Introduction
Intensive care has grown substantially over the past decades worldwide [
1‐
3] to be one of the most resource-intensive acute hospital services [
4]. Heavy utilization of the intensive care unit (ICU) over the disease course contributes to high costs of health care [
1] and raises concern over optimal use of ICU resources [
5], especially for end-of-life (EOL) care, which has increased in the last decade [
6]. Landmark studies highlighted the improving but still poor quality of EOL care in ICUs [
7‐
9]. Thus, improving EOL-care quality in ICUs is essential [
10] for improving the quality of death and dying [
11], facilitating bereavement adjustment among family members [
10], and counteracting unsustainable ICU care expenditures [
10,
11].
Family members of ICU patients are an integral part of ICU care and a critical target for improving EOL-care quality in ICUs [
12,
13]. The uncertain trajectory of critical illnesses, the frightening nature of aggressive life-prolonging treatments [
14], and the beloved’s unexpected and eventual death are traumatic events for family members of ICU patients [
15]. Therefore, family members are at increased risk for post-intensive care syndrome (PICS-F)—new or worsening impairments in physical, cognitive, or mental health status arising after a beloved’s critical illness and persisting beyond acute care hospitalization [
16], including posttraumatic stress disorder (PTSD) [
14]. PTSD takes a toll on physical [
17,
18] and mental [
17‐
19] health, personal relationship/social functioning [
20] and poses a considerable economic burden for individuals, health care systems, and societies [
21,
22]. These striking characteristics make PTSD a public mental health priority [
19].
A better understanding of the modifiable risk factors for developing posttraumatic stress-related symptoms (hereafter as PTSD symptoms) may allow health care professionals to preemptively identify family members who are vulnerable to PTSD symptoms during bereavement and develop actionable high-quality, family-centered EOL care in ICUs. However, most studies [
14] that examined potential predictors of PTSD symptoms among family members of ICU patients were done primarily on those of ICU survivors [
9,
14,
23‐
27], while only a handful were done on those of deceased patients [
7‐
9,
28‐
34]. Furthermore, only a few of these studies found significant associations, and the majority of them focused on immutable family and patient characteristics [
7,
9,
14,
23‐
26,
28,
29]. From the few [
28,
29,
33] that examined process-based quality indicators of EOL care in ICUs, it was found that participation in early family meetings and presence at the time of patient’s death were associated with worse PTSD symptoms among family members [
28], whereas withdrawal of life-sustaining treatments (LSTs) had no association [
29,
33]. The association between family’s perception/satisfaction of the quality of death in ICUs and PTSD was inconclusive [
8,
9,
25,
26,
30].
Studies on PTSD among family members of ICU patients [
7‐
9,
23‐
34] also remain limited by methodological insufficiencies. First, most studies were cross-sectional [
7‐
9,
23,
25,
27‐
29,
32], with PTSD symptoms measured across a wide range of time (1–9 months post-loss), which makes it difficult to compare prevalence of PTSD across studies and precludes the possibility of establishing temporal associations with predictors. Second, while most of the six longitudinal studies [
24,
26,
30,
31,
33,
34] assessed PTSD symptoms at 3 [
30,
31,
33,
34] and 6 [
26,
30,
31,
33,
34] months post-loss, only few studies had data at 1–2 [
24,
26] and 12 [
30,
31,
33] months post-loss. Third, sample sizes tended to be small (
N = 30–475 and five [
8,
23,
25,
26,
29] out of 15 studies had < 100 subjects), limiting the power to detect potential predictors of PTSD symptoms. Finally, none were from Asian countries which may have different cultural reactions toward grief and only one study accounted for family members’ preexisting mental health and medical conditions [
7]. Thus, this study aimed to comprehensively examine the course and predictors of PTSD over the first bereavement year among a large cohort of Taiwanese family members whose beloved died in the ICU. This study specifically focused on investigating the modifiable factors that contribute to EOL-care quality in ICUs.
Discussion
In this longitudinal study, the prevalence of clinically significant PTSD symptoms among family members of deceased ICU patients decreased significantly over the first bereavement year (from 11.0% at 1 month to 1.6% at 13 months post-loss). Even though a similar trend was seen in prior studies [
30,
31,
33], the observed prevalence of clinically significant PTSD symptoms in our study was substantially lower than that reported in other longitudinal (19–46.2% at 3–12 months post-loss) [
30,
31,
33,
34] and cross-sectional studies (14–51% at 3–18 months post-loss) [
7,
28,
29,
32]. Of note, the reported prevalence of clinically significant PTSD symptoms measured by self-reported screening instruments (including the IES-R) was 20.2% (95% CI 16.6–24.0%) among adult critical care survivors [
43], 3.3–35.1% among burn survivors during hospitalization [
44], 7.3% (4.5–11.7%)–13.8% (9.5–19.6%) among cancer survivors [
45]. All these populations experience traumatizing diseases and treatments. Our observed prevalence of clinically significant PTSD symptoms was closer to other reports: 3–62% among family caregivers of ICU survivors [
14] and 11.6% (9.6–11.6%) among parents of children with cancer [
46].
The lower rates of clinically significant PTSD symptoms among the bereaved family caregivers in our study, as compared to those reported in studies from Western countries [
7‐
9,
28‐
34], may be due to the cultural difference in grief reactions between Taiwanese and Western cultures. In Taiwanese culture, caring for an ill relative is viewed as a natural part of family life based on the concept of filial duty rooted in Confucian doctrines [
47]. Family members in Asian cultures, which are more family-oriented, are willing to help each other adjust to the loss of a beloved by providing stronger financial and emotional support [
48] than those from Western cultures, which are more individualistic. Thus, bereaved Taiwanese family members’ risk of significant PTSD symptoms may be lower.
The high rate (50.2%) of cancer patients in our study may be another reason for our low prevalence of significant PTSD symptoms. Azoulay et al. report cancer diagnosis as a factor predisposing family members to more severe PTSD symptoms [
9]. However, this finding may be confounded by other factors, such as patient’s death in the ICU and the need for family to make EOL-care decisions, which were both associated with higher rates of PTSD [
9]. The terminal-illness trajectory for cancer patients is commonly anticipated and predictable than the sudden and unexpected illness trajectory from cardiovascular or pulmonary diseases. In our study, family members of cancer patients had a lower likelihood of significant PTSD symptoms than those of patients with non-cancer diagnoses. However, this did not reach a statistical significance, likely due to the low prevalence of significant PTSD symptoms in our study. The specific characteristics of dying and death from cancer may reduce the overwhelming and uncontrollable trauma due to a beloved’s death.
We intended to focus on associations between quality indicators of EOL care in ICUs and clinically significant PTSD symptoms among family members during the first bereavement year. We did not find many significant associations for either process-based indicators [
28] of or family satisfaction with EOL care in ICUs [
8,
25,
26], which is consistent with prior studies. We observed low family satisfaction with care (mean [SD] = 67.14 [16.30]) and decision-making process (72.60 [14.76]) in ICUs which may be attributable to different expectations of care in differing health care contexts as reported [
49], but further cross-country and -cultural comparisons of SF-ICU are warranted.
However, having a DNR order documented before death was associated with a lower likelihood of clinically significant PTSD symptoms among family members (AOR [95% CI] = 0.073 [0.011, 0.490]). Taiwan’s government has launched multiple nationwide projects to facilitate dissemination of hospice philosophy and palliative care services over the past three decades with the help from non-governmental organizations. The goal of these efforts was to improve the quality of EOL care, with an emphasis on promoting DNR orders and avoidance of LSTs that do not benefit patients at EOL. The Taiwanese government promotes DNR orders by legislation and by educating health care professionals. This facilitates prognostic disclosure and respects patient’s and family’s wishes to forgo LSTs, minimizing suffering and allowing patients to die in peace. Therefore, having a DNR order may help prevent bereaved family members from experiencing a traumatized loss and developing significant PTSD symptoms.
However, despite the high prevalence of having a DNR order documented before death, Taiwanese patients dying in ICUs heavily used life support until death (79.31%) with few decisions to withdraw LSTs (20.69%). This paradox may be attributable to Taiwanese physician and family factors in withdrawing LSTs. ICU physicians in Asia, including in Taiwan, report being less likely than Western physicians to withdraw LSTs, but more likely to “do everything” for a patient with an irreversible condition like severe septic shock [
50]. Most importantly, Taiwanese family members have a strong cultural obligation based on the Confucian doctrine of filial duty [
47] to use every possible means to keep a beloved alive. Therefore, even if family members were fully informed of their beloved’s poor prognosis and were advised by physicians to withdraw LSTs, they might hesitate to initiate or accept the suggestion, especially for those whom withdrawal is interpreted as inducing immediate death, e.g., mechanical ventilation support or vasopressors. Such physician-family disagreement on the use of LSTs may partially contribute to our observed low satisfaction with care and decision-making in ICUs but warrants further cross-cultural investigation.
ICU length of stay was the only patient factor that was associated with bereaved family members’ likelihood of significant PTSD symptoms. This was contrary to the lack of association previously reported [
24]. However, McGiffin and colleagues [
15] suggested that if the ICU environment was itself a traumatic stressor, then incremental increases in ICU exposure (e.g., longer ICU stays) may be expected to approximate a dose–response relationship to worse psychological outcomes. Our study provides empirical support to this notion: spending more time in an ICU increased family members’ likelihood of significant PTSD symptoms during the first bereavement year.
We found that reported financial insufficiency increased family member’s risk of significant PTSD symptoms [
14,
33], whereas stronger perceived social support decreased the risk. Furthermore, family members who reported receiving medical care in the year prior to the patient’s critical illness were generally more likely to develop significant PTSD symptoms, consistent with a prior study [
7]. This trend reached significance in family members who reported using medications for pain, which is dissimilar from Gries and colleagues’ finding [
7] that use of medication for mood is positively associated with PTSD symptoms. Our finding may be attributable to the fact that Asian people tend to somatize their emotional distress by reporting pain [
51] to avoid stigmatization from psychiatric disorders [
52]. Further investigation of PTSD symptoms across cultures and socioeconomic class is needed.
Our study has several limitations. Participants were sampled from only two hospitals in Taiwan, which may limit generalizability of our findings to national/international populations, especially considering cultural variations in grief reactions during bereavement in Western and Eastern countries [
47]. Our findings need to be replicated in countries where cultural, societal, and health care characteristics are substantially different. Our results do not apply to family members of patients who died within 3 days of ICU admission, or to those who withdrew from post-loss surveys. PTSD symptoms were measured with the IES-R, which is not diagnostic of PTSD, thereby likely overestimating the prevalence of PTSD. However, this characteristic also helps prevent providers from overlooking the family’s need for psychological support. Our study cannot infer a causal relationship or exclude the possible impact of unmeasured factors, e.g., concordance between family members’ preferred and actual decision-making roles and physician-family agreement on appropriateness of treatment which may be more powerful than the current identified indicators of high-quality EOL care in ICUs in association with bereaved family members’ PTSD symptoms. We did not include anxiety or depressive symptoms as potential predictors but we will explore associations among these three types of psychological distress in a forthcoming study.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.