Study population

Our cohort study included participants sourced from the National Health and Nutrition Examination Survey (NHANES), a regular cross-sectional sampling conducted by the National Center for Health Statistics, under the auspices of the Centers for Disease Control and Prevention. NHANES is renowned for its ability to provide a nationally representative sample of the noninstitutionalized US civilian population. Detailed information regarding NHANES’ sampling methodology and data collection techniques is available on its official website.

Data collection

Our trained medical personnel collected five categories of data from the participants, covering demographics, dietary habits, physical examination findings, and laboratory test results. This robust dataset allows for the comprehensive analysis of various health-related factors.

Institutional approval

The NHANES study was conducted with the approval of the Institutional Review Board of the National Center for Health Statistics. All participants provided informed written consent at the time of enrollment, ensuring that ethical standards were upheld throughout the study.

Inclusion criteria

In our study, we included patients with diabetes who were at least 18 years old, adhering to established criteria for Type 2 diabetes (T2D) diagnosis, as outlined by the American Diabetes Association. These criteria encompassed various indicators, including physician diagnosis, HbA1c levels, fasting glucose levels, random blood glucose levels, oral glucose tolerance test (OGTT) results, and diabetes medication or insulin usage.

Mortality data

We collected mortality data up to December 31, 2019, by linking our cohort database to the National Death Index. This linkage allowed us to obtain comprehensive statistics on the outcomes of interest.

Study cohort

In summary, out of the initial 10,1316 participant pool, which included 11,082 individuals with diabetes, 1670 self-reported pregnancy, leaving 9412 participants meeting the T2D and non-pregnant criterion. Among these, 8697 participants had completed the full blood test, enabling us to extract the necessary SII data. These 8697 participants formed the final cohort for our SII analysis.

Study reporting

Our research adhered to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines, ensuring transparent and comprehensive reporting of our cohort study.

Laboratory testLaboratory processes

Detailed information regarding the laboratory procedures, including the complete blood count, is available on the NHANES website. In brief, the blood specimens collected from participants underwent a standardized process. After collection, the samples were promptly processed and then frozen at −20 °C. Subsequently, these frozen samples were analyzed by the experts at the National Center for Environmental Health.

Blood sample collection

The collection of blood samples for laboratory analysis occurred at a specific juncture during the enrollment process of study participants in NHANES. Notably, these samples were drawn after a designated fasting period. This fasting state necessitated that participants abstain from both food and drink for a specified duration before the blood draw.

Timing and standardization

The precise timing of blood sample collection within the NHANES assessment was aligned with the recommended fasting duration required for accurate diagnostic measurements. This careful synchronization ensured consistency across participants and different NHANES survey cycles. This uniformity in timing allows for standardized comparisons, thereby enhancing the reliability and validity of the study’s findings.

Calculation of SII

To compute the SII, comprehensive data from the complete blood count were utilized. This dataset included crucial components such as peripheral neutrophil, lymphocyte, and platelet counts. The SII is calculated as follows: platelet count multiplied by the neutrophil count, divided by the lymphocyte count.

Assessment of covariatesAssessment of demographic parameters

In our study, we assessed and categorized several essential demographic parameters through participant interviews based on self-report. These included age, gender, race, ethnicity, education levels, and family income-to-poverty ratio.

Body mass index (BMI)

BMI was calculated as the individual’s weight in kilograms divided by the square of their height in meters. BMI values were then categorized into three groups: <25, 25–30, or ≥30.

Alcohol consumption classification

Participants were categorized into three groups based on their self-reported daily alcohol consumption. Specifically, they were classified as nondrinkers, moderate drinkers, or heavy drinkers. Moderate drinkers were defined as those consuming less than two drinks per day for men and less than one drink per day for women, while heavy drinkers were those consuming two or more drinks per day for men and one or more drinks per day for women [23].

Physical activity (PA) assessment

PA was defined as participating in moderate- to vigorous-intensity sports, fitness programs, or recreational activities for more than 10 min per week. Participants who did not engage in such activities for more than 10 min per week were classified as inactive [24]. The assessment of PA was conducted using the Metabolic Equivalent (MET), a widely recognized measure that represents the relative energy metabolism level during various activities.

Healthy eating index (HEI) 2015

We used the Healthy Eating Index (HEI) 2015, developed in alignment with the US Dietary Guidelines for Americans (DGA) 2015–2020, to evaluate dietary patterns among participants [25].

Health conditions and medication data

Participants provided self-reported information regarding physician-diagnosed hypertension, hypercholesterolemia, and CVD. Trained personnel collected data on drug consumption over the previous 30 days by comparing participants’ supplied information with drug and dietary supplement databases.

Diabetes duration

Participants reported the time of their initial diabetes diagnosis, and by considering their age, we ultimately calculated the duration of diabetes, categorizing it into three groups: <3.0 years, 3.0–10.0 years, or >10.0 years.

Clinical assessments

At the time of recruitment, various clinical assessments were conducted, including measurements of HbA1c, triglycerides, total cholesterol, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol levels and renal function (serum creatinine levels).

Assessment of mortality

All-cause mortality encompassed fatalities resulting from any cause. Cardiovascular mortality was identified using codes I00-I09, I11, I13, I20-I51, and I60-I69 in the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision. Cancer mortality was designated by codes C00-C97. Chronic kidney disease (CKD) mortality was defined by codes N00-N07, N17-N19, and N25-N27.

Statical analysis

Considering the intricacies of the NHANES examination design, our analyses incorporated weighted adjustments, accounting for clustering and stratification. Person-years were computed for each participant, starting from their enrollment date until the date of death or the conclusion of follow-up on December 31, 2019, whichever occurred first. The assigned weights followed NHANES database criteria, with particular utilization of the mobile examination center (MEC) exam weight (WTMEC2YR) for this study.

To explore the baseline SII quartiles, all 8697 individuals were divided into four groups. For normally distributed data, one-way ANOVA was applied, while the Kruskal-Wallis test was used for data with abnormal distributions. Hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) were calculated using multivariable Cox proportional hazards regression models to assess the associations between SII and all-cause mortality as well as cause-specific mortality risks. The assumption of proportional hazards was evaluated using Schoenfeld residuals [26].

Two multivariable models were constructed. Model 1 adjusted for age (continuous, in years), sex (male or female), and self-reported race and ethnicity (Mexican American, non-Hispanic Black, non-Hispanic White, or other Hispanic). Model 2, an extension of Model 1, further incorporated educational level (less than high school, High School Grad/GED or Equivalent, more than college), family income-to-poverty ratio (<1.0, 1.0–3.0, or ≥3.0), BMI (<25.0, 25.0–29.9, or ≥30.0), drinking status (nondrinker, moderate, or heavy), physical activity (inactive or active), smoking status (never smoker, former smoker, or current smoker), HbA1c level (<7.0% or ≥7.0%), HEI 2015 (in quartiles), diagnosed cardiovascular disease (CVD), hyperlipidemia, self-reported hypertension, diabetes medication use (none, oral glucose-lowering medication, only insulin, oral glucose-lowering medication and insulin), creatinine (continuous) and diabetes duration (<3, 3–10, or ≥10.0). Variables with missing data were subjected to multiple imputation.

To explore the nonlinear relationship between SII levels and mortality, restricted cubic spline analysis (RCS) was performed using four knots (5th, 35th, 65th, and 95th percentiles). Extreme SII values (5% and 95%) were excluded to mitigate the potential influence of outliers, and nonlinearity was assessed via the likelihood ratio test. The associations between SII quartiles and mortality were investigated based on the results of the restricted cubic spline analyses, with the main quartile serving as the reference group. Weighted Kaplan–Meier plots were employed to compare SII levels with all-cause and cause-specific mortality.

Further stratified analyses were conducted by age (<60 or ≥60), sex (male or female), BMI (<30.0 or ≥30.0), and HbA1c level (<7.0% or ≥7.0%). The association between these stratified components was assessed using the P-value.

Sensitivity analyses were conducted to assess the robustness of our findings. Firstly, individuals who died within the initial 24 months of follow-up were excluded to reduce the potential for reverse causation bias. Secondly, individuals with a history of CVD were excluded from the primary analysis, as were participants with a history of cancer. Additionally, to control for confounding effects, we applied the multiple propensity scores adjusted technique. Sensitivity analyses also included weighted Kaplan-Meier plots and RCS analysis for all values. All statistical analyses were conducted using R 4.2.1, and statistical significance was defined as a 2-sided P-value < 0.05. Data evaluation occurred between May 1, 2022, and November 15, 2022.