Evaluating Systemic Immune-Inflammation Index and Mortality in Diabetic Retinopathy: A NHANES 2005–2018 Cohort Study

Abstract

Introduction

Diabetic retinopathy (DR) is a leading cause of preventable blindness and is linked to excess mortality. Systemic inflammation plays a central role in DR pathogenesis, and the systemic immune-inflammation index (SII) has emerged as a novel marker associated with adverse outcomes. However, its prognostic value in DR remains unclear.

Methods

Using nationally representative data from National Health and Nutrition Examination Survey 2005–2018, we investigated the association between the systemic immune-inflammation index (SII) and all-cause mortality among 209 US adults aged ≥ 50 years with DR. Mortality status was ascertained through linkage to the National Death Index through 2019.

Results

Weighted Cox regression models showed that higher SII was modestly but significantly associated with increased mortality risk (HR = 1.01, 95% CI 1.00–1.002, p < 0.001), with restricted cubic splines indicating a non-linear relationship. SII exhibited moderate predictive ability for mortality, with AUCs declining from 0.762 at 24 months to 0.576 at 60 months. Subgroup analyses suggested that the association between SII and mortality persisted across most demographic and clinical strata.

Conclusion

These findings suggest that elevated systemic immune inflammation may independently predict long-term mortality risk in adults with DR, highlighting its potential value as a prognostic biomarker beyond traditional risk factors.

Key Points

Why carry out this study?

Diabetic retinopathy (DR) is a major cause of preventable blindness and is associated with substantially increased long-term mortality

Systemic inflammation contributes to DR pathogenesis, yet reliable inflammatory biomarkers for predicting mortality risk in DR remain an unmet clinical need

We evaluated whether the systemic immune-inflammation index (SII), a novel composite inflammatory marker, predicts all-cause mortality in US adults with DR

What was learned from the study?

Higher SII levels were significantly associated with increased mortality risk (HR = 1.01, 95% CI 1.00–1.002), and this relationship was non-linear

SII demonstrated moderate short-term predictive ability for mortality but declined over longer follow-up, suggesting it may be most useful for early risk stratification

These findings highlight systemic inflammation as a meaningful prognostic factor in DR and support further exploration of inflammatory biomarkers to guide future clinical assessment and intervention strategies

Introduction

Diabetic retinopathy (DR) represents the most prevalent microvascular complication of diabetes and increasingly is recognized as a condition with systemic implications rather than an isolated ocular disorder [1, 2]. The association between DR and mortality has been observed in individuals with non-proliferative diabetic retinopathy (NPDR) having a 1.41-fold increase in all-cause mortality compared with age-matched controls [3]. In addition, multimodal retinal imaging has shown that patterns of capillary nonperfusion correlate closely with early, often subclinical, cardiovascular remodeling, reinforcing the concept that DR mirrors systemic vascular dysfunction [4].

The pathophysiology of DR involves chronic hyperglycemia-induced microvascular damage, oxidative stress, and low-grade inflammation [5]. Traditional risk factors such as glycemic control, hypertension, and dyslipidemia only partly explain the heterogeneity in DR outcomes [6]. Recent evidence has highlighted the role of systemic inflammation in mediating disease progression and mortality [7].

Several hematologic inflammatory indices have emerged as accessible markers of systemic immune activation. The systemic immune-inflammation index (SII), calculated as platelet count × neutrophil count/lymphocyte count, integrates thrombosis, inflammation, and immune status. SII has been associated with adverse outcomes in ocular disorders [8–13]. A recent hospital-based cohort also suggested a potential role of SII in predicting short-term mortality among patients with DR [14]. However, that study was limited by its single-center design and relatively short follow-up, leaving uncertainty regarding the long-term prognostic value of SII in a representative DR population.

Using nationally representative data from the National Health and Nutrition Examination Survey (NHANES) 2005–2018, this study aimed to evaluate the association of SII with all-cause mortality among US adults with DR. We hypothesized that higher levels of SII would be associated with increased long-term mortality risk, independent of traditional risk factors.

Methods

Study Population

We used data from seven NHANES cycles (2005–2018) due to full variables. NHANES applies a complex, multistage, probability sampling design to obtain a representative sample of the US non-institutionalized population. The data generated by this effort are available at the following link: https://wwwn.cdc.gov/nchs/nhanes/Default.aspx. We included adults (≥ 18 years) with diabetes (defined by self-report, use of glucose-lowering medication, fasting plasma glucose ≥ 126 mg/dl, or HbA1c ≥ 6.5%) who had DR. Participants with missing blood count data or mortality linkage were excluded.

This study involves human participants. The survey was approved by the National Center for Health Statistics (NCHS) ethics review board. The NCHS IRB/ERC protocol number in this survey covers 2005–2008 (protocol #2005‐06 and continuation of protocol #2005‐06). The participants provided informed consent to participate in the study before taking part. This study was performed in accordance with the Helsinki Declaration of 1964 and its later amendments.

Assessment of Diabetic Retinopathy

The diagnosis of DR was determined based on professional confirmation; specifically, participants who answered “yes” to the question “Has a doctor ever told you that diabetes has affected your eyes or that you had retinopathy?” were classified as having DR.

Inflammatory Indices

Complete blood counts were measured by automated hematology analyzers in NHANES laboratories. The systemic immune-inflammation index was computed as platelets*neutrophils/lymphocytes.

Mortality Outcomes

Vital status and cause-of-death information were identified through matched mortality records in the National Death Index (NDI) up to December 31, 2019. NHANES participants were linked to the NDI using standardized probabilistic matching procedures to obtain confirmed death outcomes [15].

Covariates

Data were collected through questionnaires, physical examinations, and laboratory tests. Demographic variables included age, sex, race, smoking history, alcohol use, family poverty-to-income ratio (PIR), and education level. Laboratory measurements comprised glycated hemoglobin (HbA1c). Clinical information included duration of diabetes and histories of hypertension, hyperlipidemia, and chronic kidney disease (CKD).

Statistical Analysis

Survey weights, strata, and primary sampling units were incorporated to accommodate the complex NHANES sampling framework. Baseline characteristics were described across quartiles of each inflammatory marker. Weighted Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for mortality, with inflammatory indices examined as both continuous and categorical variables divided into tertiles. Two models were specified: Model 1 included no covariate adjustment, whereas Model 2 adjusted for age, sex, race/ethnicity, PIR, smoking status, education level, alcohol intake, diabetes duration, HbA1c, hypertension, hyperlipidemia, and chronic kidney disease. Dose-response patterns were evaluated using restricted cubic spline functions. Stratified analyses were performed by age, sex, HbA1c level, and hypertension status. Additionally, mediation analyses explored the potential mediating role of HbA1c in the relationship between inflammatory indices and mortality. All analyses were conducted in R version 4.2.2 using the survey package, with statistical significance defined as a two-sided p value < 0.05.

Results

Data from NHANES 2005–2018 were linked with the 2019 mortality file. Of 70,190 participants, 57,515 with available blood biomarker data were screened. After excluding those without DR and limiting the group to individuals aged ≥ 50 years, 209 participants with DR were included in the final analysis (133 survivors and 76 deaths) (Fig. 1). Table 1 presents the baseline characteristics of participants according to survival status. Compared with survivors, participants who died were older (71.28 ± 7.52 vs. 64.22 ± 8.59 years, p < 0.001), had a lower family income-to-poverty ratio (1.91 ± 1.36 vs. 2.35 ± 1.69, p = 0.041), and had a longer duration of diabetes (21.72 ± 13.32 vs. 16.59 ± 12.42 years, p = 0.007). The mortality group also showed a higher systemic immune-inflammation index (SII) (669.35 ± 549.12 vs. 533.4 ± 280.82, p = 0.047) and a greater prevalence of chronic kidney disease (78.95% vs. 52.63%, p < 0.001). Other demographic and lifestyle variables, including sex, race, education level, marital status, smoking, alcohol use, and hyperlipidemia, did not differ significantly between groups.

Fig. 1.

Flowchart of participant selection from National Health and Nutrition Examination Survey (NHANES) 2005–2018 linked with 2019 mortality data

Table 1.

Baseline characteristics of participants by survival status

Variable	Level	Survival	Death	p value
Age		64.22 (8.59)	71.28 (7.52)	< 0.001
Gender	Female	73 (54.89%)	32 (42.11%)	0.102
	Male	60 (45.11%)	44 (57.89%)
Race	Mexican American	24 (18.05%)	11 (14.47%)	0.491
	Non-Hispanic Black	40 (30.08%)	19 (25%)
	Non-Hispanic White	32 (24.06%)	27 (35.53%)
	Other Hispanic	21 (15.79%)	12 (15.79%)
	Other race	16 (12.03%)	7 (9.21%)
Education	9th–11th Grade (includes 12th grade with no diploma)	21 (15.79%)	20 (26.32%)	0.258
	College graduate or above	14 (10.53%)	8 (10.53%)
	High school grad/GED or equivalent	28 (21.05%)	16 (21.05%)
	< 9th grade	30 (22.56%)	18 (23.68%)
	Some college or AA degree	40 (30.08%)	14 (18.42%)
Marital status	Divorced	21 (15.79%)	15 (19.74%)	0.666
	Living with partner	3 (2.26%)	1 (1.32%)
	Married	67 (50.38%)	30 (39.47%)
	Never married	8 (6.02%)	5 (6.58%)
	Separated	6 (4.51%)	3 (3.95%)
	Widowed	28 (21.05%)	22 (28.95%)
Smoking	Every day	29 (21.8%)	12 (15.79%)	0.065
	Yes	13 (9.77%)	2 (2.63%)
	No	91 (68.42%)	62 (81.58%)
Drinking	Yes	75 (56.39%)	53 (69.74%)	0.079
	No	58 (43.61%)	23 (30.26%)
Chronic kidney disease	Yes	70 (52.63%)	60 (78.95%)	< 0.001
	No	63 (47.37%)	16 (21.05%)
Hyperlipidemia	Yes	63 (47.37%)	43 (56.58%)	0.255
	No	70 (52.63%)	33 (43.42%)
Family PIR		2.35 (1.69)	1.91 (1.36)	0.041
HbA1c (%)		7.52 (1.44)	7.83 (1.96)	0.234
Diabetes duration		16.59 (12.42)	21.72 (13.32)	0.007
SII		533.4 (280.82)	669.35 (549.12)	0.047

PIR poverty income ratio, HbA1c glycated hemoglobin (hemoglobin A1c), SII systemic immune-inflammation index

Table 2 demonstrates the association between SII and the risk of all-cause mortality among patients with DR. Specifically, SII demonstrated a modest but significant association (HR = 1.01, 95% CI 1.00–1.002, p < 0.001). By contrast, SII quartiles were not significantly associated with mortality in either unadjusted or adjusted analyses, although the HR for mortality showed a nonsignificant increasing trend with increasing SII concentrations (p for trend > 0.05) (Table 3).

Table 2.

Association between systemic immune-inflammation index and mortality in diabetic retinopathy

Variable	HR	95% CI	p-value
Model 1
SII	1.01	1.00–1.01	< 0.001
Model 2
SII	1.02	1.01–1.02	< 0.001

Model 1. Adjusted without confounders

Model 2. Adjusted with age, gender, race, education, marital status, smoking, drinking, chronic kidney disease, hyperlipidemia, family poverty income ratio, hba1c (%), diabetes duration

Table 3.

Association between systemic immune-inflammation index and mortality in diabetic retinopathy

Variable	Group	Model 1 HR (95% CI)	Model 2 HR (95% CI)
SII	Q1	Reference	Reference
	Q2	0.74 (0.37–1.49)	0.86 (0.41–1.80)
	Q3	1.17 (0.62–2.21)	0.80 (0.39–1.64)
	Q4	1.52 (0.83–2.81)	1.11 (0.53–2.30)
	P for trend	0.06	0.708

Model 1. Adjusted without confounders

Model 2. Adjusted with age, gender, race, education, marital status, smoking, drinking, chronic kidney disease, hyperlipidemia, family poverty income ratio, glycated hemoglobin (%), and diabetes duration

Restricted cubic splines demonstrated a linear association between SII and all-cause mortality (Fig. 2). SII (p_overall < 0.001, p_nonlinear = 0.024) exhibited significant non-linear associations.

Fig. 2.

The dose-response relationship between systemic immune-inflammation index (SII) and mortality in diabetic retinopathy (DR)

SII demonstrated the moderate predictive ability for mortality among participants with DR, with area under curve (AUCs) declining from 0.762 at 24 months to 0.576 at 60 months. Overall, SII showed a gradual reduction in predictive accuracy over time (Fig. 3).

Fig. 3.

ROC curves for mortality in DR using SII: indicator after adjusting for confounding factors

Subgroup analyses showed that higher SII values were generally associated with increased mortality across most demographic and clinical strata in patients with DR. SII showed smaller effect sizes but remained significantly associated with mortality in several subgroups, including younger age, males, Mexican Americans, low poverty, less than high school education, widowed status, HbA1c < 7.0%, longer diabetes duration, with smoking, with alcohol abstainers, with CKD, but without hyperlipidemia (Fig. 4).

Fig. 4.

Subgroup analyses of the association between SII and mortality in DR. HbA1c glycated hemoglobin, CKD chronic kidney disease

Mediation analysis indicated that the total effect of SII on mortality was statistically significant (β = 0.001, 95% CI 1.000–1.002, p = 0.0028). However, the indirect (mediated) effect through HbA1c was not significant (β = 0.000, 95% CI −  0.000 to 0.001, p = 0.60), while the direct effect remained significant (β = 0.227, 95% CI 1.077–1.461, p = 0.0036). The proportion of the total effect mediated by HbA1c was estimated at 3.3% (data not shown).

Discussion

In this large, nationally representative cohort of adults with DR, elevated SII was significantly associated with increased risk of all-cause mortality. Unlike prior single-center studies that primarily assessed short-term outcomes, our analysis with a median follow-up of 8.6 years demonstrated that SII also carries long-term prognostic information. Although quartile-based analyses did not show significant associations, continuous and spline-based analyses indicated a robust non-linear relationship, suggesting possible threshold effects. Time-dependent ROC analyses further revealed that SII had modest but persistent discriminative ability over time.

DR is an irreversible microvascular complication of diabetes that can progress to vision loss, with chronic low-grade inflammation playing a central role in its pathogenesis. Experimental studies have shown that blocking inflammatory pathways slows DR progression in diabetic mice [16], while proteomic analyses in humans demonstrated elevated inflammatory proteins in patients with DR [17]. This inflammatory milieu is characterized by increased vascular permeability and the recruitment of neutrophils, monocyte-macrophages, and platelets, which together sustain vascular injury. Neutrophils, through persistent activation and formation of neutrophil extracellular traps, contribute to vascular occlusion and tissue damage [18, 19], while monocytes and macrophages release pro-inflammatory cytokines such as IL-1β, TNF-α, and IL-6 [20], and platelets amplify thrombosis and leukocyte activation [21]. Meta-analyses further confirm higher circulating counts of neutrophils, monocytes, lymphocytes, and other hematologic parameters in diabetes compared to healthy controls [22, 23].

SII, one of the most frequently used inflammatory markers in recent years, is considered more reliable and sensitive to systemic inflammation, as it is less affected by physiopathologic changes and can be easily calculated from routine hemogram results [24]. Given that SII integrates neutrophil, platelet, and lymphocyte counts, elevated SII values likely reflect this chronic inflammatory and pro-thrombotic state. These mechanisms may explain why higher SII was associated with increased mortality in patients with DR, as persistent systemic inflammation not only exacerbates microvascular injury but also contributes to cardiovascular and renal complications that drive excess mortality. Previous studies have reported SII was suggested as a marker for early DR diagnosis in type 2 DM [25], and higher SII values have also been associated with diabetic complications [26]. Furthermore, chronic systemic immune activation may represent a shared pathologic substrate linking retinal microvascular injury and excess mortality in DR. Persistent inflammation promotes endothelial dysfunction, leukocyte adhesion, and microvascular occlusion in the retina, while simultaneously accelerating atherosclerosis, renal dysfunction, and other systemic complications. This shared inflammatory pathway provides a biologic explanation for the observed association between elevated SII and long-term mortality in patients with DR.

Compared with earlier work that highlighted the short-term predictive role of SII in DR prognosis [14], our findings extend this knowledge by showing that SII remains associated with mortality over longer follow-up in a representative US population. Importantly, subgroup analyses indicated that the association was stronger in vulnerable groups, including younger adults, males, Mexican Americans, and those with low socioeconomic status or comorbidities. This novel observation underscores the potential heterogeneity of inflammation-related mortality risk across different demographic and clinical strata.

Clinically, SII may serve as a simple adjunctive tool for risk stratification in patients with DR. Individuals with persistently elevated SII could be considered for closer ophthalmic surveillance, more frequent systemic evaluation, or intensified management of cardiometabolic risk factors. Although our spline analyses suggest potential non-linear and threshold effects, clinically meaningful cutoff values for high-risk SII remain to be established and warrant validation in prospective studies.

Strengths of our study include its nationally representative design, large sample size, long follow-up, and comprehensive analytic approach using Cox regression, restricted cubic splines, and time-dependent ROC analysis. There are several potential limitations worth considering. First, limitations include observational design, potential residual confounding, and reliance on a single baseline SII measurement. Second, data from NHANES 2005–2008 are over a decade old and may not fully reflect current population characteristics. Third, DR status in NHANES was not uniformly defined using detailed retinal imaging or standardized clinical grading across all participants, which may have led to some degree of outcome misclassification. Future studies incorporating standardized fundus imaging and severity grading will be important to further refine disease classification and validate our findings. Finally, the generalizability of our findings may be limited, as the study sample consisted exclusively of US adults. Future research should validate our findings in diverse cohorts and examine whether interventions targeting systemic inflammation could improve survival outcomes in DR.

Conclusion

Elevated SII was independently associated with increased long-term all-cause mortality in adults with DR. By extending previous single-center findings to a nationally representative population and incorporating advanced statistical approaches, this study highlights the prognostic utility of SII and underscores the role of systemic inflammation in DR-related mortality.

Author Contributions

Conceptualization: Shuyang Guo and Bozhou Zhang. Methodology: Shuyang Guo and Yunlei Pang. Data acquisition: Shuyang Guo, Yunlei Pang, and Tong Bao. Formal analysis: Shuyang Guo and Songfu Feng. Investigation: Shuyang Guo, Juan Wang, and Baoyu Huang. Visualization: Shuyang Guo. Writing-original draft: Shuyang Guo. Writing-review & editing: Shuyang Guo, Yunlei Pang, Tong Bao, Juan Wang, Bozhou Zhang, Songfu Feng, and Baoyu Huang. Supervision: Bozhou Zhang, Songfu Feng, and Baoyu Huang. All authors reviewed and approved the final manuscript for submission.

Funding

This study was funded by the Fujian Provincial Special Fund (grant no. BPB-2023GJ) from Professor Jian Guo from the First Affiliated Hospital of Fujian Medical University. The Rapid Service Fee was funded by the authors.

Data Availability

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

Declarations

Conflict of Interest

Shuyang Guo, Yunlei Pang, Tong Bao, Juan Wang, Bozhou Zhang, Songfu Feng, and Baoyu Huang confirm that they have no competing interests to declare.

Ethical Approval

This study involves human participants. The survey was approved by the National Center for Health Statistics (NCHS) ethics review board. The NCHS IRB/ERC protocol number in this survey covers 2005–2008 (protocol #2005‐06 and continuation of protocol #2005‐06). The participants provided informed consent to participate in the study before taking part. This study was performed in accordance with the Helsinki Declaration of 1964 and its later amendments.