Subsyndromal Delirium in Elderly Critical Care Patients: A Meta‐Analysis of Prevalence and Risk Factors

ABSTRACT

Aims

To determine the pooled prevalence of subsyndromal delirium (SSD) in elderly critical care patients and synthesize evidence on its modifiable risk factors through meta‐analysis.

Study Design

Meta‐analysis.

Methods

This meta‐analysis was conducted using six databases: PubMed, Web of Science, Embase, Cochrane Library, Scopus, and EBSCO. We assessed the quality of included studies using the Newcastle‐Ottawa Scale and the Agency for Healthcare Research and Quality recommendation checklist. A meta‐analysis of the predictive performance was conducted using the forest plot package in R 3.6.1.

Results

This review included 12 studies involving 2428 participants. The pooled prevalence of SSD ranged from 11.7% to 71.4%. 0.12 studies reported 19 independent risk factors associated with SSD: older age, male sex, absence of a partner, education level ≤ 9 years, lower cognitive function, comorbidities, anemia, hyponatremia, higher C‐reactive protein levels, pain, fall history, complicated procedures, previous brain failure in the ICU, preoperative fasting time, number of medications, decreased grip strength, low ADL, smoking, deafness.

Conclusion

This meta‐analysis demonstrates that SSD is a common condition among elderly critical‐care patients. Future investigations should employ multicenter designs with larger, more diverse patient cohorts and standardized SSD assessment protocols to enhance the robustness of evidence.

Implication of Practice

This review provides updated evidence that modifiable risk factors are significantly associated with SSD in critically ill older adults. Consequently, it offers solid information to inform targeted prevention strategies and underscores the importance of early recognition by healthcare professionals in geriatric psychiatry and critical care.

Reporting Method

The research results were reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta‐analysis checklist.

No Patient or Publication Contribution

This study is a meta‐analysis, and such details do not apply to our work.

This meta‐analysis of 12 studies (2428 participants) found that subsyndromal delirium (SSD) is common in elderly critical care patients, with a pooled prevalence of 32.4%. Nineteen risk factors were identified, including older age, anemia, and polypharmacy, and so forth. Early recognition and targeted management of modifiable risk factors may help reduce SSD and improve outcomes.

1. Introduction

Delirium is an acute brain dysfunction characterized by disturbances in attention, orientation, and clarity of consciousness, often accompanied by cognitive impairment and disruptions in the sleep–wake cycle [1]. It typically develops rapidly over a short duration. With the growing body of research on delirium, it has been observed that many patients exhibit one or more delirium‐related symptoms without fulfilling the full diagnostic criteria for clinical delirium. This condition is referred to as Subsyndromal Delirium (SSD) [2]. Currently, SSD lacks a standardized definition. The fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM‐V) describes it as an “attenuated delirium syndrome.” Moreover, SSD is recognized as a risk factor for the development of full‐syndrome delirium [3].

Among older critically ill patients, SSD is more prevalent than delirium, with reported incidence rates ranging from 11.7% to 68% [4, 5, 6]. Altered levels of consciousness and cognitive decline primarily characterize it. Previous studies have demonstrated associations between SSD and negative outcomes in elderly patients, including cognitive deterioration, diminished functional capacity in activities of daily living, prolonged hospital stays, and increased mortality [7]. Clinical practice guidelines emphasize that early diagnosis of SSD facilitates the timely identification and correction of underlying causes, potentially alleviating symptoms and improving prognosis [8]. Despite its clinical importance, SSD is frequently under‐recognized or misdiagnosed in practice [9].

Currently, routine screening for SSD is not widely implemented across relevant departments, and awareness among healthcare professionals remains limited [10]. Therefore, this study aimed to systematically review the epidemiological characteristics and risk factors of SSD. By identifying high‐risk patients at an early stage, clinicians may be better equipped to implement targeted interventions, reduce the incidence and adverse outcomes of SSD, enhance the management of critically ill patients, and alleviate the associated health‐care burden.

2. Methods

This meta‐analysis was registered in the International Prospective Register of Systematic Reviews and performed based on the Meta‐analysis of Observational Studies in Epidemiology (MOOSE) and the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA).

2.1. Study Strategies

Comprehensive searches were conducted in PubMed, Web of Science, Embase, Cochrane Library, Scopus, EBSCO, January 01, 2004–December 31, 2024. The following key words were used: “intensive care unit,” “critical care,” “critical illness,” “critically ill,” “subsyndromal delirium,” “sub‐syndromal delirium,” “subclinical delirium”, “aged,” “geriatrics.” The keyword combinations were modified according to the specific requirements of each database. Supplementary studies were obtained through a review of relevant references. Rayyan was used to remove the duplicate studies and facilitate the screening process.

Two authors independently reviewed the titles and abstracts to remove studies that did not meet the inclusion criteria and then read the full text to select studies included in this meta‐analysis. Any disagreements were resolved through discussion with the other two authors.

2.2. Inclusion and Exclusion Criteria

The inclusion criteria: (1) Population: participants were in a critical care setting (≥ 60 years old) [11]; (2) Exposure: use validated SSD assessment tools (ICDSC, CAM‐ICU, CAM, etc.) and report modifiable/non‐modifiable risk factors; (3)Comparator: critically ill elderly patients without subsyndromal delirium or those with low‐risk exposures; (4) Outcome: prevalence or risk factors for SSD were reported; (5) Observational study design; (6) Full‐text published articles.

The exclusion criteria: (1) Reviews, letters, conference abstracts, case reports, duplicate publications, and incomplete articles; (2) Non English‐language; (3) Studies involving patients with terminal illnesses will be excluded; (4) Studies that fail to clearly distinguish between delirium and subsyndromal delirium or provide vague definitions of risk factors; (5) Studies with non‐extractable data, duplicate publications, or preprints will be excluded. However, references from systematic reviews will be screened separately.

2.3. Data Extraction and Quality Appraisal

Two authors independently extracted data using a standardized data collection sheet, and any disagreements were resolved by discussion. The items of the data collection sheet included: first author, publication year, country, study design, participant, mean age, men, assessment tool, frequency, evaluator, sample size, number of SSD cases, and risk factors.

The quality of included studies was assessed independently by two authors; any disagreements were resolved through consultation with the other two authors. Cross‐sectional studies were evaluated using the 11‐item criteria recommended by the Agency for Healthcare Research and Quality (AHRQ) [12]. A score of 0–3 indicates low quality, 4–7 indicates medium quality, and 8–11 indicates high quality. Cohort studies and case–control studies were evaluated using the Newcastle‐Ottawa Scale (NOS) [13]. The NOS evaluates three aspects of original studies: the selection of study groups, the comparability of the groups, and the ascertainment of either the exposure or outcome. A score of 0–3 indicates low quality, 4–6 indicates medium quality, and 7–9 indicates high quality.

2.4. Statistical Analysis

All statistical analyses were performed using the forest plot package in R 3.6.1. We plotted AUC to summarize the predictive performance by a forest plot. The Q test was used to determine whether there was heterogeneity between studies. If I² ≤ 50%, it was considered that the studies did not show statistical homogeneity, and a fixed‐effects model was used for analysis. If I² > 50%, it was considered that the studies were statistically heterogeneous, and the data were analyzed by using a random‐effects model; the sensitivity analysis was used to examine the robustness of the findings. Descriptive analysis was used for studies where the data could not be merged.

3. Results

3.1. Study Selection, Inclusion, and Study Characteristics

A total of 5201 studies were initially obtained through comprehensive searches (Figure 1). After removing duplicates (n = 1755) and excluding irrelevant studies (n = 3317) by reading titles and abstracts, 27 full‐text articles were reviewed, of which 12 studies were included. In addition, one study was supplemented through citation searching. Finally, 12 studies involving 2428 participants were included in this review.

FIGURE 1.

Flow diagram of study selection.

The characteristics of the included studies are summarized in Table 1. These studies were conducted in ten countries with the publication period ranging from 2004 to 2024. The mean age of participants was 74.5 years. Most of the included studies were cross‐sectional studies (n = 6, 50.0%), four were cohort studies (33.3%), and only two (16.7%) were case–control studies.

TABLE 1.

Characteristics of the included studies (n = 12).

First author (year)	Country	Study design	Participant	Mean age (SD)	Men n (%)	SSD assessment			Sample size	SSD, n (%)	Risk factors studied	Quality
						Tool	Frequency	Evaluator
Qian et al. [14]	China	Cross‐sectional study	SI	75.15 (7.5)	124 (63.3)	CAM	1, 1/d (1–7d)	Research nurses	181	75 (41.4)	Decreased grip strength, Advanced age, Lower cognitive function, and fall history	8
Mailhot et al. [15]	Canada	Case‐control study	ICU	70.17 (9.1)	259 (74.5)	ICDSC	4d, 3/d	Nurse	346	62 (17.9)	Older age; Lower ADL; Walking aid; History of neurological disorders; Cerebral oximetry; Hemoglobin; Creatinine; Length of procedure; Lowest rSO2 value; Fluid balance.	8
Kanno et al. [6]	Japan	Cohort study	SI	74.7 (6.2)	89 (88.1)	CAM	5d, 3/d	Surgical ward and Nurses	101	15 (14.9)	Independent daily living, Use of bed sensors; Dementia, Number of medications	9
Efraim [16]	Israel	Case‐control study	MI	75.3 (7.4)	177 (53.3)	CAM	1/d (week day)	Nurse	332	23 (14.6)	Absence of a partner, Pain, Anemia, Hyponatremia, and the use of drugs with an anticholinergic burden	9
Denny et al. [17]	USA	Cross‐sectional study	SI	—	17 (36)	CAM	—	Nurse	47	37 (72)	Age, Cognitive status	7
Hwang et al. [18]	Korea	Cross‐sectional study	MI	70.11 (7.5)	105 (65.2)	DRS‐R‐98	1, 1/1, 2, 3, 7 postoperatively	Nurse; Psychiatrist	163	19 (11.7)	Age ≥ 70 years, Education level ≤ 9 years	8
Yamada et al. [19]	Japan	Cohort study	ICU	76 (66–82)	218 (57.4)	ICDSC	1/8 h	Nurse	380	129 (33.9)	Older age, Predisposing cognitive impairment, Blood transfusion, Evaluation II (APACHE II) score, Deafness, Low red blood cell counts, and high C‐reactive protein levels	8
Denny and Such [20]	USA	Cross‐sectional study	SI	74 (7.3)	16 (37)	CAM	3d	Nurse	49	34 (69)	Increased age, Cognitive impairment, and Current smoking	7
Denny and Lindseth [21]	USA	Cross‐sectional study	SI	74 (6.2)	23 (43)	CAM	3d, 1/d	Nurse	53	36 (37.9)	Fall history (The number of falls that had occurred in the six months before enrollment), Preoperative fasting time	6
Zuliani et al. [22]	Italy	Cross‐sectional study	MI	81.8 (6.0)	175 (39.9)	DSM‐IV	2d	Trained physicians	438	166 (37)	Older individuals, more frequently widows, Higher comorbidity, fewer years of education, Lymphocytes/Hemoglobin counts were also lower	7
Cole et al. [23]	Canada	Cohort study	LTC	—	46 (44.2)	CAM	1/week	Nurse	104	68 (65.4)	Male sex, More‐severe cognitive impairment	7
Ceriana et al. [24]	Italy	Cohort study	ICU	73.8 (8.3)	13 (56.8)	ICDSC	1/d	Physician; Nurse	234	47 (20)	Previous brain failure in the ICU, Older, mechanically ventilated	6

Abbreviations: ADL: activities of daily living; APACHE II: Acute Physiology and Chronic Health Evaluation II; CAM: Confusion Assessment Method; CRP: Creactive protein; DAM‐IV: Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; DRS‐R‐98: Delirium Rating Scale; EuroSCORE II: European system for cardiac operative risk evaluation II; GI: general inpatient; HGB: hemoglobin; ICDSC: Intensive Care Delirium Screening Checklist; LCT: long‐term care; MI: medical inpatient; RBC: red blood cell count; SI: surgical inpatient.

Different diagnostic criteria and assessment tools for SSD were used in the 12 included studies. The most commonly used assessment tool was the CAM (n = 6). These studies defined SSD as the presence of one or two (or more) core symptoms of delirium that did not meet the criteria for delirium; the symptoms were acute onset and fluctuation, inattention, disorganized thinking, and altered level of consciousness. Three studies reported that SSD was assessed by ICDSC, defined SSD as the presence of 1–3 symptoms of delirium on the ICDSC; the symptoms included altered level of consciousness, inattention, disorientation, hallucinations or delusions, psychomotor agitation or retardation, inappropriate speech or mood, sleep–wake cycle disturbance, and symptom fluctuation. One study defined SSD as DSM‐IV, indicating the presence of delirium, but the DRS‐R‐98 score was < 15. All studies, SSD assessments were performed by trained nurses or physicians.

The quality assessment of 12 articles indicated that 2 studies (16.7%) demonstrated high methodological quality [6, 14] and 10 studies (83.3%) had moderate methodological quality [15, 16, 17, 18, 19, 20, 21, 22, 23, 24]. Most moderate‐quality studies did not assess SSD at baseline and did not control for confounders by design or analysis.

3.2. Risk of Bias Across Studies

Based on Newcastle‐Ottawa Scale and AHRQ assessments (Table 1, Quality column), 2 studies (16.7%) demonstrated low risk of bias (NOS/AHRQ scores ≥ 8), while 10 studies (83.3%) had moderate risk of bias (scores 4–7), primarily due to lack of baseline SSD assessment in 8 studies and inadequate control for confounders in 9 studies.

3.3. Prevalence of SSD

All included studies reported the prevalence of SSD, which ranged from 11.7% to 71.4%. Based on a random‐effects model, the pooled prevalence of SSD was 32.4% (95% CI: 23.0%–53.0%; I² = 95.6%, p < 0.0001) (Figure 2).

FIGURE 2.

The prevalence of SSD in elderly critical care patients.

3.4. Subgroup Analysis

We conducted subgroup analyses based on participant, assessment tool, study type, and region. The results indicated that the prevalence of SSD was 41.7% in the surgical inpatient, both medical inpatient units and the ICU were 25%, and 8.3% in the long‐term care inpatient. In subgroup analyses by region, the pooled prevalence in Europe, Asia, and America was 41.7%, 33.3%, and 25.0%, respectively. When stratified by assessment tool, studies used CAM (58.3%), ICDSC (25%), with DSM‐IV and DRS‐R‐98 each accounting for 8.3%. The results of the subgroup analyses are summarized in Table 2.

TABLE 2.

Subgroup analysis of the prevalence of SSD.

Subgroup	No. of studies	Total sample	Patients with SSD, n	Heterogeneity		Statistical model	Prevalence,95% CI
				I2	p
Participant				95.6%	< 0.0001	Random effects	37% [0.23–0.53]
MI	3	933	208	92.9%	< 0.0001	Random effects	26% [0.22–0.25]
ICU	3	960	238	95.9%	< 0.0001	Random effects	33% [0.22–0.28]
SI	5	431	197	95.6%	< 0.0001	Random effects	26% [0.41–0.51]
LTC	1	104	68	—	—	Random effects	65% [0.55–0.74]
Assessment tool				96.4%	< 0.0001	Random effects	37% [0.23–0.53]
CAM	7	1139	473	95.9%	< 0.0001	Random effects	39% [0.18–0.64]
ICDSC	3	960	238	98.4%	< 0.0001	Random effects	33% [0.10–0.68]
DSM‐IV	1	166	166	—	—	Random effects	38% [0.33–0.43]
DRS‐R‐98	1	163	19	—	—	Random effects	34% [0.29–0.39]
Study type				95.6%	< 0.0001	Random effects	37% [0.23–0.53]
Cross‐sectional	6	931	367	94.9%	< 0.0001	Random effects	50% [0.27–0.72]
Case control	2	678	85	0%	0.350	Random effects	17% [0.14–0.20]
Cohort	4	819	259	95.9%	< 0.0001	Random effects	31% [0.14–0.56]
Region				95.6%	< 0.0001	Random effects	37% [0.23–0.53]
Europe	5	1454	366	96.9%	< 0.0001	Random effects	42% [0.20–0.68]
Asia	4	825	238	92.6%	< 0.0001	Random effects	38% [0.18–0.62]
America	3	149	107	96.8%	< 0.0001	Random effects	26% [0.66–0.53]

3.5. Risk Factors

Twelve studies reported 19 risk factors for SSD based on multivariate analysis. Among these, five factors (older age, cognitive impairment, comorbidities, fall history, and low educational level) were considered eligible for quantitative synthesis. The remaining 14 risk factors (male sex, absence of a partner, anemia, hyponatremia, higher C‐reactive protein levels, pain, complicated procedures, previous brain failure in the ICU, preoperative fasting time, number of medications, decreased grip strength, low ADL, smoking, and deafness) were identified through narrative synthesis.

The results of the quantitative synthesis are summarized in Table 3, which includes only risk factors reported with odds ratios (ORs) in at least two studies. However, only older age, cognitive impairment, and comorbidities provided sufficient and comparable data for meta‐analysis. Although fall history and low educational level were reported in more than one study, complete and extractable OR data were available in only one study each; therefore, they were excluded from the meta‐analysis.

TABLE 3.

Meta‐analysis results of risk factors for SSD.

Risk factors	No. of studies	Total sample	Patients with SSD, n	Heterogeneity		Statistical model	OR, 95% CI
				I2	p
Older age	5	1304	332	79.6%	0.0006	Random effects	1.08 [0.89–1.31]
Cognitive impairment	2	481	144	0%	0.8824	Random effects	14.49 [2.38–88.26]
Comorbidities	2	726	191	65%	0.0912	Random effects	1.29 [0.10–17.45]

3.6. Publication Bias

Egger's test (t = 0.52, p = 0.615) indicated no significant evidence of publication bias. The scatter distribution in the funnel plot was uniform and symmetrical (see Figure S1 Funnel plot). Sensitivity analysis revealed that heterogeneity primarily stemmed from variations in SSD diagnostic thresholds and setting‐specific risk profiles.

3.7. Sources of Heterogeneity

10/12 studies were single‐center with small samples, potentially limiting external validity. Methodological variations in design and inconsistent SSD definitions (CAM vs. ICDSC cutoffs) may also contribute to classification bias.

4. Discussion

SSD is a common neuropsychiatric syndrome among critically ill older adults and is associated with multiple adverse outcomes. To enhance interpretive clarity, risk factors were categorized by evidence type. Five factors (older age, cognitive impairment, comorbidities, fall history, and low educational level) were eligible for quantitative synthesis, whereas the remaining factors (male sex, anemia, pain, and polypharmacy, among others) were identified through narrative synthesis. Based on these findings, the etiological characteristics and key risk factors of SSD are summarized as follows:

4.1. Incidence Characteristics

In our study, the reported incidence of SSD among critically ill elderly patients ranged from 11.7% to 71.4%. This wide variation may be attributed to heterogeneity in population characteristics, differences in diagnostic tools, and varying threshold criteria across studies. Additionally, one primary study that used assessment instruments such as the DRS‐R‐98 reported lower SSD incidence rates, potentially lowering the overall pooled estimate [18]. And, the high heterogeneity underscores the need for standardized SSD diagnostic criteria in future studies.

The substantial heterogeneity observed in this meta‐analysis (I² = 92.6%–98.4%) may be largely attributed to the lack of a standardized definition of SSD and the variability in assessment tools used across studies. In this review, several instruments were identified for SSD assessment, including the CAM, the ICDSC, the DRS‐R‐98, and DSM‐IV.

These tools differ substantially in diagnostic thresholds, symptom domains, and scoring approaches. Variations in diagnostic thresholds and symptom definitions across tools may lead to differences in the scope of case identification, thereby influencing both prevalence estimates and the identification of associated risk factors. Specifically, CAM is widely used due to its good international validity and ease of implementation, allowing rapid identification of delirium; however, it lacks a scoring system to reflect symptom severity [25, 26]. In contrast, ICDSC adopts a score‐based approach that enables continuous monitoring of delirium and SSD symptoms, providing a more comprehensive evaluation of symptom progression [4]. DRS‐R‐98 may reduce the likelihood of missed SSD diagnoses, but it is relatively complex and typically requires trained mental health professionals [27]. DSM‐IV provides an internationally recognized diagnostic framework; however, they are time‐consuming and less suitable for patients with impaired communication [28]. Such inconsistencies in diagnostic approaches may lead to variability in case identification, thereby contributing to differences in both prevalence estimates and risk factor identification.

Therefore, clinicians often need to integrate findings from multiple tools, each capturing different aspects of subsyndromal symptoms. Future research should prioritize the development of standardized diagnostic criteria and validated assessment tools for SSD, particularly in elderly critically ill populations, to enhance comparability across studies and improve clinical applicability.

4.2. Risk Factors of SSD

Most studies consider SSD to be a transitional state, either preceding the onset of delirium or reflecting a stage of partial remission, typically characterized by milder and more subtle symptoms. Consistent with previous findings, our study also indicates that the risk factors for SSD closely mirror those associated with delirium.

4.2.1. Non‐Modifiable Risk Factors

4.2.1.1. Demographic Characteristics

Advanced age and cognitive decline have consistently been identified as significant risk factors for SSD [14, 15, 17, 18, 19, 20, 22, 24, 29, 30]. This association may be attributed to age‐related neurodegenerative changes, along with disruptions in stress‐regulating neurotransmitter systems and intracellular signaling pathways, which collectively increase susceptibility to neuropsychiatric disturbances [31, 32].

In addition, lower educational attainment has been reported as a contributing factor in the development of SSD [18, 22]. Individuals with limited educational backgrounds are more likely to have reduced cognitive reserve, which has been linked to a higher incidence of delirium [33]. Conversely, individuals with higher intelligence quotients (IQs) are less likely to develop cognitive impairment [34, 35]. Higher levels of education are also known to mitigate the influence of occupational and socioeconomic status on health‐related behaviors [36]. However, the relationship between educational attainment and SSD remains unclear and warrants further investigation to determine whether it constitutes an independent risk factor in older critically ill patients.

4.2.1.2. Disease Burden

Our analysis indicates that both illness severity and the number of comorbid conditions are significant risk factors for SSD [6, 15, 19, 23, 24], which aligns with established findings in delirium research [37, 38]. In previous studies, disease severity was frequently evaluated using the APACHE II score, with higher scores indicating more severe clinical conditions [39, 40, 41]. One possible explanation is that critically ill patients frequently undergo intensive medical interventions and are subjected to heightened physical and psychological stress, which may increase their susceptibility to SSD.

Moreover, several studies have identified an association between pre‐existing psychiatric comorbidities and a higher incidence of postoperative SSD [14, 24]. Chronic illness and organic brain dysfunction may elevate SSD risk both through impaired psychological resilience and by inducing neurochemical disturbances that disrupt brain function. As these factors are often non‐modifiable, we recommend enhanced mental status monitoring for high‐risk patients.

4.2.2. Treatment‐Related Modifiable Risk Factors

4.2.2.1. Mechanical Ventilation

Analysis of the included studies suggests that mechanical ventilation is a significant risk factor for SSD, as it is frequently associated with complications such as unplanned extubation, prolonged hospitalization, and increased mortality [42, 43]. Furthermore, previous studies have demonstrated that extended durations of mechanical ventilation significantly increase the risk of progression from SSD to delirium [44, 45]. While mechanical ventilation is essential for the management of critically ill patients, it may also lead to significant discomfort and negative emotional responses, which could contribute to the development of SSD [46]. Given that mechanical ventilation is a potentially modifiable risk factor, healthcare providers should remain vigilant regarding its adverse effects, particularly in cases of prolonged use. The implementation of evidence‐based bundled interventions aimed at minimizing the duration of mechanical ventilation may improve patient outcomes and help reduce the incidence of SSD.

4.2.2.2. History of Falls

SSD may present with marked fluctuations in arousal and attention, accompanied by behavioral changes that increase the risk of falls. Conversely, a history of falls may also elevate the likelihood of developing SSD [14, 21, 47]. Falls are associated with a range of serious consequences, including disability, loss of independence, severe neurological injury, and increased mortality risk [48, 49]. Therefore, healthcare professionals should incorporate this information, along with other known high‐risk factors, into comprehensive risk assessment protocols to guide targeted preventive interventions for patients at elevated risk.

4.2.2.3. Smoking History

Smokers exhibited higher rates of SSD [20], an increased incidence of analgesia‐related complications, and greater opioid consumption—all of which may contribute to the progression and severity of both delirium and its associated complications [50, 51]. These adverse outcomes are believed to be driven by the direct neurotoxic effects of nicotine, as well as nicotine‐induced microvascular changes in the brain, which impair executive function and diminish cognitive reserve [52]. Smoking history is associated with adverse outcomes in elderly SSD patients and should be integrated into clinical screening and management pathways.

4.2.2.4. Other Factors

Several other treatment‐related factors, including the length of preoperative fasting time and blood transfusion, have also been associated with an increased risk of SSD [19, 22]. Prolonged fasting may result in dehydration, unnecessary intravenous fluid administration, and other perioperative complications—all of which can contribute to increased cognitive vulnerability [21, 53].

Additionally, our findings suggest that patients who received blood transfusions were more likely to develop SSD [19, 22, 47]. This association may be attributable to transfusion‐related acute systemic inflammatory responses, which are believed to contribute to the pathophysiology of SSD [54].

4.2.3. Disease‐Related Modifiable Factors

4.2.3.1. Pain Management

Our meta‐analysis identified pain as a risk factor for SSD [16]. Additionally, SSD itself may impair a patient's ability to self‐report pain. Pain is an inherently distressing experience that can trigger a range of physiological and psychological disturbances, including impaired sleep quality, heightened tension, and increased irritability [55, 56]. It is widely recognized as a major source of stress in the ICU and a key contributor to patient distress and agitation [57]. Studies have demonstrated that severe postoperative pain—particularly following orthopedic fracture surgery, is associated with a significantly higher incidence of neuropsychiatric disturbances and delirium symptoms [58]. Therefore, clinicians should adopt a stepwise approach to pain management, aiming to maintain a Numerical Rating Scale (NRS) score of ≤ 3 while minimizing opioid use.

The fluctuating course of delirium, combined with impairments in consciousness and attention, often hinders clinicians' ability to accurately assess and manage pain [40]. This can result in the under‐recognition and undertreatment of pain, further exacerbating the risk of SSD. Our findings suggest a dose–response relationship between increasing pain intensity and the risk of SSD. However, further research is needed to clarify the causal nature of this association, as it may be influenced by confounding variables.

4.2.3.2. Nutritional Status

Our studies also found that patients with hypoproteinemia or low red blood cell counts were more likely to develop SSD [16, 19], aligning with other results [59, 60]. These patients also tended to experience a higher incidence of complications and longer ICU stays [61, 62]. Notwithstanding inconsistent albumin‐SSD correlations, proactive nutritional screening and intervention are warranted. By enhancing immune function and physiological resilience, appropriate nutritional support may help reduce the incidence of SSD and shorten ICU length of stay [63].

4.3. Implications for Clinical Practice and Research

Given the high prevalence of SSD in critically ill older adults and its association with multiple risk factors and adverse outcomes, enhancing nurses' capacity for early recognition and intervention is essential. The modifiable risk factors identified, such as pain, nutritional deficits, and polypharmacy, provide actionable targets for a multidisciplinary approach. We recommend incorporating SSD screening into routine clinical assessments in the ICU and implementing collaborative protocols between critical care clinicians and geriatric specialists for early intervention. These strategies may contribute to reduced SSD incidence and improved cognitive outcomes in this vulnerable population.

From a research perspective, the lack of unified diagnostic criteria and reliance on tools originally developed for SSD remain key limitations. These inconsistencies have contributed to variability in prevalence estimates and hindered cross‐study comparisons. Future research should develop validated SSD assessment tools for older critical care populations using standardized diagnostic protocols. This will reduce methodological heterogeneity and facilitate clinical implementation.

5. Limitations

This study has several limitations that should be considered. First, the reliance on published data without additional author input may introduce reporting bias, and substantial heterogeneity (I² = 95.6%) was observed due to variations in SSD assessment methods and follow‐up durations. Second, the analysis was limited by the use of unadjusted estimates and inconsistent reporting formats across studies, with some important risk factors (e.g., smoking, ADL) reported too infrequently for quantitative synthesis.

To address these limitations, future research should employ multicenter designs with standardized SSD assessment protocols and uniform outcome measures (e.g., odds ratios). Greater emphasis should be placed on collecting adjusted estimates and systematically investigating potentially important but understudied risk factors to strengthen the evidence base for clinical practice.

6. Conclusion

This study provides clinically relevant evidence regarding SSD prevalence and modifiable risk factors (e.g., older age, male sex, absence of a partner, education level ≤ 9 years, lower cognitive function, comorbidities, anemia, hyponatremia, higher C‐reactive protein levels, pain, fall history, complicated procedures, previous brain failure in the ICU, preoperative fasting time, polypharmacy, decreased grip strength, low ADL, smoking, and deafness) in elderly critical care patients, which may guide early clinical interventions. Early and accurate recognition of SSD is essential, as it enables timely and effective prevention and management strategies [64].

Moreover, current clinical attention to SSD assessment is often passive and disengaged, which can potentially hinder early identification and timely intervention. To address these gaps, future research should focus on standardizing diagnostic criteria and validating risk factors in larger, more representative populations. Simultaneously, enhancing education and training for healthcare professionals is critical. We recommend integrating routine SSD screening into clinical workflows, developing tailored education modules, and adopting simplified bedside assessment tools. These strategies may improve early detection, promote consistent management, and enhance care quality for critically ill older adults.

Funding

The authors have nothing to report.

Disclosure

The authors have nothing to report.

Ethics Statement

The authors have nothing to report.

Supporting information

Figure S1: Funnel plot of SSD prevalence.

Data S1: Search strategies.

Data Availability Statement

The data are available from the corresponding author upon reasonable request.