Real-world injection burden, treatment switching, and visual outcomes in diabetic macular edema: a 563-eye cohort study

Abstract

Background

To evaluate real-world injection burden, treatment switching patterns, and visual outcomes in patients with diabetic macular edema (DME) receiving intravitreal anti–vascular endothelial growth factor (anti-VEGF) therapy.

Materials and methods

This retrospective observational cohort study included 563 eyes with DME treated with anti-VEGF agents and followed for at least 12 months. All eyes received an initial loading phase of three monthly bevacizumab injections followed by a pro re nata regimen. Primary outcomes were the number of injections and outpatient visits during the first year. Secondary outcomes included change in best-corrected visual acuity (BCVA), distribution of visual outcomes, and treatment switching patterns. Multivariable linear regression and generalized estimating equation (GEE) analyses were performed to identify independent predictors of visual outcomes.

Results

Eyes received a mean of 5.4 ± 2.4 injections and attended 8.0 ± 2.2 visits during the first year. Mean BCVA improved from 0.75 ± 0.43 to 0.59 ± 0.42 logMAR (mean change − 0.16 ± 0.35; P < 0.001). Visual improvement was observed in 60.2% of eyes, stability in 21.7%, and deterioration in 18.1%. Treatment switching occurred in 29.7% of eyes, typically within the first 4–5 months. In multivariable linear regression analysis, baseline BCVA was associated with visual acuity change (β = −0.33, P < 0.001). Treatment switching was associated with visual outcomes in unadjusted analysis but was not independently associated after adjustment. In the GEE analysis accounting for inter-eye correlation, baseline BCVA remained significantly associated with visual acuity change (P < 0.001), whereas injection number (P = 0.527) and treatment switching (P = 0.976) were not associated.

Conclusions

In real-world clinical practice, anti-VEGF therapy for DME was associated with meaningful visual improvement despite a moderate injection frequency. Visual outcomes were primarily associated with baseline visual acuity rather than treatment intensity. Treatment switching was not independently associated with visual acuity change after adjustment, suggesting that it likely reflects underlying disease refractoriness rather than acting as an independent factor associated with visual outcomes; however, causal inference cannot be established due to the observational study design. These findings underscore the importance of individualized treatment strategies in DME.

Introduction

Diabetic macular edema (DME) is a leading cause of vision loss among individuals with diabetes mellitus and represents a major global health burden. The prevalence of DME continues to rise in parallel with the increasing incidence of diabetes worldwide, particularly in aging populations. Vision impairment due to DME significantly affects patients’ quality of life and imposes substantial socioeconomic costs on healthcare systems [1, 2].

The pathophysiology of DME is primarily driven by disruption of the blood–retinal barrier and increased vascular permeability mediated by vascular endothelial growth factor (VEGF). Elevated intraocular VEGF levels lead to retinal thickening, fluid accumulation, and progressive visual decline [3]. As a result, intravitreal anti-VEGF therapy has become the first-line treatment for center-involving DME, largely replacing macular laser photocoagulation in routine clinical practice [4].

The efficacy of anti-VEGF therapy in DME has been firmly established in multiple randomized controlled trials. The RISE and RIDE studies demonstrated significant and sustained visual acuity improvements with ranibizumab compared with sham injections [5], while the VIVID and VISTA trials confirmed the superiority of aflibercept over laser photocoagulation [6]. Furthermore, the DRCR.net Protocol T study showed that treatment response varies according to baseline visual acuity, highlighting the importance of individualized therapeutic approaches [7]. These landmark trials have established anti-VEGF therapy as the standard of care for DME.

Despite these favorable outcomes under controlled trial conditions, treatment patterns in real-world clinical practice often differ substantially from protocol-driven regimens. Real-world studies consistently report lower injection frequencies, irregular follow-up intervals, and heterogeneous retreatment strategies compared with randomized trials [8, 9]. Consequently, visual outcomes achieved in routine practice may be less pronounced than those reported in clinical trials, reflecting a gap between efficacy and effectiveness. In real-world clinical practice, additional challenges such as variability in patient adherence, irregular follow-up intervals, differences in retreatment criteria, and healthcare system constraints may further contribute to heterogeneity in treatment delivery and outcomes.

Treatment switching represents another important aspect of DME management in real-world settings. A considerable proportion of patients exhibit suboptimal anatomical or functional response to initial anti-VEGF therapy, necessitating a switch to alternative agents. Although previous studies suggest that switching may provide additional benefit in selected cases, evidence regarding optimal timing, patient selection, and associated visual outcomes remains limited and heterogeneous [10, 11].

In addition, the treatment burden associated with repeated intravitreal injections and frequent clinical visits represents a major challenge for both patients and healthcare systems. Understanding real-world injection patterns and their relationship with functional outcomes is therefore essential for optimizing long-term management strategies in DME [9, 12].

Given these considerations, real-world data derived from large clinical cohorts are essential to complement findings from randomized trials and to better reflect routine clinical practice. Therefore, the aim of the present study was to evaluate real-world injection burden, treatment switching patterns, and visual outcomes in a large cohort of patients with DME treated with intravitreal anti-VEGF therapy. Specifically, we sought to determine whether visual outcomes are more strongly associated with treatment intensity or baseline visual acuity.

Materials and methods

Study design and setting

This retrospective observational cohort study was conducted at a tertiary referral retina clinic. Medical records of patients who received intravitreal anti–vascular endothelial growth factor (anti-VEGF) therapy between January 2023 and December 2025 were reviewed. The study evaluated real-world treatment patterns and visual outcomes during the first year following treatment initiation. The study protocol was approved by the Institutional Review Board (approval no: 800) and adhered to the tenets of the Declaration of Helsinki.

Study population

Eyes with a diagnosis of diabetic macular edema (DME), a minimum follow-up duration of 12 months, and no prior treatment for DME were included in the analysis. Eyes that had previously received anti-VEGF therapy, intravitreal corticosteroid treatment, or focal/grid laser photocoagulation for DME were excluded. The diagnosis of DME was based on clinical examination and optical coherence tomography (OCT) findings. Eyes were excluded if baseline best-corrected visual acuity (BCVA) data were unavailable or if significant ocular comorbidities that could affect visual acuity were present, including advanced glaucoma, dense cataract, or corneal opacity. Both eyes from the same patient were included if they met the eligibility criteria, reflecting routine clinical practice.

Treatment protocol

All eyes initially received three consecutive monthly intravitreal injections of bevacizumab as a loading phase, in accordance with institutional treatment protocols and reimbursement policies. Following the loading phase, patients were managed using a pro re nata (PRN) regimen. Retreatment decisions were based on OCT findings and functional assessment. Criteria for retreatment included the presence of persistent or recurrent intraretinal or subretinal fluid, increased central macular thickness, or deterioration in BCVA compared with the previous visit. Patients were followed at monthly intervals, during which OCT imaging was performed to guide retreatment decisions.

Treatment switching

Treatment switching from bevacizumab to aflibercept was considered in eyes demonstrating inadequate anatomical or functional response. Inadequate response was defined as persistent retinal fluid, insufficient anatomical improvement on OCT assessment, or lack of satisfactory functional improvement, as judged by the treating physician, despite repeated injections. Because of the retrospective real-world design, no rigid quantitative anatomical or visual acuity thresholds were predefined; treatment switching decisions were based on the treating physician’s overall clinical judgment integrating both anatomical and functional response. For eyes undergoing treatment switching, the timing of switching, BCVA before switching, and BCVA after switching were recorded. Treatment switching was unidirectional from bevacizumab to aflibercept during the 12-month study period; no eyes switched back to bevacizumab or underwent additional therapeutic modifications.

Data collection

Demographic and clinical data were extracted from medical records. Variables included age, sex, number of outpatient visits, total number of intravitreal injections, baseline and 12-month BCVA, and treatment switching status. BCVA was measured using Snellen charts and converted to logarithm of the minimum angle of resolution (logMAR) units using standardized conversion methods. Visual acuity change (ΔlogMAR) was calculated as the difference between 12-month and baseline values, with negative values indicating visual improvement. Visual outcomes were additionally categorized as improvement (decrease in BCVA of ≥ 0.1 logMAR), stability (change in BCVA between − 0.1 and + 0.1 logMAR), or deterioration (increase in BCVA of ≥ 0.1 logMAR).

Outcome measures

The primary outcomes were treatment burden parameters, including the number of intravitreal injections and outpatient visits during the first year of follow-up. Secondary outcomes included change in best-corrected visual acuity (BCVA), expressed as ΔlogMAR; distribution of visual outcomes; treatment switching frequency and timing; and the association between injection number and visual acuity change.

Statistical analysis

Statistical analyses were performed using SPSS software (version 25.0; IBM Corp., Armonk, NY, USA). Continuous variables were presented as mean ± standard deviation, and categorical variables as frequencies and percentages. Normality of distribution was assessed using the Shapiro–Wilk test. Within-group comparisons of baseline and 12-month BCVA were performed using paired t-tests. Categorical variables were compared using the chi-square test. The association between injection number and visual acuity change was evaluated using Spearman’s rank correlation coefficient.

Multivariable linear regression

To identify independent predictors of visual outcomes, a multivariable linear regression analysis was performed. The dependent variable was change in BCVA (ΔlogMAR). Independent variables included baseline BCVA, total number of injections during the first year, and treatment switching status. Variables were selected based on clinical relevance and prior literature. Regression coefficients (β) and corresponding P values were reported.

Generalized estimating equation analysis

To account for potential inter-eye correlation arising from inclusion of both eyes from some patients, a generalized estimating equation (GEE) model was performed using an exchangeable working correlation structure with robust standard errors. Patient identifier was specified as the clustering variable. The dependent variable was change in BCVA (ΔlogMAR), and independent variables included baseline BCVA, number of injections, treatment switching status, age, and sex.

Sensitivity analysis

As a sensitivity analysis, the regression model was repeated after restricting the dataset to one randomly selected eye per patient to assess the robustness of the findings. For subgroup analysis, eyes were stratified according to the median baseline BCVA value (1.0 logMAR). Separate multivariable linear regression analyses were performed within each subgroup to evaluate the association between injection number and change in BCVA.

All statistical tests were two-sided, and P < 0.05 was considered statistically significant.

Results

A total of 563 eyes with diabetic macular edema were included in the analysis. The mean age of the patients was 61.2 ± 9.3 years, and 60.1% were female. All eyes completed at least 12 months of follow-up. Baseline demographic characteristics and treatment burden are summarized in Table 1. During the first year of follow-up, eyes received a mean of 5.4 ± 2.4 intravitreal injections and attended 8.0 ± 2.2 outpatient visits (Table 1).

Table 1.

Baseline demographic characteristics and first-year treatment burden in eyes with diabetic macular edema treated with intravitreal anti-VEGF therapy

Variable	Value
Number of eyes (n)	563
Age (years), mean ± SD	61.2 ± 9.3
Female, n (%)	339 (60.1%)
Follow-up duration, months (minimum)	12
Number of injections, mean ± SD	5.4 ± 2.4
Outpatient visits, mean ± SD	8.0 ± 2.2

BCVA: best-corrected visual acuity; SD: standard deviation

Mean best-corrected visual acuity (BCVA) improved from 0.75 ± 0.43 logMAR at baseline to 0.59 ± 0.42 logMAR at the 12-month assessment, representing a mean change of − 0.16 ± 0.35 logMAR (P < 0.001). Visual improvement was observed in 339 eyes (60.2%), stability in 122 eyes (21.7%), and deterioration in 102 eyes (18.1%) (Table 2).

Table 2.

Changes in best-corrected visual acuity (BCVA) and distribution of visual outcomes after 12 months of anti-VEGF treatment in diabetic macular edema

Parameter	Value
Baseline BCVA (logMAR), mean ± SD	0.75 ± 0.43
12-month BCVA (logMAR), mean ± SD	0.59 ± 0.42
Change in BCVA (ΔlogMAR), mean ± SD	−0.16 ± 0.35
P-value	< 0.001
Improvement, n (%)	339 (60.2%)
Stability, n (%)	122 (21.7%)
Deterioration, n (%)	102 (18.1%)

BCVA: best-corrected visual acuity; SD: standard deviation; ΔlogMAR: change in logarithm of the minimum angle of resolution

Visual outcomes were categorized as improvement (ΔlogMAR ≤ − 0.1), stability (− 0.1 < ΔlogMAR < + 0.1), or deterioration (ΔlogMAR ≥ + 0.1)

P-value represents the comparison between baseline and 12-month BCVA values

An association was observed between the total number of intravitreal injections and change in BCVA in univariate analysis (Fig. 1).

Fig. 1.

Association between injection number and change in visual acuity. Scatter plot illustrating the relationship between the total number of intravitreal injections and change in best-corrected visual acuity (ΔlogMAR), with a fitted linear regression line and corresponding 95% confidence interval

Treatment switching from bevacizumab to aflibercept occurred in 167 of 563 eyes (29.7%). The mean time to switching was 4.5 ± 0.7 months after treatment initiation. The distribution of switching timing is shown in Fig. 2, and additional characteristics, including subgroup analyses, are summarized in Table 3.

Fig. 2.

Distribution of time to treatment switching from bevacizumab to aflibercept. Histogram illustrating the distribution of time to switching from bevacizumab to aflibercept in eyes with diabetic macular edema

Table 3.

Treatment switching characteristics and subgroup analysis of visual outcomes in diabetic macular edema

Parameter	Value
Eyes with switching, n (%)	167 (29.7%)
Time to switching (months), mean ± SD	4.5 ± 0.7
Subgroup analysis according to baseline BCVA
Parameter	β	P-value
Better baseline BCVA (≤ 1.0 logMAR)	−0.011	0.046
Worse baseline BCVA (> 1.0 logMAR)	−0.030	0.002

Subgroup analysis according to baseline BCVA

BCVA: best-corrected visual acuity; ΔlogMAR: change in logarithm of the minimum angle of resolution; SD: standard deviation

Time to switching was calculated from treatment initiation to the first injection of aflibercept. Subgroup stratification was based on the median baseline BCVA value (1.0 logMAR). β represents the regression coefficient for the association between injection number and change in BCVA within each subgroup

Subgroup analyses stratified by baseline BCVA showed that in eyes with better baseline visual acuity, injection number was associated with change in BCVA (β = −0.011, P = 0.046), whereas a larger effect size was observed in eyes with worse baseline visual acuity (β = −0.030, P = 0.002).

In multivariable linear regression analysis, baseline BCVA was significantly associated with change in visual acuity (β = −0.33, P < 0.001). Treatment switching was associated with visual acuity change (β = 0.17, P = 0.004), whereas the number of injections was not associated with visual acuity change (β = −0.009, P = 0.27).

In the generalized estimating equation (GEE) model accounting for inter-eye correlation, baseline BCVA remained significantly associated with visual acuity change (β = −0.395, P < 0.001). Injection number (β = −0.0035, P = 0.527) and treatment switching (β = −0.0010, P = 0.976) were not associated with visual acuity change after adjustment for age and sex. The results of both analyses are presented in Table 4.

Table 4.

Multivariable linear regression and generalized estimating equation (GEE) analyses of factors associated with change in best-corrected visual acuity (ΔlogMAR) in eyes with diabetic macular edema

Variable	Linear Regression β (95% CI)	P-value	GEE β (95% CI)	P-value
Baseline BCVA	−0.33 (− 0.39 to − 0.27)	< 0.001	−0.395 (− 0.47 to − 0.32)	< 0.001
Number of injections	−0.009 (− 0.025 to 0.007)	0.27	−0.0035 (− 0.014 to 0.007)	0.527
Treatment switching	0.17 (0.05 to 0.29)	0.004	−0.0010 (− 0.09 to 0.09)	0.976

BCVA: best-corrected visual acuity; ΔlogMAR: change in logarithm of the minimum angle of resolution; GEE: generalized estimating equation; CI: confidence interval

Linear regression model included baseline BCVA, number of injections, and treatment switching

The GEE model was adjusted for baseline BCVA, number of injections, treatment switching, age, and sex, with patient identifier specified as the clustering variable

Sensitivity analysis restricted to one eye per patient showed similar results (Table 4).

Discussion

In this large real-world cohort of patients with diabetic macular edema (DME), intravitreal anti-VEGF therapy was associated with improvement in visual acuity despite a moderate injection frequency. These findings appear to be consistent with the established role of VEGF in the pathophysiology of diabetic retinopathy and macular edema and support the effectiveness of anti-VEGF therapy in routine clinical practice [3, 4].

The magnitude of visual improvement observed in this study can be contextualized in relation to pivotal randomized controlled trials. The RISE and RIDE studies demonstrated substantial visual gains with ranibizumab [5], while the VIVID and VISTA trials confirmed similar efficacy with aflibercept [6]. In addition, the DRCR.net Protocol T study highlighted the importance of baseline visual acuity in determining treatment response [7]. In contrast to these trials, which employed fixed and intensive dosing regimens, patients in the present study received a mean of 5.4 injections during the first year. This lower injection frequency reflects real-world practice patterns and is consistent with previous observational studies reporting reduced treatment intensity compared with clinical trials [8, 12, 13]. In landmark randomized controlled trials such as RISE/RIDE and Protocol T, greater visual gains were achieved with more intensive and protocol-driven treatment regimens. In comparison, the magnitude of visual improvement observed in our cohort was more modest, which may reflect lower injection frequency, variability in follow-up adherence, and differences in retreatment strategies in real-world practice. This comparison underscores the gap between efficacy observed in controlled clinical trials and effectiveness in routine clinical care. In addition, practical challenges such as patient non-adherence, missed follow-up visits, and healthcare system or reimbursement constraints may further limit the consistent implementation of PRN treatment regimens in real-world settings. Because only treatment-naïve eyes were included, heterogeneity related to prior therapeutic exposure was minimized, allowing a more consistent evaluation of treatment burden and visual outcomes in this real-world cohort.

A key finding of this study is that the association between injection number and visual outcomes observed in univariate analysis did not persist after adjustment in multivariable models and after accounting for inter-eye correlation using GEE. These findings suggest that treatment intensity alone may not independently determine functional outcomes in DME. Instead, baseline visual acuity emerged as the strongest independent predictor of visual change, consistent with findings from randomized trials and real-world studies [7, 10, 11]. This relationship likely reflects a ceiling effect, whereby eyes with better baseline vision have less potential for measurable improvement. The subgroup analysis further demonstrated that the association between injection number and visual outcomes was more pronounced in eyes with worse baseline visual acuity. This finding suggests that treatment intensity may have a greater relative impact in eyes with worse baseline visual acuity, while its effect may be limited in eyes with better baseline vision. Importantly, the lack of an independent association between injection number and visual outcomes after adjustment should not be interpreted as evidence that treatment intensity is unimportant. Rather, this finding likely reflects heterogeneity in baseline visual status, variability in retreatment criteria, and differences in patient adherence, all of which may confound the relationship between treatment intensity and functional outcomes. In addition, the relatively lower injection frequency observed in this cohort compared with randomized controlled trials may further contribute to this finding, highlighting the gap between protocol-driven efficacy and real-world effectiveness.

Another important observation relates to treatment switching. Although treatment switching was associated with visual outcomes in unadjusted analysis, this association was no longer significant after adjustment, suggesting that switching is unlikely to be an independent determinant of visual outcomes. Instead, treatment switching should be interpreted with caution, as it is inherently influenced by clinical decision-making and underlying disease severity. In real-world practice, eyes requiring switching often represent a subgroup with persistent edema or suboptimal response to initial therapy, making switching highly susceptible to confounding by indication. Previous real-world studies have similarly reported that patients requiring treatment modification tend to have more refractory disease and less favorable outcomes [10, 11, 14]. Therefore, treatment switching may be better interpreted not as a modifiable determinant of visual outcomes, but rather as a surrogate marker of more severe or treatment-resistant disease. Given the retrospective observational design of this study, causal inferences regarding the association between switching and visual outcomes cannot be established. From a clinical perspective, the need for switching may help identify patients at higher risk of poor visual outcomes who may benefit from closer monitoring and individualized treatment strategies. However, the optimal timing for treatment switching and the most appropriate criteria for patient stratification remain uncertain in real-world practice and warrant further prospective investigation.

The difference between the conventional multivariable regression and the GEE analysis is likely attributable to their differing underlying statistical assumptions. Conventional regression assumes that each eye represents an independent observation, whereas the GEE model accounts for the within-patient correlation arising from the inclusion of both eyes from some patients. Failure to account for this correlation may underestimate standard errors, thereby increasing the likelihood of statistically significant findings. Because our study included bilateral observations, the GEE model was considered to provide the more appropriate population-averaged estimates and was therefore regarded as the primary inferential analysis.

The treatment burden associated with anti-VEGF therapy remains a major challenge in the management of DME. Frequent intravitreal injections and clinic visits can negatively affect patient adherence and long-term treatment persistence. Previous studies have demonstrated that non-adherence and irregular follow-up are associated with poorer visual outcomes [9]. In this context, the moderate injection frequency observed in our study likely reflects a balance between treatment efficacy and real-world constraints, including healthcare system capacity and patient compliance.

Structural retinal factors may also contribute to variability in treatment response. Biomarkers such as disorganization of retinal inner layers (DRIL) and disruption of the ellipsoid zone have been shown to be associated with limited visual recovery despite ongoing anti-VEGF therapy [15, 16]. Although these parameters were not evaluated in the present study, they may partly explain the lack of an independent association between injection number and visual outcomes.

Emerging therapeutic strategies aim to reduce treatment burden while maintaining efficacy. Approaches such as treat-and-extend regimens and longer-acting anti-VEGF agents have shown promise in reducing injection frequency while preserving visual outcomes [17, 18]. However, their comparative effectiveness in real-world settings may vary depending on patient characteristics, adherence, and healthcare system factors, and further real-world evidence is needed to better define their optimal use. Real-world data, such as those presented in this study, are essential for evaluating how these strategies may be implemented in clinical practice.

The strengths of this study include the large sample size and the comprehensive evaluation of treatment burden, switching patterns, and visual outcomes within a single real-world clinical setting. In addition, the use of advanced statistical methods, including GEE analysis and sensitivity analysis restricted to one eye per patient, allowed for more robust estimation of treatment effects by accounting for inter-eye correlation and potential sources of bias.

Several limitations should be acknowledged. First, the retrospective design limits causal inference and may introduce selection bias. Second, treatment decisions were not standardized and were made at the discretion of treating physicians, which may have influenced treatment patterns and outcomes. In particular, treatment switching is subject to confounding by indication, as it is preferentially applied to eyes with inadequate response. This limits the ability to disentangle its independent effect on visual outcomes. Although inter-eye correlation was addressed using generalized estimating equation analysis, residual correlation and unmeasured confounding cannot be completely excluded. This may include unmeasured differences in patient demographics, systemic comorbidities, and disease-related characteristics that could influence both treatment patterns and visual outcomes. Residual confounding inherent to the observational nature of the study cannot be fully ruled out, particularly with respect to treatment switching, which may reflect underlying disease severity and physician decision-making. The timing of treatment switching was not modeled as a time-dependent variable, which may have influenced the observed associations. Furthermore, important systemic factors such as duration of diabetes, glycemic control, and other vascular comorbidities were not included in the analysis. The severity stage of diabetic retinopathy (e.g., non-proliferative versus proliferative diabetic retinopathy) was also not consistently available in the retrospective records and therefore could not be incorporated into the analysis. Because retinopathy severity may influence ischemic burden, treatment requirements, and visual prognosis, its absence limits assessment of disease heterogeneity within the cohort. Structural and quantitative OCT biomarkers, including disorganization of retinal inner layers (DRIL), ellipsoid zone disruption, central retinal thickness, and fluid compartment characteristics, were also not evaluated. Because retreatment and treatment-switching decisions in routine clinical practice were substantially guided by OCT findings, the absence of these parameters may have influenced both physician decision-making and observed visual outcomes, thereby introducing additional unmeasured confounding. This also limits mechanistic interpretation of the relationship between anatomical and functional outcomes. Moreover, functional outcomes beyond best-corrected visual acuity, such as contrast sensitivity, reading performance, and vision-related quality of life, were not routinely available in the retrospective records and therefore could not be evaluated. As a result, reliance on BCVA alone may incompletely reflect the broader functional impact of diabetic macular edema or the real-world patient-centered benefits of treatment. Finally, although efforts were made to include complete datasets, potential missing or unrecorded data for certain variables cannot be entirely excluded, which may represent an additional source of bias.

In conclusion, in real-world clinical practice, anti-VEGF therapy for diabetic macular edema was associated with meaningful visual improvement despite a moderate injection frequency. Visual outcomes appear to be more strongly associated with baseline visual acuity than with treatment intensity. Treatment switching was not independently associated with visual acuity change after adjustment, suggesting that it likely reflects underlying disease refractoriness rather than acting as an independent factor associated with visual outcomes; however, causal inference cannot be established due to the observational study design. These findings underscore the importance of individualized treatment strategies in DME.

Author contributions

H.Ö. and H.D. performed the clinical procedures. H.Ö. and U.D. conceived the study. U.D. and M.F.A. designed the study. H.D. and U.D. collected the data. H.Ö. and M.F.A. performed the statistical analysis and interpreted the data. M.F.A. and U.D. conducted the literature review. H.Ö. and H.D. drafted the manuscript. All authors read and approved the final manuscript.

Funding

The authors received no specific funding for this work.

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

Ethical approval was obtained from the Institutional Review Board (Ethics Committee) of Health Sciences University Diyarbakır Gazi Yaşargil Training and Research Hospital (Protocol code: 800; Approval date: 19 December 2025). The study adhered to the tenets of the Declaration of Helsinki. Due to the retrospective nature of the study, the requirement for informed consent was waived by the Ethics Committee.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.