.6)
2 (prefrail)	62 (23.3)
3 (frail)	68 (25.6)
4 (frail)	41 (15.4)
5 (frail)	11 (4.1)
Cognitive function†‡	26.0 ± 3.3
Cognitively intact (n = 97)	28.9 ± 0.8
Not cognitively intact (n = 169)	24.4 ± 3.0
Number of chronic conditions§ (0–9)	2.5 ± 1.3
Pre-ICU disability count|| (0–13)	4.6 ± 3.5
ICU length of stay	2.9 ± 4.1
Mechanical ventilation	31 (11.7)
Shock	12 (4.5)

^*The five Fried frailty criteria are weight loss, exhaustion, muscle weakness, slow gait speed, and low physical activity. Scores reflect the frailty assessment before ICU admission.

^†The median time between the assessment and the ICU admission was 9.5 months (interquartile range, 5–14 months).

^‡Folstein Mini-Mental State Exam: Scores range from 0 to 30, with higher scores representing better cognitive function. Cognitively intact is defined as a score of 28–30, whereas not cognitively intact is defined as score of ≤27. Scores reflect the cognitive assessment prior to ICU admission.

^§Hypertension, myocardial infarction, heart failure, stroke, diabetes, arthritis, hip fracture, chronic lung disease, and cancer.

^||Four activities of daily living (bathing, dressing, walking, transferring), five instrumental activities of daily living (shopping, housework, meal preparation, taking medications, managing finances), and four mobility activities (walk a quarter mile, climb a flight of stairs, lift or carry 10 pounds, and driving in the past month).

The adjusted mean number of disabilities according to frailty count and cognitive status are presented in Figure 1, with the pre-ICU disability count included at month zero as a reference point. For four of the five frailty groups, participants with and without cognitive impairment experienced an increase in disability immediately after an ICU admission, but participants with cognitive impairment suffered greater post-ICU disability in the subsequent months (relative to cognitively intact participants), and did not recover to their pre-ICU disability levels. This pattern was not observed for the maximally frail group (Fried scores of 4 or 5, combined because of the small group sizes). In the multivariable analysis, the association between frailty and post-ICU disability count differed between participants who were cognitively intact and those who were cognitively impaired (i.e., the 95% credible interval [CI] for the interaction rate ratio [RR] excluded 1). Among participants with cognitive impairment, each additional frailty criterion (from 0 to 5) was associated with a 54% higher average disability count over the course of the 6 months after a critical illness (RR, 1.54; 95% CI, 1.37, 1.74). Among those who were cognitively intact, the corresponding increase was only 18% (RR, 1.18; 95% CI, 1.13, 1.23). In a series of sensitivity analyses (8), these results were robust to the competing risk of death during follow-up. In the supplementary analysis, the mean absolute disability counts increased as frailty increased and across strata of cognitive status. The author has made available a related supplementary figure. This figure may be viewed outside of ATS journal content on the author’s institutional website: https://medicine.yale.edu/intmed/pulmonary/about/lauren_ferrante-2.profile. Interestingly, the mean absolute disability count of frail participants in the cognitively intact group approximated that of the mean disability count of nonfrail participants in the cognitively impaired group.

Figure 1.

Adjusted least-squares mean number of disabilities (0–13) according to Fried frailty count (0–5) and cognitive status during the 6 months of follow-up after an ICU stay. The pre-ICU disability count from the month before ICU admission is included at month 0 as a reference point. Sequentially, the panels represent increasing Fried frailty counts, where higher scores indicate greater frailty. Within each panel, participants are stratified by cognitive status, with the n (%) of observations from the first month after discharge noted in the figure key. Fried frailty counts 4 and 5 were combined because of the small numbers in each group.

Our findings demonstrate, for the first time, that preexisting frailty and cognitive impairment interact to adversely affect the course of post-ICU disability in a cohort of older ICU survivors. We found that the relative association of each increase in frailty count (from 0 to 5) with post-ICU disability was about three times larger among persons with any cognitive impairment relative to those who are cognitively intact.

It is widely known that the prevalence of frailty and cognitive impairment each increase substantially with age (3, 10). Hence, with the aging of the population, a greater proportion of older adults will present to the ICU with varying degrees of frailty and cognitive impairment. Our study suggests that these patients are at increased risk for adverse post-ICU functional outcomes, and that this risk increases with severity of frailty. The mechanisms underlying this increased risk are currently unclear. Potential mediators of our findings should be explored, such as the illness/injury leading to ICU admission (e.g., sepsis) or complications of the ICU stay (e.g., immobility or delirium). In addition, future studies of ICU and postacute rehabilitation interventions should explore whether interventions are equally effective across these strata of vulnerability. The finding that disability did not worsen in the subgroup of participants who were maximally frail and cognitively impaired is likely attributable, at least in part, to a ceiling effect (i.e., in and immediately after the ICU), as demonstrated in our prior work (4, 6).

The strengths of our study include prospective assessments of pre-ICU frailty and cognition and the availability of high-quality, longitudinal data on disability with minimal loss to follow-up. Because frailty and cognitive impairment were assessed every 18 months, it is possible that these factors could have changed between their assessment and ICU admission, although neither would likely have improved. In addition, although we did adjust for mechanical ventilation, shock, and ICU length of stay, severity-of-illness scores were not available. Finally, our participants were drawn from one geographic region, so the results may not be generalizable to other regions.

In summary, prehospitalization cognitive impairment and prehospitalization frailty interact to amplify posthospitalization disability among older ICU survivors. Additional research is warranted to confirm these findings and to elucidate the underlying mechanisms.