Sex Disparities and Functional Outcomes after a Critical Illness

To the Editor:

Enhanced identification of patients at risk of post-ICU morbidity is essential to improve care delivery and outcomes (1, 2). Females are at higher risk than males for worse health, function, and healthcare-use outcomes in a variety of health conditions, but no research has yet examined this risk among survivors of a critical illness (3, 4). To mitigate this gap, we conducted a secondary analysis of the BRAIN-ICU (Bringing to Light the Risk Factors and Incidence of Neuropsychological Dysfunction in ICU Survivors) cohort study (5). We tested the hypothesis that female sex is associated with more long-term physical, cognitive, and psychological dysfunction after a critical illness.

Methods

Participants were ≥18 years old and had been admitted to medical or surgical ICUs in two hospitals (one academic and one community) with acute respiratory failure or shock. The exclusion criteria included >72 hours of organ dysfunction before screening, recent ICU exposure, severe baseline functional impairment, inability to speak English, homelessness, or residence >200 miles from the enrollment site. Institutional review boards at both sites approved the study protocol. The participants or their proxies provided informed consent.

The exposure variable was self-reported sex (male or female). At 3- and 12-month follow-ups, we measured 1) self-reported disability and physical function using the Katz Activities of Daily Living (ADL) scale, Functional Activities Questionnaire (FAQ), and Short Form (36) Health Survey (SF-36) physical function index (PFI); 2) self-reported psychological function using the Beck Depression Inventory II (BDI-II), post-traumatic stress disorder checklist (PCL), and SF-36 mental component score (MCS); 3) global cognitive function using the Repeatable Battery for the Assessment of Neuropsychological Status; and 4) healthcare use according to the number of self-reported rehospitalizations and whether the subject was living at home (yes/no).

Statistical analysis

Using SAS version 9.3 (SAS Institute, Inc.), we computed descriptive statistics at baseline and at 3 and 12 months, stratified by sex. We compared the sexes at baseline using independent-samples t-test and Wilcoxon signed-rank, chi-square, and Fisher’s exact tests. We fitted analysis of covariance models for continuous and logistic regression models for dichotomous 3- and 12-month outcomes, with sex as the predictor and with age, race, education, Charlson index, and closest available proxy-reported baseline measure of the outcome as covariates (see Table 2 legend for details). We used dichotomization before modeling for categorical measures. We conducted sensitivity analyses at both 3 and 12 months with additional covariates, including Sequential Organ Failure Assessment score at enrollment, history of psychiatric illness (in analyses that did not already include it, such as the SF-36 MCS), whether the participant experienced any delirium (yes/no), and socioeconomic status score (6).

Table 2.

Outcomes at 3 and 12 Months

Outcome	Male	Female	Female vs. Male Adjusted* Difference (95% CI)	Female vs. Male Adjusted* Odds Ratio (95% CI)	P Value
At 3 mo:	n = 222	n = 216
Physical function
Katz ADL† score ≥1	55 (24.8)	79 (36.6)	—–	1.67 (1.05 to 2.63)	0.03
FAQ‡ score ≥1	141 (63.5)	154 (71.3)	—–	1.62 (1.01 to 2.57)	0.04
SF-36 PFI§	34.43 ± 12.99	29.46 ± 11.25	−4.50 (−6.79 to −2.22)	—–	0.0001
Psychological function
BDI-II‖	11.08 ± 8.53	14.30 ± 10.97	2.09 (0.18 to 3.99)	—–	0.03
PCL¶	24.32 ± 8.49	27.23 ± 9.92	2.72 (0.92 to 4.51)	—–	0.003
SF-36 MCS**	51.28 ± 11.02	46.27 ± 13.41	−3.41 (−5.76 to −1.04)	—–	0.005
Cognitive function
RBANS††	78.17 ± 12.07	77.80 ± 12.72	0.84 (−1.36 to 3.03)	—–	0.45
Healthcare use
Living at home‡‡	211 (95.1)	191 (88.4)	—–	0.41 (0.19 to 0.88)	0.02
Rehospitalized	81 (36.5)	87 (40.3)	—–	1.19 (0.80 to 1.77)	0.39
At 12 mo:	n = 193	n = 189
Physical function
Katz ADL† score ≥1	44 (22.8)	58 (30.7)	—–	1.14 (0.69 to 1.89)	0.6120
FAQ‡ score ≥1	121 (62.7)	135 (71.4)	—–	1.36 (0.86 to 2.16)	0.1892
SF-36 PFI§	38.2 ± 13.6	33.8 ± 12.4	−3.10 (−5.63 to −0.57)	—–	0.0165
Psychological function
BDI-II‖	11.1 ± 9.1	12.8 ± 10.8	0.77 (−1.32 to 2.86)	—–	0.4691
PCL¶	24.2 ± 9.1	26.4 ± 9.4	1.97 (0.08 to 3.87)	—–	0.0415
SF-36 MCS**	51.3 ± 10.5	48.0 ± 12.0	−1.99 (−4.30 to 0.31)	—–	0.0900
Cognitive function
RBANS††	79.6 ± 12.4	78.7 ± 12.8	0.28 (−2.15 to 2.70)	—–	0.8234
Healthcare use
Living at home‡‡	186 (96.4)	177 (93.7)	—–	0.61 (0.23 to 1.62)	0.3233
Rehospitalized	83 (43.0)	86 (45.5)	—–	1.02 (0.67 to 1.55)	0.9331

Definition of abbreviations: ADL = activities of daily living; BDI-II = Beck Depression Inventory II; CI = confidence interval; FAQ = Functional Activities Questionnaire; MCS = mental component score; PCL = post-traumatic stress checklist; PFI = physical function index; RBANS = Repeatable Battery for the Assessment of Neurological Status; SF-36 = 36-Item Short Form Survey.

Data are shown as n (%), median (interquartile range), or mean ± SD unless otherwise specified.

^*Adjusted for age, race, education, Charlson index, and baseline measure of the outcome (or a close available marker of it). The analysis of SF-36 physical component score was adjusted for baseline Katz ADL, BDI for history of depression, PCL for history of post-traumatic stress disorder, SF-36 MCS for any history of psychiatric illness, RBANS for baseline Informant Questionnaire on Cognitive Decline in the Elderly, and whether the patient was rehospitalized or living at home for their discharge disposition.

^†Katz ADL: range 0–7; higher scores indicate greater dependence in ADLs.

^‡FAQ: range 0–30; higher scores indicate greater dependence in instrumental ADLs.

^§SF-36 PFI: mean score in U.S. populations of 50, SD = 10; higher scores indicate better physical function–related quality of life.

^‖BDI-II: range 0–63; higher scores indicate more depressive symptoms (≥17 is consistent with clinical depression).

^¶PCL: range 17–85; higher scores indicate more symptoms of trauma (≥30 suggests post-traumatic distress disorder in civilian samples).

^**SF-36 MCS: mean score in U.S. populations of 50, SD = 10; higher scores indicate better mental health–related quality of life.

^††RBANS: mean education-corrected index score = 100 with SD = 15; higher scores indicate better cognitive performance.

^‡‡Participants reported living in locations other than home as follows: males at 3 months: assisted living (n = 2), skilled rehabilitation (n = 8), and nursing home (n = 1); females at 3 months: assisted living (n = 3), skilled rehabilitation (n = 12), nursing home (n = 5), and other (n = 5); males at 12 months: assisted living (n = 1), skilled rehabilitation (n = 2), and other (n = 4); females at 12 months: assisted living (n = 4), skilled rehabilitation (n = 5), nursing home (n = 1), and other (n = 2).

Results

Of 821 participants enrolled in the study, 569 survived to the 3-month follow-up and 510 survived to the 12-month follow-up. Of 319 males enrolled, 222 (69.6%) and 193 (60.5%) underwent 3- and 12-month assessments, respectively. Of 311 females enrolled, 216 (69.5%) and 189 (60.8%) underwent 3- and 12-month assessments, respectively. Thus, follow-up rates did not differ by sex (3-mo difference P = 0.97, 12-mo difference P = 0.94). Compared with males, fewer females had private insurance, and females had lower Sequential Organ Failure Assessment scores and higher baseline ADL disability and depression (Table 1).

Table 1.

Baseline Characteristics

Characteristic	All (N = 438)	Sex
		Male (n = 222)	Female (n = 216)	P Value
Age	59.0 ± 14.8 (60.3 [20.2])	59.0 ± 14.1 (60.9 [17.6])	59.0 ± 15.6 (59.4 [22.7])	0.95
Black	49 (11.2)	27 (12.2)	22 (10.2)	0.51
Years of education	12.6 ± 2.5 (12 [2])	12.8 ± 2.6 (12 [2])	12.3 ± 2.4 (12 [2])	0.08
History of depression	153 (34.9)	60 (27.0)	93 (43.1)	0.0004
History of PTSD	15 (3.4)	4 (1.8)	11 (5.1)	0.06
History of other psychiatric disease	43 (9.8)	11 (5.0)	32 (14.8)	0.0005
Charlson comorbidities score	2.45 ± 2.24 (2 [3])	2.61 ± 2.30 (2 [3])	2.28 ± 2.17 (2 [3])	0.10
Katz ADL score*	0.76 ± 1.81 (0 [1])	0.64 ± 1.76 (0 [0])	0.88 ± 1.86 (0 [1])	0.006
Katz ADL score ≥1	113 (25.8)	44 (19.8)	69 (31.9)	0.008
FAQ†	2.35 ± 4.79 (0 [2])	2.15 ± 4.80 (0 [2])	2.55 ± 4.79 (0 [3])	0.26
FAQ score ≥1	166 (37.9)	80 (36.0)	86 (39.8)	0.62
Short IQCODE score‡	3.12 ± 0.29 (3.0 [0.06])	3.11 ± 0.29 (3.0 [0.06])	3.13 ± 0.29 (3.0 [0.13])	0.12
SOFA score	9.17 ± 3.31 (9 [4])	9.78 ± 3.29 (10 [5])	8.55 ± 3.22 (8 [5])	<0.0001
Days of delirium	3.76 ± 4.95 (2 [5])	3.59 ± 4.73 (2 [4])	3.93 ± 5.17 (2 [5])	0.65
Any delirium	327 (74.7)	167 (75.2)	160 (74.1)	0.78
ICU length of stay	8.13 ± 9.45 (4.82 [7.57])	7.79 ± 9.19 (4.56 [7.63])	8.47 ± 9.71 (4.87 [7.42])	0.45
Private insurance	282 (64.4)	155 (69.8)	127 (58.8)	0.02
Government insurance	259 (59.1)	126 (56.8)	133 (61.6)	0.31
SES score	50.6 ± 4.9 (49.8 [5.3])	50.9 ± 4.8 (50.5 [5.6])	50.3 ± 4.9 (49.6 [4.9])	0.16

Definition of abbreviations: ADL = activities of daily living; FAQ = Functional Activities Questionnaire; IQCODE = Informant Questionnaire on Cognitive Decline in the Elderly; PTSD = post-traumatic stress disorder; SES = socioeconomic status; SOFA = Sequential Organ Failure Assessment.

Data are shown as mean ± SD (median [interquartile range]) or n (%).

^*Katz ADL: range 0–7; higher scores indicate greater dependence in ADLs.

^†FAQ: range 0–30; higher scores indicate greater dependence in instrumental ADLs.

^‡Short IQCODE: range 0–5; scores >3 indicate that a family member rated the patient’s memory or intelligence to be worse than it was 10 years before the assessment.

At the 3-month follow-up, females had more physical and psychological dysfunction and lower odds of living at home than males (Table 2). After adjusting for age, race, Charlson comorbidity score, education, and baseline Katz ADL or FAQ, female sex was associated with greater odds of ADL disability (odds ratio [OR], 1.67; 95% confidence interval [CI] 1.05 to 2.63; P = 0.03), greater odds of instrumental ADL disability (OR, 1.62; 95% CI, 1.01 to 2.57; P = 0.04), and worse physical function–related quality of life (SF-36 PFI adjusted difference −4.50; 95% CI, −6.79 to −2.22; P = 0.0001). Females reported more depressive symptoms (BDI-II adjusted difference 2.09; 95% CI, 0.18 to 3.99; P = 0.03), more post-traumatic stress disorder symptoms (PCL adjusted difference 2.72; 95% CI, 0.92 to 4.51; P = 0.003), and worse global mental health (SF-36 MCS adjusted difference −3.41; 95% CI, −5.76 to −1.04; P = 0.005). Despite similar rates of discharge home and rehospitalization, females were less likely than males to live at home at the 3-month follow-up (OR, 0.41; 95% CI, 0.19–0.88; P = 0.02). At the 12-month follow-up, only the differences in global physical function (SF-36 PFI adjusted difference −3.10, P = 0.02) and trauma symptoms (PCL adjusted difference 1.97; P = 0.04) remained significant. In the 3-month sensitivity analysis with adjustment for additional covariates, females still had significantly worse SF-36 PFI (adjusted difference −3.72; 95% CI, −6.05 to −1.40, P = 0.002) and SF-36 MCS (adjusted difference −2.75; 95% CI, −5.13 to −0.37, P = 0.023), and were less likely to be living at home (OR, 0.37; 95% CI, 0.16 to 0.83, P = 0.016) than males. The rest of the associations assessed in the 3- and 12-month sensitivity analyses showed the same direction as observed in the original analyses, although the differences were not statistically significant.

Discussion

In this study of over 400 survivors of a critical illness, females had more physical dysfunction and were less likely to be living at home at the 3-month follow-up than males. Although the statistical significance of specific mood symptoms varied, females also had worse mental health–related quality of life. Most of these differences were no longer statistically significant at 12 months.

These results are internally coherent and likely to be clinically significant. Females had worse physical function than males on the Katz ADL, FAQ, and SF-36 physical component score at 3 months, indicating a broad spectrum of ADL, instrumental ADL, and global impairment that is clearly clinically significant. At 12 months, the females’ mean SF-36 remained 3.10 points below that of the males, exceeding the minimum clinically important difference (MCID) of 3 (7) and indicating persistently worse global physical dysfunction. Females had more symptoms of depression and trauma, and worse global mental health at 3 months. A lack of baseline BDI-II and PCL measurements limits assessment of MCIDs for these outcomes, but the 3.41-point difference in 3-month SF-36 MCS exceeds the 3-point MCID. Differences in physical and mental function by sex were attenuated at 12 months, a finding that may be explained by loss to follow-up or by differential trajectories of recovery by sex; both possibilities warrant additional study. The literature on sex differences is consistent with these results, as studies have demonstrated that physical and psychological impairments mutually influence each other (8), and females are more susceptible to both (9); females disproportionately lose the ability to live at home as a result of multiple factors, including lower socioeconomic status (e.g., less private insurance, as in this study), more ADL disability, and lack of an available caregiver (9, 10); and all of these factors impact quality of life (7). The main conclusion is that sex-related disparities in the functional outcomes of critical illness deserve to be examined in further studies, particularly confirmatory population-based studies.

The main limitation of this study is that it was a secondary analysis. It may have lacked power to detect the statistical significance of some of the suggestive observations from the magnitude and direction of the estimates. In addition, although we adjusted for validated baseline proxies of the outcomes, we cannot rule out residual confounding (11).

In conclusion, females appear to be at risk of worse physical dysfunction and increased short-term institutionalization after a critical illness compared with males. Further research is needed to confirm these findings and ensure equitable healthcare delivery and outcomes for survivors of a critical illness.