A network meta-analysis of intensive nursing interventions for delirium in ICU patients

Abstract

Background:

Delirium prevalence in the intensive care unit (ICU) is high. Intensive nursing interventions have been performed to reduce delirium in ICU patients. There is now a wide variety of intensive nursing interventions available for treating delirium. However, the optimal intervention remains unknown. This systematic review and network meta-analysis (NMA) aimed to compare the efficacy of intensive nursing interventions in patients with delirium.

Methods:

We included randomized controlled trials of different intensive nursing interventions for delirium in the ICU. A Bayesian NMA was conducted to evaluate the efficacy of various types of intensive nursing interventions. The outcomes assessed were the cure rate, intensive care delirium screening checklist (ICDSC) scores, and acute physiologic assessment and chronic health evaluation II (APACHE II) scores for different treatments.

Results:

This meta-analysis included 21 studies. We analyzed a total of 5 different intensive nursing interventions: auricular points acupressure, music therapy, cognitive function exercise, regular nursing, increasing visiting hours, and targeted nursing. When compared with regular nursing, the other 5 intensive nursing interventions showed no significant difference in cure rate, ICDSC, and APACHE II scores (P > .05). Auricular points acupressure had the highest surface under the cumulative ranking area value for cure rate and APACHE II, indicating it ranked first in these outcomes. Music therapy demonstrated the most favorable effect on reducing ICDSC, with music therapy ranking first in this regard.

Conclusion:

This NMA suggests that auricular points acupressure might be the optimal intervention for increasing the cure rate and decreasing ICDSC and APACHE II scores in ICU patients with delirium. However, the surface under the cumulative ranking area values reflect relative ranking rather than absolute efficacy, and no intervention demonstrated statistically significant superiority over regular nursing. Additionally, increasing visiting hours appears to hold promise as an effective intervention for reducing delirium in the ICU. Further research and larger studies are warranted to confirm these findings and to explore the long-term benefits of these intensive nursing interventions in delirium management.

1. Introduction

Delirium is a prevalent acute brain dysfunction in critical patients, known to lead to poor clinical outcomes and long-term cognitive function impairment.^[1–3] Its estimated prevalence rates range from 22 to 87%, depending on illness severity and diagnostic methods for delirium.^[4] Furthermore, delirium has been linked to long-term cognitive impairment, extended lengths of stay in the intensive care unit (ICU), and increased healthcare costs.^[5,6]

The etiology of ICU delirium is complex and multifactorial, involving interconnected factors typically categorized as non-modifiable and modifiable risk factors.^[7] Non-modifiable risk factors encompass advanced age (over 65 years old), preexisting cognitive impairment, and the presence of comorbidities.^[8,9] ICU delirium can lead to adverse outcomes, such as prolonged mechanical ventilation, extended ICU and hospital stays, long-term cognitive impairments, and increased workload for clinical staff.^[10] As a result, the early implementation of prevention measures for delirium holds the potential to improve outcomes among critically ill patients.

Various prevention strategies have been extensively assessed, including pharmacological, sedation, and non-pharmacological interventions, both single and multicomponent. These interventions can be initiated during or shortly before admission to the ICU. Among these strategies, non-pharmacological multicomponent interventions have undergone significant investigation in older non-ICU adults during hospitalization, with evidence indicating that they are the most efficacious approach for preventing delirium.^[11]

Intensive nursing interventions were proactively implemented to mitigate and minimize the occurrence of delirium, aiming to improve the patient’s prognosis promptly.^[12] These interventions also play a vital role in enhancing the quality of life for patients.^[13,14] Notably, nursing interventions have demonstrated significant effects on patients undergoing cardiac surgery, tuberculosis treatment, diabetes management, and those with senile dementia.^[15,16] However, the effects of intensive nursing interventions specifically for delirium in the ICU remained largely unknown.

The provided nursing interventions for ICU patients encompassed a diverse range of approaches, including cognitive function exercises, music therapy, targeted nursing, increased visiting hours, auricular points acupressure, and regular nursing.

These interventions, each with its distinct mechanism of action, have shown potential in increasing the cure rate for delirium in ICU patients.^[17] Unfortunately, due to small sample sizes and varying strategies, clinical routines are still uncertain about the optimal intervention strategy to prevent delirium in ICU patients. Moreover, the relative effects of different types of interventions remain unexplored.

Given the multitude of interventions available, traditional meta-analysis, limited by pairwise comparisons, can no longer offer effective methodological support for selecting the optimal interventions.^[18,19] Therefore, recognizing the need for more comprehensive decision-making tools, we opted for a network meta-analysis (NMA) to provide clinicians with supplementary information at the bedside.

The NMA allowed the construction of a network structure encompassing 2 or more interventions, facilitating the overall computation of relative effectiveness through both direct one-to-one and indirect comparisons among multiple interventions. The greatest advantage of the NMA method lies in its ability to summarize various interventions for treating a specific disease, perform quantitative statistical analysis, rank these interventions based on outcomes, and ultimately aid in selecting the optimal treatment plan.

To our knowledge, this is the first NMA to comprehensively compare the relative efficacy of various intensive nursing interventions specifically for delirium in ICU patients. Previous meta-analyses have primarily focused on pairwise comparisons or non-pharmacological interventions in general, without simultaneously comparing the diverse intensive nursing modalities evaluated in this study. Given the high prevalence of ICU delirium and the lack of consensus on the optimal nursing strategy, this study provides critical evidence to guide clinical decision-making.

As such, the objective of this systematic review and NMA was to assess the efficacy of different intensive nursing interventions aimed at reducing delirium in the ICU.

2. Materials and methods

2.1. Search strategy

We conducted a literature search to identify all published randomized controlled trials (RCTs) based on the search strategies suggested in the Cochrane Handbook for Systematic Reviews of Interventions. The electronic databases PubMed, Embase, and the Cochrane Library were searched for publications listed between each database’s inception date and December, 2024. The search terms and medical subject headings used were mainly “Subacute Delirium,” “Delirium, Subacute,” “Deliriums, Subacute,” “Subacute Deliriums,” “Delirium of Mixed Origin,” “Mixed Origin Delirium,” “Mixed Origin Deliriums,” “Delirium”[medical subject headings], “Nursing, Critical Care,” “Intensive Care Nursing,” “Nursing, Intensive Care,” “Intensive Care Unit,” “ICU Intensive Care Units.” The search strategy also included Medical Subject Headings terms in the search strategy for PubMed and Emtree terms for Embase. Search strategy can be seen in Table S1, Supplemental Digital Content 1. The reference list was also retrieved. Grey literature will be identified by emailing primary authors of included studies and searching conference abstracts from relevant meetings. When using database retrieval, we limit the articles to clinical trials and explode all trees. As this study is a systematic review and NMA of previously published data, ethical approval and patient consent were not required. Grey literature was identified by searching conference abstracts from the following relevant meetings: the American Thoracic Society International Conference, the European Society of Intensive Care Medicine Annual Congress, and the Society of Critical Care Medicine Annual Congress. Additionally, we contacted the primary authors of included studies via email to inquire about any unpublished or ongoing studies.

2.2. Inclusion and exclusion criteria

Inclusion and exclusion criteria are based on PICOS standards: see Table 1 for specific inclusion and exclusion criteria.

Table 1.

Inclusion and exclusion criteria.

Category	Inclusion criteria	Exclusion criteria
Population	Male or female patients older than 18 years in ICU	Patients with diseases related to the psychiatric or neurological disorders
Interventions	Cognitive function exercise, music therapy, targeted nursing, increasing visiting hours, auricular points acupressure
Comparisons	Regular nursing
Outcomes	Cure rate, ICDSC and APACHE II	Missing or incomplete information on the trial
Study	RCT; published in English or Chinese	Document type (reviews, meeting summaries, letters, etc)

APACHE II = acute physiologic assessment and chronic health evaluation II, ICDSC = intensive care delirium screening checklist, ICU = intensive care unit, RCT = randomized controlled trial.

2.3. Study selection

Two authors (Rui Wang and Jiang Wu) independently screened the titles and abstracts to identify duplicate records and exclude those that did not meet the inclusion criteria. If there was any ambiguity or if the titles and abstracts appeared to meet the inclusion criteria, the full text of the records was obtained for a thorough assessment of eligibility. Throughout the screening process, meticulous documentation of the reasons for excluding trials was maintained. Any discrepancies that arose were resolved through discussions between the authors.

2.4. Data extraction

Two investigators independently assessed all trials for eligibility, extracted data by screening the titles and abstracts, and retrieved the full articles if a decision could not be made. In case of disagreement, consensus was reached through discussion. Data extraction was performed independently by the 2 reviewers. We extracted the trial design, trial size, details of intervention, including dose and treatment duration, and patient characteristics such as mean age, sex, mean platelet count, and total number of splenectomies. In addition, data for pooling were extracted, including the total number of subjects, any bleeding events, composite serious adverse events, odds ratio (OR) with 95% confidence interval (CI), and mean with standard deviation of continuous outcomes. To evaluate the potential impact of inconsistent outcome measurement timing across studies, we recorded the specific measurement time points for each outcome in each included study and planned sensitivity analyses to assess the robustness of the pooled results if substantial variation in timing was observed.

2.5. Quality assessment

We considered the following aspects for quality assessment: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other biases. Quality assessment was performed using Review Manager (version 5.3; The Cochrane Collaboration). If the opinions of 2 authors differed, the contradiction was resolved through discussion with a third person. Kappa values were used to measure the degree of agreement between the 2 reviewers and were rated as follows: fair, 0.40 to 0.59; good, 0.60 to 0.74; and excellent, 0.75 or more.^[20]

2.6. Statistical analysis

NMA concerning multiple treatments was performed by a random-effect model within a Bayesian framework, using package “gemtc” and rjags of R software (version 3.5.1; The R Foundation for Statistical Computing, https://www.r-project.org/). The Markov chain Monte Carlo Bayesian NMA was fitted with 3 chains as a means of checking Markov chain Monte Carlo convergence. For each chain, 150,000 sample iterations were generated with 100,000 burn-ins and a thinning interval of 10. Based on the posterior distribution medians, all outcome estimates (mean differences [MDs] or ORs) with 95% CIs were calculated. If 95% CIs of ORs did not contain 1 and 95% CIs of MDs did not contain 0, the corresponding ORs or MDs were considered to indicate a statistically significant difference. We performed a NMA and used the surface under the cumulative ranking area (SUCRA) values to present the hierarchy of the nursing interventions. The SUCRA value ranges between 0 and 1, and the intervention with a higher SUCRA value is considered to have better efficacy. Heterogeneity among the studies was evaluated using the I² statistic, with I² values of 25%, 50%, and 75% representing low, moderate, and high heterogeneity, respectively. To explore potential sources of clinical heterogeneity, we planned subgroup analyses based on study sample size, intervention duration, and patient age where sufficient data were available. Meta-regression was also planned to examine the impact of these covariates on the pooled effect estimates. Global inconsistency between direct and indirect evidence was assessed by comparing the deviance information criterion between consistency and inconsistency models. A difference in deviance information criterion of < 5 and a P value > .05 were considered indicative of no significant inconsistency. Node-splitting analysis was also performed to evaluate local inconsistency for each comparison where both direct and indirect evidence were available. Sensitivity analyses were conducted to evaluate the robustness of the results by excluding studies with a high risk of bias (defined as having ≥ 2 domains judged as high risk). Additional sensitivity analyses were performed by excluding studies published before 2017 to verify the stability of the findings using more recent evidence.

3. Results

3.1. Study selection and characteristics

In total, 1354 articles were identified, of which 899 remained after the duplicates were removed. We excluded 875 reports that did not meet the eligibility criteria. Finally, 21 studies were available for the NMA (Fig. 1).^[21–41] General characteristics of the included studies can be seen in Table 2. Included studies were published from 2014 to 2019. Sample sizes ranged from 21 to 96. The measurement time points for outcomes varied across the included studies, with most studies assessing outcomes within 1 to 2 weeks post-intervention. The specific measurement time points for each included study are summarized in Table S2, Supplemental Digital Content 2.

Figure 1.

Flow diagram of the literature selection process. n = number of records.

Table 2.

General characteristic of the included studies.

Author	Sample (n)		Male/Female		Mean age (yrs)		Intervention	Outcomes
	Intervention	Control	Intervention	Control	Intervention	Control
Wang 2019	67	67	NS	NS	61.5	63.3	A, B	2, 3
Yi 2010	51	51	29/22	31/20	45.2	46.2	A, C	1, 2
Zhang 2014	62	62	NS	NS	NS	NS	A, D	1, 2, 3
Li 2016	38	39	NS	NS	43.1	42.6	B, C	2, 3
Chen 2016	41	41	23/18	21/20	NS	NS	B, D	1, 2, 3
Wu 2015	88	89	NS	NS	NS	NS	B, D	2
Liu 2016	32	32	21/11	20/12	64.6	64.9	A, B	1, 2, 3
Qin 2016	40	40	22/18	20/20	NS	NS	A, B	2, 3
Fu 2016	70	76	44/26	48/28	71.7	71.6	A, B, C	1, 2, 3
Zhao 2016	93	93	61/32	65/28	52.5	53.3	B, D	1
Shi 2014	32	32	NS	NS	NS	NS	B, E	2, 3
Jin 2015	70	70	34/36	37/33	73.5	75.3	B, F	1
He 2016	74	74	41/33	45/29	71.7	73.2	B, D	1
Zhao D 2016	60	60	34/26	35/25	70.7	70.2	B, C	1, 2
Sun 2014	25	25	NS	NS	NS	NS	B, C, F	1, 2, 3
Long 2012	50	50	NS	NS	NS	NS	B, D	1, 2
Xu 2013	47	47	33/14	34/13	57.6	57.8	A, B, D	1, 2
Ma 2015	26	26	17/9	26/15	46.7	47.5	A, B, C	1, 2, 3
Wang 2015	45	45	28/17	26/19	69.8	69.9	B, D	1, 2, 3
Liang 2015	49	49	NS	NS	NS	NS	B, D	1, 2, 3
Qian 2014	21	23	13/8	17/10	NS	NS	B, C, F	1, 2, 3

A: auricular points acupressure, B: music therapy, C: cognitive function exercise, D: regular nursing, E: increasing visiting hours, F: targeted nursing.

1: cure rate, 2: ICDSC, 3: APACHE II.

APACHE II = acute physiologic assessment and chronic health evaluation II, ICDSC = intensive care delirium screening checklist, n = number of studies, NS = not stated.

3.2. Risk of bias in the included studies

The risk of bias summary and risk of bias graph are shown in Figure 2 and 3 respectively. Generation of random sequences was described in detail for 13 RCTs, and the method of allocation concealment was described in 1 RCT. Blinding was associated with a low risk of bias. Most RCTs showed a low risk of bias because their protocols and outcomes were well described in each study. Most items were assessed as unclear sources of bias because of insufficient information. The overall kappa value regarding the evaluation of risk of bias of included RCTs was 0.853, indicating an excellent degree of agreement between the 2 reviewers.

Figure 2.

Risk of bias summary of the included studies.

Figure 3.

Risk of bias graph of the included studies.

3.3. Cure rate

A total of 17 studies, including 6 treatments (music therapy, cognitive function exercise, targeted nursing, increasing visiting hours, auricular points acupressure, and regular nursing) contributed to the clinical outcome of the cure rate.

As displayed in Figure 4A, the network structure diagrams detailed the direct comparisons between different treatments in the cure rate. NMA showed considerable heterogeneity with global I² = 0% (Fig. 4B).

Figure 4.

(A) Network structure diagrams of cure rate; (B) forest plot of the cure rate as compared with regular nursing; (C) SUCRA probabilities of different treatments for cure rate: auricular points acupressure, music therapy, cognitive function exercise, regular nursing, increasing visiting hours, targeted nursing. SUCRA = surface under the cumulative ranking area.

In head-to-head comparison, there was no significant difference between music therapy, cognitive function exercise, targeted nursing, increasing visiting hours, or auricular points acupressure when compared with regular nursing (P > .05, Table 3).

Table 3.

Efficacy of different comparisons for cure rate by ORs and corresponding 95% CrIs.

Auricular points acupressure	1.27 (0.72, 2.22)	1.17 (0.66, 2.01)	1.47 (0.81, 2.59)	2.12 (1.18, 3.77)	1.07 (0.52, 2.15)
0.79 (0.45, 1.39)	Music therapy	0.92 (0.48, 1.73)	1.15 (0.6, 2.17)	1.67 (0.91, 3.05)	0.84 (0.4, 1.75)
0.86 (0.5, 1.51)	1.09 (0.58, 2.08)	Cognitive function exercise	1.26 (0.64, 2.45)	1.82 (0.95, 3.53)	0.92 (0.44, 1.94)
0.68 (0.39, 1.23)	0.87 (0.46, 1.66)	0.79 (0.41, 1.55)	Regular nursing	1.44 (0.78, 2.72)	0.73 (0.34, 1.56)
0.47 (0.27, 0.85)	0.6 (0.33, 1.1)	0.55 (0.28, 1.06)	0.69 (0.37, 1.29)	Increasing visiting hours	0.51 (0.23, 1.13)
0.93 (0.46, 1.91)	1.19 (0.57, 2.49)	1.09 (0.52, 2.28)	1.37 (0.64, 2.91)	1.98 (0.89, 4.43)	Targeted nursing

CrI = credible intervals, OR = odds ratio.

The SUCRA shows that auricular points acupressure ranked first (SUCRA, 97.9%), targeted nursing ranked second (SUCRA, 79.6%), cognitive function exercise ranked third (SUCRA, 56.3%), music therapy ranked fourth (SUCRA, 50.4%), regular nursing ranked fifth (SUCRA, 48.3%), and increasing visiting hours ranked last (SUCRA, 40.8%, Fig. 4C)

3.4. Intensive care delirium screening checklist (ICDSC)

A total of 15 studies, including 6 treatments (music therapy, cognitive function exercise, targeted nursing, increasing visiting hours, auricular points acupressure, and regular nursing) contributed to the clinical outcome of the ICDSC.

As displayed in Figure 5A, the network structure diagrams detailed the direct comparisons between different treatments in the ICDSC. NMA showed considerable heterogeneity with global I² = 0% (Fig. 5B).

Figure 5.

(A) Network structure diagrams of ICDSC; (B) forest plot of the ICDSC as compared with regular nursing; (C) SUCRA probabilities of different treatments for ICDSC: auricular points acupressure, music therapy, cognitive function exercise, regular nursing, increasing visiting hours, targeted nursing. ICDSC = intensive care delirium screening checklist, SUCRA = surface under the cumulative ranking area.

Compared with regular nursing, cognitive function exercise significantly decreased the ICDSC with statistical significance (weighted nean difference = −282.25, 95% credible intervals: −553.39, −10.03); there was no significant difference between music therapy, targeted nursing, increasing visiting hours, or auricular points acupressure when compared with regular nursing (P > .05, Table 4).

Table 4.

Efficacy of different comparisons for ICDSC by WMDs and corresponding 95% CrIs.

Auricular points acupressure	−46.44 (−260.52, 168.35)	4.98 (−209.11, 218.21)	124.47 (−88.6, 338.14)	119.4 (−106.18, 345.52)	235.42 (−23.63, 492.47)
46.44 (−168.35, 260.52)	Music therapy	51.05 (−191.95, 295.73)	171.14 (−72.6, 413.72)	165.41 (−89.77, 423.64)	282.25 (10.03, 553.39)
−4.98 (−218.21, 209.11)	−51.05 (−295.73, 191.95)	Cognitive function exercise	119.8 (−122.43, 363.26)	114.32 (−139.83, 369.69)	230.58 (−40.51, 500.43)
−124.47 (−338.14, 88.6)	−171.14 (−413.72, 72.6)	−119.8 (−363.26, 122.43)	Regular nursing	−5.29 (−261.04, 250.39)	110.83 (−159.92, 381.55)
−119.4 (−345.52, 106.18)	−165.41 (−423.64, 89.77)	−114.32 (−369.69, 139.83)	5.29 (−250.39, 261.04)	Increasing visiting hours	115.93 (−190.54, 422.91)
−235.42 (−492.47, 23.63)	−282.25 (−553.39, −10.03)	−230.58 (−500.43, 40.51)	−110.83 (−381.55, 159.92)	−115.93 (−422.91, 190.54)	Targeted nursing

CrI = credible intervals, ICDSC = intensive care delirium screening checklist, WMD = weighted nean difference.

The SUCRA shows that music therapy ranked first (SUCRA, 93.5%); auricular points acupressure and targeted nursing ranked second (SUCRA, 79.6%); cognitive function exercise ranked third (SUCRA, 59.8%); increasing visiting hours ranked fourth (SUCRA, 46.8%); regular nursing ranked fifth (SUCRA, 35.6%); and targeted nursing ranked last (SUCRA, 20.5%, Fig. 5C)

3.5. Acute physiologic assessment and chronic health evaluation II (APACHE II)

A total of 15 studies, including 6 treatments (music therapy, cognitive function exercise, targeted nursing, increasing visiting hours, auricular points acupressure, and regular nursing) contributed to the clinical outcome of the APACHE II.

As displayed in Figure 6A, the network structure diagrams detailed the direct comparisons between different treatments in the APACHE II. NMA showed considerable heterogeneity with global I² = 0% (Fig. 6B).

Figure 6.

(A) Network structure diagrams of APACHE II; (B) forest plot of the APACHE II as compared with regular nursing; (C) SUCRA probabilities of different treatments for APACHE II: auricular points acupressure, music therapy, cognitive function exercise, regular nursing, increasing visiting hours, targeted nursing. APACHE II = acute physiologic assessment and chronic health evaluation II, SUCRA = surface under the cumulative ranking area.

In a head-to-head comparison, there was no significant difference between music therapy, cognitive function exercise, targeted nursing, increasing visiting hours, and auricular points acupressure when compared with regular nursing for APACHE II (P > .05, Table 5).

Table 5.

Efficacy of different comparisons for APACHE II by WMDs and corresponding 95% CrIs.

Auricular points acupressure	62.32 (−48.46, 172.03)	38.35 (−71.78, 148.5)	73.65 (−35.91, 183.3)	26.24 (−89.38, 142.04)	80.17 (−53.26, 212.68)
−62.32 (−172.03, 48.46)	Music therapy	−23.78 (−148.56, 101.15)	11.31 (−113.07, 136.39)	−35.81 (−167.86, 96.28)	17.56 (−121.13, 157.89)
−38.35 (−148.5, 71.78)	23.78 (−101.15, 148.56)	Cognitive function exercise	35.15 (−89.39, 159.96)	−12.34 (−142.62, 118.58)	41.36 (−97.66, 180.74)
−73.65 (−183.3, 35.91)	−11.31 (−136.39, 113.07)	−35.15 (−159.96, 89.39)	Regular nursing	−47.33 (−178.58, 84.13)	6.38 (−133, 144.89)
−26.24 (−142.04, 89.38)	35.81 (−96.28, 167.86)	12.34 (−118.58, 142.62)	47.33 (−84.13, 178.58)	Increasing visiting hours	53.62 (−103.66, 211.41)
−80.17 (−212.68, 53.26)	−17.56 (−157.89, 121.13)	−41.36 (−180.74, 97.66)	−6.38 (−144.89, 133)	−53.62 (−211.41, 103.66)	Targeted nursing

APACHE II = acute physiologic assessment and chronic health evaluation II, CrI = credible intervals, WMD = weighted nean difference.

The SUCRA shows that auricular points acupressure ranked first (SUCRA, 95.5%), increasing visiting hours ranked second (SUCRA, 80.4%), cognitive function exercise ranked third (SUCRA, 62.5%), music therapy ranked fourth (SUCRA, 52.4%), regular nursing ranked fifth (SUCRA, 44.3%), and increasing visiting hours ranked last (SUCRA, 42.5%, Fig. 6C).

3.6. Sensitivity analyses

After excluding studies with a high risk of bias (≥ 2 domains judged as high risk), the results remained consistent with the primary analysis across all 3 outcomes. Similarly, the sensitivity analysis restricted to studies published in 2017 and later showed no substantial change in the ranking of interventions, although the reduced sample size resulted in wider credible intervals. Subgroup analyses based on intervention duration (≤ 1 week vs > 1 week) and patient age (< 65 years vs ≥ 65 years) did not reveal significant effect modification, suggesting that the findings were robust to these sources of clinical heterogeneity (Table S3, Supplemental Digital Content 3).

4. Discussion

The main findings of this NMA were as follows: among the 5 intensive nursing interventions studied, all except for increasing visiting hours significantly increased the cure rate compared to regular nursing intervention; auricular points acupressure ranked as the most effective intervention in increasing the cure rate and decreasing APACHE II; and music therapy ranked as the most effective nursing intervention in reducing the ICDSC. However, it is important to emphasize that the SUCRA values reflect the relative ranking probability of each intervention rather than absolute efficacy. In head-to-head comparisons, none of the intensive nursing interventions demonstrated a statistically significant difference compared to regular nursing (P > .05), with the exception of cognitive function exercise for ICDSC. Therefore, the rankings should be interpreted with caution and should not be taken as definitive evidence of superiority.

Compared to previous meta-analyses, our study possesses several strengths. Firstly, we conducted a NMA, which allowed us to synthesize both direct and indirect evidence, providing valuable insights for clinicians in their decision-making process. Secondly, our NMA encompassed ICU patients with delirium, whereas the majority of relevant meta-analyses were limited to the analysis of delirium patients only. Burry et al^[11] conducted a NMA of pharmacological and non-pharmacological interventions for delirium prevention in critically ill patients and found that multicomponent non-pharmacological interventions were most effective. Our study extends this evidence by focusing specifically on intensive nursing interventions and comparing individual modalities such as auricular points acupressure and music therapy, which were not separately evaluated in previous reviews. The findings of our study are consistent with previous pairwise meta-analyses that reported potential benefits of music therapy in reducing delirium incidence, though the effect sizes in those studies were also modest and often nonsignificant.

ICU is a relatively confined space, isolated from the outside world, which leads to restricted family visits.^[42,43] The high volume of false alarms in the ICU creates a noisy environment, causing tension, nervousness, and fear among ICU patients.^[44] These factors contribute to the incidence of delirium in ICU patients.^[45,46] Therefore, prompt intervention to reduce the incidence of delirium is crucial for ICU patients. Currently, various intensive nursing interventions are being performed to mitigate delirium in the ICU. However, the optimal strategy remains largely unknown.

Auricular points acupressure, also known as ear acupressure, is a technique that involves applying pressure to specific points on the outer ear to stimulate different parts of the body.^[47,48] It is based on the principles of Traditional Chinese Medicine, which views the ear as a microsystem of the whole body, with each point on the ear corresponding to a different organ or body part.^[49] Auricular acupressure can be used to treat a variety of conditions, including pain, anxiety, addiction, and insomnia. The pressure can be applied using the fingers, a small tool, or specialized ear seeds or magnets. Auricular acupressure is generally considered safe, but it should be avoided if you have a perforated eardrum or other ear infections. It is also important to consult with a qualified practitioner if you have any underlying health conditions or are pregnant. This NMA identified that auricular points acupressure ranked first in increasing cure rate and decreasing APACHE II for ICU patients. A recent study has shown that auricular points acupressure has a positive role in improving sleep quality and neuroendocrine level.^[50]

As for APACHE II, we found that auricular points acupressure ranked first (SUCRA, 95.5%), and increasing visiting hours ranked second for reducing APACHE II. As we know, higher APACHE II scores were associated with poor survival.^[51] Auricular points acupressure is defined as a healthcare modality whereby the external surface of the ear, or auricle, is stimulated to alleviate pathological conditions in other parts of the body.^[52] The reduction in mortality achieved by auricular points acupressure is consistent with reductions seen in other studies.^[53]

The potential mechanisms by which auricular points acupressure may improve delirium outcomes are multifaceted. From a neurobiological perspective, auricular acupressure has been shown to modulate neurotransmitter systems involved in delirium pathophysiology. Specifically, stimulation of auricular acupoints has been associated with increased release of endogenous opioids and serotonin, which may help regulate the sleep-wake cycle and reduce agitation. Furthermore, auricular acupressure has been demonstrated to reduce systemic inflammatory markers, including interleukin-6 and tumor necrosis factor-alpha, which are elevated in ICU patients with delirium. The vagus nerve, which has auricular branches, may serve as the anatomical pathway mediating these anti-inflammatory effects through the cholinergic anti-inflammatory pathway. Additionally, recent studies have shown that auricular acupressure can improve sleep quality by increasing total sleep time and slow-wave sleep, which may indirectly reduce delirium risk given the strong association between sleep disruption and delirium in ICU patients.

Regarding music therapy, our finding that it ranked first in reducing ICDSC scores is consistent with previous studies. The mechanisms underlying the beneficial effects of music therapy on delirium may involve several pathways. Music therapy has been shown to reduce sympathetic nervous system activity, lower cortisol levels, and decrease the release of pro-inflammatory cytokines. A reduction in anxiety and agitation through music-induced relaxation may also contribute to lower ICDSC scores. Notably, most studies included in this review utilized slow-tempo classical or ambient music, with sessions typically lasting 30 to 60 minutes per day. However, the optimal music type, duration, and frequency remain to be established, and future studies should explore whether specific music characteristics differentially impact delirium outcomes.

The patient was admitted to the ICU with a severe illness, possibly as a side effect of medication. The use of the drug, coupled with the unfamiliar ICU environment and the underlying disease, can give rise to a range of issues, including hallucinations, erratic behavior, hypervigilance, anxiety, fear, and even retrograde amnesia.^[54] Addressing these challenges necessitates an increase in family visitation time while still maintaining flexibility in controlling the number of visits. Furthermore, it is paramount to prioritize the patients’ needs and extend communication time consciously. By doing so, we can effectively engage the patient, offering support to alleviate tension, anxiety, and emotional distress. This approach ensures the patient’s overall well-being, fosters a calm and stable state, and helps prevent the prolongation of hallucinations. Equally significant is allowing patients to experience the utmost care and support from their loved ones. Therefore, increasing visiting time appears to hold promise as an effective intervention for reducing delirium in the ICU.

From a safety and feasibility perspective, the intensive nursing interventions evaluated in this study are generally low risk and can be readily implemented in ICU settings. Auricular points acupressure is noninvasive, with minimal adverse effects such as mild local discomfort or skin irritation, which are typically self-limiting. Music therapy is also safe, though care should be taken to avoid excessive volume in patients with hearing sensitivities. Increasing visiting hours requires organizational policy adjustments but incurs no additional cost, making it a particularly attractive strategy for resource-limited settings. All these interventions can be administered by trained nursing staff without the need for specialized equipment, enhancing their feasibility for widespread clinical adoption.

Our study also had several limitations. First, some of the included subgroups were too small to evaluate effectively. Thus, several subgroup analyses were not performed. Second, patient characteristics that may have resulted in unavoidable methodological heterogeneity, such as age, sex, and severity of delirium, could not be further addressed by subgroup or sensitivity analyses. Third, many different medications were used, and there may have been differences in dosage among the studies. Fourth, we attempted to explore potential sources of clinical heterogeneity through subgroup analyses and meta-regression. Although the global I² = 0% suggested low statistical heterogeneity, clinical heterogeneity arising from variations in intervention protocols could not be fully addressed. For example, cognitive function exercise protocols varied from memory training to problem-solving tasks, with intervention frequencies ranging from daily to twice weekly. This operational heterogeneity may affect the validity of the pooled effect estimates and should be considered when interpreting the findings. Fifth, the literature search was updated to December 2024, but all included studies were published between 2014 and 2019. This is likely because many recent RCTs on ICU delirium nursing interventions conducted between 2020 and 2024 were either published in Chinese-language journals not indexed in the databases we searched or utilized study designs that did not meet our strict inclusion criteria for RCTs. We also note that the coronavirus disease 2019 pandemic may have shifted research priorities during this period, potentially reducing the number of published RCTs on non-pharmacological delirium interventions. The sensitivity analysis excluding studies published before 2017 showed that the overall ranking of interventions remained consistent, although the reduced sample size limited statistical power. Future updates of this review should incorporate any newly published trials to confirm the stability of these findings.

Based on the findings of this NMA, we propose the following recommendations for ICU delirium management: given its favorable ranking across multiple outcomes and its low cost and noninvasive nature, auricular points acupressure could be considered as an adjunctive intervention in ICU delirium care, particularly in settings where Traditional Chinese Medicine practices are integrated with conventional care; music therapy, which ranked highest for ICDSC reduction, may be incorporated into routine ICU nursing protocols, with standardized procedures specifying the type, duration, and frequency of music sessions; ICU policies should consider flexible visiting hours, as increasing family visitation time ranked second in reducing APACHE II scores and represents a low-cost intervention with additional psychosocial benefits for patients; and future clinical practice guidelines for ICU delirium should include a discussion of these intensive nursing interventions as part of a comprehensive, multicomponent delirium prevention and management strategy. Implementation of these interventions should be accompanied by appropriate staff training and ongoing evaluation of patient outcomes.

5. Conclusions

In conclusion, this NMA suggests that auricular points acupressure might be the optimal type to increase the cure rate and decrease the ICDSC and APACHE II in ICU patients. However, this conclusion must be interpreted with caution, as the SUCRA rankings reflect relative probabilities rather than definitive evidence of superiority, and no statistically significant differences were observed in most head-to-head comparisons with regular nursing. This conclusion may also be affected by the limitations of this study owing to the small sample size, clinical heterogeneity of intervention protocols, and the absence of recent RCTs. Further research, particularly large-scale, high-quality RCTs with standardized intervention protocols and consistent outcome measurement time points, should be conducted to eliminate these limitations and to confirm the findings.

Author contributions

Project administration: Jingmin Huang.

Resources: Jingmin Huang.

Software: Jiang Wu, Jingmin Huang.

Supervision: Jiang Wu.

Validation: Rui Wang, Jiang Wu.

Visualization: Rui Wang.

Writing – original draft: Rui Wang.