The Effects of Family Voice Intervention on Delirium in Adult Critical Care Patients: A Systematic Review and Meta‐Analysis

ABSTRACT

Background

Family voice interventions have emerged as a non‐pharmacological approach for delirium prevention; however, evidence regarding their effectiveness remains inconsistent.

Aim

The aim of this study was to evaluate the effectiveness of family voice interventions in reducing delirium among adult intensive care unit (ICU) patients.

Study Design

We conducted a systematic review and meta‐analysis. A comprehensive literature search was conducted across PubMed, EMBASE, CINAHL and the Cochrane Library from inception through 4 March 2025. Study quality was assessed using the Cochrane Risk of Bias II tool, with analyses performed using Comprehensive Meta‐Analysis version 4.0.

Results

Six randomised controlled trials (reported across seven publications) involving 598 adult ICU patients were included. Family voice interventions significantly reduced delirium incidence (OR = 0.282, 95% CI: 0.126–0.630, p = 0.002) and duration (Hedges' g = −1.929, 95% CI: −2.980 to −0.879, p < 0.001). No significant effects were observed for mechanical ventilation duration (Hedges' g = −0.826, 95% CI: −1.696 to 0.044, p = 0.063) or ICU length of stay (Hedges' g = −0.534, 95% CI: −1.371 to 0.303, p = 0.211). Subgroup analysis showed multicomponent interventions significantly reduced both delirium incidence (OR = 0.178, 95% CI: 0.046–0.681, p = 0.012) and duration (Hedges' g = −1.888, 95% CI: −2.917 to −0.859, p < 0.001), while single‐component interventions showed no significant effects.

Conclusions

Family voice interventions effectively reduce delirium incidence and duration in adult ICU patients, with multicomponent approaches demonstrating superior efficacy.

Relevance to Clinical Practice

Critical care nurses may incorporate family voice interventions as a practical alternative within delirium prevention strategies. Although direct family presence is generally regarded as the preferred form of engagement, recorded voice messages can provide a feasible and meaningful option to enhance patient orientation and emotional connection when families cannot always be present.

Protocol Registration

The study protocol was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO: CRD42025628074)

Impact Statements.

• What is known about the topic

  • ○
    Delirium is common among adult ICU patients and is associated with serious adverse outcomes, including prolonged hospitalisation, increased mortality and long‐term cognitive decline.
  • ○
    Although no pharmacological treatments are currently recommended for ICU delirium prevention, non‐pharmacological interventions—including exercise, sleep promotion and family involvement—are widely endorsed by clinical guidelines.
  • ○
    Family involvement has emerged as a promising delirium prevention strategy; however, evidence regarding the effectiveness of specific family voice interventions has shown inconsistency across studies.
• What this paper adds

  • ○
    This study provides evidence that family voice interventions significantly reduce both delirium incidence and duration in adult ICU patients.
  • ○
    Multicomponent family voice interventions demonstrate superior effectiveness compared to single‐component approaches, suggesting synergistic benefits when combined with other sensory stimulation or environmental modifications.
  • ○
    Family voice interventions do not significantly impact mechanical ventilation duration or ICU length of stay.
  • ○
    Standardised family voice intervention protocols are needed for consistent clinical implementation, along with large‐scale multicentre trials to validate long‐term effectiveness.

1. Introduction

Delirium is an acute and fluctuating disturbance of attention and awareness, often accompanied by cognitive impairments such as disorganised thinking and perceptual disturbances [1]. This condition is highly prevalent in intensive care units (ICUs), where multiple risk factors—including advanced age, sedative administration, mechanical ventilation and disease severity—frequently converge to disrupt normal neurocognitive function [2]. A large meta‐analysis involving more than 28 000 ICU patients reported a delirium incidence of approximately 22% [3]. ICU delirium is associated with serious adverse outcomes, including prolonged hospital stays, increased mortality and long‐term cognitive decline following discharge [4, 5, 6]. These consequences highlight the urgent need for effective prevention strategies in critical care nursing.

The Society of Critical Care Medicine's PADIS guidelines [7] and recent umbrella reviews [8] emphasise the importance of non‐pharmacological approaches for delirium prevention. Among these strategies, family involvement has emerged as a particularly promising intervention [9, 10]. Such interventions encompass various formats, including extended visitation hours, real‐time or virtual family interactions and the use of pre‐recorded family voice messages [11]. Multicomponent programmes incorporating family education, emotional support, cognitive reorientation and integration into the ABCDE bundle have also demonstrated benefits [12].

One specific and innovative form of family involvement is the family voice intervention, which involves playing pre‐recorded messages from family members to ICU patients—particularly those who are non‐communicative or sedated [13]. ICU patients commonly experience sensory overload or deprivation, exposure to unfamiliar routines and invasive procedures and social isolation—all factors that can precipitate delirium [14]. Although traditional approaches have primarily focussed on minimising sensory input (such as reducing light and noise exposure and promoting sleep), family voice interventions provide familiar and emotionally meaningful auditory stimulation. This type of input may help modulate stress responses, support cognitive orientation and provide psychological comfort [13, 15]. The intervention is not only simple and cost‐effective but also readily implementable by bedside nurses, even for unconscious or sedated patients [13, 15]. Moreover, by encouraging family participation in patient care, it aligns with humanised nursing principles and may enhance satisfaction for both patients and families [16, 17].

Despite its potential benefits, family voice interventions remain underutilised in clinical practice, and evidence regarding their effectiveness shows inconsistency. For example, Kasapoğlu [18] found that a multicomponent intervention including family voice recordings significantly reduced delirium incidence. Similarly, Adineh et al. [19] reported reductions in delirium duration, ICU length of stay and mechanical ventilation duration among patients receiving family‐based sensory stimulation. Conversely, Liang et al. [20] observed no significant effect on delirium incidence. These discrepancies may arise from differences in study design, patient populations, intervention content, frequency and duration of auditory exposure and outcome assessment methods. Furthermore, most existing studies have been conducted at single centres with relatively small sample sizes, limiting the generalisability of their findings.

Given these inconsistencies, a comprehensive synthesis of existing literature is needed. A systematic review and meta‐analysis can integrate findings across diverse studies, provide quantitative estimates of overall effect sizes and inform future research directions and clinical guidelines.

2. Aim

This study aimed to systematically review and meta‐analyse available evidence on the effectiveness of family voice interventions in preventing and reducing delirium among ICU patients.

3. Design and Methods

3.1. Design

This study employed a systematic review and meta‐analysis approach to evaluate the effects of family voice interventions on delirium in ICU patients. The review was conducted in accordance with methodological guidance from the Cochrane Handbook for Systematic Reviews of Interventions [21] and reported in line with the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) guideline [22]. The study protocol was prospectively registered with the International Prospective Register of Systematic Reviews (PROSPERO: CRD42025628074).

3.2. Data Source and Search Strategy

A comprehensive literature search was conducted across PubMed, Excerpta Medica Database (EMBASE), Cumulative Index to Nursing and Allied Health Literature (CINAHL) and the Cochrane Library from database inception through 4 March2025 to identify studies examining family voice interventions for ICU patients. The primary search strategy, developed in consultation with an experienced medical librarian, combined Medical Subject Headings (MeSH) with free‐text terms. It was adapted for each database (see Data S1) and included keywords related to critical care (e.g., “Critical Care,” “ICU,” “SICU,” “MICU,” “CCU”), family involvement (e.g., “family,” “caregiver,” “relative”) and voice or sensory stimulation (e.g., “voice,” “reorientation,” “sensory,” “sensory stimulation”).

3.3. Study Selection

Studies were included based on the following eligibility criteria: (1) Participants: adult patients (≥ 18 years) admitted to the ICU; (2) Interventions: family voice interventions, either as stand‐alone interventions or as part of multicomponent approaches; (3) Comparisons: standard care as the control condition; (4) Outcomes: primary outcomes included delirium incidence and duration, whereas secondary outcomes encompassed duration of mechanical ventilation and ICU or hospital length of stay; and (5) Study Design: randomised controlled trials (RCTs). Exclusion criteria included the following: (1) studies not published in English or Korean; and (2) studies for which full‐text articles were unavailable (e.g., conference abstracts or proceedings).

After removing duplicates using reference management software (EndNote 20), two reviewers (JL and JK) independently screened titles and abstracts. Discrepancies were resolved through discussion until consensus was reached. Full texts of potentially eligible studies were subsequently reviewed, and final inclusion decisions were determined by agreement between the two reviewers.

3.4. Data Extraction

For studies included in the final analysis, data were initially extracted by one researcher (JL) using a standardised format and subsequently verified for accuracy by a second researcher (JK). The extracted information included: (1) study characteristics—first author, year of publication and country; (2) participant characteristics—patient population types, age and sample size per group; (3) intervention and control descriptions; (4) delirium assessment tools; and (5) main study findings.

3.5. Quality Assessment

Two reviewers (JL and JK) independently evaluated the methodological quality of included studies using the Cochrane Risk of Bias Tool 2.0 [23]. The assessment encompassed the following domains: bias arising from the randomisation process, deviations from intended interventions, missing outcome data, bias in outcome measurement, bias in selection of reported results and overall risk of bias. Each domain was rated as ‘low risk’, ‘high risk’ or ‘some concerns’ according to predefined signalling questions and decision algorithms. For trials reported in multiple publications [22, 24], risk of bias was assessed once at the study level by integrating information from both reports. Discrepancies between reviewers were resolved through discussion.

3.6. Data Synthesis and Analysis

To compare and analyse the characteristics of family voice interventions used in the included trials, we categorised them by intervention type and documented specific content, duration, timing of initiation, frequency and time of day provided. Primary outcomes included delirium incidence and duration, whereas secondary outcomes encompassed mechanical ventilation duration and ICU length of stay. None of the included trials reported length of hospital stay as an outcome. Delirium incidence was not restricted to specific assessment tools. Delirium duration was extracted only when explicitly reported in hours or days.

Meta‐analysis was performed using Comprehensive Meta‐Analysis version 4.0. Heterogeneity among studies was assessed using Cochran's Q test and I ² statistics [25]. A random‐effects model was applied when I ² was ≥ 30%, indicating significant heterogeneity [26]. Odds ratios (ORs) were calculated for dichotomous outcomes, whereas standardised mean differences (Hedges' g) were used for continuous outcomes. All effect sizes were calculated with 95% confidence intervals and presented visually using forest plots.

Subgroup analysis was conducted based on family voice intervention type. However, due to the limited number of included studies, neither evidence grading (e.g., GRADE approach) nor sensitivity analyses were performed. Additionally, publication bias was not examined using funnel plots or Egger's test because fewer than 10 studies were included, consistent with current Cochrane Handbook guidance for small study numbers [26].

4. Results

4.1. Literature Selection Process

Our initial search identified 6769 records from four major databases. After removing 1218 duplicates, 5551 records remained for title and abstract screening. Of these, 5529 records were excluded for not meeting inclusion criteria, leaving 22 full‐text publications for detailed review. Two publications could not be retrieved (abstract‐only articles) despite multiple attempts and were classified as ‘reports not retrieved’.

Following full‐text review of the remaining 20 publications, 2 were excluded for not being RCTs, 10 did not report any prespecified outcomes, and 3 did not involve family voice interventions. Ultimately, five publications were deemed eligible for inclusion. Two of these originated from the same trial, and two additional publications were identified through citation searching. As a result, six unique RCTs, reported across seven publications, were included in the review (Figure 1). All included publications provided sufficient statistical data for quantitative synthesis. One trial generated two separate publications that reported different outcomes (delirium vs. duration of mechanical ventilation duration and ICU days). For outcome‐specific meta‐analyses, each publication was included according to the relevant outcome, without duplication. Intervention characteristics and quality assessment were summarised at the study level (six RCTs).

FIGURE 1.

PRISMA 2020 flow diagram of study selection: A total of seven publications were identified, corresponding to six randomised controlled trials (RCTs). One RCT generated two publications reporting different outcomes, and both were included in the outcome‐specific analyses without overlap.

4.2. Risk of Bias

Based on quality assessment using the ROB II tool, Domain 1 (randomisation process) was rated as having ‘some concerns’ in three studies due to insufficient descriptions of random sequence generation or allocation concealment. In Domain 2 (deviations from intended interventions), three studies received ‘some concerns’ ratings owing to limited information on blinding procedures or intervention compliance rates. All studies were rated as ‘low risk’ in Domain 3 (missing outcome data) and Domain 4 (outcome measurement). However, in Domain 5 (selection of reported results), three studies were assessed as having ‘some concerns’ due to unclear pre‐registration of analysis plans or potential for selective reporting. Notably, one study was rated as ‘high risk’ in this domain due to concerns about selective reporting. Overall, one study was judged to have ‘low risk’ of bias, four were rated as having ‘some concerns’, and one was assessed as ‘high risk’ (Figure 2).

FIGURE 2.

Risk of bias of studies.

4.3. Characteristics of Individual Studies

A total of seven publications from six RCTs published between 2017 and 2025 were included in this review. The trials were conducted in the United States, Turkey, Iran, China and Germany, enrolling 598 patients with mean ages ranging from 36.7 to 76.1 years.

All trials were conducted in adult ICUs and included populations such as surgical and emergency critically ill patients and those with traumatic brain injury. Delirium, the primary outcome, was assessed using the CAM‐ICU across all RCTs. Depending on the trial, delirium‐related outcomes included incidence, duration, severity and delirium‐free days. Some trials also reported secondary outcomes such as mechanical ventilation duration and ICU length of stay (Table 1).

TABLE 1.

Characteristics of included studies.

Author year	Country	Participants	Sample size	Mean age (SD)	Intervention	Control	Delirium Measurement	Results
Adineh 2023 [19]	Iran	Patients with traumatic brain injury, GCS 6–12, intubated and on mechanical ventilation	N = 66 IG: n = 33 CG: n = 33	36.7 (13.9)	Recorded family voice+sensory stimulation	Usual care	CAM‐ICU	Delirium incidence IG: 24.2%, CG: 84.8% Mean delirium days (SD) IG: 3.00(0.75) CG: 8.57(3.09) Mean MV days (SD) IG: 7.54 (3.17) CG: 16.36 (6.46) Mean ICU days (SD) IG: 14.3 (3.9) CG: 24.24 (7.02)
Kasapoglu 2022	Turkey	Adult ICU patient	N = 94 IG 1: n = 30 IG 2: n = 31 CG: n = 33	76.1 (11.7)	IG 1: Recorded non‐family voice IG 2: Recorded family voice+cognitive stimulation+sleep focussed intervention	Usual care	CAM‐ICU	Mean delirium days (SD) IG: 1.41 (1.11) CG: 3.62 (1.64) Delirium incidence IG: 6.5%, CG: 27.3%
Liang 2023a	China	Postoperative adult patients in surgical ICU	N = 152 IG: n = 76 CG: n = 76	56.2 (16.2)	Recorded family voice+sensory stimulation	Usual care	CAM‐ICU CAM‐ICU‐7	Delirium incidence IG: 13.2%, CG:25.0% Mean delirium‐free days (SD) IG: 3.66(2.55) CG: 2.84(2.63) Mean delirium days (SD) IG:1.70 (0.82) CG: 4.50(2.70) Mean delirium severity (SD) IG: 3.70 (1.25) CG: 5.68 (1.57)
Liang 2023b	China	Postoperative adult patients in surgical ICU	N = 152 IG: n = 76 CG: n = 76	56.2 (16.2)	Same as Liang 2023a	Usual care	CAM‐ICU	Median MV days (IQR) IG: 9.50 (16.00) CG: 16.50 (20.75) Mean ICU days (SD) IG:3.88(2.61) CG:3.96(2.85)
Ma 2025	China	Adult patients in emergency ICU	N = 213 IG 1: n = 71 IG 2: n = 71 CG: n = 71	57.3 (17.4)	IG 1: Recorded structured family voice IG 2: Recorded unstructured family voice	Usual care	CAM‐ICU CAM‐ICU‐7	Delirium incidence IG1: 22.5% CG: 49.3% Median delirium duration (IQR) IG: 2.0 (2.0–2.0) CG: 4.0 (3.0–4.0) Median delirium‐free days (IQR) IG: 3.0 (3.0–3.0) CG: 1.0 (1.0–2.0) Mean delirium severity (SD) IG 1: 3.1 (0.4) IG 2: 4.0 (0.3) CG: 5.1 (0.3)
Munro 2017	USA	Adult ICU patients	N = 30 IG 1: n = 10 IG 2: n = 10 CG: n = 10	59.5 (17.0)	IG 1: Recorded unknown voice IG 2: Recorded family voice	Usual care	CAM‐ICU	Mean delirium‐free days (SD) IG: 1.9 (0.99) CG: 1.6 (1.13) Mean delirium days (SD) IG: 0.3 (0.48) CG: 0.9 (1.28)
Sprügel 2025	Germany	Intubated neurological ICU patients on mechanical ventilation ≥ 48 h	N = 43 IG: n = 21 CG: n = 22	60.0 (11.5)	Recorded family voice	Muted audio recording	CAM‐ICU	Delirium incidence IG: 57.1% CG: 59.1% Ventilator weaning failure IG: 52.4% CG:63.6%

Abbreviations: CAM‐ICU = Confusion Assessment Method for the Intensive Care Unit; CG = control group; GCS = Glasgow Coma Scale; ICU = intensive care unit; IG = intervention group; MV = mechanical ventilation.

Family voice interventions varied in type, content, duration, frequency and timing. Three RCTs employed single‐component interventions consisting solely of family voice recordings, whereas three used multicomponent interventions that combined additional elements such as cognitive stimulation, sleep promotion and sensory stimulation. The content of family voice interventions generally focussed on providing simple orientation, explanation of the ICU setting, emotional support or reassurance. Some trials also included treatment‐related information, personalised messages, encouragement for recovery and reinforcement of family presence.

In most RCTs, interventions were provided from the day of ICU admission, and all interventions were delivered via headphones or open speakers. Intervention duration ranged from 2 to 45 min and was administered between 6:00 a.m. and 8:00 p.m. Intervention frequency varied from once to eight times per day, and intervention periods ranged from three to 7 days (Table 2).

TABLE 2.

Intervention characteristics.

Type	Study	Content	Duration	Timing of initiation	Frequency and period	Time of day provided
Multicomponent	Adineh 2023	‐ Family voice: recorded emotional support and personalised messages, verbal orientation in person—Sensory stimulation: visual (photographs, videos), olfactory (familiar scents), tactile (massage, touch), motor (passive limb movements)	1 h	At ICU admission	Once‐daily during ICU stay	4 PM–5 PM
	Kasapoglu 2022	‐ Family voice: recorded orientation, explanation of ICU setting and devices, patient condition and emotional support. ‐ Cognitive stimulation: newspaper reading—Sleep‐focussed intervention: eye patch at night	45 min	Not reported	Twice‐daily for 3 days	10 AM and 4 PM
	Liang 2023a Liang 2023b	‐ Family voice: recorded orientation, ICU setting and devices explanation, breathing‐tube notice, reassurance about care and family presence. ‐ Sensory stimulation: auditory (radio player), visual (clock & calendar, family photos, vision aids, book/magazine)	45 min	At ICU admission	Once‐daily up to 7 days	10 AM–12 PM or 4 PM–6 PM
Single component	Ma 2025	‐ Family voice: recoded orientation, ICU environment & condition, reassurance and support, family & social recall, encouragement and hobbies	5 min	≤ 24 h after ICU admission	Five times daily for 5 days	8 AM–4 PM
	Munro 2017	‐ Family voice: recoded orientation, reassurance and support, patient condition, ICU settings and devices, family presence	≤ 2 min	≤ 24 h after ICU admission	8 times daily for 3 days	9 AM–4 PM
	Sprügel 2025	‐ Family voice: recoded orientation, medical condition, reassurance and support, breathing guidance	10 min	On the first day of ventilator weaning	Three times daily	6 AM–8 AM 12 PM–2 PM 6 PM–8 PM

Abbreviation: ICU = intensive care unit.

4.4. Meat‐Analysis on the Effect of Family Voice Intervention

Five included publications evaluated the effect of family voice interventions on delirium incidence. As significant heterogeneity was observed among these publications (I ² = 65%, p = 0.022), a random‐effects model was applied. The meta‐analysis (Figure 3A) revealed that family voice interventions significantly reduced delirium incidence (OR = 0.282, 95% CI: 0.126–0.630, p = 0.002).

FIGURE 3.

Forest plots of the overall effects of family voice interventions on (A) delirium incidence, (B) delirium duration, (C) duration of mechanical ventilation and (D) Length of ICU stay in adult ICU patients: The line of no effect (red vertical dotted line) is indicated at OR = 1 for incidence and g = 0 for duration. Diamonds represent pooled effect sizes, which demonstrate a significant reduction in both delirium incidence and duration.

Four included publications reported on the effect of interventions on delirium duration. Given the substantial heterogeneity observed (I ² = 93%, p < 0.001), a random‐effects model was also used. The meta‐analysis (Figure 3B) demonstrated that family voice interventions significantly reduced delirium duration compared to standard care (Hedges' g = −1.929, 95% CI: −2.980 to −0.879, p < 0.001).

Three publications reported mechanical ventilation duration as an outcome. Due to high heterogeneity among these publications (I ² = 87%, p < 0.001), a random‐effects model was applied. The meta‐analysis (Figure 3C) showed no statistically significant difference in mechanical ventilation duration between family voice intervention and control groups (Hedges' g = −0.826, 95% CI: −1.696 to 0.044, p = 0.063).

Four publications examined the effect of interventions on ICU length of stay. Given substantial heterogeneity across publications (I ² = 91%, p < 0.001), a random‐effects model was also used. The meta‐analysis (Figure 3D) indicated that family voice interventions did not result in statistically significant reductions in ICU length of stay (Hedges' g = −0.534, 95% CI: −1.372 to 0.303, p = 0.211).

4.5. Subgroup Analysis

We conducted subgroup analysis based on intervention type (single‐component vs. multicomponent). For delirium incidence (Figure 4A), multicomponent interventions demonstrated statistically significant reductions (OR = 0.178, 95% CI: 0.046–0.681, p = 0.012), whereas single‐component interventions did not yield statistical significance (OR = 0.471, 95% CI: 0.160–1.392, p = 0.173). Regarding delirium duration (Figure 4B), a statistically significant reduction was observed with multicomponent interventions (Hedges' g = −1.888, 95% CI: −2.917 to −0.859, p < 0.001), whereas the effect of single‐component interventions was not statistically significant (Hedges' g = −1.909, 95% CI: −4.437 to 0.618, p = 0.139).

FIGURE 4.

Subgroup analysis of family voice interventions by type (multicomponent vs. single‐component intervention): The line of no effect (red vertical dotted line) is indicated at OR = 1 for incidence and g = 0 for duration. Diamonds represent pooled effect sizes within each subgroup. Labels for subgroup categories are displayed within the figure. (A) delirium incidence, (B) delirium duration, (C) duration of mechanical ventilation and (D) Length of ICU stay.

For mechanical ventilation duration (Figure 4C), neither single‐component interventions (Hedges' g = −0.192, 95% CI: −1.034 to 0.649, p = 0.654) nor multicomponent interventions (Hedges' g = −1.089, 95% CI: −2.267 to 0.089, p = 0.070) showed statistically significant effects. Similarly, for ICU length of stay (Figure 4D), no statistically significant differences were found for either single‐component interventions (Hedges' g = −0.669, 95% CI: −1.533 to 0.196, p = 0.130) or multicomponent interventions (Hedges' g = −0.500, 95% CI: −1.522 to 0.523, p = 0.338).

5. Discussion

This systematic review analysed six RCTs (reported in seven publications) encompassing 598 ICU patients to evaluate the effectiveness of family voice interventions in preventing delirium. These interventions were implemented as either stand‐alone strategies or combined with other components, with variations in both content and delivery methods. Meta‐analysis demonstrated that family voice interventions significantly reduced both delirium incidence and duration but showed no significant effects on mechanical ventilation duration or ICU length of stay. Subgroup analyses further revealed that multicomponent interventions maintained significant reductions in delirium incidence and duration, whereas single‐component interventions no longer demonstrated statistically significant effects.

PADIS guidelines recommend non‐pharmacological interventions for delirium prevention [7], and subsequent meta‐analyses [10, 27] have reported that non‐pharmacological strategies involving family participation are effective in preventing ICU delirium. A recent meta‐analysis by Li et al. [11] further confirmed the significant role of family involvement in delirium prevention among critically ill patients. However, their analysis categorised interventions broadly into caregiving participation, visitation and companionship, and indirect contact without delineating specific effects of each approach. In contrast, the present study focusses specifically on recorded family voices played for ICU patients and provides evidence supporting their effectiveness. This contributes valuable guidance for evidence‐based practice in ICU settings.

Family voices are meaningful auditory stimuli that activate the sympathetic nervous system, enhancing consciousness and arousal [28]. This activation, accompanied by increased heart rate, cerebral blood flow and metabolic rate, may promote brain function, orientation and attention. Recognising familiar voices provides psychological stability, serving as emotional cues that help restore orientation and reduce stress in confused states [29]. Beyond cognitive benefits, family voice interventions can reduce anxiety, depression and PTSD [24, 29]. Additionally, a systematic review by Dijkstra et al. [30] demonstrated that family involvement reduces anxiety and PTSD symptoms among family members. Facilitating communication and participation in patient care fosters trust in medical teams and reduces feelings of alienation. These family‐centred approaches humanise care by benefiting both patients and families.

Subgroup analysis findings suggest that family voice interventions may have more synergistic effects in alleviating delirium when combined with other sensory stimuli or environmental modifications. These synergistic effects may be explained by neural mechanisms underlying multisensory stimulation. Unlike single‐sensory input, multisensory stimulation activates broader neural circuits—including attention, mood and frontoparietal networks—thereby enhancing cognitive processing and emotional integration [31, 32, 33]. It also promotes synapse formation and neural regeneration, facilitating brain recovery. This neuroplasticity enhancement stimulates the reticular activating system and elevates consciousness levels [34, 35]. Accordingly, multisensory stimulation including family voices is expected to promote recovery from attention deficits, disorientation and confusion by enhancing the interplay between cognitive restoration and emotional stability.

In this study, no statistically significant effects were observed on mechanical ventilation duration or ICU stay. This may be attributed to these outcomes being influenced by complex interactions of clinical factors—including age, disease severity, comorbidities and overall physiological recovery—beyond delirium presence [36, 37]. Nevertheless, a previous study reported that family‐centred auditory and tactile stimulation shortened ICU length of stay in patients with traumatic brain injury [38]. Similarly, a recent meta‐analysis by Li et al. [11] found that family involvement significantly reduced ICU stay. The absence of such effects in our meta‐analysis may be due to relatively low intervention intensity in included trials or insufficient sample size to detect meaningful differences. These findings highlight the need for larger, more rigorous trials to evaluate the impact of family voice interventions on these clinically important outcomes.

Family voice interventions examined in included RCTs varied in types, content, duration, timing and frequency. To facilitate broader implementation of these interventions in clinical practice, developing and adopting standardised protocols is essential. Standardisation offers key advantages, including systematic education and training for healthcare professionals and consistent quality control of interventions [39]. Johnson et al. [17] developed the Family Members' Voice Reorientation Intervention for delirium prevention through a co‐design approach. This intervention encompasses four main domains—ICU orientation, ICU routine, ICU procedures and a free category—and includes detailed guidance on scripting, recording methods and frequency, providing a practical framework for clinical application. Implementation of such standardised interventions is expected to support the establishment of protocols regarding participant selection criteria, core components, optimal timing and intensity of interventions and integration of additional sensory stimuli.

Our findings suggest that family voice interventions can reduce delirium in critically ill patients. Nevertheless, direct family presence and verbal interaction remain the optimal approach for patient‐centred critical care [40]. The interventions included in this review were limited to recorded family voices, which should not be considered a replacement for family presence. Rather, they may serve as a supplementary strategy to provide patients with a sense of connection and familiarity when direct interaction is not feasible. Further research is needed to determine the comparative effectiveness of direct family engagement versus recorded messages across different ICU contexts.

6. Limitation

This study has several limitations. First, only seven publications were included, reflecting the limited number of RCTs investigating family voice intervention effectiveness for delirium. Consequently, some subgroup analyses were based on a single publication, depending on the variable examined. Meta‐ANOVA, which tests for statistical differences in effects between subgroups, requires at least two to three studies per group [41]; therefore, this analysis could not be conducted.

Second, quality assessment of included RCTs revealed methodological concerns. One trial was rated as having a high risk of bias, and four were judged to raise some concerns. Bias was particularly evident in domains related to randomisation procedures and selective outcome reporting.

Third, considerable heterogeneity existed among included RCTs regarding intervention content, duration, timing and frequency. This variability may limit clinical application feasibility. To promote the adoption of family voice interventions in practice, developing and validating standardised protocols across key dimensions is essential.

Fourth, long‐term intervention effects remain unclear. Most trials assessed outcomes only during ICU stays, with limited evaluation of longer term impacts on cognitive function or quality of life following discharge.

7. Implications for Future Practice and Research

Family voice interventions, which are simple, effective and easily implementable, may serve as valuable non‐pharmacological strategies for reducing delirium in ICU patients. Their effectiveness appears to increase when combined with other forms of sensory stimulation or environmental modifications, suggesting that multicomponent approaches may be optimal for clinical practice. Therefore, developing specific and standardised intervention protocols could facilitate the delivery of humanised care in intensive care settings while ensuring consistent implementation across healthcare institutions.

7.1. Clinical Implications

Critical care nurses may incorporate recorded family voice interventions as a practical approach to support delirium prevention in everyday practice. To enhance their effectiveness and sustainability, healthcare institutions could develop protocols that include family education, recording guidelines and standardised implementation procedures.

7.2. Research Implications

Given current study limitations, future research should prioritise large‐scale, multicentre randomised controlled trials employing standardised protocols. These studies should evaluate not only immediate delirium‐related outcomes but also long‐term effects on cognitive function, quality of life and healthcare utilisation following discharge. Additionally, research investigating optimal intervention characteristics—including duration, frequency, timing and content—would provide valuable guidance for evidence‐based practice development.

8. Conclusions

This systematic review and meta‐analysis of six RCTs (reported in seven publications) encompassing 598 ICU patients demonstrated that family voice interventions effectively prevent delirium and reduce its duration, with multicomponent interventions demonstrating superior efficacy compared with single‐component approaches. However, heterogeneity in intervention content and delivery methods underscores the urgent need for standardised protocols. Future large‐scale, rigorously designed multicentre trials are essential to validate the effectiveness of standardised family voice interventions and investigate their long‐term impacts on patient outcomes beyond delirium, including cognitive function and quality of life following ICU discharge.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Data S1: Supporting Information.

Lee J. and Kang J., “The Effects of Family Voice Intervention on Delirium in Adult Critical Care Patients: A Systematic Review and Meta‐Analysis,” Nursing in Critical Care 30, no. 6 (2025): e70208, 10.1111/nicc.70208.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.