Study characteristics

Figure 1 displays the study flowchart. After eliminating duplicate entries from 1567 publications, 1378 titles and abstracts were carefully evaluated. Subsequently, 35 articles that showed probable relevance were chosen for a detailed examination of their full texts, according to the defined inclusion and exclusion criteria. In total, 18 cross-sectional studies [13, 14, 25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40] and 1 case-control study published between 2013 and 2024 [41].

Fig. 1

PRISMA flow diagram of the literature search and study selection process

Using forward and backward citation tracking of the included studies, no new articles were found. Table 1 provides a concise overview of the key features of the studies that were included.

The studies were done entirely in Asia.

16 studies were evaluated for optical coherence tomography (OCT) findings, while 4 studies were assessed for optical coherence tomography angiography (OCT-A) findings.

Only six research focused only on pregnant women, whereas four studies exclusively focused on children.

The studies included in the meta-analysis had a quality score ranging from 2 to 8 out of 8.(Table 3).

A total of 833 individuals with IDA and 894 individuals without IDA were included in these studies. The mean age of the patients was 30.55 ± 7.45 years, and 31.25% of them were male.

Association between iron deficiency anemia (IDA) and retinal nerve fiber layer thickness (RNFLT)

The average retinal nerve fiber layer thickness (RNFLT) and iron deficiency anemia (IDA) were found to be significantly correlated when pooled standardized mean differences (SMDs) were compared (SMD = -0.76, 95% CI: -1.09 to -0.44 ; p-value = 0.001, I2 = 86.88%). In patients with IDA and without IDA, the mean and standard deviation of average RNFLT were, respectively, 98.454 ± 8.99 μm and 105.44 ± 11.41 μm, respectively (WMD = -6.99 μm, 95% CI: -10.35 to -3.64; p-value = 0.001, I2 = 92.56%). Figure 2 displays the results of the meta-analysis.

Fig. 2

Forest plot (random-effects model) depicting: (a) the association of iron deficiency anemia(IDA) and average retinal nerve fiber layer thickness(RNFLT), (b) the association of iron deficiency anemia(IDA) and cup to disc(C/D) ratio

supplementary Fig. 1 demonstrates the association between iron deficiency anemia and the retinal nerve fiber layer thickness (RNFLT) by differentiating it into four regions: superior, inferior, nasal, and temporal. The most significant reduction in the Retinal Nerve Fiber Layer Thickness (RNFLT) was observed in the inferior region, whereas the smallest reduction was observed in the superior region.

Association between iron deficiency anemia (IDA) and the cup to disc ratio (C/D ratio)

The results indicated that there was no significant association between iron deficiency anemia (IDA) and cup to disc ratio (SMD = 0.07, 95% CI: -0.13 to 0.28; p value = 0.49, I2 = 0.0%).

The C/D ratios for patients with and without IDA had respective means and standard deviations of 0.393 ± 0.0325 and 0.385 ± 0.0307 (WMD = 0.008, 95% CI: -0.025 to 0.042; p-value = 0.628, I2 = 0.02%). The meta-analysis’s findings are presented in Fig. 2.

Association between iron deficiency anemia (IDA) and foveal avascular zone (FAZ) parameters

The available data regarding foveal avascular zone (FAZ) parameters in the included studies were the FAZ area, FAZ perimeter, and foveal vessel density in the SCP. We perform a meta-analysis on these items.

A weak negative association was observed between IDA and both the FAZ area (SMD =-0.35, 95% CI: -0.67 to -0.02; p value = 0.04, I2 = 59.76%) and the FAZ perimeter (SMD = -0.34, 95% CI: -0.65 to -0.03; p value = 0.03, I2 = 55.98%). Additionally, there was a negative association between IDA and foveal VD (SMD = -0.73, 95% CI: -1.28 to -0.18; p value = 0.01, I2 = 72.66%). The results of the meta-analysis are displayed in Fig. 3.

Fig. 3

Forest plot (random-effects model) depicting the association of iron deficiency anemia(IDA) and foveal avascular zone(FAZ) parameters in superficial capillary plexus(SCP): (a) FAZ area, (b) FAZ perimeter, (c) foveal density

Association between iron deficiency anemia (IDA) and vascular density (VD)

Our results demonstrated that IDA patients have significantly lower whole image vascular density in the superficial capillary plexus (SCP) than controls (SMD = -1.12, 95% CI: -1.85 to -0.39; p-value = 0.001, I2 = 83.15%). Similar findings were observed in the deep capillary plexus (DCP), however the difference was not statistically significant (SMD = -0.30, 95% CI: -0.89 to 0.29; ,p-value = 0.32 ,I2 = 77.20%). The results of the meta-analysis are shown in Fig. 4.

Fig. 4

Forest plot (random-effects model) depicting the association of iron deficiency anmeia(IDA) and vascular denisty : (a) superficial capillary plexus(SCP), (b) deep capillary plexus(DCP)

Subgroup analysis and meta-regression

All the results of the subgroup analyses are shown in Table 2. The subgroup analysis by study population revealed no significant difference between strata (p-value = 0.814, Supplementary Fig. 2a, SMD = -0.82, 95% CI: -1.47 to -0.17 for studies consisted only women, SMD = -0.98, 95% CI: -2.01 to 0.04 for studies consisted only children and SMD = -0.67, 95% CI: -1.00 to -0.34 for studies with no specified population).

Table 2 Subgroup analyses for association between IDA and various outocmes

There were also no significant differences between studies with low risk of bias (ROB) compared to high to moderate ROB (p-value = 0.902, Figure S2b).

Studies that reported right eye retinal nerve fiber layer thickness (RNFLT) did not differ significantly from those reported mean RNFLT of both eyes (p-value = 0.754, Figure S2c).

Subgroup analysis by OCT/OCTA devices revealed a significant difference between them (p-value = 0.03, Figure S2d). Moreover, the statistical heterogeneity in the ZEISS subgroup was considerably decreased compared with all study I2 (30.78 vs. 86.88, respectively). For other objectives, subgroup analyses were not feasible due to the small number of studies.

Table 3 and supplementary Fig. 6 present the meta-regression results for the role of mean age and male to female ratios. Among all of the conducted meta-regressions, only the male-to-female ratio in association between IDA with FAZ perimeter and FD in SCP resulted in a considerable reduction in residual I2 (0.00 residual I2 for both).

Table 3 Meta-regressions for different variablesRisk of bias assessment

Table 4A and 4B summarizes the results of the JBI checklist for cross-sectional and case-control studies. In cross sectional studies the total quality score of the included articles varied from 3 to 8. Two studies had a low risk of bias, twelve had a moderate risk, and one had a high risk of bias. The association between iron deficiency anemia and RNFLT doesn’t differ significantly among high to moderate and low-quality studies, as previously mentioned. All investigations used a valid and reliable method for RNFLT assessment.

Table 4 A. Detailed results of the risk of bias assessment of the included studies based on the JBI for cross-sectional studiesTable 4 B. Detailed results of the risk of bias assessment of the included studies based on the JBI for case-control studiesPublication bias

Egger’s tests, funnel plots, and the trim and fill method were used to evaluate the publication bias (Fig. 5). Since using a funnel plot is not appropriate when there are fewer than 10 studies are available, we opted to only utilize it for assessing the association between IDA and RNFLT [42]. For the association between IDA and RNFLT, the funnel plot revealed some asymmetry, and the Egger test yielded significant results. However, the trim and fill method did not detect considerable publication bias. For all other objectives, the Egger’s test was nonsignificant, and the trim and fill method did not indicate publication bias. The trim and fill findings are included in Supplementary Fig. 3, along with the p value for each objective egger test.

Fig. 5

Publication bias assessment. Association between IDA and average RNFLT

Sensitivity analysis

Following the use of the one-out, one-remove approach, no single study had a considerable impact on the pooled estimates (Supplementary Fig. 4).