Search results and study selection are depicted in the study flow diagram (Fig. 1).
The initial search returned 11,243 articles and the updated search returned 992 articles, yielding 12,235 articles. Of these, 3,084 were duplicates, leaving 9,151 articles. Nine thousand forty-one were excluded on examination of their abstracts. Articles excluded at this stage included non-human studies, commentaries, letters, and studies in which patient participants were < 18 yr of age.
The remaining 110 articles were assessed for eligibility by examination of their full-text format. Of these, 98 were excluded. Reasons for exclusion were as follows: 41, no intervention tested; 15, outcomes of interest not addressed; three, systematic reviews (these were examined for additional eligible studies and none were identified); one, duplicate study; one, pre-hospital study; 37, intervention did not have substantial potential to increase physical or emotional proximity to the patient. Details of excluded interventional studies are provided (Table 2). 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81 The remaining 12 studies were included in the review.82,83,84,85,86,87,88,89,90,91,92,93
Summary of included studies
The characteristics of included studies are described (Table 3) with further details provided
(available as ESM; Appendix 3). Only two were randomized-controlled studies,82,83 five were pre- and post-intervention cohort studies,84,85,86,87,88 four were prospective cohort studies with non-randomized controls,89,90,91,92 and one was a retrospective study.93 Four discrete humanizing interventions were assessed: diaries,83,84 liberalization of visitation,87,88,89 witnessed resuscitation,82,92,93 and family participation in basic care.85,90,91 One study of a mixed intervention with both humanizing and non-humanizing initiatives was included on the basis that some of the interventions tested had potential to increase physical and emotional proximity to the patient (liberal visitation and family involvement in basic care).86 Ten studies measured anxiety among a total of 1,055 family members.82,83,84,85,88,89,90,91,92,93 Two studies measured burnout among a total of 288 critical care professionals86,87 No studies addressed empathy or compassion fatigue. Two studies of witnessed resuscitation sought unspecified adverse psychologic effects among relatives and reported none.82,93
Risk of bias in individual studies
Eleven of the 12 included studies had an overall high risk of bias.83,84,85,86,87,88,89,90,91,92,93 Details of risk of bias assessment for each study are provided (Table 3) (Fig. 2). Commonly occurring themes were lack of randomization, lack of valid control groups, unclear outcome priorities, and failure to provide sample size justification.
Study quality in individual cohort studies
For the ten included cohort studies, none achieved a high-quality rating; five studies received a score of 5,84,85,87,90,93 putting them at the lower end of a moderate rating, and five had a low quality rating86,88,89,91,92 (Table 4).
No eligible studies were found that measured empathy or compassion fatigue in healthcare professionals or relatives.
Burnout among healthcare professionals and relatives
Two studies addressed burnout among healthcare professionals; none addressed burnout among relatives.86,87 Both studies used the Maslach-Jackson Burnout Inventory (MBI) to measure burnout.94 Gianni studied liberal visitation (a minimum of eight hours a day).87 Locally organized staff training sessions were provided at each centre before the change. Burnout among healthcare professionals was measured before and at six and 12 months after the policy change. Baseline levels of burnout were higher among nurses than physicians. A small but significant increase in burnout levels was seen in the year following the policy change, with the increase being greater for nurses than physicians. Of note, staff perceptions of liberal visitation reflected burnout levels with those with high burnout scores expressing more negative opinions than those with low burnout scores.
Quenot also used a before-and-after cohort study to examine the effects of an ‘intensive communication strategy’ that included unrestricted visiting hours, greater family involvement in basic patient care, educational sessions, more frequent family meetings, staff debriefing, role playing, and working groups.86 The strategy was designed to improve communication among healthcare professionals, patients, and families and was designed in response to information obtained by psychologists, who interviewed staff beforehand to identify specific areas for intervention and who remained available for consultation during the study period. The prevalence of severe burnout and depression among staff decreased significantly after the intervention, with all three components of the MBI (emotional exhaustion, depersonalization, and personal accomplishment) showing a significant change. As both studies had an overall high risk of bias and investigated interventions with important differences in their content, a pooled estimate of effect was not calculated.
Anxiety among relatives of the critically ill
Ten studies addressed this outcome.82,83,84,85,88,89,90,91,92,93
Of these ten studies, two measured the effects of liberal visitation on anxiety among relatives,88,89 using the State-Trait Anxiety Inventory.95 Only one reported the actual figures for this outcome and the timing of outcome measurement.89 One study found a significant decrease89 and the other a non-significant decrease in anxiety.88
Two studies evaluated the effects of family-maintained diaries on anxiety among relatives.83,84 One measured the prevalence of anxiety at three months after patient discharge from critical care, using a score of > 8 on a self- administered questionnaire as an indicator of anxiety.84 The other measured mean anxiety levels on the third postoperative day, using the State-Trait Anxiety Inventory.83 Both found a non-significant decrease in anxiety in the diary groups.
Three studies measured the effects of family involvement in basic patient care on anxiety among relatives.85,90,91 Two used the State-Trait Anxiety Inventory,85,90 and one used the Hospital Anxiety and Depression Scale.91 Timing of outcome measurement was reported by only one study.91 All reported significant results.
Three studies investigated the effects of family presence during resuscitation on anxiety among relatives.82,92,93 Two used the State-Trait Anxiety Inventory,92,93 and one used the Hospital Anxiety and Depression Scale and Beck Anxiety Inventory.82 Timing of outcome measures ranged from 48 hr to nine months. None reported significant results.
No pooled estimates of effect were calculated for this outcome as no two studies (each with less than a high risk of bias) of any of the included interventions measured the same outcome using comparable scales at comparable time points.
No adverse events were reported by the two studies that sought them; hence, no pooled estimate of effect was calculated.82,93
Overall quality of evidence
The overall quality of the available evidence was poor with evidence being either low or very low quality. Details and reasons why the evidence was downgraded are provided (summary of findings in Table 5).