What this is
- This systematic review evaluates the effectiveness of in neonatal and pediatric intensive care units.
- It compares single-component interventions, like cycled lighting, to complex multicomponent strategies.
- The review synthesizes data from 41 studies involving 2,548 participants, focusing on physiological outcomes and sleep quality.
Essence
- significantly improve physiological parameters and sleep duration in critically ill neonates. No significant differences were found between single and multicomponent interventions.
Key takeaways
- reduced heart rate and respiratory rate while improving oxygen saturation and sleep duration. Meta-analysis of 16 studies showed significant effects with standardized mean differences of -0.70 for heart rate and -0.75 for respiratory rate.
- Subgroup analysis revealed no statistically significant differences between single and multicomponent interventions across all outcomes. This suggests that simpler interventions may be as effective as complex bundles in neonatal populations.
- The overall certainty of evidence was rated low due to high risk of bias and statistical heterogeneity. This limits confidence in the precise magnitude of effects, although the interventions themselves were effective.
Caveats
- High heterogeneity (72–84%) among studies limits the reliability of pooled estimates. This reflects true variations in intervention protocols and populations.
- The predominance of studies from middle-income countries may restrict the generalizability of findings to other healthcare contexts.
- Publication bias could not be formally assessed due to the small number of studies per outcome, although visual inspection suggested potential asymmetry.
Definitions
- circadian interventions: Clinical strategies aimed at restoring biological rhythms, such as light-dark cycling and sensory modulation.
Simplified
Introduction
During early life, sleep constitutes a dynamic, active biological process that is fundamental to the maturation of the central nervous system (1, 2). During infancy, the brain undergoes rapid structural organization, which includes synaptogenesis, synaptic pruning, and myelination processes that are inextricably linked to the preservation of sleep architecture and circadian rhythmicity (3–7). Unlike adults, developing infants possess a unique sleep physiology characterized by high proportions of Rapid Eye Movement (REM) and Slow-Wave Sleep (SWS), which facilitate critical functions such as memory consolidation, sensory processing, and emotional regulation (8, 9). Consequently, the integrity of these cycles is a prerequisite for proper neurodevelopment; conversely, sleep fragmentation induces a state of non-restorative sleep that triggers neuroinflammation and has been linked to adverse long-term cognitive and behavioral trajectories (9, 10).
Physiologically, the human circadian system is organized as a hierarchical network governed by the Suprachiasmatic Nucleus (SCN) of the hypothalamus. As the central pacemaker, the SCN orchestrates internal timing by synchronizing peripheral clocks located in nearly every tissue and organ system, including the liver, heart, and immune cells (11). This internal alignment ensures that diverse physiological processes ranging from metabolic flux to hormone secretion, occur in a coherent temporal order (12). The formation of this complex system is a sequential process that begins long before birth. During gestation, the fetal SCN is structurally present but functionally dependent on maternal signals. Among these, the transplacental rhythms of melatonin, cortisol, and body temperature are highly relevant to synchronize its molecular clockwork (6, 7). Birth marks a critical biological transition: the developing brain must switch from this maternal regulation to autonomous rhythmicity, a process that requires exposure to high-amplitude environmental zeitgebers (time givers), particularly robust light–dark cycles and predictable social routines (6, 13). These external cues are not merely triggers for sleep; they are the essential drivers that wire the SCN to the rest of the brain, enabling the emergence of stable cortisol rhythms and endogenous melatonin secretion (2, 7). The successful establishment of these robust hormonal rhythms is critical, as it lays the foundation for physiological resilience and optimal health that extends well into adulthood (14, 15).
However, this delicate developmental program is frequently altered by non-normative life events such as hospital admissions due to preterm birth or critical illness. For the preterm infant, birth represents an abrupt, premature severance from the maternal circadian pacemaker. This separation strips the fetus of the transplacental signals that are essential for organizing the SCN, occurring precisely at a time when the brain is most vulnerable and normally shielded by the uterine environment (13). Instead of completing neurodevelopment in the rhythmic protection of the womb, these infants are thrust into the Neonatal or Pediatric Intensive Care Unit (NICU/PICU), environments that fundamentally disrupt this developmental trajectory. Instead of providing the rhythmic cues necessary for SCN maturation, these units present a “sensory mismatch” (8, 13). The immature nervous system is subjected to continuous artificial illumination, excessive auditory stimuli, and invasive 24-h care routines that lack temporal logic (16–18), producing a phenomenon termed circadian disruption or misalignment (12, 19, 20). Recent data confirms that light and noise levels in these units consistently exceed international recommendations. Organization guidelines suggest that noise levels in hospital ward rooms should not exceed a sound pressure level (SPL) of 35 decibels (dB-A) during the day and 30 dB-A at night, recommending that noise in hospital treatment rooms “remain as low as possible” (21). As a consequence, infants lack the environmental conditions required for proper circadian oscillations (10, 13).
To mitigate these adverse effects, the emerging field of Circadian Medicine has proposed specific clinical interventions (22, 23). Crucially, the literature distinguishes these from traditional sleep interventions. While sleep interventions primarily aim to increase the duration or depth of quiescence (often through sedation or environmental silencing) (24), circadian interventions are defined by their intent to restore the timing and amplitude of the biological rhythm. To systematically evaluate these approaches, we established an a priori conceptual framework distinguishing between two main categories. The first encompasses primary circadian interventions (e.g., light–dark cycling, chrononutrition), which are specifically designed to entrain the biological clock by mimicking lost maternal or environmental rhythmicity (8, 25). The second category involves supportive sensory modulation strategies (e.g., earmuffs, facilitated tucking, swaddling). While these supportive strategies primarily aim to reduce stress and sensory overload, they may exert indirect circadian benefits by preventing stimulus-induced sleep fragmentation. Together, these approaches aim to consolidate sleep into the biological night and promote active wakefulness during the biological day (26).
Despite this clear physiological rationale, translating these principles into clinical practice presents significant challenges. As highlighted in recent research, the ICU is a complex hospital environment where the implementation of circadian interventions is often hindered by competing clinical priorities and a lack of standardized protocols (27, 28). Consequently, efforts to protect sleep and circadian rhythms are frequently applied inconsistently or in isolation (e.g., using eye masks without controlling noise), failing to address the multisensory nature of the disruption (28).
This implementation gap raises a critical question regarding the complexity of the required solution. Recent systematic reviews suggest that single-component interventions often yield inconsistent results, potentially because they are insufficient to override the multifaceted circadian disruption of the ICU environment (29). Conversely, evidence from adult critical care suggests that combined or multicomponent bundles, which integrate environmental modification with care coordination, may be more effective in reducing delirium and improving sleep quality (30, 31).
Therefore, the objective of this systematic review is to evaluate the comparative effectiveness of combined circadian interventions versus single-component interventions on clinical recovery, sleep quality, and physiological stability in patients admitted to NICUs and PICUs. The results of this review will highlight priority areas for future research and clinical protocols. The goal is to shift current practice toward interventions that not only improve immediate recovery but also preserve the circadian foundations associated with better health outcomes in later life.
Methods
To write this systematic review, we conducted a literature search following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 (PRISMA 2020) (32).
Protocol registration
The protocol for this systematic review was registered a priori in the International Prospective Register of Systematic Reviews (PROSPERO) (Registration No. CRD4201091137) and posted on the Open Science Framework (OSF)1. Additionally, the protocol has been archived as a preprint in medRxiv (33) and was submitted to a peer-reviewed journal, where it currently remains under review. Deviations from the original protocol were implemented prior to study selection to minimize bias and improve the review’s scope. Specifically, language and date restrictions were removed, and eligibility criteria were refined to explicitly include both neonatal (NICU) and pediatric (PICU) intensive care populations.
Research question
This systematic review will answer the focused question: “In neonatal and pediatric patients hospitalized in intensive care units, are combined circadian interventions more effective than single circadian interventions for improving clinical recovery and well-being during the ICU (NICU/PICU) stay?”. The question was developed using the PICO acronym (34) as follows: (1) Population (P): Neonates (preterm or term) and pediatric patients (up to 18 years) admitted to Neonatal or Pediatric Intensive Care Units (NICU/PICU). (2) Interventions (I): Multicomponent circadian-based interventions (strategies combining multiple elements, such as light/dark cycles, noise reduction, and feeding schedules) aimed at enhancing recovery or well-being. (3) Comparators (C): Single-component circadian interventions (e.g., eye masks only, earplugs only) or standard care. (4) Outcomes (O): Clinical recovery or well-being, including length of hospital stay, sleep quality, feeding tolerance, and physiological circadian markers (e.g., cortisol, melatonin).
Eligibility criteria
Study eligibility criteria were established based on the structured PICOS framework (Population, Intervention, Comparator, Outcomes, and Study Design). The inclusion and exclusion criteria were as follows: (1) Population: The review included neonates (preterm or term) and pediatric patients (up to 18 years of age) admitted to intensive care settings, specifically Neonatal Intensive Care Units (NICUs) and Pediatric Intensive Care Units (PICUs). Studies involving mixed ages or mixed units were included only if data for the neonatal/pediatric intensive care population could be extracted or if the study population was predominantly neonatal/pediatric. (2) Interventions: We included studies evaluating circadian interventions aimed at enhancing patient recovery or well-being. These interventions encompassed light exposure regulation, sleep–wake cycle optimization, timing of feeding, and other strategies promoting circadian alignment. (3) Comparators: Eligible studies compared circadian interventions to standard care, no intervention, or alternative circadian approaches (e.g., single vs. multicomponent interventions). (4) Outcomes: Studies were required to report at least one clinical outcome related to patient recovery or well-being. Primary and secondary outcomes of interest included, but were not limited to, hospital length of stay, feeding tolerance, blood glucose regulation, sleep quality, circadian-regulated hormonal levels (e.g., cortisol, melatonin), and physiological parameters (e.g., blood pressure, body temperature, heart rate). (5) Study Design: We included randomized controlled trials (including cluster designs), quasi-experimental studies (e.g., interrupted time series, controlled before–after), and observational designs (cohort, case–control, cross-sectional) evaluating the impact of circadian interventions. (6) Language and Availability: No language or publication date restrictions were applied. Records in languages unfamiliar to the review team were screened using machine translation and, if potentially eligible, assessed by a native or fluent speaker. Only full-text articles (peer-reviewed or preprint) with sufficient methodological detail to support risk-of-bias assessment and data extraction were included. We excluded systematic, scoping, or narrative reviews; qualitative studies lacking quantitative patient-level outcomes; case reports; case series; editorials; and opinion pieces. Studies focusing exclusively on healthcare providers (e.g., nurses or physicians) or adult populations without a distinct neonatal/pediatric subgroup were also excluded. Furthermore, studies focusing solely on environmental factors (e.g., noise or lighting levels) without reporting patient-level clinical or physiological outcomes were not eligible.
Information sources
A comprehensive literature search was conducted, using the following electronic databases: MEDLINE (via PubMed), ScienceDirect (Elsevier), Cumulative Index to Nursing and Allied Health Literature (CINAHL, EBSCO), Cochrane Central Registry of Controlled Trials (CENTRAL, Wiley), Web of Science Core Collection, Scopus, LILACS, SciELO, and Epistemonikos. These databases were selected to provide broad coverage of published studies, encompassing both global and Latin American research contexts. In addition to the electronic database search, we searched the ClinicalTrials.gov↗ and WHO International Clinical Trials Registry Platform (ICTRP) databases to identify ongoing or completed trials. We also screened the reference lists of included studies and performed forward and backward citation chasing to identify additional relevant records. All sources were searched from inception to the date of search, with no language limits applied. The retrieval of all databases was conducted between October 01st-6th, 2025.
Search strategy
The search strategy combined controlled vocabulary from MeSH (Medical Subject Headings) and DeCS (Health Sciences Descriptors) with relevant free-text terms. The core search concepts included: (1) neonatal/pediatric intensive care (e.g., NICU, PICU, intensive care, critical care) and (2) circadian/sleep/chronobiology (e.g., circadian rhythm, chronotherapy, sleep, melatonin, light, noise, day–night). Boolean operators (AND, OR) were applied to refine the results. The strategy was adapted for each database to account for specific indexing systems, syntax, and available filters. The full, database-specific search strings are provided in. Supplementary Material 1
Selection process
All records retrieved from electronic databases were imported into Covidence (Veritas Health Innovation, Melbourne, Australia), a systematic review management platform, which was used to for initial duplicate detection and removal and streamline the screening process as recommended in the Cochrane Handbook for Systematic Reviews of Interventions (35). The study selection was conducted in three stages (1) title screening, (2) abstract screening, and (3) full-text review. Prior to formal screening, a pilot calibration exercise was performed to ensure inter-rater reliability. All reviewers independently screened a random sample of 10 records. As the disagreement rate was less than 10% (indicating >90% concordance), the team proceeded to the formal screening phase. In the first stage, all reviewers independently distributed the data and screened titles and abstracts against the eligibility criteria. In the second stage, full-text articles of potentially relevant records were retrieved and assessed independently as described. Disagreements at any stage were resolved through arbitration by a third reviewer (L. B. G.). Reasons for exclusion at the full-text stage were documented and are reported in the PRISMA flow diagram (Figure 1).
PRISMA 2020 flow diagram illustrating the literature search and study selection process.
Data collection process
Data extraction was performed using a standardized electronic form developed specifically for this review within the Covidence platform. Prior to full implementation, the form was piloted on a subset of eligible studies to ensure clarity and consistency among the review team. Following this calibration, the included studies were distributed among the reviewers for data extraction. Each study was processed by a single reviewer, and any uncertainties or ambiguities regarding the extracted data were resolved through adjudication by a third reviewer (L. B. G.). Although a protocol was established to contact authors for missing or unclear data, this step was not required as all necessary information was available in the published reports.
Data items
The following variables were extracted from each included study: (1) General characteristics: Study title, lead author, year of publication, country or region, original language, and sponsorship or funding source. (2) Methods: Study design (e.g., Randomized Control Trials (RCT), crossover, retrospective cohort), group configuration, total sample size, follow-up duration, and methodological notes relevant to bias assessment. (3) Population: Inclusion and exclusion criteria, primary clinical diagnosis/condition, age (including gestational age), sex distribution, care unit setting (NICU/PICU), and participant withdrawals. (4) Intervention details: Type of circadian intervention (categorized as single vs. multicomponent), detailed description of the procedure, frequency, intensity (e.g., light lux levels, noise decibels), and total duration of the intervention. (5) Comparators: Description of the control group conditions, such as standard care or alternative environmental settings. (6) Outcomes: Clinical and physiological outcomes categorized as follows: Clinical Recovery: Hospital length of stay. Growth and Nutrition: Weight gain, head circumference, length, and feeding tolerance. Sleep: Quantity (total sleep time) and quality of sleep. Physiological Parameters: Heart rate, respiratory rate, oxygen saturation (SpO2), blood pressure, and heart rate variability. Biomarkers: Hormonal levels such as cortisol or melatonin.
Risk of bias assessment
The risk of bias was assessed using tools appropriate to each study design. For RCTs, we used the Cochrane Risk of Bias tool version 2 (RoB 2) (36). For the crossover trial included in the review, the specific RoB 2 for crossover trials variant (36) was applied by two independent reviewers (C. C. M. and J. M. M). These tools evaluate bias across five domains: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. To manage the review workload, the assessment of RCTs was distributed among the review team, with each study assessed by a single reviewer. Any ambiguities regarding the judgment were resolved through consultation with a third reviewer (L. B. G.). For non-randomized studies (including quasi-experimental and cohort designs), the risk of bias was evaluated using the ROBINS-I tool (Risk Of Bias In Non-randomized Studies – of Interventions) (37). This tool assesses bias across seven domains, encompassing confounding, participant selection, and classification of interventions. Unlike the RCTs, the assessment of non-randomized studies was conducted independently by two reviewers (C. C. M. and L. B. G.) to ensure rigor given the higher inherent risk of bias in these designs. Any discrepancies in the ROBINS-I assessments were resolved through consensus between the two reviewers.
Data synthesis
Data synthesis was conducted in accordance with the Synthesis Without Meta-analysis (SWiM) reporting guidelines (38). A quantitative synthesis (meta-analysis) was performed for outcomes where sufficient data were available and clinical homogeneity was established. For outcomes or studies not suitable for pooling, a structured narrative synthesis was conducted.
Meta-analysis
Studies were considered eligible for pooling if they reported continuous outcomes as mean and standard deviation (SD), or provided sufficient data (e.g., standard error, 95% confidence intervals) to estimate these values. We excluded studies reporting only medians and interquartile ranges (IQR), longitudinal data without summary statistics, or categorical data that could not be converted.
Meta-analyses were performed using a custom script in R statistical software (v4.3.1; R Foundation for Statistical Computing, Vienna, Austria). Effect sizes for continuous outcomes were expressed as standardized mean differences (SMD) with 95% confidence intervals (CI), allowing for the comparison of results across studies using different measurement scales. A random-effects model was applied for all analyses to account for the anticipated methodological and clinical heterogeneity across interventions. Statistical heterogeneity was assessed using the I2 statistic, with an I2 > 50% considered indicative of substantial heterogeneity.
Narrative synthesis
For studies or outcomes that could not be pooled quantitatively, due to insufficient data format (e.g., median/IQR), high heterogeneity (I2 > 85%), or disparate outcome measures, a narrative synthesis was conducted. Results were grouped by outcome domain (e.g., physiological parameters, sleep metrics) and intervention type (single vs. multicomponent). To standardize the reporting of these findings, we prioritized the direction of effect and statistical significance (p < 0.05) as reported in the primary studies. Where available, effect estimates were presented alongside their confidence intervals to facilitate interpretation.
Reporting bias assessment
Due to the limited number of studies included in each meta-analysis (k < 10 for all outcomes), publication bias could not be formally assessed using funnel plots or Egger’s regression test. According to Cochrane guidelines, tests for funnel plot asymmetry are generally not recommended when there are fewer than 10 studies, as the test power is insufficient to distinguish chance from real asymmetry (35). Consequently, the potential impact of publication bias on the pooled effect sizes remains unknown.
Certainty of evidence
The certainty of the evidence for each outcome was assessed using the GRADE approach (Grading of Recommendations Assessment, Development and Evaluation) (39, 40). The evidence was graded as high, moderate, low, or very low based on five domains: risk of bias, inconsistency (heterogeneity), indirectness, imprecision, and publication bias. For narrative synthesis outcomes, certainty ratings were primarily informed by the consistency of effects and the precision of reported estimates across included studies.
Results
Study selection
The comprehensive search strategy yielded a total of 11,899 records from electronic databases. After removing 8,049 duplicates using Covidence, 3,850 records remained for title and abstract screening. Of these, 3,768 were excluded for not meeting eligibility criteria.
A total of 82 full-text articles were assessed for eligibility. Following this review, 41 studies were excluded, primarily due to No full text availability (n = 18) and Wrong outcome and intervention (n = 7, respectively for each reason). Finally, 41 studies met the inclusion criteria and were included in the systematic review. Of these, 28 studies provided sufficient data for quantitative synthesis (meta-analysis). The selection process is detailed in the PRISMA flow diagram (Figure 1).
Study characteristics
A summary of the 41 studies included is presented in Table 1. The publication dates ranged from 1986 to 2025, with a notable increase in research activity in the last 5 years (2020–2025), which accounted for nearly 46% of the included studies (n = 19) (41–59). The studies represented a diverse global distribution, with a significant concentration in the Middle East and Asia, being the most represented countries: Iran (n = 10) (41, 55, 60–67), Turkey (n = 7) (43, 44, 49, 51, 54, 56, 58) and India (n = 3) (46, 57, 68). Another important proportion of studies came from the USA (n = 7) (47, 48, 50, 69–72) and Brazil (n = 3) (53, 73, 74). Finally, European studies (UK, Sweden) were less frequent (n = 1, respectively) (75, 76). Thirty-nine studies were published in English and the remaining studies were in Korean (n = 2). Regarding the study design, the majority of included studies were Randomized Controlled Trials (RCTs) (n = 23) (41, 43–46, 51, 53, 54, 56, 58, 60–64, 66–68, 70–72, 75, 76). Other designs included: crossover trials (n = 3) (73, 74, 77); quasi-experimental studies (n = 2) (52, 78) and retrospective or prospective cohorts (n = 5) (47, 48, 57, 59, 65). The study population consisted predominantly of preterm infants admitted to NICU, with gestational ages ranging typically from 26 to 36 weeks. Specific clinical subgroups included Very Low Birth Weight (VLBW) infants (59). Only two studies focused on pediatric populations outside the neonatal period: one involving infants up to 12 months (48) and one PICU study including patients up to 17 years of age (50). The interventions were categorized into two main groups: The first corresponded to (1) Single-Component Interventions (n = 26) (45–49, 53, 54, 57, 58, 60, 61, 63, 64, 68–75, 77–81). These studies evaluated the isolated effect of a specific zeitgeber. The most common strategies were lighting modifications like cycled lighting or incubator covers (n = 15) (45, 49, 53, 54, 61, 64, 69, 70, 72, 75, 77–81), noise reduction using ear protectors or earmuffs (n = 7) (47, 57, 60, 63, 71, 73, 74), and chrononutrition (timed feeding) (58). The other group included (2) Multicomponent Interventions (n = 15) (41–44, 50–52, 55, 56, 59, 62, 65–67, 76). These studies utilized “bundles” combining multiple strategies to minimize sensory mismatch (50, 66). Common combinations included light and noise reduction protocols (42–44, 62, 65, 67, 76) often paired with clustered care (52, 55), facilitated tucking (41), or music therapy (59). Regarding the timing and duration of these interventions, significant heterogeneity was observed. Enrollment typically occurred early in the ICU stay, often within the first week of life for preterm infants, though specific eligibility windows varied. The duration of interventions ranged from discrete, short-term applications (e.g., earmuffs applied only during a 2-h quiet time or during specific procedures) to continuous environmental modifications maintained for the entirety of the ICU stay (e.g., cycled lighting continued until discharge). Consequently, the timing of outcome assessments also varied, ranging from immediate post-intervention physiological measurements to cumulative outcomes like hospital length of stay. The most frequently reported outcomes, either by single or combined interventions, were physiological parameters (heart rate, respiratory rate, oxygen saturation) (41–44, 46, 47, 49, 53, 54, 57–61, 63, 67, 71, 76, 79) and sleep metrics (duration, efficiency, and sleep states) (46, 47, 51, 55, 56, 59, 62, 64–66, 68, 73–75, 77, 78, 80). Other key outcomes included weight gain (52, 54, 58, 62, 69, 71, 72, 75, 79, 80), length of hospital stay (LOS) (45, 52, 71, 72, 81), and circadian biomarkers like cortisol and/or melatonin (47, 48, 73, 78).
| Reference | Country | Year | Study design | Mean age (SD) | Sample size | Intervention type | Intervention | Outcomes measured | Outcome impact |
|---|---|---|---|---|---|---|---|---|---|
| Mann et al. () [75] | UK | 1986 | RCT | GA < 36 weeks | 41 | Single, Primary Circadian | Night/day cycle (light and noise reduction) | Sleep, feeding time, weight gain | Increased sleep; reduced feeding time; 0.5 kg heavier at 3 months (< 0.02)p |
| Miller et al. () [69] | USA | 1995 | Longitudinal interventional study with parallel groups | GA 28 weeks, BW 1100 g | 41 | Single, Primary Circadian | Cycled lighting | Weight gain, feeding, motor development | Greater weight gain (14% vs. 7.4%/week,< 0.05); earlier oral feedingp |
| Hellström-Westas et al. () [77] | Sweden | 2001 | Crossover | GA 26–34 weeks | 18 | Single, Primary Circadian | Incubator covers (artificial night) | Sleep states (EEG) | Increased quiet sleep duration; no effect on active sleep |
| Rivkees et al. () [70] | USA | 2004 | RCT | GA < 32 weeks | 38 | Single, Primary Circadian | Cycled lighting (LD 12:12) | Rest-activity patterns (actigraphy) | Earlier development of circadian rest-activity patterns |
| Jung () [79] | South Korea | 2005 | Non-randomized controlled trial | LBW infants | 38 | Single, Primary Circadian | Cycled lighting | Weight, physiological variables, behavioral states | Improved weight gain and behavioral states |
| Abou Turk et al. () [71] | USA | 2009 | RCT | GA < 32 weeks, VLBW | 34 | Single, Supportive Sensory | Silicone earplugs | Weight gain, LOS, physiological stability | No significant difference in weight gain or LOS |
| Lee et al. () [80] | South Korea | 2012 | Quasi-experimental | GA 32–36 weeks | 38 | Single, Primary Circadian | Cycled lighting | Sleep time, weight gain | Increased total sleep time; improved weight gain (< 0.05)p |
| Aita et al. () [76] | Canada | 2013 | RCT | GA 26–32 weeks | 108 | Multicomponent, Primary Circadian | Light and noise reduction program | HR, SpO2 | No significant difference in HR or SpO2 |
| Abdeyazdan et al. () [60] | Iran | 2014 | RCT | GA 28–34 weeks | 64 | Single, Supportive Sensory | Earmuffs for noise reduction | HR, RR, SpO2 | Significant reduction in HR and RR; improved SpO2 (< 0.05)p |
| Reyhani et al. () [61] | Iran | 2014 | RCT | GA 30–34 weeks | 38 | Single, Primary Circadian | Artificial night (incubator cover) | RR, SpO2 | Reduced RR; improved SpO2 (< 0.05)p |
| Vásquez-Ruiz et al. () [81] | Mexico | 2014 | Randomized Interventional Study | GA 31–35 weeks | 37 | Single, Primary Circadian | Artificial night (head cover) | LOS | Reduced length of stay (< 0.05)p |
| Esmaeilizadeh et al. () [62] | Iran | 2016 | RCT | GA < 37 weeks | 74 | Multicomponent, Primary Circadian | Cycled light and noise reduction | Weight gain, LOS, sleep time | Increased weight gain; reduced LOS; increased sleep time (< 0.05)p |
| Kaneshi et al. () [78] | Japan | 2016 | Quasi-experimental | GA 26–33 weeks | 22 | Single, Primary Circadian | Red light during night procedures | Melatonin rhythm, sleep, body growth | Earlier melatonin rhythm development; better nocturnal sleep |
| Araújo et al. () [73] | Brazil | 2017 | Crossover RCT | GA 28–36 weeks | 12 | Single, Supportive Sensory | Hearing protection (earmuffs) | Salivary cortisol, sleep | Lower cortisol; improved sleep quality |
| Brandon et al. () [72] | USA | 2017 | RCT | GA < 28 weeks | 121 | Single, Primary Circadian | Cycled light (early vs. late introduction) | Weight gain, LOS, ventilator days | Early introduction: faster weight gain to 36 weeks PMA |
| Khalesi et al. () [63] | Iran | 2017 | RCT | GA 28–34 weeks | 72 | Single, Supportive Sensory | Earmuffs for noise reduction | HR, RR, SpO2 | Significant reduction in HR and RR; improved SpO2 (< 0.05)p |
| Valizadeh et al. () [64] | Iran | 2017 | RCT | GA 28–34 weeks | 60 | Single, Primary Circadian | Light reduction at night | Sleep duration | Increased sleep duration (< 0.05)p |
| Karami and Marofi () [65] | Iran | 2018 | Prospective, randomized, crossover clinical trial | GA 28–34 weeks | 66 | Multicomponent, Primary Circadian | Noise reduction and light attenuation | Quiet sleep | Increased quiet sleep duration (< 0.05)p |
| Mony et al. () [68] | India | 2018 | RCT | GA < 37 weeks | 42 | Single, Supportive Sensory | Nesting (environmental stimuli reduction) | Sleep duration, quiet sleep | Increased sleep duration and quiet sleep (< 0.05)p |
| Sato et al. () [74] | Brazil | 2018 | Crossover | Preterm | 20 | Single, Supportive Sensory | Earmuffs during quiet time | Sleep stages (polysomnography) | No significant difference in sleep stages |
| Bazregari et al. () [66] | Iran | 2019 | RCT | GA 28–34 weeks | 64 | Multicomponent, Supportive Sensory | Clustered nursing care | Quiet sleep, active sleep | Increased quiet sleep duration (< 0.05)p |
| Zeraati et al. () [67] | Iran | 2019 | RCT | GA 30–34 weeks | 120 | Multicomponent, Supportive Sensory | Quiet time protocol | RR | Reduced respiratory rate (< 0.05)p |
| Ezabadi et al. () [41] | Iran | 2020 | RCT | GA 28–34 weeks | 40 | Multicomponent, Primary Circadian | Artificial night with facilitated tucking | HR, RR, SpO2 | Significant reduction in HR and RR; improved SpO2 (< 0.05)p |
| Abdel Hamid et al. () [42] | Egypt | 2021 | Pretest-post test interventional study | GA 28–36 weeks | 100 | Multicomponent, Primary Circadian | Cycled light and noise reduction | HR, RR, SpO2 | Significant reduction in HR and RR; improved SpO2 (< 0.001)p |
| Akarsu and Balci () [43] | Turkey | 2022 | RCT | GA 28–34 weeks | 60 | Multicomponent, Primary Circadian | Light and noise reducing hat | HR, RR, SpO2 | Significant reduction in HR and RR; improved SpO2 (< 0.05)p |
| Íncekar et al. () [44] | Turkey | 2022 | RCT | GA 28–34 weeks | 80 | Multicomponent, Primary Circadian | Light and noise reduction protocol | HR, SpO2 | No significant difference in HR; trend toward improved SpO2 |
| Sánchez-Sánchez et al. () [45] | Mexico | 2022 | RCT | GA 30–35 weeks | 73 | Single, Primary Circadian | Light–darkness cycle | LOS | Reduced length of stay (< 0.05)p |
| Vadakkan et al. () [46] | India | 2022 | RCT | GA 32–36 weeks | 76 | Single, Supportive Sensory | Nesting and swaddling | Sleep duration, HR, RR | Increased sleep duration; reduced HR and RR (< 0.05)p |
| Bloch-Salisbury et al. () [47] | USA | 2023 | Prospective Cohort | GA 28–34 weeks | 20 | Single, Supportive Sensory | Hearing protection device | HRV, sleep efficiency | Improved HRV; trend toward better sleep |
| Bradford et al. () [48] | USA | 2023 | Retrospective cohort | Infants <12 months | 55 | Single, Primary Circadian | Exogenous melatonin | Opioid use, pain/sedation scores, adverse events | Reduced opioid use (= 0.049); no adverse eventsp |
| Çetin and Ekici () [49] | Turkey | 2023 | Repeated measures | GA 37 weeks (median) | 64 | Single, Primary Circadian | Incubator cover | HR, RR, SpO2, temperature | Improved SpO2 (< 0.05); no significant HR/RR differencep |
| Curley et al. () [50] | USA | 2024 | Two-phase cohort | PICU patients (2 weeks-17 years) | 56 | Multicomponent, Primary Circadian | RESTORE Resilience Bundle (7 components) | Daytime activity ratio (DARE), melatonin | Improved daytime activity pre-extubation (= 0.04); NS post-extubationp |
| Düken and Yayan () [51] | Turkey | 2024 | RCT | GA 32–37 weeks | 120 | Multicomponent, Supportive Sensory | Massage therapy and white noise | Sleep duration, sleep efficiency | Increased sleep duration and sleep efficiency in both intervention groups (< 0.05)p |
| Hendy et al. () [52] | Egypt | 2024 | Quasi-experimental | GA 28–34 weeks | 53 | Multicomponent, Primary Circadian | Clustered care and healing environment (light/noise reduction) | Behavioral outcomes, discharge weight, LOS | Improved behavioral outcomes; higher discharge weight; reduced LOS (< 0.05)p |
| Odebrecht et al. () [53] | Brazil | 2024 | RCT | GA 30–35 weeks | 40 | Single, Primary Circadian | Light protection equipment (eye masks) at night | Days until discharge, HR variability | Earlier discharge (8 vs. 12 days,= 0.025); greater HR day-night variationp |
| Olgun et al. () [54] | Turkey | 2024 | RCT | GA 28–34 weeks | 30 | Single, Primary Circadian | Light–dark cycle | Weight gain, HR | Improved weight gain; no significant HR difference |
| Sabagh and Ghaljaei () [55] | Iran | 2024 | Non-randomized controlled trial | Preterm | 62 | Multicomponent, Primary Circadian | Quiet time protocol (light/noise reduction + clustered care) | Sleep status (categorical) | 58.1% vs. 0% achieved deep sleep (< 0.001)p |
| Suna Dağ et al. () [56] | Turkey | 2024 | RCT | GA 26–32 weeks | 99 | Multicomponent, Supportive Sensory | White noise and facilitated tucking | Sleep efficiency | Improved sleep efficiency (< 0.05)p |
| Ray et al. () [57] | India | 2025 | Prospective observational, non-randomized, controlled cohort | GA 28–34 weeks | 100 | Single, Supportive Sensory | Earmuffs for noise reduction | HR, RR, SpO2 | Significant reduction in HR and RR; improved SpO2 (< 0.001)p |
| Temizsoy et al. () [58] | Turkey | 2025 | RCT | GA 26–32 weeks | 62 | Single, Primary Circadian | Chronobiological feeding model | Weight gain | Increased weight gain (< 0.05)p |
| Zhao et al. () [59] | China | 2025 | Retrospective cohort | VLBW infants | 100 | Multicomponent, Supportive Sensory | Music therapy and multidimensional nursing | HR, RR, SpO2, sleep time | Reduced HR and RR; improved SpO2; increased sleep time (< 0.05)p |
Risk of bias in studies
To assess the risk of bias in the included studies, we used tools appropriate to each study design. The 23 included RCTs (41, 43–46, 51, 53, 54, 56, 58, 60–64, 66–68, 70–72, 75, 76) were evaluated using the Cochrane RoB 2 tool (36), while the 3 crossover studies were assessed using the specific RoB 2 tool for cross over studies (73, 74, 77). The 15 non-randomized, prospective, retrospective, cohort or quasi-experimental studies (42, 47–50, 52, 55, 57, 59, 65, 69, 78–81) were evaluated using ROBINS-I (37). Among the 23 RCTs, the overall risk of bias was rated as High in 18 studies (78.3%) (41, 43, 44, 46, 51, 54, 56, 58, 60–64, 66–68, 70, 75, 76), Some Concerns in 4 studies (17.3%) (45, 53, 71, 72), and Low in only 1 study (3.8%) (58). The primary driver of this assessment was deviations from intended interventions (Domain 2, 95% rated as High). Due to the nature of environmental interventions (e.g., cycled lighting, noise reduction), it was often impossible to blind participants or care providers, leading to a high risk of performance bias in 84.6% (n = 22) of trials (41, 43–46, 51, 53, 54, 56, 60–64, 66–68, 70–72, 75, 76). However, selection bias (Domain 1) was generally well-managed, with 69.5% of studies showing Low risk or Some Concerns (21.7%) regarding the randomization process. Similarly, reporting bias (Domain 5) was low, with 73.9% (n = 17) of studies showing no evidence of selective reporting (43–46, 51, 53, 54, 56, 58, 63, 64, 67, 68, 71, 72, 75, 76). In addition to the parallel-group trials, the included 3 crossover trials (73, 74, 77) were similarly classified as High risk due to the inherent performance bias described above.
For the 15 non-randomized studies, the overall risk of bias was rated as Moderate in 9 studies (60.0%) (52, 55, 57, 59, 69, 78–81) and Serious in 6 studies (40.0%) (42, 47–50, 65). No studies were classified as Critical or Low risk. The most prevalent sources of bias were confounding (Domain 1) and measurement of outcomes (Domain 6). Confounding was a moderate issue in 66% (n = 10) (47, 49, 52, 55, 57, 69, 78, 79, 81, 82) of studies due to the lack of randomization and insufficient adjustment for variables such as illness severity. Measurement bias was identified as a moderate-to-serious risk in all 9 studies (60%), primarily because behavioral outcomes (e.g., sleep state, crying) were assessed by unblinded observers aware of the intervention status (42, 48, 49, 55, 59, 65, 79–81). Conversely, biases related to participant selection and missing data were minimal. Remarkably, in the domain of classification of interventions (Domain 3), 100% of non-randomized studies were evaluated as low risk. This reflects the unambiguous nature of environmental circadian interventions (e.g., lighting protocols, physical noise reduction devices), which are structurally implemented and documented in the intensive care setting, leaving little room for misclassification of the intervention status.
The results of the risk of bias judgments for each study and domain are presented in Figures 2–4.
Risk of bias assessment for randomized controlled trials using the Cochrane RoB 2 tool.Weighted summary plot of the overall risk of bias across domains (= 23).Traffic light plot showing the risk of bias judgment for each individual study. (A) (B) n
Risk of bias assessment for crossover randomized controlled trials using the Cochrane RoB 2 tool.Weighted summary plot illustrating the overall proportion of studies classified as low risk, some concerns, or high risk of bias across each domain.Traffic light plot presenting the specific risk of bias judgments for each individual crossover study included in the review. (A) (B)
Risk of bias assessment for non-randomized studies using the ROBINS-I tool.Weighted summary plot illustrating the overall proportion of studies classified as low, moderate, serious, or critical risk of bias across each domain.Traffic light plot presenting the specific risk of bias judgments for each individual non-randomized study included in the review. (A) (B)
Results of syntheses
A total of 41 studies involving 2,548 participants met the inclusion criteria. Due to variations in data reporting and outcome metrics, the synthesis was conducted in two parts: a quantitative meta-analysis of 16 studies (41, 43, 46, 51, 52, 54, 56, 57, 59–61, 63–65, 67, 80) that provided comparable mean and standard deviation data, and a narrative synthesis (SWiM) (38) of 16 studies that utilized non-parametric or incompatible metrics (47–50, 53, 55, 69–75, 77–79). The remaining 9 studies (42, 44, 45, 58, 62, 66, 68, 76, 81) were excluded from the outcome synthesis due to insufficient statistical reporting (e.g., missing variance) but contributed to the qualitative description of study characteristics. The distribution of studies across these synthesis groups is detailed in Supplementary Material 2.
Effect of circadian interventions on physiological parameters
The effect of circadian-tailored interventions on physiological parameters was evaluated in 11 studies. For heart rate, the pooled analysis of 8 studies (n = 542) (41, 43, 46, 54, 57, 59, 60, 63) demonstrated a significant reduction in the intervention group compared to standard care (SMD -0.70; 95% CI [−1.10, −0.30]; p < 0.001). As shown in Figure 5A, significant heterogeneity was observed among the studies (I2 = 79.6%). Similarly, the analysis of respiratory rate in 9 studies (n = 670) (41, 43, 46, 57, 59–61, 63, 67) revealed a significant decrease in infants receiving circadian care (SMD = −0.75; 95% CI: [−1.16 to −0.35]; p < 0.001), with high heterogeneity (I2 = 84.1%) (Figure 5B). Regarding oxygenation, the pooled results from 7 studies (n = 474) (41, 43, 57, 59–61, 63) indicated that oxygen saturation (SpO2) was significantly higher in the intervention group (SMD = 1.33; 95% CI: [0.95 to 1.72]; p < 0.001). Heterogeneity for this outcome was also substantial (I2 = 71.7%), as illustrated in Figure 5C. Sensitivity analyses were performed by sequentially removing individual studies, but the statistical significance of the pooled effect sizes remained robust across all three physiological outcomes. A summary of the analysis can be found in Supplementary Material 3.
Forest plots illustrating the effects of circadian-tailored interventions versus standard care on physiological stability and sleep outcomes in hospitalized infants. Panels show the pooled meta-analysis results forheart rate,respiratory rate,oxygen saturation (SpO), andtotal sleep duration. In each plot, individual study effect sizes are shown as squares with horizontal lines representing 95% confidence intervals (CIs). The diamond at the bottom represents the pooled standardized mean difference (SMD) and its 95% CI calculated using a random-effects model. For heart rate and respiratory rate (A, B), values to the left of the zero line indicate a reduction favoring the circadian intervention. Conversely, for SpOand sleep duration, values to the right of the zero line indicate an increase favoring the intervention. Thestatistic denotes the degree of statistical heterogeneity across studies. For multi-arm trials sharing a control group, the control sample was split to avoid unit-of-analysis errors, as recommended by the Cochrane Handbook. (A) (B) (C) (D) (C,D) 2 2 I 2
Effect of circadian interventions on sleep duration
The impact of circadian interventions on total sleep duration was assessed in 3 studies involving 256 participants. The meta-analysis showed that infants in the intervention group had significantly longer sleep duration compared to those in the control group, with a large effect size (SMD = 0.92; 95% CI:[0.66–1.19]; p < 0.0001). In contrast with the heterogeneity for the physiological outcomes, sleep duration exhibited no heterogeneity (I2 = 0%) as described in Figure 5D, which shows consistency among the studies.
Subgroup analysis: single vs. multicomponent interventions
To investigate potential sources of heterogeneity and compare intervention strategies, a subgroup analysis was conducted. We found no statistically significant difference between single-component interventions and multicomponent bundles for any of the measured outcomes (p > 0.05) (see Table 2). However, it is crucial to note that this review was not designed or powered to formally demonstrate non-inferiority or equivalence. The absence of a statistically significant difference may reflect limited statistical power, high clinical heterogeneity, and the low certainty of the current evidence, rather than true formal equivalence.
| Outcome | Single ()K | Single SMD [95% CI] | Multi ()K | Multi SMD [95% CI] | p-subgroup |
|---|---|---|---|---|---|
| Heart rate | 5 | −0.61 [−1.26, 0.04] | 3 | −0.85 [−1.14, −0.55] | 0.518 |
| Respiratory rate | 5 | −0.67 [−1.43, 0.08] | 4 | −0.82 [−1.11, −0.52] | 0.73 |
| SpO2 | 4 | +1.45 [0.98, 1.92] | 3 | +1.19 [0.55, 1.82] | 0.502 |
| Sleep duration | 2 | +0.93 [0.49, 1.38] | 2 | +0.95 [0.50, 1.39] | 0.965 |
Narrative synthesis (SWiM)
Sixteen studies (n = 786) (47–50, 53, 55, 69–75, 77–79) were synthesized narratively using vote counting based on the direction of effect. Consistent with the meta-analysis results, 75% (12/16) of these studies reported findings favoring the intervention group (47–49, 53, 55, 69, 70, 73, 75, 77–79), while 18.8% (3/16) found no significant difference (71, 72, 74), and 6.2% (1/16) reported partial benefits (50). No study reported results favoring the control condition. Specifically, interventions using cycled lighting (n = 3) consistently reported improved weight gain and earlier oral feeding (69, 75, 79). In contrast, results for individual noise reduction (e.g., earmuffs) were mixed, with some studies noting reduced cortisol levels (73) but others finding no effect on weight gain (71). These findings are summarized in in Table 3.
| Study | Intervention type | Sample (N) | Key outcome measured | Main finding | Direction |
|---|---|---|---|---|---|
| Light cycling and reduction | |||||
| Mann et al. (1986) () [75] | Cycled Light/Noise | 41 | Sleep time; Weight gain | Increased sleep duration; 0.5 kg greater weight gain at 3 mo (p < 0.02). | Favors (↑) |
| Miller et al. (1995) () [69] | Cycled Light | 41 | Weight gain rate | Higher rate of weight gain (14% vs. 7.4%/week; p < 0.05). | Favors (↑) |
| Rivkees et al. (2004) () [70] | Cycled Light | 38 | Rest-activity patterns | Earlier development of circadian rest-activity distinct patterns. | Favors (↑) |
| Jung (2005) () [79] | Cycled Light | 20 | Weight; SpO2 | Increased weight gain and SpO2; decreased heart rate. | Favors (↑) |
| Hellström-Westas et al. (2001) () [77] | Incubator Cover | 18 | Sleep states (EEG) | Increased quiet sleep duration during covered periods. | Favors (↑) |
| Brandon et al. (2017) () [72] | Cycled Light (Early) | 121 | Growth velocity | No significant difference in growth between early vs. late cycling. | No Diff (↔) |
| Çetin and Ekici (2023) () [49] | Incubator Cover | 60 | SpO2; Vital signs | Higher SpO2 with cover (median 98% vs. 97%;< 0.05).p | Favors (↑) |
| Odebrecht et al. (2024) () [53] | Eye Masks (Night) | 40 | Length of stay | Earlier hospital discharge (median 8 vs. 12 days;= 0.025).p | Favors (↑) |
| Noise reduction | |||||
| Abou Turk et al. (2009) () [71] | Silicone Earplugs | 34 | Weight gain; LOS | No significant difference in weight gain or length of stay. | No Diff (↔) |
| Sato et al. (2018) () [74] | Earmuffs | 24 | Sleep states | No significant difference in total sleep time or sleep efficiency. | No Diff (↔) |
| Araújo et al. (2017) () [73] | Earmuffs | 14 | Salivary cortisol | Significant reduction in salivary cortisol levels during use. | Favors (↑) |
| Bloch-Salisbury et al. (2023) () [47] | Hearing Protection | 10 | Sleep percentage | 14.1% increase in sleep duration with protection (= 0.02).p | Favors (↑) |
| Multicomponent and Other | |||||
| Kaneshi et al. (2016) () [78] | Red Light (Night) | 48 | Sleep/Wake Rhythm | Facilitated development of sleep–wake rhythm (< 0.01).p | Favors (↑) |
| Bradford et al. (2023) () [48] | Melatonin | 55 | Opioid use | Significant reduction in opioid utilization and pain scores (p = 0.049). | Favors (↑) |
| Curley et al. (2024) () [50] | RESTORE Bundle | 56 | Activity Ratio (DARE) | Improved daytime activity ratio pre-extubation (= 0.04).p | Partial (↑) |
| Sabagh and Ghaljaei (2024) () [55] | Quiet Time Protocol | 62 | Deep Sleep | 58% of intervention group achieved deep sleep vs. 0% controls (< 0.001).p | Favors (↑) |
Reporting bias assessment
As the number of included studies in our meta-analysis was fewer than 10 for all primary outcomes. Specifically, Heart Rate (k = 8), Respiratory Rate (k = 9), Oxygen Saturation (k = 7), and Sleep Duration (k = 3), funnel plots and Egger’s regression test were not performed to assess publication bias. According to Cochrane guidelines, tests for funnel plot asymmetry are generally not recommended when there are fewer than 10 studies, as the test power is insufficient to distinguish chance from real asymmetry (35). Consequently, the potential impact of publication bias on the pooled effect sizes could not be formally evaluated.
Certainty assessment
The certainty of the evidence was assessed using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach (40). Overall, the certainty of the body of evidence was judged to be low for all primary outcomes. For the physiological parameters (heart rate, respiratory rate, and oxygen saturation), the evidence was downgraded from high to low primarily due to serious risk of bias. Specifically, performance bias related to the lack of blinding in environmental interventions and serious inconsistency, as indicated by the substantial statistical heterogeneity (I2 > 70%) observed across the included studies. Regarding sleep duration, although inconsistency was not a concern, the evidence was similarly graded as low due to serious risk of bias and serious imprecision, attributable to the limited number of studies (k = 4) and the relatively small pooled sample size (N = 296). Despite the low certainty ratings, the direction of effect was consistent across all outcomes, with statistically significant benefits observed in the intervention groups. A detailed summary of the GRADE assessment, including the rationale for downgrading each outcome, is presented in Table 4.
| Outcome | Studies | Participants | Effect (SMD) | 95% CI | Certainty | Level | Interpretation |
|---|---|---|---|---|---|---|---|
| Heart Rate | 8 | 542 | −0,7 | [−1.10, −0.30] | ⊕⊕◯◯ | LOWa,b | Circadian interventions may reduce heart rate in preterm infants (low certainty evidence) |
| Respiratory Rate | 9 | 670 | −0,75 | [−1.16, −0.35] | ⊕⊕◯◯ | LOWa,b | Circadian interventions may reduce respiratory rate in preterm infants (low certainty evidence) |
| Oxygen Saturation (SpO2) | 7 | 474 | 1,33 | [0.95, 1.72] | ⊕⊕◯◯ | LOWa,b | Circadian interventions may improve oxygen saturation in preterm infants (low certainty evidence) |
| Sleep Duration | 4 | 256 | 0,92 | [0.66, 1.19] | ⊕⊕◯◯ | LOWa,c | Circadian interventions may increase sleep duration in preterm infants (low certainty evidence) |
Discussion
Summary of main findings
This systematic review and meta-analysis represents the first attempt to directly compare single-component versus multicomponent circadian interventions in neonatal and pediatric intensive care units. Analyzing 41 studies spanning nearly four decades (1986–2025), our meta-analysis demonstrated that circadian interventions significantly modulated all four primary physiological outcomes: reducing heart rate and respiratory rate, while increasing oxygen saturation, and sleep duration (p < 0.001). Importantly, while our GRADE assessment rated the certainty of evidence as “low” primarily due to the inherent difficulty of blinding environmental interventions (36) and heterogeneity, this rating indicates limited confidence in the precise magnitude of effects, not in the effectiveness of the interventions themselves (39).
Single vs. combined interventions in neonates: an ontogenic perspective
The central finding of this review is the absence of significant differences between single-component and multicomponent interventions across all outcomes (p > 0.05). This challenges the intuitive assumption that comprehensive “bundles” yield superior results compared to focused approaches. A critical consideration in interpreting this finding is the developmental stage of the circadian system. The suprachiasmatic nucleus (SCN) is present by 18 weeks of gestation but undergoes substantial postnatal maturation (83, 84). Critically, the neonatal SCN contains only ~13% of the adult number of vasopressin-expressing neurons, and adult levels are not attained until 2–3 years of life (6, 85). This developmental immaturity has profound implications. During the sensitive window of prematurity, the plastic circadian system may respond more robustly to a single, strong zeitgeber (like light) than to multiple, competing signals (7, 86). This “less is more” hypothesis is supported by Van Gilst et al. (13), who demonstrated that preterm infants exposed to cycled light showed earlier emergence of sleep–wake rhythms compared to term infants, suggesting a unique sensitivity to photic entrainment. Furthermore, Govindan et al. (81) showed that circadian rhythm amplitude increases as a function of postnatal age, emphasizing the role of ex-utero maturation. Therefore, in an immature system, providing a clear, consistent light–dark cycle may be more effective than introducing multiple zeitgebers simultaneously, which could potentially overwhelm an underdeveloped entrainment pathway. The attractive “less is more” hypothesis (that simple interventions may yield similar benefits to complex ones) appears plausible for the immature neonatal circadian system. However, this finding must be strictly restricted to neonatal populations. It should not be extrapolated to older pediatric patients (PICU) or adults, whose mature circadian systems likely require more comprehensive, multicomponent approaches to overcome ICU disruption.
Complexity does not guarantee effectiveness: the case for cycled light
Consistent with the developmental hypothesis, cycled light emerged as a particularly robust intervention. Despite heterogeneity in implementation including variations in light intensity (ranging from <30 lux to 300–580 lux during daytime), duration of exposure, timing of introduction, and follow-up periods, studies consistently demonstrated favorable outcomes (45, 49, 53, 54, 61, 64, 69, 70, 72, 75, 77, 79–81). The 2024 Cochrane review by Morag et al. (87), while noting the need for larger studies, cautiously supports cycled light implementation in NICUs. The biological rationale for cycled light’s effectiveness is well-established. Light is the primary zeitgeber for the SCN, acting through melanopsin-containing intrinsically photosensitive retinal ganglion cells (ipRGCs) that project directly to the circadian pacemaker (11, 22). Hazelhoff et al. (8) emphasized that while preterm infants may not yet have fully developed photic entrainment pathways, the basic structures of the eye and central clock are present by 24 weeks of gestation. Environmental light input can therefore be conveyed to an SCN that is – at least partly – functional, supporting circadian system maturation during the critical period of NICU hospitalization. This consistency across diverse protocols suggests that the fundamental principle of establishing light–dark cycles may be more important than specific parameters in neonatal populations. For mature circadian systems, however, the quality and spectral composition of light may be more critical. Wasden et al. (88) demonstrated that circadian rhythms are best entrained by light wavelengths between 446 and 477 nanometers – a specificity that may be less relevant to immature systems but crucial for adult circadian alignment.
Challenges in standardization and clinical implications
A significant challenge encountered in this review was the lack of standardized definitions. We identified circadian interventions ranging from simple earmuff application to comprehensive chronotherapeutic bundles. Some included interventions may not be purely circadian in nature; noise reduction with earmuffs primarily serves as a stress-reduction intervention with secondary circadian benefits (73, 89), and nesting/swaddling have multiple physiological effects beyond circadian modulation (56). At present, many authors are acknowledging this challenge and called for standardized definitions and outcomes measures (27, 28). The minimal reporting guidelines developed by Spitschan et al. (90) for lighting interventions provide a useful starting point. Until such consensus is achieved, systematic reviews in this field will continue to face difficulties in comparing results across studies and populations. Despite these definition challenges, the clinical implications for nursing are clear. Circadian interventions are inherently nursing-driven, positioning nurses as primary implementers (91, 92). The finding that simpler interventions are as effective as complex bundles in neonatal populations has direct implications for nursing workload. Altimier and Phillips (93) developed the Neonatal Integrative Developmental Care Model with seven neuroprotective core measures. Our findings suggest that prioritizing light–dark cycles may yield substantial benefits for neonates without requiring simultaneous implementation of all components. However, this recommendation is age-specific; adult and pediatric populations may require multicomponent approaches.
Strengths and limitations
This review has several strengths. We followed PRISMA 2020 guidelines and registered our protocol prospectively. Our comprehensive search across multiple databases identified 41 studies spanning nearly four decades. The use of both meta-analysis and SWiM narrative synthesis allowed inclusion of studies with diverse reporting formats. Risk of bias was assessed using validated tools (RoB 2 and ROBINS-I), and certainty of evidence was evaluated using GRADE methodology.
However, several limitations must be acknowledged. The substantial heterogeneity observed (I2 = 72–84% for physiological outcomes) limits confidence in pooled estimates and likely reflects true variation in intervention protocols, populations, and outcome measurement timing. Nine studies were excluded from synthesis due to incompatible outcome definitions or insufficient statistical reporting, potentially introducing selection bias. The predominance of studies from middle-income countries may limit generalizability to other healthcare contexts. Publication bias could not be formally assessed due to the small number of studies per outcome, though funnel plot inspection suggested possible asymmetry.
Future directions
Based on our findings, we propose several priorities for future research. First, standardized definitions and reporting guidelines for circadian interventions are urgently needed. The minimal reporting guidelines developed by Spitschan et al. (90) for lighting interventions provide a useful starting point that should be adopted and extended to other circadian domains. Second, adequately powered randomized trials specifically addressing cycled light implementation are warranted. Given the consistent positive findings despite protocol heterogeneity, pragmatic trials testing real-world implementation strategies may be more valuable than efficacy trials with strict protocols. Third, research in pediatric intensive care populations is critically needed. The near-absence of PICU studies represents a significant gap that cannot be addressed by extrapolating neonatal findings. Fourth, studies should incorporate objective circadian biomarkers such as melatonin rhythms, cortisol profiles, or actigraphy to verify that interventions actually achieve circadian alignment rather than relying solely on clinical outcomes. Finally, long-term neurodevelopmental follow-up should be prioritized. While immediate physiological changes are encouraging, the ultimate goal of circadian interventions is to optimize long-term outcomes for this vulnerable population (16).
Conclusion
This systematic review and meta-analysis demonstrates that circadian-tailored interventions effectively modulate physiological parameters and improve sleep duration in critically ill neonates. A key finding is that single-component interventions, such as cycled lighting alone, showed no statistically significant difference in outcomes compared to complex multicomponent bundles in the NICU setting. This supports a potential ‘less is more’ approach specifically for the developing neonatal circadian system, suggesting that solitary, robust environmental cues may be sufficient to promote entrainment. Clinicians in neonatal units should prioritize the feasible implementation of standardized light–dark cycles and sensory protection. However, due to the critical scarcity of evidence from older pediatric populations, these findings cannot be extrapolated to the PICU. Future research must strictly stratify by age, prioritize primary studies in pediatric intensive care, and focus on standardizing environmental reporting.
Circadian Biology