Non-pharmacological ICU nursing interventions and melatonin concentrations: a systematic review and meta-analysis of randomized controlled trials

Abstract

Background

Sleep disruption and circadian dysregulation are common in patients admitted to the intensive care unit (ICU) and are associated with adverse recovery outcomes. Non-pharmacological nursing strategies such as environmental control and clustered care are widely implemented to reduce nocturnal disturbance, yet their physiological effects remain uncertain. Melatonin may serve as a potential marker of circadian processes; however, its interpretation depends on temporal dynamics that are not consistently captured across studies.

Methods

This systematic review and meta-analysis of randomized controlled trials evaluated the effects of non-pharmacological ICU nursing interventions on melatonin concentrations, with secondary analyses of cortisol levels and sleep quality. PubMed, Embase, Web of Science, Scopus, and the Cochrane Library were searched from inception to February 7, 2026. Random-effects meta-analysis using Hedges’ g was performed. Risk of bias was assessed using the Cochrane RoB 2.0 tool and certainty of evidence using GRADE. The review was prospectively registered in PROSPERO (CRD420261303901) in February 2026.

Results

Five randomized trials were included. Non-pharmacological nursing interventions were associated with higher measured melatonin concentrations compared with usual care (Hedges’ g = 0.28, 95% CI 0.08–0.49, p = 0.01; I² = 5.4%). However, in the absence of temporal profiling, this finding cannot be interpreted as evidence of circadian alignment. Cortisol outcomes showed no significant difference (Hedges’ g = − 0.11, 95% CI − 0.83–0.61, p = 0.76; I² = 86.8%). Sleep quality outcomes were inconclusive, with substantial heterogeneity across studies (Hedges’ g = 1.22, 95% CI − 0.05–2.48, p = 0.06; I² = 94.3%). Certainty of evidence was low for melatonin and very low for other outcomes due to risk of bias, imprecision, and inconsistency.

Conclusions

Non-pharmacological ICU nursing care was associated with modest increases in melatonin concentrations; however, the clinical significance of this finding remains uncertain, and no consistent benefits were observed for perceived sleep quality or cortisol concentrations. These findings suggest that such interventions may be associated with higher nocturnal melatonin concentrations, although their effects on patient-centered clinical outcomes remain unclear. Further large, well-designed trials are required to confirm these findings.

PROSPERO registration number

Registered in PROSPERO (CRD420261303901).

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12912-026-04938-9.

Introduction

Sleep disturbance and circadian dysregulation are common among patients admitted to the intensive care unit (ICU). Continuous exposure to artificial light, elevated noise levels, frequent clinical and nursing activities, and limited exposure to natural day–night cues may interfere with normal sleep physiology. Previous studies have associated poor sleep in ICU patients with adverse outcomes including delirium, impaired immune responses, prolonged mechanical ventilation, and delayed recovery trajectories.

Melatonin is a neuroendocrine hormone involved in the regulation of circadian rhythms and the sleep–wake cycle. Under physiological conditions, melatonin concentrations follow a nocturnal pattern; however, this rhythm is often attenuated, delayed, or fragmented in critically ill individuals [2]. Several ICU-related factors—such as continuous light exposure, sedative medications, mechanical ventilation, and disruption of daily routines—have been proposed as contributors to abnormal melatonin profiles [7]. Consequently, melatonin has been considered a potential objective marker reflecting interactions between ICU care processes and circadian regulation.

To address sleep disturbance without increasing sedative exposure, various non-pharmacological ICU nursing interventions have been introduced. These include the use of earplugs and eye masks, environmental noise reduction, light adjustment, clustering of nursing activities, and circadian-oriented nursing care. Individual trials suggest that such interventions may influence physiological markers, while evidence for patient-reported sleep improvement remains inconsistent in ICU patients [8]. However, reported effects vary across studies and their consistency remains uncertain.

Although sleep-related outcomes are commonly evaluated using subjective scales or observational measures, fewer studies have examined hormonal indicators that may help explain underlying mechanisms. Cortisol, another circadian-related hormone associated with stress responses, has been investigated in some ICU studies, but findings remain inconsistent across populations and intervention types. To date, evidence has not been systematically synthesized to clarify whether non-pharmacological ICU nursing interventions influence melatonin concentrations, cortisol regulation, and sleep quality simultaneously.

Accordingly, this systematic review and meta-analysis aimed to quantitatively evaluate the effects of non-pharmacological ICU nursing interventions on melatonin concentrations, with secondary analyses of cortisol levels and sleep quality outcomes. By integrating physiological and clinical indicators, this study seeks to clarify the potential role of routine nursing care in influencing melatonin concentrations, cortisol levels, and sleep-related outcomes in critically ill patients.

Methods

Eligibility criteria

According to the PICOS framework, studies were considered eligible if they met the following criteria: (P) adult patients admitted to an intensive care unit (ICU); (I) non-pharmacological nursing interventions aimed at sleep or circadian support, including sleep facilitation strategies, environmental modification, circadian-oriented nursing care, or multicomponent nursing protocols; (C) comparison with usual ICU care or standard nursing practice; (O) reporting at least one of the following outcomes: melatonin concentrations, cortisol levels, or sleep quality; (S) randomized controlled trials. The review protocol predefined eligibility criteria, outcomes, and subgroup analyses, and no post hoc modifications were made after study selection.

Exclusion criteria

Studies were excluded if they involved pediatric populations or animal models. Trials evaluating pharmacological interventions, including exogenous melatonin or other sleep-related medications, were not eligible. Non-comparative designs (single-arm studies, case reports, reviews, conference abstracts, and editorials) were excluded. Studies were also excluded when outcome data could not be extracted or converted into summary measures. Articles focusing on outcomes unrelated to sleep or circadian regulation, duplicate publications, or overlapping populations were removed.

Information sources and search strategy

The review followed PRISMA-S guidelines. The following databases were searched from inception to February 7, 2026: PubMed, Embase, Web of Science, Scopus, and the Cochrane Library.

Search strategy

The search strategy combined controlled vocabulary (e.g., MeSH terms in PubMed and Emtree terms in Embase) and free-text keywords across four predefined conceptual domains.

1. Nursing and environmental interventions, including terms related to nursing care, environmental modification, light therapy, dynamic lighting, light exposure, and sleep hygiene.

2. Intensive care settings and populations, including Intensive Care Units (ICUs), critical care, and critically ill patients.

3. Circadian rhythm–related outcomes, including circadian rhythm, circadian regulation, chronobiology, melatonin, and biological rhythms.

4. Randomized study designs, including randomized controlled trials, randomization, randomized, randomised, and trial-related terms.

The PubMed search incorporated both MeSH terms (e.g., “Intensive Care Units” and “Circadian Rhythm”) and free-text keywords. Equivalent controlled vocabulary and database-specific indexing terms were adapted for Embase (Emtree), while keyword-based strategies were used for Scopus, Web of Science, and the Cochrane Library. Search syntax was modified as appropriate for each database while maintaining the same conceptual framework.

The full electronic search strategies for all databases are provided in the Supplementary Materials to ensure transparency and reproducibility.

No language or publication status restrictions were applied. Reference lists of included studies and relevant reviews were manually screened. Full search strategies are provided in Supplementary Appendix 1. To enhance completeness, reference lists of included studies and relevant review articles were manually examined. The literature search was updated prior to the final analysis to capture any newly published eligible trials. Completed PRISMA 2020 and PRISMA-S checklists are provided in the Supplementary Appendix 2 and Appendix 3.

Study selection

Two reviewers independently screened titles and abstracts, followed by full-text assessment of potentially eligible studies. Disagreements were resolved by discussion; when consensus was not reached, a third reviewer adjudicated. Inter-reviewer agreement during full-text screening was quantified using Cohen’s kappa coefficient.

Data extraction

Two reviewers independently extracted data using a standardized form. Extracted information included study characteristics, ICU population, sample size, intervention duration, and biological sample type for melatonin measurement. Interventions were categorized by components (light management, noise control, workflow modification, or multicomponent nursing care). For each outcome, mean values and dispersion measures were collected. For hormonal outcomes, only overall nocturnal or summary measurements representing the night period as defined by the original study were extracted. Studies reporting isolated single time-point concentrations (e.g., measurements at a specific clock time without an aggregated nocturnal estimate) were not included in quantitative synthesis for that outcome, because melatonin exhibits strong circadian variation, and single measurements without temporal context are not directly comparable across studies. The number of studies excluded for this reason is reported in the PRISMA flow diagram. For multiple intervention arms, groups were combined according to Cochrane recommendations. When dispersion measures were missing, they were calculated from available statistics where possible. Graphical data were digitized using WebPlotDigitizer. Outcome units were extracted whenever reported. For outcomes where measurement units were not specified in the original publication, this was indicated in Table 1 using an asterisk (*).

Table 1.

Patient characteristics and data of outcomes

Study ID	Country	ICU population	melatonin Type	specimen	duration (days)	Group	Sample Size (n)	Age (years)	Male (%)	cortisol	melatonin	sleep quality
Pelin [8]	US	Unselected	melatonin	saliva	1	circadian intervention	22	67.57 ± 12.53	72.7	96,236 ± 51,259*	65.9 ± 51.9*	22.41 ± 6.67 (SQ-CC)
						usual	22	60.36 ± 11.28	77.3%	140,186 ± 100,925*	83.1 ± 73.7*	50.45 ± 10.63 (SQ-CC)
Lin [7]	Taiwan	Unselected	MT6s	urine	2	sleep facilitation intervention	55	57.92 ± 12.61	34.6	-	279.04 ± 409.59 (pg/ml)	60.64 ± 25.56 (RCSQ)
						usual	52	60.13 ± 13.21	41.1	-	172.08 ± 226.42 (pg/ml)	56.40 ± 25.35 (RCSQ)
Khoddam [6]	Iran	Unselected	MT6s	urine	4	sleep facilitation intervention	17	51.5 ± 9.2	47.1	−69.62 ± 147*	25.5 ± 31.8*	–13.06 ± 25.5 (VSH)
						usual	17	58.4 ± 8.5	53.3	31.19 ± 35.4*	4.25 ± 25.6*	-0.18 ± 25.5 (VSH)
Guo [2]	China	underwent surgery	melatonin	urine	3	circadian intervention	81	73.3 ± 6.1	38.2	4.93 ± 2.59 (ug/kg)	23.76 ± 15.26 (ug/kg)	-
						usual	79	73.7 ± 5.2	43	3.34 ± 1.83 (ug/kg)	18.15 ± 13.51 (ug/kg)	-
Hu [5]	China	underwent surgery	MT6s	urine	2	Sleep-environment interventions	20	56.8 ± 11.2	55	82.5 ± 47.3 (ug/kg)	8.3 ± 12.69 (ug/12 h)	23.7 ± 20.6 (sleep score)
						usual	25	56.6 ± 11	64	74.9 ± 56.3 (ug/kg)	7.1 ± 9.8 (ug/12 h)	54.0 ± 25.5 (sleep score)

*: unit not reported; VSH: Verran and Snyder-Halpern; SQ-CC: Sleep Quality Scale in Coronary Intensive Care Patients; RCSQ: Richards–Campbell Sleep Questionnaire; MT6s: urinary 6-sulfatoxymelatonin

Study risk of bias assessment

Risk of bias was assessed independently by two reviewers using the Cochrane Risk of Bias tool (RoB 2.0) for randomized controlled trials. The following five domains were evaluated: (1) bias arising from the randomization process, (2) bias due to deviations from intended interventions, (3) bias due to missing outcome data, (4) bias in measurement of the outcome, and (5) bias in selection of the reported result. Each domain was judged as “low risk”, “some concerns”, or “high risk”, and an overall risk-of-bias judgement was assigned for each outcome according to the RoB 2.0 algorithm. Risk of bias was assessed at the outcome level. Discrepancies between reviewers were resolved through discussion and consensus. Inter-rater agreement was evaluated using Cohen’s kappa statistic. The certainty of evidence for each outcome was assessed using the GRADE approach, considering risk of bias, inconsistency (I² > 50%), imprecision (e.g., wide confidence intervals or small sample size), indirectness, and publication bias.

Data synthesis

Meta-analysis was performed when at least two studies reported the same outcome. A random-effects model was selected a priori due to anticipated differences in patient populations, intervention components, and outcome assessment methods across ICU settings. Pre-specified subgroup analyses examined intervention type (sleep facilitation vs. circadian-oriented nursing care) and intervention duration. Sensitivity analyses were conducted using a leave-one-out approach to evaluate robustness of the pooled estimates. Clinical heterogeneity was expected because ICU nursing interventions are typically delivered as bundled care rather than isolated components.

Effect measures

Continuous outcomes were summarized using standardized mean differences (Hedges’ g) with 95% confidence intervals following Cochrane recommendations [3]. Effect direction was aligned so that positive values indicated benefit of the intervention. When different scales were used, data were standardized prior to pooling. Hormonal concentrations reported in different biological matrices were pooled using standardized mean differences; however, this approach does not fully account for physiological differences between urine, serum, and saliva measurements, and may introduce additional variability in interpretation.

Data synthesis and statistical analysis

All statistical analyses were performed using STATA (version 19). Random-effects meta-analyses were conducted using the DerSimonian–Laird method to account for expected clinical and methodological variability across studies. Statistical significance was defined as a two-sided p value < 0.05. When substantial heterogeneity was observed, pooled estimates were interpreted cautiously and considered together with subgroup and sensitivity analyses.

Assessment of heterogeneity

Statistical heterogeneity was assessed using Cochran’s Q test and quantified using the I² statistic. I² values were interpreted as low (25%), moderate (50%), and high (75%) heterogeneity. Given the limited number of included studies, heterogeneity estimates were interpreted cautiously [10].

Assessment of publication bias

Because fewer than 10 studies were available per outcome, funnel plot asymmetry and Egger testing were not performed according to Cochrane recommendations [4].

Results

Study selection

A total of 375 records were identified through database searching, including PubMed (n = 41), Web of Science (n = 89), Embase (n = 59), the Cochrane Library (n = 78), and Scopus (n = 54). After removal of 219 duplicate records, 156 records remained and were screened based on titles and abstracts. Of these, 145 records were excluded. Full texts of 11 reports were sought for retrieval and subsequently assessed for eligibility. All identified reports were successfully retrieved. Following full-text review, six studies were excluded, including five studies due to inability to extract relevant outcome data and one study that did not meet the inclusion criteria. Ultimately, five studies met the eligibility criteria and were included in the systematic review and meta-analysis. Some studies were not included in specific quantitative syntheses because only isolated single time-point hormonal measurements were reported, which were not comparable across studies according to the predefined extraction criteria. The study selection process is presented in the PRISMA 2020 flow diagram (Fig. 1).

Fig. 1.

The PRISMA flow diagram. PRISMA, preferred reporting items for systematic reviews and meta-analyses

Characteristics of included studies

A total of five randomized controlled trials were included in the final analysis, published between 2015 and 2025, and conducted in the United States [8], China [2, 5], Iran [6], and Taiwan [7]. All studies enrolled adult patients admitted to intensive or coronary care units, including general ICU populations as well as patients undergoing surgical procedures requiring postoperative intensive care. All interventions were delivered as part of routine bedside nursing care rather than investigator-driven procedures.

Sample sizes across studies ranged from 34 to 160 participants, with mean participant age generally in the mid-50s to early-70s. The proportion of male participants varied across trials, ranging from approximately 35% to 77%. Intervention durations ranged from 1 to 4 days, depending on study design and ICU stay.

The non-pharmacological nursing interventions were heterogeneous but shared a common focus on sleep promotion and circadian regulation. Interventions included combinations of light management (e.g., eye masks or circadian rhythm–aligned lighting), noise reduction strategies (e.g., earplugs, music), and workflow-related nursing adjustments aimed at minimizing nighttime interruptions. Several studies incorporated multicomponent approaches integrating both environmental modification and nursing care activities. Some studies implemented circadian-oriented strategies, combining eye masks with dynamic or circadian-aligned lighting to reinforce day–night cues [2, 8]. Others focused primarily on sleep facilitation, using noise-reduction measures such as earplugs, music, or eye masks without explicit circadian light modulation [5–7]. Overall, interventions represented modifications of routine night-time nursing practice rather than additional therapeutic procedures.

Melatonin was measured using either urine or saliva samples, while cortisol outcomes were reported in a subset of studies using urinary or serum assays. Sleep quality was assessed using validated instruments, including the Richards–Campbell Sleep Questionnaire and the Sleep Quality Scale in Coronary Care Units, or sleep score–based assessments, depending on the study. Detailed characteristics of the included studies, intervention components, and outcome measures are summarized in Tables 1 and 2.

Table 2.

Details of intervention

Study ID	Light management	Noise control	Enteral nutrition	Workflow and care activities
Pelin [8]	eye masks +circadian rhythm light	earplugs+ circadian rhythm sound	mimic normal circadian patterns	minimize nighttime interruptions
Lin [7]	-	earplugs	-	-
Khoddam [6]	eye masks	earplugs	-	minimize nighttime interruptions
Guo [2]	eye shades+ circadian rhythm light	earplugs+ circadian rhythm sound	mimic normal circadian patterns	minimize nighttime interruptions
Hu [5]	eye masks	earplugs +music	-	-

Risk of biases assessment

The visual representation of bias assessment is shown as a risk of bias graph (Fig. 2). Overall, most domains were judged as low risk, particularly for the randomization process, deviations from intended interventions, and missing outcome data. Some concerns were mainly observed in the measurement of the outcome and selection of the reported results.

Fig. 2.

Graph for the risk of bias

Across outcomes, the risk-of-bias pattern was generally consistent for sleep quality, cortisol, and melatonin, although slightly more concerns were noted in outcome measurement for subjective outcomes such as sleep quality. No domains were judged as high risk, although several domains were rated as “some concerns”. A detailed study-level “traffic light” (print view) of the risk-of-bias assessment is provided in the Supplementary Appendix 4.

Outcome

Melatonin

Five studies were included in the meta-analysis of melatonin concentrations. Non-pharmacological ICU nursing interventions were associated with higher measured melatonin concentrations compared with usual care (Hedges’ g = 0.28, 95% CI 0.08 to 0.49, p = 0.01), with low heterogeneity (I² = 5.4%) (Fig. 3A).

Fig. 3.

Forest plots of nursing intervention effects on outcomes. Standardized mean differences (Hedges’ g). (A) Melatonin; (B) Cortisol; (C) Sleep quality

Sensitivity analysis showed that the direction of effect remained unchanged after removal of individual studies. Subgroup analysis suggested a larger effect in sleep facilitation interventions than circadian-aligned care (Table 3).

Table 3.

sensitivity analysis and subgroup for melatonin outcome

Leave-one-out sensitivity
Study omitted	Estimate	[95% Conf.	Interval]	I²	tau²
Pelin [8]	0.360	0.149	0.572	0	0
Guo [2]	0.212	-0.130	0.554	41.7	0.0465
Lin [7]	0.242	-0.111	0.596	48	0.0604
Khoddam [6]	0.228	-0.011	0.467	26.9	0.0120
Hu [5]	0.299	0.030	0.569	44.1	0.0232
subgroup analysis by intervention
subgroup (study numbers)	Estimate	[95% Conf.	Interval]	I²	tau²
circadian (2)	0.110	-3.987	4.207	73.26	0.156
sleep facilitation (3)	0.338	-0.267	0.942	0	0

Cortisol

Four studies reported cortisol outcomes. No statistically significant difference was observed between intervention and control groups (Hedges’ g = − 0.11, 95% CI − 0.83 to 0.61, p = 0.76), with substantial heterogeneity (I² = 86.8%) (Fig. 3B). Subgroup and sensitivity analyses did not materially change the direction of the pooled estimate (Table 4).

Table 4.

sensitivity analysis and subgroup for cortisol outcome

Leave-one-out sensitivity for cortisol
Study omitted	Estimate	[95% Conf.	Interval]	I²	tau²
Pelin [8]	0.014	-0.911	0.939	89	0.5915
Guo [2]	-0.419	-1.025	0.188	65.2	0.1874
Lin [7]	0.136	-0.581	0.853	85.6	0.3358
Hu [5]	-0.214	-1.201	0.772	92.2	0.6816
subgroup analysis for cortisol by intervention
subgroup (study numbers)	Estimate	[95% Conf.	Interval]	I²	tau²
circadian (2)	0.11	-7.78	8	32.41	0.71
sleep facilitation (2)	-0.37	-7.12	6.38	81.28	0.46

sleep quality

Four studies assessed sleep quality. The pooled estimate did not reach statistical significance (Hedges’ g = 1.22, 95% CI − 0.05 to 2.48, p = 0.06) and showed considerable heterogeneity (I² = 94.3%), indicating that the evidence remains inconclusive (Fig. 3C). Subgroup and sensitivity analyses yielded consistent findings (Table 5).

Table 5.

sensitivity analysis and subgroup for sleep quality outcome

Leave-one-out sensitivity
Study omitted	Estimate	[95% Conf.	Interval]	I² (%)	tau²
Pelin [8]	0.602	-0.049	1.253	76.7	0.2493
Khoddam [6]	1.476	-0.190	3.142	95	2.0568
Lin [7]	1.591	0.086	3.096	91	1.6302
Hu [5]	1.216	-0.586	3.018	94.6	2.4210
subgroup analysis for by intervention
subgroup (study numbers)	Estimate	[95% Conf.	Interval]	I² (%)	tau²
circadian (1)	3.103	2.233	3.973	-	-
sleep facilitation (3)	1.217	-0.857	3.292	76.01	0.249
subgroup analysis by duration
subgroup (study numbers)	Estimate	[95% Conf.	Interval]	I² (%)	tau²
2 days (2)	0.69	-6.31	7.69	88.88	0.54

GRADE level

According to GRADE assessment, certainty of evidence was low for melatonin and very low for cortisol and sleep quality, primarily due to risk of bias and imprecision, and additionally due to inconsistency for cortisol and sleep quality outcomes. Detailed assessments are presented in Table 6.

Table 6.

GRADE level

	study limitation	imprecision	inconsistency	indirectness	publication bias	GRADE
Melatonin	downgrade due to high risk of bias in blinding and allocation concealment	no downgrade	no downgrade	no downgrade	downgraded as small group studies	low
Cortisol	downgrade due to high risk of bias in blinding and allocation concealment	Downgraded due to wide confidence intervals and small number of contributing studies.	downgrade as I2= 86.78%	no downgrade	downgraded as small group studies	very low
Sleep quality	downgrade due to high risk of bias in blinding and allocation concealment	Downgraded due to wide confidence intervals and small number of contributing studies.	downgrade as I2= 94.20%	no downgrade	downgraded as small group studies	very low

Discussion

This review evaluated whether routine ICU nursing care is associated with measurable changes in melatonin concentrations, cortisol concentrations, and sleep-related outcomes. Across the included trials, non-pharmacological nursing interventions were associated with a modest increase in melatonin concentrations.

Melatonin is a dynamic circadian signal characterized by timing, phase, and amplitude; therefore, increases in concentration alone cannot be interpreted as evidence of circadian recovery or alignment. These findings suggest a possible association between nursing care organization and melatonin-related measures, which should be interpreted cautiously [9].

Melatonin reflects aspects of circadian physiology but cannot by itself establish circadian alignment in the absence of temporal information. Interventions such as light minimization, eye masks, and clustering of nighttime activities may be associated with higher nocturnal melatonin concentrations, although the underlying mechanisms remain uncertain. This distinction may explain why physiological markers demonstrated more consistent effects than subjective sleep reports in critically ill populations exposed to continuous care-related stimuli.

From a nursing practice perspective, the results support the role of routine bedside care organization in optimizing the nighttime care environment. Structured night-time care, reduction of avoidable awakenings, and environmental modification can be implemented without additional equipment or pharmacological treatment and may be incorporated into standard supportive care protocols [1]. The findings also suggest that consistency of implementation may be more important than individual intervention components, supporting bundled sleep-promotion approaches in ICU nursing practice. Similar observations in other nursing populations indicate that non-pharmacological care may produce measurable physiological effects beyond comfort improvement [11]. The absence of consistent improvements in subjective sleep outcomes may reflect competing arousal stimuli inherent to critical illness rather than ineffectiveness of the interventions. Sleep perception in ICU patients is influenced by disease severity, sedation, and recall bias, whereas hormonal markers may more directly reflect environmental circadian organization. Future research should determine whether changes in melatonin concentrations are associated with clinically meaningful outcomes such as delirium reduction, ventilator duration, or recovery trajectory. The substantial heterogeneity observed in cortisol and sleep quality outcomes likely reflects differences in outcome measurement characteristics rather than inconsistent biological effects. Sleep quality was assessed using multiple subjective instruments with varying scoring ranges and recall periods, and ICU patients often have impaired perception due to sedation and illness severity. In contrast, melatonin demonstrated low heterogeneity, possibly because it represents a direct physiological marker less influenced by perception or reporting bias. Additionally, hormonal sampling schedules differed across studies, and circadian hormones exhibit pronounced temporal variation, which may further contribute to variability in cortisol estimates across trials. The absence of statistical significance should not be interpreted as absence of effect, but rather as insufficient or inconsistent evidence.

Several limitations should be acknowledged. Melatonin concentrations represent only one aspect of circadian physiology. Circadian phase, amplitude, and temporal patterns were not consistently captured across studies and therefore could not be evaluated. The number of included trials was small, and most studies were single center with limited sample sizes, reducing precision of pooled estimates. Some concerns were mainly related to the randomization process and measurement of the outcome, particularly due to limited reporting of allocation concealment and lack of blinding inherent to the interventions. Hormonal measurements were not standardized across biological matrices, and pooling across urine, saliva, and serum may introduce physiological variability that is not fully addressed by statistical standardization.Clinical heterogeneity may also arise from differences in intervention composition and ICU populations. Some trials implemented bundled circadian-oriented nursing care, whereas others applied single sleep-facilitation components such as earplugs or eye masks. Furthermore, included populations ranged from postoperative patients to medically critically ill patients, who differ in sedation exposure, illness severity, and baseline circadian disruption. These factors may attenuate observable effects on subjective outcomes while preserving measurable physiological changes. Overall, non-pharmacological ICU nursing interventions were associated with higher melatonin concentrations but not with consistent improvements in patient-reported sleep outcomes. Given the small number of included studies, I² estimates should be interpreted cautiously, as heterogeneity statistics are unstable in meta-analyses with few trials. These findings suggest that organization of routine night-time nursing care may be considered as a potential supportive nursing strategy, although evidence for clinical benefit remains limited. However, the current evidence does not support definitive conclusions regarding clinical benefit, and interpretations should remain cautious.

Conclusion

Non-pharmacological ICU nursing interventions were associated with a small increase in melatonin concentrations, while no consistent effects were observed for cortisol concentrations or perceived sleep quality. These findings suggest a possible association between such interventions and nocturnal melatonin concentrations; however, the clinical significance of this observation remains uncertain. Further well-designed trials are needed to determine whether these interventions lead to meaningful improvements in patient outcomes, including sleep, delirium, and recovery.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

ICU

Intensive Care Unit

RCT

Randomized Controlled Trial

SMD

Standardized Mean Difference

CI

Confidence Interval

GRADE

Grading of Recommendations Assessment, Development and Evaluation

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

PRISMA-S

PRISMA Search Extension

VSH

Verran and Snyder-Halpern

SQ-CC

Sleep Quality Scale in Coronary Intensive Care Patients

RCSQ

Richards–Campbell Sleep Questionnaire

MT6s

Urinary 6-sulfatoxymelatonin

Author contributions

Conceptualization: Dan Cai, Bo Zheng Data curation: Dan cai, Xiaojing Zheng, Hui Ye Formal analysis: Xiaojing Zheng, Hui Ye Investigation: Xiaojing Zheng, Hui Ye Project administration: Xiaojing Zheng, Hui Ye Resources: Xiaojing Zheng Software: Xiaojing Zheng Supervision: Xiaojing Zheng Validation: Hui Ye Visualization: Hui Ye Writing – original draft: Dan cai, Hui Ye, Xiaojing Zheng, Bo Zheng Writing – review & editing: Dan cai, Hui Ye, Xiaojing Zheng, Bo Zheng

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Data availability

All data analyzed during this study are included in the published articles and supplementary materials of the studies included in this review. Additional data supporting the findings of this study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Not Applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.