This study utilized a cross-sectional research approach, utilizing data from the NHANES, a comprehensive, multistage, stratified, and complex survey conducted nationally by the Centers for Disease Control and Prevention (CDC). Participants underwent initial interviews in their homes, followed by a series of clinical and laboratory tests at Mobile Examination Centers (MEC). Detailed information about the survey and associated research data is available on the NHANES website at https://www.cdc.gov/nchs/nhanes/.

The study selected participant data from two NHANES survey cycles (2015–2016 and 2017–2018). Initially, 19,225 female participants were included. In the second step, 15,720 samples lacking PID data were excluded. In the third step, 168 samples with missing dietary magnesium data were excluded. Additionally, 303 participants with missing demographic information, hypertension data, and smoking data were considered ineligible. Ultimately, a total of 3,034 women aged 20–59 were included in our final analysis, of which 183 had PID. Additionally, our study employed weighted adjustments to accurately reflect the demographic composition of the United States population. The flowchart detailing the inclusion and exclusion criteria is shown in Figure 1.

The NHANES initiative received ethical clearance from the Ethical Review Committee of the National Center for Health Statistics, with all participants providing written informed consent. Our study rigorously adhered to the protocols outlined by the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria (15).

The NHANES study utilized a 24-h dietary recall questionnaire, administered to all study participants. This questionnaire collected comprehensive information on all foods and beverages consumed in the past 24 h, detailing the types and quantities of each item (16). For each NHANES participant from the 2015–2018 cycles, two 24-h recalls were conducted. The first dietary recall was performed at the NHANES Mobile Examination Center (MEC), and the second recall was conducted by trained interviewers via telephone 3–10 days after the MEC interview. We calculated the average dietary magnesium intake from the two recall periods; if the second recall was unavailable, the data from the first recall was used. To accurately estimate the dietary magnesium consumption, the Food Intake Analysis System developed by the University of Texas was used, in conjunction with the USDA’s National Nutrient Database for Standard Reference.

The diagnosis of PID was assessed using data obtained from the NHANES Reproductive Health questionnaire. Specifically, question RHQ078 asked: “Have you ever been treated for an infection in your fallopian tubes, uterus, or ovaries, also called a pelvic infection, pelvic inflammatory disease, or PID?” Participants who answered “Yes” to this question were considered to have PID.

Consistent with previous pertinent research (4, 16–18), demographic factors (age, race, educational background, marital status), socioeconomic status [Poverty Income Ratio (PIR)], health and lifestyle indicators [Body Mass Index (BMI), smoke status, presence of chronic conditions such as diabetes and hypertension], and reproductive health metrics (regularity of menstrual periods) were included as covariates in our analysis. Age was classified into two groups: 20–40 years and 41–59 years. Race was classified into four categories: Non-Hispanic White, Mexican American, Non-Hispanic Black, and other races. Meanwhile, educational background was organized into three tiers: below high school, high school, and above high school (19). Marital status was categorized as married or living with a partner or living alone (20). NHANES meticulously recorded participants’ smoking behaviors, including status, duration, and associated activities (21). Smoking status was categorized as follows (22): never smokers are adults who have not smoked or have smoked less than 100 cigarettes in total; former smokers have smoked 100 or more cigarettes but no longer smoke; current smokers have smoked over 100 cigarettes and continue to smoke, whether occasionally or daily. For the evaluation of diabetes, participants were asked, “Have you ever been told by a doctor or health professional that you have diabetes or sugar diabetes?” Respondents affirming this question were classified as diabetic (23). Hypertension was diagnosed by averaging three measurements of resting brachial arterial pressure, with the criteria being a systolic pressure of 140 mmHg or higher and a diastolic pressure of 90 mmHg or higher (24). The PIR compares household income to the poverty line, indicating socioeconomic status. Categories include low (PIR under 1.35), middle (PIR 1.35–3.0), and high (PIR over 3.0) (25, 26). BMI is classified into normal weight (below 25.0 kg/m²), overweight (25.0–29.9 kg/m²), and obesity (above 30.0 kg/m²) (27). Regular period data were assessed through the reproductive health questionnaire that asked, “Had regular periods in past 12 months (eligible for participants aged 12 years older)?,” and women who were pregnant or who had bleeding as result of medical issues, hormonal treatments, or surgical procedures were excluded.

All statistical analyses in this study were conducted using dietary weights, ensuring precision and integrity in line with the least common denominator strategy guided by NHANES sampling weights (28). Continuous data were presented as mean values with standard deviations (SD), while categorical data were presented as counts (n) and proportions (%). To compare the baseline characteristics between the PID and non-PID groups, we employed the Mann–Whitney U test or t-test for continuous variables, and the chi-square test or Fisher’s exact test for categorical variables.

To elucidate the relationship between dietary magnesium intake and PID, weighted multivariable logistic regression analyses were conducted. Initially, magnesium intake was evaluated as a continuous variable and then divided into quartiles for detailed analysis: Q1 (below 133.12 mg/day), Q2 (133.12–214.93 mg/day), Q3 (214.93–287.19 mg/day), and Q4 (above 287.19 mg/day), with Q1 serving as the baseline comparison group. We tested different models (Models 1–3) by progressively adjusting for various risk factors. Model 1 remained unadjusted, Model 2 adjusted for key confounders including age and race, and Model 3 further adjusted for education level, marital status, PIR, BMI, regularity of menstrual periods, smoking status, and presence of chronic conditions including diabetes and hypertension. After converting dietary magnesium intake from a continuous to a categorical variable, trend tests were conducted to examine the linear relationship trend between dietary magnesium consumption and PID. Restricted cubic splines (RCS) were utilized to evaluate the nonlinear and dose–response relationships between the intake of dietary magnesium and PID. Additionally, stratified analyses and the inclusion of interaction terms were employed to examine the consistency of this association across different demographic groups. Multiple imputation (MI) is a method for addressing missing data by creating multiple datasets (29). The MI program uses the chained equations method to impute missing values for various variables, including education, PIR, BMI, and diabetes.

The association between magnesium intake and PID is expressed through odds ratios (ORs). Analyses were performed with R (version 4.2) and EmpowerStats (version 5.0). A two-sided p-value below 0.05 was considered statistically significant.

Table 1 presents the weighted demographic and baseline characteristics of the study participants. This study included 3,034 participants, with an average age of 40.01 years. The majority were Non-Hispanic White, with a high school education or higher, and either married or living with a partner. Compared to participants without PID, those diagnosed with PID were older, more likely to smoke and experience irregular menstrual cycles, had lower family income, higher BMI, and lower dietary magnesium intake levels (229.60 mg/day, p < 0.001) (Figure 2).

Table 2 illustrates the association between dietary magnesium intake and the risk of developing PID. When dietary magnesium was treated as a continuous variable, a significant correlation was observed across all three models (Models I–III). Specifically, for each 1 mg/day increment in dietary magnesium intake, there was a notable reduction in PID risk, with odds ratios (OR) of 0.996 (0.994–0.998), 0.996 (0.994–0.999), and 0.997 (0.995–1.000) in Models I, II, and III, respectively. These findings suggest a robust relationship between higher dietary magnesium consumption and a reduced incidence of PID.

Additionally, when categorizing dietary magnesium intake into quartiles, the fully adjusted model (Model III) revealed that compared to individuals in the lowest quartile Q1 (less than 133.12 mg/day), the adjusted ORs for participants in Q2 (133.12–214.93 mg/day), Q3 (214.93–287.19 mg/day), and Q4 (more than 287.19 mg/day) were 0.480 (95% CI: 0.281–0.821, p = 0.012), 0.639 (95% CI: 0.323–1.265, p = 0.175), and 0.395 (95% CI: 0.177–0.879, p = 0.027), respectively (Table 2). Trend analysis showed that the likelihood of PID decreased by 23.4% with each quartile increase in dietary magnesium intake (adjusted OR = 0.766, 95%CI: 0.605–0.968, p = 0.046).

Restricted cubic spline (RCS) plots were employed to visualize the correlation between dietary magnesium intake and PID. Figure 3A elucidates the non-adjusted dose–response curve between dietary magnesium intake and PID, while Figures 3B,C show the RCS curve adjusted for covariates, similar to Model II and Model III, respectively. These findings demonstrate a gradual decline in the incidence of PID with higher dietary magnesium intake. Overall, there is a linear negative relationship between dietary magnesium intake and PID, with the relationship being particularly pronounced in Subfigure C (P for non-linearity = 0.865).

To further explore the potential link between the intake of dietary magnesium and PID, stratified analysis was performed for various subgroups within Model III (Figure 4). The analysis revealed a significant negative correlation between dietary magnesium consumption and PID among several subgroups. Specifically, elderly women, individuals with lower to middle household incomes, those with normal weight or overweight, non-smokers, individuals without diabetes, and those experiencing irregular menstrual cycles all showed a notable inverse relationship between dietary magnesium intake and PID (all p-values < 0.05). Of particular interest, the analysis identified a statistically significant interaction effect with age (interaction p-value = 0.023), indicating that the impact of age on this association is nuanced and warrants further investigation. To elucidate this interaction, we depicted the age-dependent relationship between dietary magnesium intake and PID using stratified smoothing curves in Figure 5. These curves provide a visual representation of how the relationship between dietary magnesium intake and PID varies across different age groups, offering deeper insights into the underlying dynamics of this association.