Determinants of health literacy in Iran: a systematic review and meta-analysis of quantitative studies (2001–2024)

Abstract

Background

Health literacy is a critical social determinant of health and a key lever for reducing health inequities. Understanding its determinants is essential for designing effective public health interventions.

Objective

This study aimed to synthesize quantitative evidence on the individual- and contextual-level determinants of health literacy among the Iranian population.

Methods

Following PRISMA 2020 guidelines, we systematically searched PubMed, Scopus, Web of Science, Cochrane Library, and Iranian databases (SID, Magiran) for observational and interventional studies published between January 2001 and December 2024. Eligible studies used quantitative designs, assessed health literacy using validated instruments (e.g., HLS-Iran, TOFHLA), were conducted among Iranian populations, and reported sufficient statistical data to calculate effect sizes. Two independent reviewers conducted screening and data extraction; disagreements were resolved by a third reviewer. Study quality was appraised using STROBE and CONSORT checklists. Effect sizes were harmonized as standardized mean differences (Hedges' g) with 95% confidence intervals (CIs) and pooled using random-effects models (REML) in Stata version 16.0. Heterogeneity was assessed using I² and Cochran's Q. Subgroup analyses, meta-regression, and sensitivity analyses were performed where applicable.

Results

Ninety-seven primary studies were included. Among demographic determinants, educational attainment showed the largest association (k=18; g=0.65, 95% CI: 0.55–0.75), followed by urban residence (k=10; g=0.35, 95% CI: 0.22–0.48); age, gender, and marital status showed small and imprecise associations. Psychological determinants showed strong positive associations, particularly self-efficacy (k=12; g=0.58, 95% CI: 0.46–0.70) and self-esteem (k=9; g=0.51, 95% CI: 0.39–0.63), whereas social-geographic isolation was negatively associated (k=7; g=−0.38, 95% CI: −0.52 to −0.24). Among cultural determinants, digital literacy (k=8; g=0.52, 95% CI: 0.40–0.64) and cultural capital (k=7; g=0.47, 95% CI: 0.36–0.58) were positively associated, while traditional lifestyle adherence showed a negative association (k=6; g=−0.31, 95% CI: −0.44 to −0.18). Economic determinants revealed that household income was positively associated (k=13; g=0.48, 95% CI: 0.36–0.60), whereas the Gini coefficient (k=8; g=−0.42, 95% CI: −0.55 to −0.29) and inflation (k=5; g=−0.28, 95% CI: −0.40 to −0.16) were negatively associated. Among social determinants, quality of life showed the strongest positive association (k=14; g=0.60, 95% CI: 0.50–0.70), followed by social support in healthcare (k=11; g=0.45, 95% CI: 0.35–0.55). Moderate-to-substantial heterogeneity was observed across most estimates (I²: 46%–63%). Sensitivity analyses and Egger tests did not reveal significant changes in estimates or consistent publication bias.

Conclusion

Health literacy in Iran is shaped by a complex interplay of psychological, socio-economic, cultural, and geographic factors, with education, quality of life, self-efficacy, and digital literacy emerging as the strongest determinants. Multilevel interventions—spanning educational development, digital inclusion, poverty reduction, psychosocial support, and community engagement—are needed to address these determinants and advance health equity.

Introduction

Health literacy is increasingly recognized as a key determinant of individual and population health. It refers to the capacity to obtain, understand, appraise, and use health information and services to make appropriate health decisions [1]. Contemporary frameworks view health literacy as a multidimensional construct that extends beyond basic reading and numeracy skills to include cognitive, social, and cultural competencies needed to navigate health systems and engage in health-promoting behaviors [2, 3].

Low health literacy has been associated with poorer self-management, reduced use of preventive services, greater hospitalization, and higher healthcare costs [4]. In contrast, higher health literacy is linked to better adherence, more effective communication with healthcare providers, and improved access to care [5]. Its importance became especially visible during the COVID-19 pandemic, when individuals had to interpret rapidly changing public health guidance and manage high levels of uncertainty and misinformation [6].

In Iran, health literacy has received growing attention because of socioeconomic inequality, cultural diversity, and the increasing burden of non-communicable diseases [7]. Existing studies suggest that health literacy levels vary across populations and are influenced by social, economic, psychological, cultural, and demographic factors. However, findings remain fragmented due to differences in study design, sampling, measurement tools, and analytical methods [8]. As a result, a comprehensive synthesis of the evidence is needed to clarify the main determinants of health literacy in this context [9, 10].

Health literacy should also be understood as more than an individual attribute. It is shaped by broader social determinants such as education, income, employment, place of residence, and cultural resources, all of which influence access to information and the ability to use health services effectively [11, 12] .Identifying the relative importance of these determinants is important for designing interventions that are not only educational but also structural, context-sensitive, and equitable [13, 14].

Therefore, this systematic review and meta-analysis synthesized quantitative evidence published between January 2001 and December 2024 on the determinants of health literacy among Iranian populations. We examined associations across social, cultural, psychological, economic, and demographic domains to provide a quantitative summary that can inform research, policy, and intervention design.

Methods

Study design

This study employed a systematic review and meta-analysis in accordance with the PRISMA 2020 guidelines. A protocol was developed a priori; however, it was not registered in a public database (e.g., PROSPERO), which is acknowledged as a limitation. The methodology comprised five sequential steps: (1) protocol development, (2) comprehensive literature search, (3) screening and eligibility assessment, (4) methodological quality appraisal, and (5) statistical meta-analysis. To minimize selection bias, two independent reviewers conducted all stages of the review process; disagreements were resolved through discussion with a third reviewer. Given the inclusion of observational and interventional studies, the analysis focused on associations rather than causal inference.

Eligibility criteria

Included studies: (1) used a quantitative design (e.g., cross-sectional, interventional, or correlational); (2) assessed health literacy as a primary or secondary outcome; (3) were conducted among Iranian populations; and (4) reported sufficient statistical data (e.g., means, standard deviations, correlation coefficients, regression coefficients, or t/F statistics) to calculate effect sizes. Exclusion criteria were: qualitative, theoretical, or review studies; non-empirical or duplicate publications; and studies lacking extractable statistical data. To ensure comparability across heterogeneous study designs, effect sizes were harmonized prior to pooling. For group comparisons, standardized mean differences (SMDs) were calculated directly. For correlational data, Pearson’s r was transformed using Fisher’s z and, where necessary, converted to SMD using established formulas (e.g., ). Regression coefficients were converted to correlation coefficients using standard approximation methods when sufficient data were available. These transformations were applied under assumptions of linearity and comparable variance and were interpreted cautiously.

Information sources and search strategy

A systematic search was conducted across international and Iranian scientific databases from January 2001 to December 2024. International databases included PubMed, Scopus, Web of Science, and the Cochrane Library; Iranian databases included Scientific Information Database (SID) and Magiran. Google Scholar was used for supplementary hand-searching. Search terms combined keywords related to “health literacy,” “determinants,” “Iran,” and specific factors (e.g., “education,” “income,” “gender,” “cultural capital,” “psychological distress”) using Boolean operators (AND/OR). The full search strategy for each database is available in the supplementary materials. Additionally, reference lists of included studies were screened for further eligible publications.

Study selection and data extraction

Titles and abstracts were screened for relevance, followed by full-text assessment against eligibility criteria. Data were extracted using a standardized form capturing author, publication year, study location, sample characteristics, research design, health literacy measurement tools, independent variables, and key statistical results. When a single study contributed data to multiple determinant domains, each effect size was extracted separately but carefully tracked to avoid unit-of-analysis errors and artificial inflation of precision. In cases of overlapping samples, only the most comprehensive dataset was retained .

Quality assessment

Study quality was assessed using the STROBE checklist for observational studies and CONSORT for interventional studies. A customized three-point scoring system (0 = not reported, 1 = partially reported, 2 = adequately reported) was applied. Total scores (range: 0–44) classified studies as poor (1–15), moderate (16–30), or good (31–44). Only studies scoring ≥ 16 were included in the meta-analysis.

These tools were used solely to assess reporting quality and not as validated risk-of-bias instruments. Therefore, quality scores were used for descriptive purposes and sensitivity analyses only, and no study was excluded solely based on these scores. Sensitivity analyses were conducted to evaluate the impact of lower-quality studies on pooled estimates.

Statistical analysis

Data were analyzed using Stata software (version 16.0). Given the diversity of measurement instruments, SMD was selected as the primary effect size to standardize results across studies. The primary effect size metric was the standardized mean difference (SMD) with 95% confidence intervals (CIs), calculated for each factor associated with health literacy. A random-effects model was used to account for expected heterogeneity across studies. Between-study variance was estimated using the restricted maximum likelihood (REML) method. Statistical heterogeneity was quantified using the I² statistic and Cochran’s Q test, (with I² > 50% or p < .05 indicating substantial heterogeneity). I² values were reported for each pooled estimate, and prediction intervals were calculated where feasible. Subgroup analyses were conducted by domain of determinants (socio-demographic, economic, cultural, psychological) and by health literacy instrument. Meta-regression analyses were performed, where sufficient studies were available, to explore sources of heterogeneity (e.g., publication year, geographic region, instrument type). Sensitivity analyses assessed the influence of study quality on pooled estimates. When ≥ 10 studies were available for a given factor, publication bias was evaluated using Egger’s regression test and funnel plots. Given the large number of statistical tests, the potential for type I error inflation was acknowledged; however, no formal multiple comparison correction was applied, and findings were interpreted cautiously.

Associations involving macro-level economic indicators (e.g., Gini coefficient, inflation) were interpreted with caution due to the potential for ecological fallacy.

The study selection process is summarized in a PRISMA 2020 flow diagram (Fig. 1).

Fig. 1.

PRISMA 2020 flow diagram of study selection

Results

Ninety‑seven unique primary studies (2001–2024) were included. Table 1 presents pooled effect sizes (Hedges’ g), 95% CIs, I², 95% PIs, direction, and significance for each determinant (see Table 1). All pooled estimates were produced using random‑effects models (REML).

Table 1.

Pooled effect sizes for all social determinants are presented in Table 1

Domain	Determinant	k	d	95% CI	I²	Prediction Interval	Direction	Sig.
Social	Quality of life	14	0.60	0.50–0.70	61%	0.04–1.16	Positive	Yes
	Social support in healthcare	11	0.45	0.35–0.55	54%	−0.02–0.92	Positive	Yes
	Public trust in health services	5	0.18	0.08–0.28	63%	−0.15–0.51	Positive	Yes*
Cultural	Digital literacy	8	0.52	0.40–0.64	58%	0.05–0.99	Positive	Yes
	Cultural capital	7	0.47	0.36–0.58	52%	0.02–0.92	Positive	Yes
	Traditional lifestyle	6	−0.31	−0.44–−0.18	49%	−0.72–0.10	Negative	Yes
	Belief in modern medicine	5	0.28	0.16–0.40	47%	−0.05–0.61	Positive	Yes
Psychological	Self-efficacy	12	0.58	0.46–0.70	55%	0.08–1.08	Positive	Yes
	Self-esteem	9	0.51	0.39–0.63	50%	0.07–0.95	Positive	Yes
	Social-geographic isolation	7	−0.38	−0.52–−0.24	57%	−0.82–0.06	Negative	Yes
	Health education belief	5	0.32	0.20–0.44	46%	0.01–0.63	Positive	Yes
Economic	Household income	13	0.48	0.36–0.60	53%	0.01–0.95	Positive	Yes
	Economic participation	6	0.35	0.24–0.46	48%	0.02–0.68	Positive	Yes
	Economic inequality (Gini)	8	−0.42	−0.55–−0.29	61%	−0.88–0.04	Negative	Yes
	Inflation	5	−0.28	−0.40–−0.16	52%	−0.60–0.04	Negative	Yes
	Unemployment	4	−0.22	−0.35–−0.09	49%	−0.55–0.11	Negative	Yes†
Demographic	Education level	18	0.65	0.55–0.75	59%	0.12–1.18	Positive	Yes
	Place of residence	10	0.35	0.22–0.48	52%	0.01–0.69	Positive	Yes
	Age	15	0.15	0.05–0.25	58%	−0.10–0.40	Positive	Yes*
	Marital status	9	0.12	0.02–0.22	50%	−0.08–0.32	Mixed	Yes*
	Gender	14	0.10	0.01–0.19	46%	−0.06–0.26	Mixed	Yes*

k = number of unique primary studies contributing to the pooled estimate for that determinant. Effect sizes are reported as Hedges’ g (adjusted standardized mean difference), not Cohen’s d. 95% CI = 95% confidence interval; I² = heterogeneity statistic; PI = 95% prediction interval. Sig. = statistical significance of pooled estimate (p < .05)

* Estimate is statistically significant (p < .05), but a wide 95% CI and a PI that crosses zero indicate limited precision and fragility; interpret with caution. † Based on k < 5 studies; pooled estimate has low precision and should be interpreted with particular caution. Between-study variance (τ²) was estimated using restricted maximum likelihood (REML); all pooled estimates were obtained from random‑effects models with REML unless stated otherwise. Macro-level economic variables (e.g., Gini coefficient, inflation) carry additional ecological-fallacy risk because most primary studies did not use multilevel modelling

Approximately 20–30 associations were tested; no formal correction for multiple comparisons was applied.

Social determinants (Table 1, rows 1–3)

Five social determinants were synthesized. Quality of life showed the largest positive association (k = 14; g = 0.60, 95% CI: 0.50–0.70; I² = 61%; 95% PI: 0.04–1.16). Social support in healthcare demonstrated a moderate positive association (k = 11; g = 0.45, 95% CI: 0.35–0.55; I² = 54%; 95% PI: −0.02–0.92). Public trust in health services showed a small positive association (k = 5; g = 0.18, 95% CI: 0.08–0.28; I² = 63%; 95% PI: −0.15–0.51; imprecise; * in Table 1).

Heterogeneity was moderate (I²: 54%–63%); some PIs cross zero.

Cultural determinants (Table 1, rows 4–7)

Digital literacy had the strongest positive association (k = 8; g = 0.52, 95% CI: 0.40–0.64; I² = 58%; 95% PI: 0.05–0.99). Cultural capital was positive (k = 7; g = 0.47, 95% CI: 0.36–0.58; I² = 52%). Traditional lifestyle was negative (k = 6; g = − 0.31, 95% CI: −0.44 to − 0.18; I² = 49%).

Belief in modern medicine was positive (k = 5; g = 0.28, 95% CI: 0.16–0.40; I² = 47%).

Heterogeneity was moderate; Egger tests did not show consistent asymmetry.

Psychological determinants (Table 1, rows 8–11)

Self‑efficacy showed the largest positive association (k = 12; g = 0.58, 95% CI: 0.46–0.70; I² = 55%; 95% PI: 0.08–1.08). Self‑esteem was positive (k = 9; g = 0.51, 95% CI: 0.39–0.63; I² = 50%). Social‑geographic isolation was negative (k = 7; g = − 0.38, 95% CI: −0.52 to − 0.24; I² = 57%). Health‑education belief was positive (k = 5; g = 0.32, 95% CI: 0.20–0.44; I² = 46%).

Heterogeneity was moderate to substantial; some PIs were wide.

Economic determinants (Table 1, rows 12–16)

Household income was positive (k = 13; g = 0.48, 95% CI: 0.36–0.60; I² = 53%). Economic participation was positive (k = 6; g = 0.35, 95% CI: 0.24–0.46; I² = 48%). Gini coefficient was negative (k = 8; g = − 0.42, 95% CI: −0.55 to − 0.29; I² = 61%). Inflation was negative (k = 5; g = − 0.28, 95% CI: −0.40 to − 0.16; I² = 52%). Macro‑level indicators are ecological and should not be interpreted as individual‑level effects.

Demographic determinants (Table 1, rows 17–21)

Education level showed the largest association (k = 18; g = 0.65, 95% CI: 0.55–0.75; I² = 59%; 95% PI: 0.12–1.18). Place of residence was positive (k = 10; g = 0.35, 95% CI: 0.22–0.48; I² = 52%). Age was small and imprecise (k = 15; g = 0.15, 95% CI: 0.05–0.25; I² = 58%; 95% PI: −0.10–0.40). Gender (k = 14; g = 0.10, 95% CI: 0.01–0.19; I² = 46%) and marital status (k = 9; g = 0.12, 95% CI: 0.02–0.22; I² = 50%) were small and imprecise; PIs include the null.

Overall pattern and heterogeneity (Table 1; Figs. 2, 3, 4 and 5 )

Fig. 2.

Forest plot displaying pooled standardized mean differences (Hedges’ g) (d) with 95% confidence intervals for all 17 health literacy determinants synthesized across 97 studies (Iranian populations, 2001–2024). Diamond markers indicate pooled estimates; horizontal lines represent 95% CIs. Dotted vertical lines indicate Cohen’s benchmarks (small = 0.20, medium = 0.50, large = 0.80). Grey shaded zone denotes the negligible effect region (|d| < 0.20). k = number of studies; I² = heterogeneity statistic. Random-effects model; DerSimonian–Laird estimator

Fig. 3.

Horizontal bar chart displaying pooled standardized mean differences (Hedges’ g) (d) ranked by absolute magnitude with 95% confidence intervals. Bars extending to the left represent negative associations (higher value of determinant associated with lower health literacy); bars extending to the right represent positive associations. Error bars show 95% CIs. Colour coding indicates determinant domain. Cohen’s small (0.20), medium (0.50), and large (0.80) benchmarks are shown as dotted vertical reference lines

Fig. 4.

Funnel plot displaying study-level effect estimates (d) against their standard errors (SE) for assessment of publication bias across all included studies. Each point represents one study-level estimate. The red vertical line marks the overall pooled effect size. Dashed lines indicate the 95% pseudo-confidence interval. Broadly symmetric distribution around the pooled estimate suggests no major small-study effect; Egger’s regression test did not reveal significant asymmetry across domains (all p > .10). Asymmetry, where present, should be interpreted cautiously given small numbers of studies in some subgroups

Fig. 5.

Bubble chart illustrating the relationship between pooled effect size (x-axis) and statistical heterogeneity I² (y-axis) for each health literacy determinant. Bubble size is proportional to the number of contributing studies (k). Colour coding indicates determinant domain. Horizontal reference lines mark I² = 50% (moderate heterogeneity threshold) and I² = 75% (high heterogeneity threshold). Vertical dotted lines indicate Cohen’s small (0.20), medium (0.50), and large (0.80) benchmarks. Determinants positioned in the upper-right quadrant (large effect, high heterogeneity) warrant subgroup analysis to identify sources of variability

Largest pooled associations: education (g = 0.65), quality of life (g = 0.60), self‑efficacy (g = 0.58), digital literacy (g = 0.52). Moderate‑to‑high heterogeneity was present (I² frequently > 50%). Figure 2 shows the forest plot; Figs. 3, 4 and 5 show ranked effect sizes and effect–heterogeneity relationships.

Publication bias and sensitivity (Fig. 4; Table 2)

Table 2.

Methodological reporting quality and publication bias assessment by domain

Domain	k studies	Good quality	Moderate quality	Poor quality	Mean score (%)	Pub. bias (Egger)	Egger p-value
Social	5	3 (60%)	2 (40%)	0%	72%	No	0.42
Cultural	7	4 (57%)	3 (43%)	0%	67%	No	0.38
Psychological	8	5 (63%)	3 (37%)	0%	71%	No	0.44
Economic	6	4 (67%)	2 (33%)	0%	68%	No	0.36
Demographic	14	9 (64%)	5 (36%)	0%	73%	No	0.51
Total	40	25 (63%)	15 (37%)	0%	70%	—	—

Quality classification based on adapted STROBE/CONSORT reporting checklist (poor: 1–15; moderate: 16–30; good: 31–44); scores reflect reporting completeness, not validated risk-of-bias assessment. Publication bias assessed via Egger’s weighted regression test (p > .10 interpreted as no significant asymmetry) and funnel plot visual inspection. ‘No’ = no significant evidence of publication bias detected. k studies = number of unique primary studies contributing to this domain (not number of effect sizes)

Egger tests and funnel plots did not show consistent evidence of publication bias for estimates with ≥ 10 studies (Egger p > .10), but power is limited in small subgroups Sensitivity analyses (alternative conversions, excluding low‑reporting studies, REML vs. DL) yielded similar directions and magnitudes.

The forest plot displaying pooled effect sizes with 95% CIs for all determinants is shown in Fig. 2.

Ranked effect sizes with 95% CIs are displayed in Fig. 3

Funnel plots for assessment of publication bias across all included studies are shown in Fig. 4

The relationship between pooled effect size and heterogeneity (I²) by determinant is illustrated in Fig. 5.

Discussion

This meta-analysis demonstrates that health literacy among Iranian populations is shaped by a multidimensional set of social, cultural, psychological, economic, and demographic determinants rather than a single dominant factor. Consistent with prior research, the findings highlight the importance of structural and psychosocial conditions over purely demographic characteristics, supporting the need for multilevel interventions [15, 16]. Given the heterogeneity of included studies, standardized mean differences (SMDs) were used to enable comparability; however, the transformation of different effect size metrics requires cautious interpretation.

Social determinants

Social determinants, particularly quality of life and perceived support in healthcare settings, were strongly associated with health literacy. Participation in health promotion programs and communication networks further emphasized the role of social engagement and system responsiveness, consistent with evidence from community-based interventions [17]. However, variability across studies suggests that differences in measurement and context contributed to observed heterogeneity.

Economic determinants

Economic factors showed mixed associations. Individual-level indicators such as income were positively associated with health literacy, whereas macro-level indicators such as inequality and inflation were negatively associated. These findings underscore the role of socioeconomic context [16, 18], but should be interpreted cautiously due to the risk of ecological fallacy, as most studies did not apply multilevel designs.

Cultural determinants

Digital literacy emerged as a critical enabler of health literacy in the contemporary context (d = 0.50), consistent with recent research on digital health literacy [19]. Cultural capital and awareness were also positively associated. These results align with sociocultural theories positing that cultural capital functions as both a cognitive and symbolic resource, enhancing individuals’ capacity for informed health decision-making [20, 21]. Notably, adherence to a traditional lifestyle was negatively associated with health literacy (d = − 0.30), potentially reflecting resistance to modern health practices in culturally conservative settings. This finding supports calls for culturally sensitive health communication strategies that bridge traditional beliefs with evidence-based care [22, 23].

Psychological determinants

Psychological factors, especially self-efficacy, demonstrated strong positive associations, consistent with established behavioral theories. Engagement with health education and trust in modern medicine further supported health literacy [24], whereas social and geographic isolation (d = − 0.48) significantly impeded health literacy—particularly in rural and underserved regions [25]. These findings underscore the need for interventions that simultaneously strengthen psychological resilience and mitigate geographic inequities in health information access.

Demographic determinants

Demographic variables demonstrated weaker and less consistent associations. Only educational attainment (d = 0.25) and urban residence (d = 0.22) reached statistical significance, suggesting that structural access to education and health services may be more consequential than intrinsic demographic characteristics. In contrast, age, gender, and marital status showed minimal or non-significant effects—a pattern consistent with other meta-analyses cautioning against overreliance on demographic profiling in health literacy interventions [26–28].

Strengths and limitations

A key strength of this study is its rigorous adherence to PRISMA guidelines and the inclusion of a diverse body of evidence spanning multiple domains of health literacy determinants.

Substantial heterogeneity was observed across analyses, reflecting differences in study design, populations, and measurement tools. Although partially explored, unexplained variability remains. Additionally, given the number of analyses conducted, the possibility of type I error cannot be excluded. Effect sizes should therefore be interpreted cautiously, as even moderate values may have limited practical significance in large observational datasets.

From a practical perspective, the findings suggest prioritizing determinants with stronger and more consistent associations—particularly education, quality of life, self-efficacy, digital literacy, and economic resources. These results support multilevel strategies that integrate educational, psychosocial, and structural interventions rather than focusing solely on individual knowledge deficits.

This study has several limitations. The predominance of cross-sectional designs limits causal inference. The use of SMD across heterogeneous measures introduces methodological uncertainty. Some studies contributed to multiple domains, raising potential non-independence.

Additionally, STROBE and CONSORT reflect reporting quality rather than validated risk-of-bias assessment. Finally, the lack of protocol registration may reduce transparency.

Abbreviations

CI

Confidence Interval

CONSORT

Consolidated Standards of Reporting Trials

g

Hedges’ g (standardized mean difference)

HL

Health Literacy

HLS

Health Literacy Scale

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta-Analyses

REML

Restricted Maximum Likelihood

SES

Socioeconomic Status

STROBE

Strengthening the Reporting of Observational Studies in Epidemiology

TOFHLA

Test of Functional Health Literacy in Adults

Authors' contributions

R. Menati: Contributed to the conception and design of the study, conducted the systematic literature search, and drafted the manuscript T.A: Contributed to study screening, data extraction, statistical analysis (meta-analysis, subgroup analysis, and meta- regression),interpretation of results, and critically revised the manuscript W. M: Contributed to study supervision, quality assessment of included studies (STROBE/CONSORT appraisal), resolution of disagreements, critical review, and final approval of the manuscript.

Funding

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

Data availability

The datasets generated and/or analyzed during the current study (including data extraction sheets, effect size calculations, and CMA analysis files) are available . Search strategies and PRISMA checklist are provided as supplementary materials.”

Declarations

Ethics approval and consent to participate

Not applicable. This study is a systematic review and meta-analysis of previously published studies and does not involve human participants or primary data collection.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.