Mental health symptoms in family members of COVID-19 ICU survivors 3 and 12 months after ICU admission: a multicentre prospective cohort study

This was a prospective multicentre cohort study conducted in ICUs of ten hospitals in the Netherlands (three university, four teaching and three non-teaching hospitals; most participating ICUs were cohort ICUs for the treatment of COVID-19 patients only and since temporary ICUs were created, for instance in the recovery or the paediatric ICU, the exact number of ICUs cannot be reported). The study was approved by the local research ethics committee (CMO region Arnhem-Nijmegen, the Netherlands, number 2020-6878). For the recruitment of family members, 254 COVID-19 ICU patient survivors, who were alive until hospital discharge, of an already established cohort as part of the MONITOR-IC study [13] were approached. These patients were admitted during the first COVID-19 surge in the Netherlands between March 1, 2020 and July 1, 2020 and were enrolled as soon as possible after ICU discharge either before hospital discharge or shortly after. There was no prespecified time point when patients had to be enrolled, but all were included within the first weeks after hospital discharge. After inclusion, the patient was asked for the contact information of the family member most closely involved during the ICU admission, who was approached subsequently. We excluded survivors and family members who were younger than 16 years. Informed consent was obtained from all ICU survivors and family members.

Family members received questionnaires at three time points: the first questionnaire regarding their perceived mental health before ICU admission (T0) immediately after inclusion and was therefore answered in retrospect, and the second (T1) and third (T2) questionnaire 3 and 12 months after ICU admission, respectively. Questionnaires could be completed on paper or online, depending on the family members’ preference. Non-responders received two reminders for all questionnaires.

Mental health outcomes were symptoms of anxiety, depression and post-traumatic stress disorder (PTSD). For symptoms of anxiety and depression, the Hospital Anxiety and Depression Scale (HADS) was used, consisting of 14 questions with a 4-point Likert scale (0–3) [14]. The HADS has two 7-question subscales (HADS-Anxiety [HADS-A] and HADS-Depression [HADS-D]) and a cut-off score of ≥ 8 on each subscale is regarded as indicative for the presence of symptoms of anxiety and depression, respectively.

For symptoms of PTSD, the abbreviated six-item version of the Impact of Event-Revised (IES-R) was used, with a five-point Likert scale (0–4) [15]. The total score was divided by the number of questions resulting in a mean score with a cut-off value of ≥ 1.75 indicating presence of symptoms of PTSD [16].

QoL was evaluated using the 12-item Short Form Health Survey (SF-12), derived from the SF-36, providing a Physical Component Summary (PCS) and a Mental Component Summary (MCS) score [17]. Both summary scores range from 0 to 100 and a higher score indicates a better perceived QoL. Additionally, work-related problems as result of the ICU admission were assessed by the question: ‘Did your work situation change as a direct consequence of the ICU admission of your relative?’ [18].

All questionnaires are validated in Dutch, recommended [19] and numerously used in family members of COVID-19 and non-COVID-19 ICU survivors [11, 17, 20‐22]. Demographics of the family members (e.g. age, education level and relation to the survivor) and ICU experiences (e.g. visiting allowance and the use of an ICU diary) were assessed as well. Additionally, family members were asked if they experienced symptoms of COVID-19 during the ICU admission of their relative and if hospital or ICU admission was required for these symptoms. Demographics of the ICU survivors and ICU admission characteristics, i.e. severity of illness, length of ICU and hospital stay, were retrieved from the NICE (Dutch National Intensive Care Evaluation) registry and medical records [23].

Associations between potential risk factors, i.e. baseline characteristics of ICU survivors and family members, ICU admission characteristics and ICU experiences, and the presence of mental health symptoms were analysed using logistic regression analysis and reported as crude odds ratios (ORs) in univariable regression and as adjusted ORs (aOR) in multivariable regression analyses, both with 95% confidence intervals (CIs). Due to the sample size, not all independent variable could be tested in multivariable analysis and therefore only factors with a p value < 0.20 in the univariable analysis were included in the multivariable regression analysis [24].

Only family members who completed the baseline and 12-month follow-up questionnaire were included in the analyses. Missing data in the HADS and SF-12 questionnaires were imputed using a subject’s means score if at least half of the items were answered (the half-rule). Missing data in the IES-6 were replaced with the mean score if only one of the six answers was missing [25, 26]. All statistical tests were two-sided and statistical significance was defined as a p value < 0.05. Data were analysed using IBM SPSS version 25 (Chicago, IL, USA).

In total, 197 family members consented (77.6% of 254 eligible survivors and family members), of whom 166 (84.3%) completed both the baseline and 12-month follow-up questionnaire and were therefore included in the analyses (Fig. 1). Of those, 153 (92.2%) completed the 3-month follow-up questionnaire.

Family members had a mean age of 57.8 years (SD 11.4) and were mostly female (Table 1). ICU survivors had a mean age of 61.1 years (SD 9.3) and were mostly male. Their length of ICU and hospital stay were median 20 days (IQR 11–33) and median 31 days (IQR 21–46), respectively, and 83 (50.3%) survivors were transferred to or from an ICU in another hospital. During follow-up, one survivor died, of whom the family member wished to stop participation and, therefore, no bereaved family members were included in the analyses.

There were no differences in family member, patient and pre-ICU characteristics between responders and non-responders, except for the relationship to the patient as 86.1% of the responders was a spouse compared to 63% of the non-responders (p = 0.01) (Supplemental Table S1).

Before ICU admission and at 3- and 12-month follow-up, symptoms of anxiety were experienced by 28/165 (17%), 48/152 (31.6%) and 47/162 (29%) family members, respectively, and symptoms of depression by 26/165 (15.8%), 43/152 (28.3%) and 37/162 (22.8%), respectively (Fig. 2 and Supplemental Table S2). Figure 2 shows that the prevalence of symptoms of anxiety and depression was significantly higher at 12-month follow-up compared to baseline, but was the highest at 3-month follow-up. At 3- and 12-month follow-up, symptoms of PTSD were experienced by 45/152 (29.6%) and 33/163 (20.2%) family members, respectively.

Compared with baseline, family members had significantly more mental health symptoms at 3 and 12 months: a positive mental composite score in 22.4% versus 46.1% (p < 0.001) and 22.4% versus 38.3% (p < 0.001), respectively. Furthermore, 22.8% of all family members experienced at least two mental health symptoms (out of anxiety, depression or PTSD) and 11.1% experienced all three mental health symptoms 12 months after ICU admission (Supplemental Fig. S1). The longitudinal course of mental health symptoms during the first year is depicted in Supplemental Fig. S2, which shows only a small proportion of family members without symptoms at 3 months is experiencing symptoms at 12 months.

Family members had significantly higher scores on the HADS-A and HADS-D at 12-month follow-up than before ICU admission (Table 2). Individual changes in HADS scores from baseline to 12-month follow-up are presented in Supplemental Fig. S3. Additionally, family members’ PCS and MCS scores of the SF-12 were significantly lower at 12-month follow-up compared with prior to ICU admission [mean difference scores, PCS: − 2.4 (95% confidence interval (CI) 0.8 to − 4.0, p = 0.004) and MCS − 5.5 (95% CI − 7.4 to − 3.6, p < 0.001)]. A total of 104 family members worked before ICU admission and 12 months after ICU admission 29 (27.9%) experienced work-related problems (7 (6.7%) worked less hours than before ICU admission, 3 (2.9%) retired early, 8 (7.7%) were still on sick leave and 11 (10.6%) stopped working completely). Mental health symptom scores between 3- and 12-month follow-up show a tendency to decrease, but are not significantly different (Supplemental Table S3).

In the multivariable analysis, only symptoms of anxiety or depression before ICU admission were significantly associated with mental health symptoms in family members 12 months after ICU admission with an aOR for symptoms of anxiety of 9.23 (95% CI 2.29–37.24; p = 0.002) and for symptoms of depression of 5.96 (95% CI 1.29–27.42; p = 0.02) (Table 3). In the multivariable analysis for mental health symptoms after 3 months, only symptoms of anxiety before ICU were significantly associated (aOR 6.74; 95% CI 1.70–26.63); p = 0.007) (Supplemental Table S4).

This study has several limitations. First, only family members of ICU survivors were included, so no conclusion can be drawn regarding bereaved family members. Bereaved family members of non-COVID-19 ICU patients experience more mental health symptoms than family members of non-COVID-19 ICU survivors [38]. Additionally, family members who did not say goodbye or witnessed terminal dyspnoea experienced more symptoms as well. Therefore, it is likely the prevalence of mental health symptoms in bereaved family members of COVID-19 ICU patients is higher compared with our results [39]. Second, there was no control group of non-COVID-19 ICU family members to compare our outcomes with. Therefore, outcomes of this study were compared with previous studies. Third, due to the unplanned ICU admission, information regarding baseline mental health was assed retrospectively, potentially causing recall bias. However, this information is vital for the interpretation of long-term outcomes. Fourth, we only studied the mental health symptoms, while non-COVID-19 ICU studies suggest that family members may experience physical and cognitive symptoms as well [40]. Additionally, we had no information regarding the type of ICU, e.g. a traditional, a cohort ICU for COVID-19 patients only, or extra created ICUs (e.g. originally recovery or paediatrics unit) or medical comorbidities, e.g. respiratory or cardiovascular insufficiencies, of the family members, which could have impacted the development of mental health symptoms [41]. Fifth, although statistically significant, factors associated with mental health outcomes (pre-ICU mental health) have broad CIs. Therefore, the results of this exploratory study should be confirmed with larger sample sizes.