Co-occurrence of and remission from general anxiety, depression, and posttraumatic stress disorder symptoms after acute lung injury: a 2-year longitudinal study

Abstract

Objective

To evaluate the co-occurrence, and predictors of remission, of general anxiety, depression, and posttraumatic stress disorder (PTSD) symptoms during 2-year follow-up in survivors of acute lung injury (ALI) treated in an intensive care unit (ICU).

Design, Setting, and Patients

This prospective cohort study enrolled 520 patients from 13 medical and surgical ICUs in 4 hospitals, with follow-up at 3, 6, 12, and 24 months post-ALI.

Measurements and Main Results

The outcomes of interest were measured using the Hospital Anxiety and Depression Scale (HADS) anxiety and depression subscales (scores ≥8 indicating substantial symptoms) and the Impact of Event Scale-Revised (IESR, scores ≥1.6 indicating substantial PTSD symptoms). Of the 520 enrolled patients, 274 died before 3-month follow-up; 186/196 consenting survivors (95%) completed at least one HADS and IESR assessment during 2-year follow-up, and most completed multiple assessments. Across follow-up time points, the prevalence of supra-threshold general anxiety, depression, and PTSD symptoms ranged from 38–44%, 26–33%, and 22–24%, respectively; more than half of the patients had supra-threshold symptoms in at least one domain during 2-year follow-up. The majority (59%) of survivors with any supra-threshold symptoms were above threshold for 2 or more types of symptoms (i.e., of general anxiety, depression, and/or PTSD). In fact, the most common pattern involved simultaneous general anxiety, depression, and PTSD symptoms. Most patients with general anxiety, depression, or PTSD symptoms during 2-year follow-up had supra-threshold symptoms at 24-month (last) follow-up. Higher SF-36 physical functioning domain scores at the prior visit were associated with a greater likelihood of remission from general anxiety and PTSD symptoms during follow-up.

Conclusions

The majority of ALI survivors had clinically significant general anxiety, depressive, or PTSD symptoms, and these symptoms tended to co-occur across domains. Better physical functioning during recovery predicted subsequent remission of general anxiety and PTSD symptoms.

Introduction

Survivors of acute lung injury (ALI) and other critical illnesses frequently suffer with substantial general anxiety, depression, and posttraumatic stress disorder (PTSD) symptoms (1–3), as well as cognitive impairment (4) and muscle weakness (5). Since these phenomena occur so frequently in survivors of critical illness (6), a task force recently suggested the term “post-intensive care syndrome” (PICS) to raise awareness (7). In previous reports, we have separately examined the prevalence or incidence, and correlates, of general anxiety, depression, and PTSD symptoms in a large cohort of ALI survivors (8–13). However, we have not previously examined co-occurrence of symptoms across different domains, and we are unaware of prior studies of factors that may influence remission from these symptoms; hence, these issues are the focus of the current report. With such information, clinicians may be better prepared to care for the growing number of survivors of critical illness.

Materials and Methods

Study population

Mechanically ventilated patients with acute lung injury (ALI) (14) were enrolled consecutively in a prospective longitudinal cohort study involving 13 medical and surgical intensive care units (ICUs) at 4 hospitals in Baltimore, Maryland (ClinicalTrials.gov Identifier: NCT00300248). Specifically, we performed daily screening and enrolled mechanically ventilated patients who had acute-onset hypoxia, bilateral pulmonary infiltrates, PaO₂/FiO₂ ≤300 mmHg, and no clinical evidence of left atrial hypertension. Enrollment started in 2004 and ended in late 2007, with 2-year follow-up visits completed in 2010 (8). To avoid inclusion of patients with primary neurologic disease or head trauma, neurologic specialty ICUs at the participating hospitals were excluded. Key exclusion criteria were 1) pre-existing illness with a life expectancy of <6 months; 2) pre-existing cognitive impairment (ascertained via medical records and/or proxy interview) or communication/language barriers; 3) no fixed address; 4) transfer to a study site ICU with pre-existing ALI of >24 hours duration; 5) >5 days of mechanical ventilation before ALI; and 6) a physician order for no escalation of ICU care (e.g., no vasopressors or hemodialysis) at the time of study eligibility. The reasons for exclusion criteria 4 and 5 were that we wished to ensure collection of potentially relevant data regarding critical illness/intensive care early in the course of ALI or prior to onset of ALI.

Informed consent for follow-up was obtained after patients regained capacity, typically around the time of hospital discharge (15). Follow-up occurred at 3, 6, 12, and 24 months after the onset of ALI (or a few weeks later – median 3 weeks, interquartile range 0 to 7 weeks). Trained research assistants conducted the follow-up interviews, ensuring that patients understood the questions and gave answers that were internally consistent. Most of the interviews (88%) were conducted face-to-face – at a Johns Hopkins Hospital research clinic (61%), at patients’ homes (26%), or at another facility (e.g., a hospital or nursing home – 13%); most patients (58%) required at least one home or other facility visit. Twelve percent of the interviews were conducted over the telephone. Patients received $25 for each visit, as well as a meal coupon and taxi ride or parking voucher/mileage reimbursement. At the 24-month follow-up visit, patients reported retrospectively on mental health treatment since ALI. The institutional review boards of Johns Hopkins University and all participating study sites approved this research.

Definition of general anxiety, depression, and PTSD symptoms

At each follow-up time point, we assessed general anxiety and depressive symptoms using the Hospital Anxiety and Depression Scale (HADS) (16, 17). The HADS anxiety and depression subscales each consist of 7 items, and each item is rated on a 4-point scale; potential scores for each subscale range from 0 to 21. Patients with scores ≥8 on each subscale are considered symptomatic with general anxiety or depressive symptoms. The ≥8 threshold on each subscale is associated with sensitivities and specificities of ~0.8 for the detection of clinically significant anxiety or depressive syndromes in general medical patients (17). We chose the HADS because it was developed for detecting psychiatric symptoms in patients with general medical problems. For example, the HADS depression items primarily assess anhedonia, rather than neurovegetative symptoms (e.g., diminished energy) that could be symptomatic of general medical conditions (e.g., muscular weakness and fatigue) rather than depressive states.

We assessed PTSD symptoms using the Impact of Event Scale-Revised (IESR) (18). The IESR consists of 22 items, each rated on a 4-point scale; scores of all items are averaged to generate a mean total score with a potential range of 0 to 4. In a previous study, we examined the measurement properties of the IESR vs. clinicians’ structured interview-based DSM-IV PTSD diagnostic assessments (using the Clinician-Administered PTSD Scale) (19). At an optimal threshold of ≥1.6, the sensitivity for detecting full DSM-IV PTSD was 100%, and the specificity was 85%; positive and negative likelihood ratios were 6.5 and 0.0, respectively (19). Thus, we chose the ≥1.6 threshold in the current study to define clinically significant PTSD symptoms.

Definitions of remission and recurrence

We defined remission from general anxiety or depression symptoms as having HADS anxiety or depression scores <8 at any follow-up time point after symptoms were evident, along with a statistically reliable decrease in score using the Reliable Change Index (RCI) (20). To calculate the RCI, we employed standard deviation and test-retest reliability estimates from previous studies (21, 22). To have a statistically significant change, patients had to have a difference in scores of ≥4; this threshold is substantially larger than the minimal clinically important difference (1.5 points) established for these subscales (23). We defined recurrence as having a HADS anxiety or depression score ≥8 at any follow-up after remission, along with an increase in score of ≥4.

We defined remission from PTSD symptoms as having an IESR score ≤1.6 at any follow-up time point after PTSD symptoms were evident, along with a statistically reliable decrease in score using the RCI. To calculate the RCI, we employed standard deviation and test-retest reliability estimates from previous studies (24, 25). To have a statistically significant change, patients had to have a difference in scores of ≥0.5; note that this difference is slightly larger than that in our previous publication (13), as we now incorporate the standard deviation from a recent large population-based study (24). We defined recurrence as having an IESR score ≥1.6 at any follow-up after remission, along with an increase in score of ≥0.5.

Predictors of remission

Based on results from prior studies of remission of PTSD and depression (26–28), we evaluated candidate baseline (pre-ALI) variables that could affect remission. These baseline variables included: [1] demographic factors (age, sex, education, employment, and 2011 median income in the patient’s zip code (29–31); [2] mental health factors (a history of depression, anxiety, heavy alcohol use, or illicit drug use); [3] general health factors (evaluated using the Charlson Comorbidity Index (32) and the Functional Comorbidity Index (33)); and [4] physical functioning, defined in two ways: [i] “impaired physical function,” defined as ≥2 dependencies in instrumental activities of daily living (IADLs) (12, 34), and [ii] the SF-36 physical function domain score (35), normed for age and sex, with a general population mean of 50 and standard deviation (sd) of 10. We abstracted mental and general health variables from patients’ medical records, and patients/proxies provided retrospective reports regarding pre-ALI physical functioning. In abstraction of pre-ALI mental health variables, “depression” included any depression diagnosis; “anxiety” included any anxiety diagnosis, including panic disorder and PTSD; “heavy alcohol use” included any indication of problem drinking, alcohol abuse, alcoholism, or alcoholic liver disease (including cirrhosis); and “illicit drug use” included non-injection and injection illicit drug use.

We were also interested in longitudinal relationships between mental health and general health status during the 2 years of follow-up. Thus, we examined new, existing, and remitted general anxiety, depression, and PTSD symptoms at the last visit as potential predictors of remission at each subsequent visit. For example, “new depressive symptoms at the last visit” means that patients were below threshold in the second-to-last visit but had a reliable increase in symptom score to above the instrument’s threshold by the last visit. “Existing depressive symptoms” indicates that patients had depressive symptoms during the last 2 visits (without a reliable decrease by the last visit). “Remitted depressive symptoms” means that patients had depressive symptoms in the second-to-last visit, but had a reliable decrease in score to below threshold by the last visit. These analyses were motivated, in part, by a prior report that anxiety and depressive disorders with onsets after PTSD predicted longer-duration PTSD episodes (26). In addition, we examined new, existing, and remitted impaired physical function (≥2 IADL dependencies) and number of IADL dependencies at the last visit as potential predictors of remission at each subsequent visit. Finally, we examined last-visit SF-36 physical function domain scores as predictors of subsequent-visit remission.

Statistical methods

We examined within-person longitudinal symptom stability using Spearman’s rho correlations. We also examined relationships among HADS anxiety scores, HADS depression scores, and IESR scores using Spearman’s rho correlations. When survivors had missing HADS anxiety, HADS depression, or IESR values during follow-up, we assumed that their prior status remained unchanged (i.e., we carried their last observation forward), as in prior work (12, 13). Since this method is conservative when estimating remission, we supplemented this analysis with additional analyses: 1) only including data from 2-year survivors, and 2) only including data from survivors with no missing data during 2-year follow-up. We employed discrete time survival analysis models (36) to examine potential predictors of remission. To avoid over-fitting of regression models (37), logical groupings of potential predictor variables were evaluated separately in nine small multivariable models. When more than one variable was associated with remission among these initial models, we then examined the variables associated with remission simultaneously in subsequent multivariable models. We defined statistical significance with a 2-sided p-value ≤0.05 and conducted all analyses using R statistical software (38).

Results

Patient flow

A total of 520 patients with ALI were enrolled in the study; 274 (53%) of the enrolled patients died before 3-month follow-up (Figure 1). A total of 186/196 consenting survivors (95%) completed at least one HADS and IESR assessment during 2-year follow-up, and most survivors completed multiple assessments (specifically, 52% had all 4 assessments, 24% had 3, 15% had 2, and 9% had only 1, with death as an important cause of missed follow-up visits – Figure 1). Table 1 shows pre-ALI demographic, health, and physical function characteristics in these survivors.

Figure 1. Flow diagram of study participants.

^a Some patients had a follow-up visit but did not have complete Hospital Anxiety and Depression Scale (HADS) or Impact of Event Scale-Revised (IESR) data for the following reasons, respectively, physically incapable/cognitively incapable/other:

At 3 months, 22 did not have complete HADS data (9/5/8), and 23 did not have complete IESR data (7/5/11).

At 6 months, 12 did not have complete HADS data (4/2/6), and 13 did not have complete IESR data (5/2/6).

At 12 months, 14 did not have complete HADS data (5/4/5), and 15 did not have complete IESR data (5/4/6).

At 24 months, 10 did not have complete HADS data (0/6/4), and 11 did not have complete IESR data (1/5/5).

Table 1.

Pre-acute lung injury descriptive statistics in 186 survivors

Age, mean (sd)	49 (14)
Male sex	56%
Caucasian race	60%
Unemployed	40%
High school or greater education	71%
Median household income (by zip code), mean (sd)	$53,000 ($20,000)
Charlson Comorbidity Index, mean (sd)	2.1 (2.5)
Functional Comorbidity Index, mean (sd)	1.6 (1.5)
Specific comorbidities, per medical records: Depression	21%
Anxiety	6%
Heavy alcohol use	25%
Illicit drug use	32%
SF-36 Physical Function normalized score, mean (sd)*	44 (14)
Number of IADL dependencies, mean (sd)	1.7 (2.3)
Impaired physical function (≥2 IADL dependencies)	38%

IADL = instrumental activities of daily living; sd = standard deviation

^*T-score (general population mean=50, sd=10, normalized based on patient age and sex)

Critical illness/intensive care characteristics of these 186 survivors have been reported previously (12). Briefly, most (86%) had been treated in medical ICUs; 30% had a maximum daily Sequential Organ Failure Assessment (39) score >10; 21% had a mean daily midazolam equivalent dose >75 mg; 29% had a mean daily prednisone equivalent dose ≥50 mg; their mean ICU length of stay was 19 days (sd 17 days); and almost all had been delirious or deeply sedated/comatose at some point during their ICU stay. Notably, more than 97% of these survivors met criteria for acute respiratory distress syndrome (PaO₂/FiO₂ ≤200 mmHg) at some point during their critical illness, and 33% had PaO₂/FiO₂ ≤100 mmHg at some point during their critical illness.

Symptom trajectories

Mean scores did not vary substantially over 2-year follow-up: the average HADS anxiety score remained around 6, the average HADS depression score remained around 5, and the average IESR score remained around 0.9 after the 3-month follow-up visit. All symptoms were highly stable longitudinally; for anxiety, within-person correlations ranged from 0.56 to 0.74 across follow-up time-points; for depression, 0.48 to 0.67; for PTSD, 0.66 to 0.70. Figure 2 shows the prevalence of supra-threshold symptoms of general anxiety, depression, and PTSD during two-year follow-up. Across all follow-up time points, general anxiety symptoms were most common (point prevalences ranged from 38 to 44%), followed by depression symptoms (26 to 33%) and PTSD symptoms (22 to 24%).

Figure 2. Prevalence of general anxiety symptoms (HADS anxiety scores ≥8), depression symptoms (HADS depression scores ≥8), and PTSD symptoms (IESR scores ≥1.6) in the first 2 years after ALI.

HADS, Hospital Anxiety and Depression Scale; IESR, Impact of Event Scale-Revised; PTSD, posttraumatic stress disorder

Figures 3, 4, and 5 illustrate individual and mean symptom trajectories for patients who never had supra-threshold symptoms (of general anxiety, depression, and PTSD, respectively), those who had supra-threshold symptoms throughout the follow-up period, those who had a remission without recurrence, and those who had a remission and recurrence. For those with supra-threshold symptoms at any time point, the most common pattern was to have symptoms remain above threshold throughout 2-year follow-up. The next most common patterns of symptom trajectory were remission without recurrence, then remission with recurrence. Notably, 105 of the 186 patients (56%) had general anxiety symptoms at any point during two-year follow-up; in 91/105 (87%), the symptoms were evident during the first year of follow-up. Eighty-five of the 186 patients (46%) had depressive symptoms at any point during follow-up; in 74/85 (87%), the symptoms were evident during the first year. Sixty-six of the 186 patients (35%) had PTSD symptoms at any point during follow-up; in 60/66 (91%), the symptoms were evident during the first year. At virtually all follow-up time points after 3-month follow-up, survivors were more likely to have remissions than onsets of general anxiety, depression, and PTSD symptoms; specifically, across follow-ups, patients were 1.2, 1.9, and 3.5 times as likely to have remissions than onsets of general anxiety, depression, and PTSD symptoms, respectively.

Figure 3.

Trajectories of Hospital Anxiety and Depression Scale (HADS) anxiety scores among ALI survivors whose general anxiety symptoms remained below threshold throughout 24-month follow-up (“No Symptoms”), those whose symptoms remained above threshold throughout (“Maintainers”), those whose symptoms remitted and did not recur (“Remitters”), and those with remissions and recurrences (“Recurrence”). Thin gray lines indicate individual trajectories, and thick black lines indicate mean trajectories. The horizontal dashed lines indicate the threshold for general anxiety symptoms (HADS anxiety score ≥8)

Figure 4.

Trajectories of Hospital Anxiety and Depression (HADS) depression scores among ALI survivors whose depression symptoms remained below threshold throughout 24-month follow-up (“No Symptoms”), those whose symptoms remained above threshold throughout (“Maintainers”), those whose symptoms remitted and did not recur (“Remitters”), and those with remissions and recurrences (“Recurrence”). Thin gray lines indicate individual trajectories, and thick black lines indicate mean trajectories. The horizontal dashed lines indicate the threshold for depression symptoms (HADS depression score ≥8).

Figure 5.

Trajectories of Impact of Event Scale-Revised (IESR) scores among ALI survivors whose posttraumatic stress disorder (PTSD) symptoms remained below threshold throughout 24-month follow-up (“No Symptoms”), those whose symptoms remained above threshold throughout (“Maintainers”), those whose symptoms remitted and did not recur (“Remitters”), and those with remissions and recurrences (“Recurrence”). Thin gray lines indicate individual trajectories, and thick black lines indicate mean trajectories. The horizontal dashed lines indicate the threshold for PTSD symptoms (IESR score ≥1.6).

Survivors reported marked reductions in physical function after ALI, with gradual improvements during 2-year follow-up. Pre-ALI, survivors’ mean SF-36 Physical Function score was 44 (sd 14); at 3-, 6-, 12-, and 24-month follow-up visits, mean scores were 34 (13), 36 (13), 37 (13), and 39 (13), respectively. Similarly, pre-ALI, survivors’ mean number of IADL dependencies was 1.7 (sd 2.3); at 3-, 6-, 12-, and 24-month follow-up visits, mean numbers were 3.4 (2.9), 2.7 (2.8), 2.2 (2.7), and 2.2 (2.5), respectively. Finally, pre-ALI, 38% of survivors had ≥2 IADL dependencies; at 3-, 6-, 12-, and 24-month follow-up visits, 64%, 56%, 48%, and 47% had ≥2 IADL dependencies, respectively.

Co-occurrence of symptoms

HADS anxiety and depression symptoms were the most highly correlated symptoms (correlations ranged from 0.57 to 0.68 across follow-up time-points), followed by HADS anxiety and PTSD symptoms (0.52 to 0.66), then HADS depression and PTSD symptoms (0.35 to 0.47); all correlations were statistically significant (p ≤ 0.001).

The majority (59%) of survivors with any supra-threshold symptoms were above threshold for 2 or more types of symptoms (i.e., general anxiety, depression, and/or PTSD symptoms). In fact, the most common pattern involved simultaneous general anxiety, depression, and PTSD symptoms (25%). The next most common patterns, in order, were: general anxiety symptoms alone (21%), general anxiety and depression symptoms simultaneously (20%), general anxiety and PTSD symptoms simultaneously (12%), depression symptoms alone (12%), PTSD symptoms alone (8%), and depression and PTSD symptoms simultaneously (2%).

At the 24-month follow-up visit, 138 survivors reported retrospectively on mental health treatment since ALI (Table 2). Of those who had no supra-threshold general anxiety, depression, or PTSD symptoms during follow-up, only 12% reported taking psychiatric medications (such as for depression or anxiety), and only 8% reported receiving specialist mental health care. In contrast, of those who had supra-threshold symptoms in all three domains, 58% reported taking psychiatric medications, and 47% reported receiving specialist mental health care. In almost all cases, there was a monotonic positive relationship between the number of domains of supra-threshold mental health symptoms and receiving treatment (Table 2).

Table 2.

Self-reported mental health treatment during the first 2 years after acute lung injury, by number of mental health domains with supra-thresholdsymptoms

	Number of mental health domains* with supra-threshold symptoms
	0 (n=40)	1 (n=34)	2 (n=27)	3 (n=36)
Tookany psychiatric medications	12%	21%	52%	58%
Received anymental health care:	8%	32%	41%	47%
Saw a psychiatrist	5%	24%	37%	42%
Saw a psychologist	2%	12%	15%	19%
Saw a counselor	2%	6%	15%	11%

^*Symptoms of 3 mental health domains (general anxiety, depression, and post-traumatic stress disorder) were assessed using validated instruments.

Predictors of remission

At least one remission occurred for a minority of patients for each of the three mental health outcomes. Specifically, remissions occurred for 34 of 105 patients with general anxiety symptoms (32%), 26 of 85 with depression symptoms (30%), and 30 of 66 with PTSD symptoms (45%). In the subset of 2-year survivors, remissions occurred for 36%, 30%, and 52%, respectively; and in the subset of patients with no missing data during 2-year follow-up, remissions occurred for 36%, 36%, and 61%, respectively. Baseline demographic and health factors were not significantly related to remission of general anxiety, depression, or PTSD symptoms (Table 3). Better SF-36 physical functioning at the last follow-up visit was independently associated with remission of both general anxiety and PTSD symptoms. Though recent IADL dependencies (evident at the last visit) were negatively associated with remission of general anxiety symptoms, this relationship was no longer statistically significant when recent-onset depressive symptoms were included in a multivariable model (Table 3, Model 10). Similarly, though recent-onset depressive symptoms (evident at the last visit) were negatively associated with remission of general anxiety symptoms, this relationship was no longer statistically significant when recent SF-36 physical functioning was included in a multivariable model (Table 3, Model 11).

Table 3.

Multivariable predictors of remission of anxiety, depression, and PTSD symptoms during the first 2 years after acute lung injury

	Anxiety			Depression			PTSD
	OR*	95% CI	p	OR*	95% CI	p	OR*	95% CI	p
Model 1
Age	0.97	0.93, 1.01	0.17	1.02	0.97, 1.07	0.39	0.99	0.95, 1.04	0.82
Male sex	1.13	0.44, 2.87	0.80	1.42	0.43, 4.67	0.56	0.55	0.21, 1.44	0.21
Unemployment	1.28	0.43, 3.76	0.65	0.75	0.23, 2.49	0.64	1.14	0.37, 3.50	0.82
High school or more education	0.66	0.25, 1.77	0.40	0.73	0.21, 2.53	0.62	0.79	0.28, 2.17	0.17
Median zip code household income	0.91	0.71, 1.16	0.43	0.91	0.65, 1.28	0.57	0.83	0.62, 1.09	0.17
Model 2
Baseline depression	0.50	0.19, 1.29	0.14	0.95	0.32, 2.80	0.92	0.62	0.20, 1.89	0.39
Baseline anxiety	3.01	0.86, 10.5	0.08	1.16	0.24, 5.63	0.85	1.41	0.27, 7.32	0.67
Baseline heavy alcohol use	1.31	0.58, 2.93	0.51	1.92	0.75, 4.91	0.16	1.36	0.53, 3.48	0.51
Baseline illicit drug use	1.15	0.52, 2.54	0.72	0.81	0.33, 2.02	0.65	1.02	0.43, 2.43	0.97
Model 3
Baseline Charlson Comorbidity Index	0.97	0.82, 1.13	0.67	0.94	0.77, 1.15	0.54	0.83	0.68, 1.03	0.08
Baseline Functional Comorbidity Index	0.83	0.63, 1.09	0.17	0.95	0.71, 1.27	0.73	1.01	0.75, 1.37	0.94
Baseline impaired PF**	0.44	0.18, 1.06	0.06	0.65	0.26, 1.66	0.36	1.13	0.46, 2.79	0.78
Model 4
Baseline normed SF-36 PF	1.01	0.97, 1.04	0.75	1.00	0.96, 1.04	0.89	0.98	0.93, 1.02	0.31
Last visit normed SF-36 PF	1.05	1.01, 1.10	0.02	1.03	0.98, 1.08	0.30	1.07	1.01, 1.13	0.02
Model 5
Last visit new anxiety symptoms	-	-	-	0.63	0.21, 1.96	0.42	0.63	0.19, 2.07	0.44
Last visit existing anxiety symptoms	-	-	-	0.59	0.17, 2.03	0.39	0.69	0.21, 2.24	0.53
Last visit remitted anxiety symptoms	-	-	-	-	-	-	0.29	0.03, 2.91	0.28
Model 6
Last visit new depressive symptoms	0.25	0.07, 0.82	0.02	-	-	-	0.81	0.28, 2.35	0.69
Last visit existing depressive symptoms	0.55	0.17, 1.71	0.29	-	-	-	0.77	0.26, 2.32	0.64
Last visit remitted depressive symptoms	2.00	0.52, 7.6	0.30	-	-	-	0.57	0.06, 5.61	0.63
Model 7
Last visit new PTSD symptoms	0.82	0.31, 2.16	0.69	0.55	0.17, 1.81	0.32	-	-	-
Last visit existing PTSD symptoms	0.50	0.15, 1.67	0.25	1.32	0.38, 4.58	0.65	-	-	-
Last visit remitted PTSD symptoms	0.44	0.05, 3.85	0.45	-	-	-	-	-	-
Model 8
Baseline impaired PF**	0.51	0.21, 1.24	0.13	0.55	0.22, 1.42	0.21	1.08	0.44, 2.60	0.86
Last visit new impaired PF**	0.53	0.16, 1.74	0.29	1.22	0.24, 6.34	0.81	0.90	0.24, 3.41	0.88
Last visit existing impaired PF**	0.53	0.17, 1.69	0.27	2.70	0.54, 13.5	0.22	1.12	0.34, 3.67	0.84
Last visit remitted impaired PF**	1.45	0.40, 5.19	0.56	0.94	0.07, 12.7	0.96	1.21	0.23, 6.27	0.82
Model 9
Baseline impaired PF**	0.57	0.23, 1.41	0.22	0.61	0.24, 1.56	0.29	1.16	0.48, 2.80	0.74
Last visit # of IADL dependencies	0.84	0.71, 0.99	0.04	1.06	0.90, 1.26	0.47	0.89	0.75, 1.05	0.15
---
Model 10
Last visit # of IADL dependencies	0.87	0.74, 1.02	0.08	-	-	-	-	-	-
Last visit new depressive symptoms	0.29	0.09, 0.98	0.04	-	-	-	-	-	-
Last visit existing depressive symptoms	0.54	0.15, 1.91	0.33	-	-	-	-	-	-
Last visit remitted depressive symptoms	2.36	0.58, 9.61	0.22	-	-	-	-	-	-
Model 11
Last visit SF-36 PF	1.05	1.01, 1.09	0.01	-	-	-	-	-	-
Last visit new depressive symptoms	0.31	0.09, 1.08	0.06	-	-	-	-	-	-
Last visit existing depressive symptoms	0.89	0.26, 3.04	0.86	-	-	-	-	-	-
Last visit remitted depressive symptoms	2.57	0.63, 10.6	0.18	-	-	-	-	-	-

CI, confidence interval; IADL, instrumental activities of daily living; OR, odds ratio; PF, physical function; PTSD, posttraumatic stress disorder;

^*calculated using multivariable discrete time survival models – bold indicates statistical significance;

^**≥2 IADL dependencies

Discussion

In this multi-site, 2-year prospective longitudinal cohort study of 186 ALI survivors, we found high levels of co-occurrence of general anxiety, depression, and PTSD symptoms, relatively low levels of remission, and that better physical function predicted subsequent remission from general anxiety and PTSD symptoms.

Supra-threshold general anxiety, depression, and PTSD symptoms co-occurred frequently; indeed, 59% of those with any supra-threshold symptoms had them in 2 or more symptom domains, and having supra-threshold symptoms in more domains was associated with more mental health treatment. Our co-occurrence findings are consistent with others’ recent findings in critical illness survivors (40–42). These findings are also consistent with our clinical experience with these patients (i.e., ALI survivors referred for one psychiatric illness often meet diagnostic criteria for at least one other) and underscore the need for clinicians to assess and treat their patients for a spectrum of distress-related psychiatric phenomena if any are evident.

Supra-threshold general anxiety, depression, and PTSD phenomena were long-lasting. Most ALI survivors with any supra-threshold symptoms during 2-year follow-up had them at the last follow-up visit; this was largely due to lack of remission, but it was also due in part to recurrence by the last follow-up visit. The most common supra-threshold symptoms were general anxiety symptoms (point prevalences 38 to 44%) – this is a new finding to add to our prior observations of the long-lasting/recurrent nature of depression and PTSD symptoms in this cohort (12, 13). Notably, remissions, especially from PTSD symptoms, were more common in patients who survived to 2-year follow-up, and particularly in the 52% of initial survivors who were able to participate in all follow-up visits.

Surprisingly, baseline demographic characteristics like older age, female sex, unemployment, and lower socioeconomic status did not adversely affect remission from general anxiety, depression, or PTSD symptoms, contrary to our expectations based on the general psychiatric literature (26–28). Further, baseline (pre-ALI) mental and general health characteristics did not predict remission, also contrary to our expectations (27, 28). Thus, prior experience regarding remission in other patient populations may not be generalizable to ALI survivors. The only robust predictor of remission was higher SF-36 physical function domain scores at survivors’ previous visit. With a five-point (i.e. half a standard deviation) higher normed SF-36 physical function score, survivors had a 28% greater odds of remission from generalized anxiety or PTSD symptoms. These results underscore the need to treat the whole patient, including a focus on both physical and mental health sequelae of critical illness (7); such attention may help ALI survivors recover more quickly. Improvements in physical functioning may promote a sense of independence/ability, as well as an increased capacity for exercise, which could positively affect emotional well-being (43, 44). Importantly, in-ICU physical and occupational therapy appears to reduce the duration of delirium and improve patients’ functional outcomes at hospital discharge (45); future studies of in-ICU rehabilitation should examine whether such early interventions improve survivors’ long-term mental health as well.

In the current study, we did not address whether or not ALI/intensive care may be causally related to general anxiety, depression, or PTSD symptoms during follow-up. Results of other studies that used identical measures of psychiatric symptoms before and after critical illness may be of interest in this regard. In a prospective longitudinal study of patients with diabetes (the Pathways Epidemiologic Follow-up Study), Davydow and colleagues found that having a critical illness/ICU stay was independently associated with depressive symptoms at follow-up (46), but having depressive symptoms at baseline also increased the risk of an ICU stay during follow-up (47). In a population-based cohort study of older adults (part of the Health and Retirement Study), the prevalence of supra-threshold depressive symptoms was similar at a median of 1.2 years before and 0.9 years after an episode of severe sepsis (48). Nevertheless, in a very large cohort study using linked data from Danish national population-based medical databases, Wunsch and colleagues found that having a psychiatric diagnosis in the prior 5 years was associated with an increased risk for critical illness (prevalence ratio 2.7), but having a critical illness increased the risk of a new psychiatric diagnosis or psychiatric medication very substantially (hazard ratios >20) (49). Thus, psychiatric symptoms may be both a risk factor for and a result of critical illness.

Several potential limitations should be considered with these results. First, we measured general anxiety, depression, and PTSD symptoms using validated questionnaires (19, 22), rather than clinical interviews incorporating multiple sources of mental health information. Because of the added respondent burden (which may have required up to 2 hours at each follow-up visit), we believe that clinical interviews would have exceeded the limits of feasibility and resulted in substantially higher losses to follow-up, especially during the first year when patients were still in early recovery and required 3 follow-up visits. Moreover, obtaining clinician diagnoses and symptom ratings also would have been logistically difficult, given the need for expert clinicians to be physically present in patients’ homes or long-term care facilities (58% of participants required at least one such visit during 2-year follow-up). Second, we used medical records to identify baseline health problems. Anxiety disorders, in particular, may be under-recognized in medical records; i.e., in general population studies, anxiety disorders are the most common mental illnesses (50, 51), but in our study anxiety disorders were recorded much less frequently than depressive disorders (6% versus 21%, respectively). Such potential bias is generally unavoidable given the infeasibility of directly assessing patients’ psychiatric history prior to ALI onset. Third, we did not take into account the possible salutary effect of treatment on symptoms, and we may have missed symptom onset or remission before 3-month follow-up or in between follow-up visits. Though we collected basic information on receipt of psychiatric treatment retrospectively at 2-year follow-up, we chose not to analyze this data as a predictor of course of psychiatric symptoms, since we were concerned about confounding by indication, as well as our lack of information regarding specific treatments, timing, duration, etc. Fourth, although we statistically controlled for potential confounders in our analysis of remission predictors, residual confounding could have influenced our findings. Nevertheless, it is not possible to randomize patients to better mental or physical function during the recovery period; thus, observational research, like this study, provides essential information regarding such associations. Fifth, we did not measure social support during patients’ critical illnesses or during 2-year follow-up; strong social support may mitigate the emotional effects of a severe stressor like ALI (52, 53). Sixth, we limited our focus to survivors of the archetypal critical illness ALI (54), and our results may not generalize well to all populations of critical illness survivors (1–3). Finally, future studies should examine a broader range of psychiatric symptoms. For example, many patients surviving a critical illness develop substantial illness anxiety (55); our clinical experience suggests that many patients develop compulsive cleaning rituals and avoidance of germs and irritants that contribute to overall dysfunction.

Conclusions

In this multi-site, 2-year prospective longitudinal cohort study of 186 ALI survivors, we found high levels of comorbidity of general anxiety, depression, and PTSD symptoms, relatively low levels of remission, and that better physical function predicted subsequent remission from general anxiety and PTSD symptoms. When caring for ALI survivors, clinicians should address the full spectrum of potential mental and physical health sequelae to maximize patient recovery.