We conducted a multicentre prospective observational cohort study as an a-priori planned sub study to a larger study , with four participating centres in tertiary care hospitals in Sweden. Adult cohabitants to ICU patients included in the larger study were approached and asked about study participation when visiting the ICU at times when research staff was available. In brief, patients ≥18 years of age and with ICU length of stay (LOS) ≥12 hours were eligible for inclusion in the larger study during January to April 2016. Study participants were excluded if they could not understand written Swedish. Patients were followed up with validated questionnaires regarding their physical and psychological state three months after the ICU stay.
The Stockholm ethical review board approved the current study (registration number 2015/1799-31), which was performed in accordance with the 1964 Declaration of Helsinki and its later amendments. All study participants (hereafter referred to as informal caregivers) and patients gave informed consent. The study was registered at clinicaltrials.gov, NCT02712541.
Data collection in-ICU
Patient data collection in-ICU consisted of age, sex, admission diagnosis (medical, surgical or trauma), pre-ICU physical function (assessed with the Barthel Index (BI)), severity of disease (APACHE II score), occurrence and duration of invasive mechanical ventilation and ICU length of stay (LOS).
Data on informal caregivers’ age, sex and relation to the patient were collected. Informal caregivers were also asked to compile the RAND-36 at inclusion, with their status two weeks prior to hospitalization of the patient in mind, as a proxy for baseline HRQL. The RAND-36 is a validated 36-item questionnaire assessing eight domains regarding HRQL. The domains can be divided into two component scores, the mental health component score (MCS) and the physical health component score. A higher score indicates better HRQL .
Data collection at follow-up
Three months after ICU discharge questionnaires were sent by postal mail to patients and their informal caregivers in parallel.
Patients received questionnaires assessing physical and psychological problems. Physical problems were assessed with the Barthel Index (BI), a 10-item questionnaire with a maximum score of 100, assessing the independency in performing activities of daily living . Depressive symptoms and anxiety were assessed with the Hospital Anxiety and Depression Scale (HADS) consisting of 14 validated questions assessing depression (7 items) and anxiety (7 items). Each question generates a score from 0-3, a higher score indicating more severe problems . Symptoms of post-traumatic stress (PTS) were assessed with the Post-traumatic Stress Syndrome 14-Questions Inventory (PTSS-14) part B, a 14-item validated questionnaire. Each item generates a score from 1 (never) to 7 (always) .
Informal caregivers received the Caregiver Burden Scale (CBS), a 22-item questionnaire assessing the perceived caregiver burden among people caring for disabled family members. Each item generates a score between 1 and 4, a higher score indicating a higher degree of burden . Questions cover practical as well as emotional issues, such as general strain, isolation, disappointment, emotional involvement and environment. They also received the RAND-36, the HADS and PTSS-14.
Two weeks after the questionnaires were posted, non-responders received a reminder phone call. A new set of questionnaires was posted if participants did not respond to the call.
Definition of patient outcomes
Patients were considered to have adverse physical and/or psychological outcome if they scored above the pre-defined cut-off in either the physical or psychological follow-up questionnaires. A ≥10 point reduction in the BI total score compared to baseline was considered an adverse physical outcome, and is close to the suggested minimal clinically important difference for the BI. In the HADS questionnaire, a subscale score ≥11 has been suggested indicative of substantial symptoms of depression or anxiety. In the PTSS-14 part B a score >45 is indicative of substantial symptoms of posttraumatic stress disorder . A score above cut-off in any of the two HADS subscales, and/or the PTSS-14 part B was considered an adverse psychological outcome.
The primary outcome was caregiver burden, assessed with the CBS, in relation to adverse psychological or physical patient outcome. Secondary outcomes were prevalence of symptoms of anxiety, depression or posttraumatic stress, measured with the HADS and the PTSS-14 among informal caregivers and associations between caregiver burden and informal caregivers’ psychological symptoms and mental HRQL (measured with RAND-36). Further, informal caregivers were asked to state the number of days they had refrained from any planned activities, and the number of sick-days due to taking care of their cohabitant ICU survivor. They were also asked an open-ended question regarding additional measures they believed could have been helpful in order to contribute to the patient’s full recovery.
Sample size estimation
An estimated difference of 10 points in CBS mean scores between informal caregivers to patients with an adverse outcome (CBS score 50, standard deviation (SD) ±15) and informal caregivers to patients without an adverse outcome (CBS score 40, SD± 15) would render a sample size of 50. With an estimated loss to follow-up of 40%, each study centre was asked to include 35 participants, rendering follow-up data on a total of 84 caregivers.
Statistical analysis methods
All statistical analyses were performed with STATA version 12.1 (StataCorp, College Station, TX, USA). Normally distributed data were presented with means and standard deviations and non-parametric data with medians and interquartile ranges (IQR). Categorical data was summarized with numbers and percentages. Two-sided significance level was set to 0.05. For comparisons of non-parametric data, the Mann Whitney’s U test was used. For comparison of categorical data the Chi-square test or Fischer’s exact test were used. Linear regression with robust estimates for the standard errors, also adjusting for possible confounding by informal caregivers’ sex and age was used to determine mean differences (MD) with 95% confidence intervals (CI). Correlation between non-parametric continuous data was analysed with Spearman’s rank correlation coefficient. No imputation was made for missing data.