2.1. Study Population

National Health and Nutrition Examination Survey (NHANES) provides a cross-sectional dataset designed to reflect the health and nutritional characteristics of the United States population. The application of sample weights accommodates the complex sampling framework, permitting estimates that are generalizable to national demographic distributions across age, sex, and ethnicity as outlined in national census definitions [16]. Information was obtained through standardized home interviews, physical examinations conducted at mobile centers, and laboratory assessments, all carried out through a multistage probability sampling method. The NHANES protocol received approval from the National Center for Health Statistics (NCHS) ethics review board, with written informed consent obtained from all participants.

An initial pool of 10,179 subjects from the 2013–2014 NHANES cycle was considered. Based on predefined exclusion criteria—including absence of AAC scores, mortality data, or inflammatory markers (n = 7136), AAC score less than 1 (n = 2129), and missing essential covariate data (n = 26)—a total of 888 eligible participants remained for analysis. The detailed data selection process is illustrated in Figure 1. The data employed in this study is publicly accessible (https://wwwn.cdc.gov/nchs/nhanes/default.aspx) and has been demographically weighted for subsequent analyses.

2.2. Measurement of LMR

According to the NHANES protocol, lymphocyte and monocyte counts were determined through a complete blood count utilizing automated hematology analyzers (specifically the Colter DxH 800 analyzer) and were expressed in units of × 10³ cells/µL. The LMR was derived by dividing the absolute count of lymphocytes by the absolute count of monocytes.

2.3. Measurement and Definition of AAC

Dual-energy X-ray absorptiometry (DXA) using the Discovery A densitometer (Hologic, Marlborough, MA, USA) was employed to scan the lumbar spine (vertebrae L1–L4), applying the Kauppila scoring system to assess and quantify AAC. Certified and trained radiologic technologists performed the DXA scans at the NHANES mobile examination centers. In this analysis, Kauppila scores ranged from 0 to 24 (AAC total 24 score), with scores of ≥1 signifying the presence of calcification, thus, identifying AAC. Further details on AAC measurement procedures can be found at https://www.cdc.gov/Nchs/Nhanes/20132014/DXXAAC_H.html.

2.4. Ascertainment of Mortality and Follow-Up

Mortality status was ascertained by linking NHANES data with the National Death Index (NDI) records, which can be accessed at https://www.cdc.gov/nchs/data-linkage/mortality-public.htm. Participants were classified as either deceased or alive based on NDI data. The follow-up period was determined by calculating the time from the date of the NHANES examination to the date of death or December 31, 2019, whichever came first.

2.5. Assessment of Covariates

Collected covariates encompassed demographic details (such as age, sex, race/ethnicity, and education), lifestyle factors (smoking and alcohol consumption), clinical and biochemical parameters (body mass index (BMI), poverty income ratio (PIR), hemoglobin A1c (HbA1c), low-density lipoprotein cholesterol (LDL-C), total cholesterol (TC), creatinine, and serum uric acid), as well as disease history (diabetes, hypertension, CHD, heart failure, and stroke). Race was categorized as non-hispanic white, non-hispanic black, mexican american, other hispanic, and other/multiracial. Education levels were classified into “Less than 9th Grade,” “9–11th Grade,” “ = High School Grad/GED,” “Some College or Associate Degree,” and “College Graduate or above” [17]. Smoking status was divided into never smokers (individuals who smoked fewer than 100 cigarets in their lifetime), former smokers (those who smoked 100 or more cigarets in their lifetime but were not currently smoking), and current smokers (individuals who smoked cigarets daily or on some days at the time of the survey). BMI was computed as weight in kilograms divided by the square of height in meters and categorized as normal (<25.0 kg/m²), overweight (25.0–29.9 kg/m²), and obese (≥30.0 kg/m²). DM was identified as either undergoing treatment for or having a medical diagnosis of hyperglycemia, with hemoglobin A1c levels of ≥6.5%, fasting blood glucose levels of ≥126 mg/dL, or a 2-h blood glucose level of ≥200 mg/dL. Hypertension was defined as the use of antihypertensive medications, a medical diagnosis of hypertension, or three consecutive measurements of systolic blood pressure of ≥140 mmHg or diastolic blood pressure of ≥90 mmHg.

2.6. Statistical Analysis

Recognizing that NHANES utilizes a complex, multistage, probability sampling design to select representative participants, we incorporated sample weights, clustering, and stratification into all analyses to achieve national estimates. Continuous variables were presented as weighted means ± standard error, while categorical variables were expressed as numbers (n) with percentages. One-way ANOVA, Kruskal–Wallis H test, or Chi-squared test were used to compare continuous and categorical variables across different groups, respectively.

To examine the associations between the LMR and the risk of all-cause mortality, we utilized a multivariate Cox regression model to estimate hazard ratios (HRs) and 95% confidence intervals (CIs). Given that LMR distribution is right-skewed, values were divided into quartiles, analyzed both as continuous and categorical variables. The first quartile (Q1) served as the reference group, with median values assigned to each category to assess linear trends. Variables with p < 0.1 in univariate analysis or those clinically relevant to prognosis were included in the multivariable Cox regression model. We developed three models: The first model was unadjusted; the second was adjusted for age, sex, and race; and the fully adjusted model additionally included education level, PIR, BMI, smoking status, alcohol consumption, HbA1c, creatinine, serum uric acid, LDL-C, DM, stroke, hypertension, CHF, CHD, and AAC total 24 score. An AAC score >6 was defined as “severe AAC” [18, 19]. Kaplan–Meier survival analysis was used to compare survival probabilities among different LMR levels in individuals with AAC, with log-rank tests employed for group comparisons.

Stratified analyses were also conducted by gender (male and female), age group (≥65 and <65), smoking status (no and yes), alcohol consumption (no and yes), DM (no and yes), CHD (no and yes), stroke (no and yes), heart failure (no and yes), and hypertension (no and yes). Possible interactions between these stratification factors and LMR were also assessed. Additionally, restricted cubic spline (RCS) and threshold effect analysis were utilized to evaluate any nonlinear relationships between LMR and all-cause mortality.

p-value < 0.05 was deemed statistically significant. Statistical analyses were performed utilizing R software (version 4.2.1).

3.1. Baseline Characteristics of Study Participants

A total of 888 individuals were included in the study, with 49.9% female participants and an average age of 64.83 ± 11.9 years. During a median follow-up period of 71 months (interquartile range (IQR), 64.0–78.0 months), the incidence of all-cause mortality among the participants was 16.33%. The weighted mean ± standard error of the natural LMR was 3.63 ± 1.49. Table 1 presented the weighted sociodemographic and medical characteristics of participants across the LMR quartiles. Participants in the lower LMR quartiles tended to be older, more frequently female, and more likely to be smokers with comorbid hypertension and heart failure. They also exhibited higher values in waist measurement, creatinine, TC, low-density lipoprotein (LDL), and AAC scores (all p < 0.05). Supporting Information 1: Table S1 provided baseline characteristics segmented by all-cause mortality.

3.2. Association Between LMR and All-Cause Mortality in US Adults With AAC

During a median follow-up period of 71 months (IQR, 64.0–78.0 months), 145 (16.3%) of the 888 participants passed away. As shown in Table 2, the LMR was significantly associated with a reduced risk of all-cause mortality in the unadjusted model (HR = 0.57, 95% CI: 0.47–0.71). This association remained strong and statistically significant after multivariable adjustments, with Model 2 showing an HR of 0.69 (95% CI: 0.59–0.81) and Model 3 showing an HR of 0.72 (95% CI: 0.57–0.91). Compared to the first quartile of LMR, the HRs for participants in the third and fourth quartiles were consistently lower across all models. In Model 1, the HRs were 0.32 (95% CI: 0.18–0.56) for the third quartile and 0.18 (95% CI: 0.09–0.33) for the fourth quartile, with a significant trend (p for trend <0.001). In Model 2, the HRs were 0.43 (95% CI: 0.24–0.77) for the third quartile and 0.34 (95% CI: 0.17–0.66) for the fourth quartile, also showing a significant trend (p for trend <0.001). Model 3 presented HRs of 0.40 (95% CI: 0.19–0.55) for the third quartile and 0.34 (95% CI: 0.13–0.90) for the fourth quartile (p for trend = 0.02). The Kaplan–Meier survival curves illustrated significantly lower all-cause survival probabilities for individuals in the lower quartile levels of LMR compared to those in the higher quartiles (p < 0.001) (Figure 2).

Subgroup analyses and interaction tests were conducted to examine the robustness of the relationship between LMR and all-cause mortality across various subgroups, including age, sex, smoking status, alcohol consumption, obesity, DM, hypertension, and history of CHD, heart failure, and stroke. The results indicated a consistent association without significant interactions between these characteristics and LMR (p for interaction >0.05). Lastly, subgroup analysis by severity of AAC showed no significant interaction (p for interaction = 0.71) (Table 3).

3.3. Nonlinear Relationship Between LMR and All-Cause Mortality in US Adults With AAC

In this study, we employed the RCS to explore potential nonlinear relationships between the LMR and all-cause mortality in US adults with AAC. The fully adjusted model revealed a nonlinear relationship between LMR and all-cause mortality (Figure 3).

Threshold effect analysis indicated that LMR was associated with all-cause mortality in a nonlinear fashion when multiple confounders were adjusted for, with a threshold value identified at an LMR of 4.49). Below this threshold, higher LMR was significantly associated with a decreased risk of all-cause mortality (HR = 0.49, 95% CI: 0.40–0.60, p < 0.0001). However, above this threshold, LMR did not show a significant association with all-cause mortality (HR = 1.14, 95% CI: 0.77–1.7, p = 0.51; Table 4).

3.4. The Predictive Ability of LMR for All-Cause in US Adults With AAC

The time-dependent receiver operating characteristic (ROC) curve analysis showed that the area under the curve (AUC) for the LMR was 0.647, 0.707, and 0.682 for predicting all-cause mortality at 2 years, 4 years, and 6 years, respectively (Figure 4). These findings indicate that the predictive ability of LMR for all-cause mortality remains consistently effective across different time intervals. Moreover, the results suggest that the predictive value of LMR for all-cause mortality is superior to that of using lymphocyte or monocyte counts alone over 2-year, 4-year, and 6-year periods (Supporting Information 2: Figure S1).