Associations Among Skeletal Muscle Health, Disability, and Self-Reported Physical Function in Survivors of Critical Illness

Abstract

BACKGROUND:

Factors associated with disability after critical illness are incompletely understood. Lower pre-existing muscle mass and muscle density on CT imaging are associated with greater mortality resulting from critical illness. Their relationship with disability and physical function in survivors of critical illness is unclear.

RESEARCH QUESTION:

We sought to understand the relationship between muscle mass and muscle density before critical illness and disability and self-reported physical function after surviving a critical illness.

STUDY DESIGN AND METHODS:

We conducted a nested cohort study of 125 survivors of critical illness with abdominal imaging between 6 months before and 4 days after ICU admission. We measured skeletal muscle cross-sectional area at the L3 vertebra, indexed by height, to obtain the skeletal muscle mass index and measured skeletal muscle density by calculating the mean Hounsfield units of the muscles. We recorded discharge location and, at 3 and 12 months after hospital discharge, assessed for disability with basic activities of daily living, instrumental activities of daily living, and self-reported physical function. We used multivariable regression to assess the relationship between baseline skeletal muscle mass index or density and outcomes.

RESULTS:

We found no association between skeletal muscle mass index and discharge to a facility or disability. However, lower skeletal muscle density was associated with greater odds of discharge to a facility, but not with disability at either time point.

INTERPRETATION:

A substantial percentage of participants (40%) demonstrated low muscle mass on abdominal imaging before seeking treatment for a critical illness. We did not find muscle mass or density to be associated with long-term disability or physical function after critical illness, although lower density was associated with increased risk of discharge to a facility. Further work is needed to understand the relationship between muscle health and physical recovery after critical illness.

TRIAL REGISTRY:

ClinicalTrials.gov; No.: NCT00392795; URL: www.clinicaltrials.gov

More than one-half of patients who survive a critical illness will be faced with new impairments in their physical function, new disabilities, or both.^1–5 Identifying patients at highest risk of poor physical recovery and new disability remains challenging because factors associated with new disability are not well characterized. Key contributing factors are hypothesized to include a combination of baseline (ie, before critical illness) health and function and events of the critical illness.^6–9

In community-dwelling adults, a decline in muscle health resulting from loss of muscle mass and muscle strength is associated with disability and impaired physical function.^10–13 However, routine assessment of muscle health is rare, and few patients with critical illness know the baseline status of their muscle health.^13,14 Further, standard functional assessments of muscle health (eg, grip strength) at admission to the ICU cannot be conducted or are confounded by critical illness, delirium, and sedation. Muscle mass and density (an indicator of muscle quality) can be measured with CT imaging and used as a surrogate assessment of muscle health.^13,15,16 Prior studies have shown that poorer muscle health at ICU admission is associated with worse short-term outcomes, including increased ICU length of stay and higher mortality.^17–20 The relationship between preexisting muscle health and long-term disability and physical function in survivors of critical illness remains unclear.

To address these gaps in knowledge, we sought to understand the association of muscle health before critical illness, measured by muscle mass and muscle density on CT imaging, with long-term disability and physical function impairments developing in survivors of critical illness. We hypothesized that poorer muscle health was associated with greater disability and worse physical function.

Study Design and Methods

We tested these hypotheses in a secondary analysis of a nested cohort of ICU survivors from the Bringing to Light the Risk Factors and Incidence of Neuropsychological Dysfunction in ICU Survivors (BRAIN-ICU) prospective cohort study. This study was conducted in accordance with the ethical standards of the Vanderbilt Human Research Protection Program and was approved by the Vanderbilt University Health Science Committee Institutional Review Board 2 (Identifier: 060593) on July 27, 2006. The hypotheses and analyses presented herein have not been published previously.

Setting and Participants

The parent study protocol was published previously²¹ and may be found in e-Appendix 1. In brief, BRAIN-ICU enrolled participants who were 18 years of age or older and were treated for respiratory failure or shock in medical and surgical ICUs. Those with organ dysfunction for > 72 hours, recent ICU exposure, severe cognitive impairment, inability to communicate in English, or a history of substance misuse, homelessness, or residence > 200 miles from Nashville, Tennessee, were excluded. Participants or their proxies provided consent. We conducted a secondary analysis of a nested cohort of survivors of critical illness. We identified participants who had survived to the 3-month follow-up and were enrolled from Vanderbilt University Medical Center. We then conducted a chart review to identify the 3-month survivors who underwent an abdominal CT imaging for clinical indications during the 6 months before ICU admission and up to 4 days after ICU admission.

Measuring Muscle Mass and Density

We selected abdominal CT scans performed within the study window for analysis based on previous work establishing a correlation between muscle at the L3 vertebral slice of abdominal CT scans and total body muscle mass.^22,23 Trained research personnel from the Vanderbilt Diet, Body Composition, and Human Metabolism Core (Core) selected axial images obtained from the midbody of the L3 vertebra, converted them to Digital Imaging and Communications in Medicine format, and analyzed them using an automated version of Slice-O-Matic version 4.3 software (TomoVision).^24,25

Using established Hounsfield unit (HU) attenuation thresholds, the software quantified the cross-sectional area for skeletal muscle, subcutaneous adipose tissue, and visceral adipose tissue. Skeletal muscle was identified at −29 to +150 HU,²² subcutaneous adipose tissue was identified at −190 to −30 HU,²² and visceral adipose tissue was identified at −150 to −50 HU.²⁶ The automated tissue depots were edited manually by Core research personnel who were masked to study aims and outcomes.

To measure the association between muscle mass and outcomes, we used the skeletal muscle mass index (SMI) as the main exposure.²⁷ We calculated the SMI by summing the cross-sectional area of the psoas, quadratus lumborum, erector spinae, lateral oblique, internal and external oblique, and rectus abdominus (in square centimeters) and dividing by height (in square meters). Lower SMI values indicate lower muscle mass.

We considered low muscle mass to be present if the SMI value was more than 2 SDs less than the sex-adjusted norm for a healthy adult (male, < 55 cm²/m²; female, < 39 cm²/m²).²⁷ We used previously defined cutpoints of BMI to define underweight, normal, overweight, and obese body habitus. We considered participants to be underweight if the BMI was < 18.5 kg/m², normal if the BMI was ≥ 18.5 and < 25kg/m², overweight if the BMI was ≥ 25 and < 30 kg/m², and obese if the BMI was ≥ 30 kg/m². For descriptive purposes, we categorized participants according to the SMI and BMI.

Intramuscular lipid infiltration compromises muscle quality and function.^16,28 Greater lipid accumulation reduces the density of muscle on CT imaging and is quantified by measuring the attenuation of the muscle tissue in HUs.^29,30 Lower HU values indicate lower skeletal muscle density (SMD) and represent potentially poorer quality of muscle function.^16,28 In secondary analyses, we determined the association between SMD and outcomes using the mean HU of all skeletal muscle tissue at the L3 level.

Outcomes

Trained BRAIN-ICU study personnel recorded discharge location, including home, rehabilitation hospital, nursing home, or long-term acute care facility, from the medical record. At 3 and 12 months after hospital discharge, trained psychology personnel who were masked to all details of hospitalization assessed participants using a battery of well-validated tests selected to capture the participants’ experiences of disability and physical function. Personnel tested for disability in basic activities of daily living (ADLs) using the Katz Index of Independence in Activities of Daily Living (Katz ADL),³¹ disability in instrumental activities of daily living (IADLs) using the Functional Activities Questionnaire (FAQ),³² and self-reported physical function using the Physical Component Score (PCS) of the 36-Item Short Form Survey (SF-36).³³ The Katz ADL is a well-validated measure of performance in 6 basic ADLs (bathing, dressing, toileting, transferring out of bed or a chair, bladder and bowel continence, and feeding).^31,34

Each activity is scored 0 (complete independence), 1 (partial dependence), or 2 (complete dependence). Scores on each activity are summed to generate a composite score, with higher scores indicating greater disability in ADLs. The FAQ is a well-validated measure of independence of 10 instrumental ADLs, including balancing a checkbook; assembling tax records; shopping for necessities; participating in a skilled activity; heating water; preparing a balanced meal; tracking current events; comprehension of television, books, and magazines; remembering appointments; and traveling outside the home.³² Each activity is scored from 0 (no assistance needed) to 3 (completely dependent). Scores on each activity then are summed to generate a composite score, with higher scores indicating greater disability in IADLs. The SF-36 is a well-validated, widely used survey of self-reported perceptions of health.³³ The survey comprises 36 questions across 8 domains that are grouped into Mental Component Scores and PCSs. Because our primary interest was in physical function and recovery, we examined associations between muscle health and PCSs. See e-Appendix 2 for detailed descriptions of these instruments.

Statistical Analysis

We present all continuous variables as medians and interquartile ranges and categorical variables as frequencies and proportions. We performed a Shapiro-Wilk test of the distribution of SMI and SMD and, based on these results, tested for a difference in SMI and SMD based on 3 age categories (≤ 49 years, > 49 years and < 65 years, and ≥ 65 years) using a Kruskal-Wallis test and a 1-way analysis of variance, respectively.

We used multivariable regression to evaluate associations between SMI or SMD and outcomes, adjusted for covariates. Our primary model examined the relationship between SMI and Katz ADL at 3 months. We also performed prespecified analyses between muscle health and secondary (exploratory) outcomes including the relationship between SMI and discharge location, Katz ADL scores at 12 months, FAQ scores at 3 and 12 months, and SF-36 PCSs at 3 and 12 months. We further examined the relationship between SMD and these outcomes. We treated SMI and SMD as continuous variables in our models to provide the most statistical power to detect an association with outcomes (ie, reduce the risk of a false-negative finding or type II error) and to increase the precision of our estimates. As covariates, we a priori selected the following for inclusion our models: age at enrollment, sex, Charlson Comorbidity Index score at enrollment, baseline Katz ADL score, baseline FAQ score, mean daily Sequential Organ Failure Assessment score, and the duration of mechanical ventilation (e-Appendix 3 provides more detail about the covariates). We used a directed acyclic graph to understand the relationships among exposures, outcomes, and covariates and to determine the minimal sufficient adjustment set (e-Fig 1).³⁵ To avoid overfitting and risk of an unstable model, we conducted an assessment for multicollinearity and included all indicated covariates before completing the final analysis. For discharge location, we used multivariable logistic regression. For ADL, IADL, and physical function outcomes, we used proportional odds logistic regression.

To avoid potential confounding because of critical illness-associated muscle wasting, we performed a sensitivity analysis excluding participants with CT scans obtained after the day of ICU admission (n = 20). To address the potential for competing risks of death and loss to follow-up, we performed additional sensitivity analyses assuming a worst possible outcome in ADLs and IADLs.^36,37 For the 3-month follow-up sensitivity analyses, we included all participants who were alive and active in the study at hospital discharge (n = 176). Participants who were lost to follow-up or withdrew before the 3-month assessment were assigned the maximum disability score (ADL = 12; IADL = 30). Participants who died before the 3-month assessment were assigned a score consistent with 1 full disability worse than the maximum disability score (ADL = 14; IADL = 33). The 12-month follow-up sensitivity analyses followed the same scoring structure, but included only those participants alive and active in the study at the 3-month assessment (n = 125).

Follow-up data were > 96% complete. To reduce bias resulting from missing outcome data, we used multiple imputation for participants with at least some follow-up (ie, we did not impute outcome values for those who completed no assessments because of death, loss to follow-up, or study withdrawal).³⁸ We allowed all associations with continuous covariates to be nonlinear using restricted cubic splines. We excluded nonlinear terms if the P value for the global test for nonlinearity was > .20. All model assumptions were met. Because all analyses were prespecified, we did not adjust for multiple comparisons. We considered P values of < .05 to be significant. We used Stata/SE version 15.1 software (StataCorp, LLC) for all analyses.

Results

Between March 2007 and June 2010, the BRAIN-ICU study enrolled 521 participants at Vanderbilt University Medical Center. Of these, 241 patients underwent abdominal CT imaging within 6 months before or 4 days after ICU admission and survived to hospital discharge. Of these, 48 participants died, withdrew, or were lost to follow-up before the 3-month evaluation. An additional 5 participants did not complete the entire 3-month evaluation, 2 participants had scans that could not be analyzed because of motion artifact and portions of subcutaneous fat outside the scan field-of-view, and 1 participant had missing height data that prevented calculation of SMI. We included 125 participants in the 3-month analyses. Between the 3-month and 12-month follow-up, 19 participants died, withdrew, or were lost to follow-up, leaving 106 participants who were included in the 12-month analysis (Fig 1).

Figure 1 –

Flow diagram showing progression of study participants through the study. VUMC = Vanderbilt University Medical Center.

Baseline participant characteristics are outlined in Table 1. Participants had a median age of 56 years (interquartile range [IQR], 47–64 years), 46% were female, 91% were mechanically ventilated for a median of 4 days (IQR, 2–8 days) and had a high severity of illness, with a mean daily Sequential Organ Failure Assessment score of 10 (IQR, 8–12). The majority of participants had scans collected before or up to the day of ICU admission (n = 105; 84%). Timing of all CT scans used is presented in e-Table 1.

TABLE 1 ].

Patient Demographic and Clinical Characteristics (N = 125)

Characteristic	Data
Female sex	56 (45)
Age at enrollment, y	56 (47–64)
Race
White	111 (89)
Black	14 (11)
BMI, kg/m2	29 (25–35)
ICU type
Medical	39 (31)
Surgical	86 (69)
Charlson Comorbidity Index	3 (1–4)
Mean SOFA in ICU	10 (8–12)
Admission diagnosis
Sepsis	36 (29)
Acute respiratory failure	25 (20)
Cardiogenic shock, myocardial ischemia, or arrhythmia	8 (6)
Hepatobiliary or pancreatic surgery	22 (18)
Gastric or colonic surgery	15 (12)
Cirrhosis or hepatic failure	4 (3)
Other surgery	9 (7)
Other diagnosis	6 (5)
Clinical frailty scale score at enrollment
1 (very fit)	4 (3)
2 (well)	25 (20)
3 (well with treated comorbid disease)	53 (42)
4 (apparently vulnerable)	18 (14)
5 (mildly frail)	7 (6)
6 (moderately frail)	17 (14)
7 (severely frail)	1 (1)
Katz ADL score at enrollment	0 (0–1)
Partial disability	40 (32)
ADL score with disability	2 (1,5)
Functional Activities Questionnaire score at enrollment	0 (0–2)
Partial disability	37 (30)
FAQ score with partial disability	7 (3–12)
Mechanical ventilation
No. of patients	114 (91)
No. of days	4 (2–8)
Delirium
No. of patients	96 (77)
No. of days	4 (2–7)
Coma
No. of patients	78 (62)
No. of days	3 (1–5)
ICU length of stay, d	5 (2–10)
Hospital length of stay, d	12 (7–18)
Discharge location
Home	70 (56)
Rehabilitation hospital	35 (28)
Nursing home	9 (7)
LTAC	10 (8)
Hospice	0 (0)
Other hospital	1 (1)
Skeletal muscle index, cm2/m2	51.9 (42.3–66.0)
Muscle density, Hounsfield unit	20.5 (14.6–27.6)

Data are presented as No. (%) or median (interquartile range). ADL = activity of daily living; FAQ = Functional Activities Questionnaire; Katz ADL = Katz Index of Independence in Activities of Daily Living; LTAC = long-term acute care; SOFA = Sequential Organ Failure Assessment.

The median baseline SMI was 51.9 cm²/m² (IQR, 42.3–66.0 cm²/m²), and the median BMI was 29 kg/m² (IQR, 25–35 kg/m²). Based on our predefined parameters, 51 of 125 participants (41%) demonstrated low skeletal muscle mass. Analyzed by sex, 34 of 69 male participants (49%) and 17 of 56 female participants (31%) demonstrated low muscle mass at baseline. Older adults (defined as aged 65 years of or older) demonstrated a lower median SMI than younger participants, and the between-group difference was significant by the Kruskal-Wallis test (H(2) = 10.665; P = .005), although low muscle mass was present in younger participants as well (Fig 2). The prevalence of low muscle mass was greatest among those with either a normal, overweight, or obese BMI. Of the 51 participants meeting the threshold for radiographic low muscle mass, only 6 participants (12%) were underweight, whereas 15 participants (29%) had a normal BMI, 16 participants (31%) were overweight, and 14 participants (28%) were obese (e-Table 2) . Baseline SMD was distributed normally with a mean (SD) of 21.1 (9.5) HU. Participants in the 2 older age groups (49–65 years and 65 years and older) showed lower median SMD than younger participants, and the between-group difference was significant by 1-way analysis of variance (F(2,122) = 3.43; P = .04) (Fig 3).

Figure 2 –

Graph showing skeletal muscle mass index before critical illness, stratified by sex and age. Skeletal muscle mass index (SMI) is calculated by summing the cross-sectional area of the psoas, quadratus lumborum, erector spinae, lateral oblique, internal and external oblique, and rectus abdominus (in square centimeters) measured at the L3 lumbar vertebra on an abdominal CT scan and dividing by height (in square meters). Pink dots indicate baseline SMI of a female patient. Blue dots indicate baseline SMI of a male patient. The cohort is stratified into 3 groups based on age: 18 to 49 years, 50 to 64 years old, and 65 years and older. The median SMI for each group was 54.4 cm²/m², 54.2 cm²/m², and 42.6 cm²/m², respectively. The difference between groups was significant by Kruskal-Wallis test (H(2) = 10.665; P = .005). The accepted cutoff values for radiographic low muscle mass are indicated as a blue bar for male participants (SMI < 55 cm²/m²) and a pink bar for female participants (SMI < 39 cm²/m²).²⁷

Figure 3 –

Graph showing skeletal muscle density before critical illness, stratified by sex and age. Skeletal muscle density (SMD) is measured by averaging the attenuation (in HUs) of the psoas, quadratus lumborum, erector spinae, lateral oblique, internal and external oblique, and rectus abdominus muscles at the L3 lumbar vertebra on an abdominal CT scan. Pink dots indicate baseline SMD for female patients and blue dots indicate baseline SMD for male patients. The cohort is stratified into 3 groups based on age: 18 to 49 years, 50 to 64 years, and 65 years and older. The mean SMD for each group was 24.4 HU, 19.7 HU, and 19.5 HU, respectively. A statistically significant difference between groups was determined by 1-way analysis of variance (F(2,122) = 3.43; P = .04). Lower HU values (y-axis) represent lower attenuation of muscle on CT imaging and were correlated with increased intramuscular adipose content of the muscle and decreased muscle quality. Cutoffs for low SMD are indicated as a blue bar for male patients (SMD < 38.5 HU) and a pink bar for female patients (SMD < 34.3 HU).⁶⁴ HU = Hounsfield unit.

Fifty-five of 125 participants (44%) were discharged to a facility, including 35 participants (28%) discharged to a rehabilitation hospital, 10 participants (8%) discharged to a long-term acute care hospital, 9 participants (7%) discharged to a nursing home, and 1 participant (1%) discharged to another acute care hospital. Median ADL scores at 3 months were 0 (IQR, 0–1) and 0 (IQR, 0–0) at 12 months, and median IADL scores were 2 (IQR, 0–8) and 1 (IQR, 0–7) at 3 and 12 months, respectively. Median SF-36 PCS scores were nearly 2 SDs less than population means at both 3 and 12 months (28 [IQR, 22–38] and 35 [IQR, 25–46], respectively). Disability and physical functional scores are presented in e-Table 3.

After adjusting for age, sex, coexisting illnesses, baseline ADLs and IADL, severity of illness, and the duration of mechanical ventilation, SMI was not associated with the primary outcome of Katz ADL score at 3 months (OR, 1.01; 95% CI, 0.96–1.06; P = .73) (Table 2). Likewise, SMI was not associated with the secondary outcomes of discharge location, Katz ADL scores at 12 months, FAQ scores at 3 or 12 months, or SF-36 PCS scores at 3 or 12 months. We found that each 1-HU increase in the secondary exposure, SMD, was associated with 6% greater odds of being discharged home (OR, 1.06; 95% CI, 1.0–1.1; P = .04) (Table 2). However, SMD was not associated Katz ADL scores, FAQ scores, or SF-36 PCS scores at 3 or 12 months of follow-up.

TABLE 2 ].

Association Between Skeletal Muscle Mass and Skeletal Muscle Quality and Outcomes at 3 and 12 Months

Assessment Timing	Association With SMI			Association With SMD
	OR	95% CI	P Value for Trend	OR	95% CI	P Value for Trend
Discharge to home vs other location
Discharge	1.00	0.97–1.02	.67	1.06	1.00–1.12	.04
Katz ADL score
3 mo	1.01	0.96–1.06	.73	1.00	0.98–1.03	.88
12 mo	0.99	0.97–1.02	.56	1.02	0.96–1.08	.62
FAQ score
3 mo	1.01	0.97–1.04	.78	1.01	0.99–1.02	.50
12 mo	1.00	0.99–1.02	.74	1.00	0.96–1.04	.99
SF-36 PCS score
3 mo	1.01	0.97–1.04	.78	0.99	0.97–1.01	.29
12 mo	0.99	0.98–1.01	.49	0.99	0.95–1.03	.51

Results are shown from regression models. For models involving SMI, the OR indicates that each 1-cm²/m² increase in SMI is associated with the indicated increase (OR > 1) or decrease (OR < 1) in the odds of having a higher outcome score or being discharged home. For models involving SMD, each 1-Hounsfield unit increase in SMD is associated with the indicated increase (OR > 1) or decrease (OR < 1) in the odds of having a higher outcome score or being discharged home. All models were adjusted for age at enrollment, sex, Charlson Comorbidity Index score at enrollment, baseline Katz ADL score, baseline FAQ score, mean modified daily Sequential Organ Failure Assessment score, and duration of mechanical ventilation. FAQ = Functional Activities Questionnaire; Katz ADL = Katz Index of Independence in Activities of Daily Living; SMD = skeletal muscle density; SMI = skeletal muscle mass index; SF-36 PCS = 36-Item Short Form Survey Physical Component Score.

Sensitivity analyses that removed the 20 participants whose scans were obtained after ICU admission (e-Table 1) did not change our results qualitatively, with the exception of 2 findings. First, higher SMI was associated with an increased odds of a higher FAQ score, indicating worse disability in IADLs at 3 months (OR, 1.03; 95% CI, 1.00–1.05; P = .03). Second, higher SMI was associated with lower odds of having a higher SF-36 PCS score indicating better physical health score at 3 months (OR, 0.98; 95% CI, 0.95–0.99; P = .02). Full results are reported in e-Table 4.

Sensitivity analyses to examine competing risks of death and loss to follow-up similarly did not change our results qualitatively, excepting for 1 finding. After assigning maximum disability and worse than maximum disability scores to participants who were lost to follow-up and died, respectively, higher SMI was associated with a decreased odds of a higher ADL score, indicating less disability in ADLs at 3 months (OR, 0.98; 95% CI, 0.96–1.00; P = .04). Characteristics of the full survivor cohort (N = 176) are reported in e-Table 5, and full results of the sensitivity analyses are reported in e-Table 6.

Discussion

In this study of 125 survivors of critical illness, we found that 2 of 5 survivors demonstrated low muscle mass before ICU admission, including a large proportion who had either normal, overweight, or obese BMIs. We did not find an association between muscle mass before critical illness and our primary outcome, disability in ADLs at 3 months. In a prespecified exploratory analysis, we found that greater muscle density before ICU admission was associated with an increased likelihood of being discharged home, an important patient-centered outcome and marker of short-term disability.³⁹ However, we did not find that either pre-critical illness muscle mass or muscle density were associated with long-term disability or self-reported physical function at 3 and 12 months after critical illness.

Survivors of critical illness often experience myriad new chronic conditions, a major component of which is new physical disability.⁶ Although a large body of prior work has focused on the loss of muscle tissue and alterations in neuromuscular and mitochondrial function that, together, commonly manifest as ICU-acquired weakness as a cause of disability in survivors of critical illness, little information surrounding the effect of pre-critical illness muscle health on these outcomes is available.^40–42 Because disability is hypothesized to develop when a vulnerable host encounters a sufficiently strong stressor, we hypothesized that poor muscle health before a critical illness results in a higher likelihood of disability in survivors. To test this hypothesis, we measured 2 independent measures of muscle health: SMI, which reflects muscle mass, and SMD, which reflects (indirectly) the lipid content—and therefore the quality—of skeletal muscle.⁴³ We did not find associations between SMI before critical illness and disability or physical function. Our findings may indicate that the drivers of impaired physical function after critical illness relate more to the rapid muscle loss during the critical illness and to intervening events after discharge, although the study simply may have been underpowered to detect differences. Larger studies that measure both muscle mass before critical illness and disability in survivors are warranted.

Our findings that greater SMD, but not SMI, was associated with greater odds of being discharged home could suggest that intramuscular lipid accumulation may be an important predictor of short-term, patient-centered outcomes in survivors of critical illness. In community-dwelling adults, SMD is associated with muscle strength and mobility.^16,44,45 Thus, it may be the case that greater SMD in this cohort reflects better muscle strength before critical illness. Because muscle strength, which is measured rarely before critical illness and after ICU admission, is subject to confounding by factors such as the acute illness, delirium, and sedation, SMD could serve as a surrogate marker of baseline muscle strength. Future studies are needed to explore the relationship further and to define better the relationship between muscle strength before critical illness and outcomes in survivors.

Increasing our understanding of which patients are most likely to experience a new disability and the drivers of new disability will allow providers to personalize care to the patient’s risk factors at each stage in the disease process. Our exploratory findings suggest that better SMD at ICU admission is associated with greater odds of being discharged home, indicating better function in the immediate period after hospitalization. These data support the idea that factors before critical illness play a role in outcomes after critical illness. Few data exist on the effects of critical illness on these measures of muscle health. Moreover, how physical rehabilitation and early mobility during the ICU stay may maintain or restore measures of muscle health are unclear because few studies to date have examined the effects of these interventions on these measures.⁴⁶ Because physical activity has been shown to prevent muscle fat infiltration while increasing strength even in the face of loss of muscle mass in community-dwelling older adults at high risk of mobility disability, an understanding of the effects of physical rehabilitation and early mobility interventions in those with critical illness on measures of muscle health is needed.^47,48

Prior studies focused on the association between low muscle mass at ICU admission and mortality. Weijs and colleagues¹⁹ measured total skeletal muscle mass at the L3 level in 240 mechanically ventilated adults in a mixed medical-surgical ICU. After deriving cutpoints for low muscle mass from a receiver operating characteristic curve analysis, they found that low muscle mass was associated with higher mortality. Further studies have found an association between low muscle mass at admission for critical illness and mortality in hospital at 30 days, 6 months, and 1 year after discharge.^49–51 Similarly, patients with SARS-CoV-2 infection requiring ICU care were observed to have increasing rates of mortality with lower baseline muscle mass.^52,53 Our findings expand on these prior studies by evaluating associations between skeletal muscle mass before critical illness and long-term disability and physical function in survivors of critical illness.

We found no associations between muscle mass before critical illness and long-term disability and physical function. These findings are supported by a recent study of 1,724 noncritically ill, hospitalized older adults from a longitudinal cohort study who underwent dual-energy radiographic absorptiometry scans within 5 years of hospitalization that found that appendicular lean mass (a marker of muscle mass) before hospitalization was not predictive of new disability after hospitalization.⁵⁴ In contrast, Cox and colleagues⁵⁵ reported that patients with sepsis who had sarcopenia at hospital admission showed worse physical function as measured by the Short Physical Performance Battery at the 3-month and 6-month follow-up. The difference in our findings may relate to different methods for assessing physical function (questionnaire vs performance-based measurement), or that Cox and colleagues⁵⁵ did not adjust for baseline physical function in their models. Future studies that incorporate measures of physical function before illness are needed to understand better the effects of muscle mass before critical illness on long-term function in survivors.

Lower skeletal muscle density, a marker of greater intramuscular adipose tissue and thus poorer muscle quality, is associated with lower muscle strength^16,44 and loss of mobility function in community-dwelling older adults.⁴⁵ Two previous studies in those with critical illness found associations between lower muscle density and greater mortality at both 90 days and at 6 months.^20,56 Our study builds on these findings by reporting an association between lower muscle density and being discharged to an institution, an indicator of short-term disability. Nevertheless, we did not find an association between muscle density and longer-term disability. Andrews and colleagues⁵⁴ reported that lower handgrip strength was predictive of new disability in older adults surviving a hospitalization. Thus, although muscle density is correlated to function in some populations, it may not be an adequate measure of muscle strength in patients requiring hospitalization. Future studies are needed to explore methods to assess muscle strength before illness in patients unable to participate in traditional strength examinations because of illness or sedation.

We found that low muscle mass was present in 2 of 5 participants, including a substantial proportion with normal, overweight, or obese BMIs. This prevalence of low muscle mass is 4 times higher than that of nonhospitalized older adults worldwide⁵⁷ and double that reported in older adults requiring hospitalization who were not admitted to the ICU.⁵⁸ The role of poor musculoskeletal health before critical illness remains unclear, despite impacting a large portion of the critically ill population. In addition, rapid muscle wasting occurs in critical illness⁵⁹ and during prolonged periods of bed rest.⁶⁰ Future studies are needed to explore the relationship between skeletal muscle health before critical illness, the dynamic changes to skeletal muscle during critical illness, and long-term recovery.

Our study has several strengths. First, we included a diverse mix of patients with medical and surgical critical illness, increasing generalizability. Second, we included CT scans performed up to 6 months before ICU admission or within 4 days of ICU admission, representing a pragmatic baseline of muscle mass and quality, thereby limiting the effects of critical illness on these exposures. We also controlled for baseline disability in ADLs and IADLs, increasing the robustness of our analyses and conclusions. Finally, we used 2 validated measures of muscle health, assessing both muscle mass and quality.

Our study also should be considered in light of several limitations. This was a single-center, nested cohort study of participants who underwent CT scans during clinical workup over a period of 6 months before or 4 days after admission for critical illness, introducing the potential for selection bias and confounding related to the effects of critical illness into our study. To reduce the chance, we adjusted for multiple covariates, including Charlson Comorbidity Index score, in our models. We included CT scans that ranged from 6 months before ICU admission until 4 days after admission because our intention was to examine patients’ baseline muscle health. Although this window could introduce variability into patients’ baseline data, the rate of age-related decline in muscle mass is 3% to 8% per decade after 30 years of age; thus, it is unlikely that change in muscle mass over that time was significant. We also included CT scans obtained up to 4 days after ICU admission; thus, our findings could have been affected by the muscle loss that frequently accompanies critical illness. A sensitivity analysis that excluded the 20 patients who had CT scans after ICU admission did not substantively alter our findings. Second, less than one-half of our original cohort underwent abdominal CT scans, highlighting the clinical limitation of using CT scans for identifying low muscle mass. Moreover, serial CT imaging to assess changes in muscle health over time is limited by the need to transport patients with acute critical illness, radiation exposure, and cost. This is a direct reflection of the current status of assessing muscle health in critical illness. Few patients seek treatment having ever received a formal assessment of muscle health, CT scans are ordered as part of clinical care, and targeted muscle health assessments through alternative methods are not performed routinely. Future studies would benefit from exploring alternative methods that are scalable to more patients in the ICU, such as ultrasound.⁶¹ Third, we analyzed associations among those participants who survived critical illness and participated in follow-up, potentially introducing survivor bias in our outcomes. The primary focus of this preliminary study was to evaluate the association between muscle mass and quality with long-term disability and physical functional outcomes among survivors. To characterize better the risk of bias introduced from the competing events of death and loss to follow-up, we conducted sensitivity analyses that assumed the worst possible outcome by assigning maximum disability scores to participants who experienced these events. Interestingly, we did not find any results that were substantially different from our original analysis. Our findings are consistent with work previously published by Murphy and colleagues⁶² examining the relationship between preexisting frailty and disability after critical illness that found the competing risk of death had only a small effect on disability count outcomes, such as those used in the present analyses, when follow-up was < 5 years. Fourth, most ICU survivors in this cohort reported no disability in ADLs at either time point and no disability in IADLs at 3 months (median score, 0) that increased to minor difficulty in 1 activity (median score, 1) at 12 months. Finally, we did not collect data on rehabilitation interventions during hospitalization or after discharge that may alter the course of the development of disability and poor physical function during critical illness and subsequent recovery. We did include data on discharge location, providing some insight into care after hospitalization, with 28% of the cohort going to a rehabilitation hospital. Most participants (56%) were discharged home, however, and most ICU survivors discharged to home receive little rehabilitation, limiting the impact on our findings.⁶³

Interpretation

In this nested cohort study of survivors of critical illness, 40% demonstrated low muscle mass before critical illness, despite most having a normal, overweight, or obese BMI. Nevertheless, muscle mass before critical illness was not associated with short-term or long-term disability or self-reported physical function. Greater skeletal muscle density, a marker of muscle quality, was associated with discharge to home vs an ongoing care facility, indicating better physical function at discharge and suggesting an important role for better muscle health in this patient-centered outcome. Further work is needed to understand better skeletal muscle health across the continuum of critical illness and survivorship to inform targeted interventions to enhance recovery of physical function after critical illness.

Take-Home Points.

Study Question:

We sought to understand if using measurements of muscle mass and muscle density on CT imaging as an indicator of muscle health before critical illness is associated with physical function and acquired disability among survivors of critical illness.

Results:

We found that higher skeletal muscle density, a measure of muscle tissue quality, was associated with an increased odds of being discharged to home, but did not find an association between pre-existing muscle mass or muscle density and long-term disability in activities of daily living, instrumental activities of daily living, or self-reported physical function at 3 and 12 months after critical illness.

Interpretation:

Our findings indicate that baseline muscle health may not be as influential on long-term physical recovery after critical illness as other factors encountered during the acute illness, although it is also possible that current reliance on clinically available CT imaging is insufficient to address this question adequately.

ABBREVIATIONS:

ADL

activity of daily living

FAQ

Functional Activities Questionnaire

HU

Hounsfield unit

IADL

instrumental activity of daily living

IQR

interquartile range

Katz ADL

Katz Index of Independence in Activities of Daily Living

PCS

Physical Component Score

SF-36

36-Item Short Form Survey

SMD

skeletal muscle density

SMI

skeletal muscle mass index