The study utilized data from the National Health and Nutrition Examination Survey (NHANES), a biennial nationwide survey ethically approved and conducted by the National Center for Health Statistics to assess the health and nutritional status of the U.S. population (22). NHANES employs a multi-stage, stratified sampling method to ensure a representative sample, and collects data via surveys, clinical evaluations, and laboratory tests in accordance with the Declaration of Helsinki.

In the NHANES 2013–2020 cycles, 6,271 reproductive-aged females (18–44 years) were initially screened. Those who were pregnant or breastfeeding (n = 1,420), lacked infertility information (n = 549), or had missing lifestyle data (n = 1,944) were ruled out. Further exclusions were made for participants without serum uric acid measurements (n = 107) and those with absent covariates (n = 184). Finally, 2,067 participants were included in the analysis (Supplementary Figure S1).

The healthy lifestyle score was calculated by aggregating the total count of individual factors, including current non-tobacco smoking, non-excessive alcohol intake, sufficient physical activity, healthy diet, ideal waist circumference, and adequate sleep duration (24, 25). The healthy lifestyle score, whose higher values indicated healthier lifestyles, ranged from 0 to 6. According to previous reports, current non-tobacco smoking and consumption of 1–14 g/day alcohol were considered as healthy levels (24). For physical activity, moderate-to-vigorous leisure time physical activity of ≥150 min/week were defined as healthy level (24). Dietary information was gathered using a computerized 24-h dietary recall method, where the first recall was carried out through in-person interviews while the second was conducted via telephone. Healthy Eating Index-2015 (HEI-2015), which consists of 9 adequacy and 4 moderation components, was applied to evaluate diet quality. Healthy diet was characterized by being in the highest 40% of the HEI-2015 score (24). In this study, waist circumference served as an indicator of obesity, with a healthy measurement considered to be under 80 cm for women (26). Sufficient sleep duration was defined as sleeping 7–9 h per day (25). Detailed definitions and cut-off values for each component are provided in Supplementary Table S1.

Infertility was ascertained with a self-reported reproductive health questionnaire (27). Specifically, participants were asked “Have you ever attempted to become pregnant over a period of at least a year without being pregnant?” and “Have you ever been to a doctor or other medical provider because you have been unable to become pregnant?” Infertility was classified in women who responded “Yes” to either question.

Demographic, socioeconomic, and medical information were collected with a computer-assisted personal interview system by trained interviewers. Race/ethnicity was categorized as non-Hispanic White or Black, Mexican American, and others. Marital status was categorized as married and others (e.g., living with partner). Self-reported education attainment was grouped as under high school, high school, and above high school. To measure income status, the poverty-income ratio (PIR) was determined by dividing the family (or individual) income by the poverty line applicable to the survey year. Blood pressure (BP) was examined with mercury sphygmomanometers by technicians in mobile center. Hypertension was characterized by a systolic BP of ≥140 mmHg, diastolic BP of ≥90 mmHg, a physician diagnosis, or current use of prescribed antihypertensive medication. Diabetes were identified based on fasting plasma glucose ≥126 mg/dL, 2-h plasma glucose ≥200 mg/dL in oral glucose tolerance test, hemoglobin A1C ≥ 6.5%, a physician diagnosis, or the use of insulin or oral hypoglycemic medication. Cardiovascular disease (CVD) was characterized as physician-diagnosed CHD, angina/angina pectoris, heart attack, or stroke. Regularity of menstrual cycle, history of pelvic infection, history of pregnancy, and hormone pills usage were collected in reproductive health questionnaire.

As the number of participants adhering to 0 or 6 factors was limited, those with 0–1 and 5–6 healthy lifestyle factors were combined to enhance statistical power (Table 1). After checking the distribution pattern with Q-Q plots, continuous data were presented as mean with standard deviation (SD), while categorical variables were presented as numbers with percentages. Population characteristics across healthy lifestyle scores were compared with one way analysis of variance and chi-square test for continuous and categorical data, respectively.

Multivariable logistic regression analysis was first conducted to examine the independent healthy lifestyle-infertility association. Multicollinearity was excluded since all variance inflation factors were below 10. In model 1, we accounted for age (categorized as below 30 or above) and race/ethnicity (classified as non-Hispanic White and other groups). In model 2, we additionally controlled for marital status (categorized as married or others), PIR (categorized as below 3.5 or above), education level (classified as above high school and high school or below), as well as hypertension, diabetes, CVD, regular menstrual cycle, history of pelvic infection, history of pregnancy, and use of hormone pills (all recorded as yes or no). For each additional healthy lifestyle factor, we also estimated its association with infertility by computing the multivariable-adjusted odds ratios (ORs) and 95% confidence intervals (CIs). To examine whether the aforementioned confounders modified the relationship, we further conducted stratified analyses and evaluated multiplicative interactions with likelihood ratio tests. To investigate the individual contribution of six lifestyle factors, we initially evaluated one factor at a time with all other factors adjusted simultaneously. Then, we reconstructed new scores by removing one healthy lifestyle factor each time from the score and adjusted the removed factor in the models.

The following sensitivity analyses were carried out to assess the robustness of results. First, women who reported having underwent an ovariectomy or hysterectomy were excluded. Second, we redefined the healthy level of alcohol intake as none to moderate alcohol drinking (≤14 g/day). Third, propensity score (PS) adjustment, another method to control for covariates (28), was utilized to address observed confounding. Fourth, the potential for residual confounding was assessed by calculating E-values, which estimate the minimum association strength that an unaccounted confounder must have with both exposure and outcome to explain away the observed relationship after adjusting for measured variables (29). Finally, we developed a weighted healthy lifestyle score system to modify the combined pattern. Briefly, the β coefficients were computed from a logistic regression model that accounted for all six lifestyle factors and relevant covariates. Each binary lifestyle factor (0 or 1) was then multiplied by its corresponding β coefficients with weighted standardization, summed, and finally multiplied by 6. We then categorized the weighted lifestyle scores into quartiles to evaluate their association with infertility risk. A restricted cubic spline (RCS) with three knots was also constructed to illustrate the dose–response relationship.

The association of healthy lifestyle score with serum uric acid was examined with a generalized linear regression model. A multivariable-adjusted logistic regression model and RCS were used to investigate the association of serum uric acid with risk of infertility. Mediation analysis was implemented to assess the mediating role of uric acid in the association between the healthy lifestyle score and infertility, utilizing the R “mediation” package (version 4.5.0).

Survey weights provided by NHANES were not applied in our analyses. This decision was made because some of the statistical models used (e.g., mediation analysis) are not compatible with complex survey weights, and for consistency we therefore reported unweighted results throughout. All statistical analyses were conducted with R version 4.3.2 (The R Foundation for Statistical Computing, Vienna, Austria). Two-sided p-values < 0.05 were considered statistically significant.

This cross-sectional study comprised 2,067 reproductive-aged women (mean age 32.64 years) (Table 1). The proportion of individuals engaging in key healthy lifestyle behaviors were as follow: current non-tabacco smoking (79.87%), low-to-moderate alcohol intake (84.18%), sufficient physical activity (41.12%), healthy diet (40.01%), optimal waist circumference (20.61%), and adequate sleep duration (65.07%) (Table 1). The distribution of population characteristics is shown in Supplementary Table S2. Females with more healthy lifestyle factors were more likely to be younger, other races/ethnics, married, well-educated, and had better income status, higher HEI-2015 scores and lower BMI, waist circumference, and serum uric acid level. In contrast, women with fewer healthy lifestyle factors were more inclined to be hypertensive and diabetic. The prevalence of CVD, irregular menstrual cycle, and pelvic infection were also higher in individuals with a reduced number of healthy lifestyle factors (Supplementary Table S2).

In the crude model, the OR for infertility among participants with 5–6 versus 0–1 healthy lifestyle factors was 0.34 (95% CI: 0.20–0.55) (Table 2). After fully adjusting for confounding factors, women with 5–6 healthy lifestyle factors were confronted with 70% (OR: 0.30, 91% CI: 0.17–0.52) lower risk of infertility when compared to those with 0–1 factor (Table 2). In addition, each additional healthy lifestyle factor was associated with 22% (OR: 0.78, 95% CI: 0.70–0.88) decreased risk of infertility (Table 2).

To assess whether the healthy lifestyle-infertility relationship differed by demographic, socioeconomic, and medical status, we conducted stratified analyses and examined interaction effects (Table 3). The inverse association of healthy lifestyle score with infertility was found to be stronger in those aged <30 years (P-interaction <0.001) and females who have never been pregnant (P-interaction = 0.01). No significant interactions were observed regarding other subgroups.

Among individual lifestyle factors, only being a current non-smoker and having an optimal waist circumference were significantly associated with a reduced risk of infertility. The ORs (95% CIs) were 0.70 (0.51–0.98) and 0.61 (0.41–0.89), respectively. (Supplementary Table S3). Marginally significant and inverse association between low-to-moderate drinking and infertility was also observed (OR: 0.72, 95% CI: 0.53–1.00) (Supplementary Table S3). The associations of five-component lifestyle scores with infertility were attenuated when current non-tobacco smoking, non-excessive drinking, sufficient physical activity, health-conscious diet, ideal waist circumference, and adequate sleep duration were removed from the score, with ORs (95% CIs) comparing 4–5 versus 0–1 healthy lifestyle factors being 0.47 (0.29–0.74), 0.49 (0.31–0.76), 0.46 (0.29–0.72), 0.36 (0.23–0.57), 0.43 (0.28–0.68), and 0.37 (0.23–0.57), respectively (Supplementary Table S4).

Multiple sensitivity analyses were carried out to test the consistency of our findings (Supplementary Tables S5–S10). After excluding women with an ovariectomy or hysterectomy history, those with 5–6 healthy lifestyle factors exhibited a 69% lower risk of infertility (OR: 0.31, 91% CI: 0.18–0.55) compared to those with only 0–1 factor (Supplementary Table S5). The inverse association persisted in sensitivity analysis redefining the healthy level of alcohol drinking, with an OR of 0.78 (95% CI: 0.69–0.87) for infertility per additional healthy lifestyle factor (Supplementary Table S6). After applying PS adjustment to cope with observed confounders, the OR for infertility comparing participants with 5–6 versus 0–1 healthy lifestyle factors was 0.35 (95% CI: 0.21–0.58) (Supplementary Table S7). The E-value was as high as 6.12 (Supplementary Table S8), indicating that only exceptionally strong confounding factors negate the observed inverse association. Finally, in the weighted healthy lifestyle score model, optimal waist circumference was found to have the largest contribution (weighted β: 0.30), followed by current non-tobacco smoking (weighted β: 0.22), non-excessive alcohol intake (weighted β: 0.20), sufficient physical activity (weighted β: 0.14), adequate sleep duration (weighted β: 0.08), and healthy diet (weighted β: 0.06) (Supplementary Table S9). After controlling for all covariates, females with the highest weighted healthy lifestyle score quartile were confronted with 56% (OR: 0.44, 95% CI: 0.29–0.66) decreased risk of infertility compared to those in the lowest quartile (Supplementary Table S10). The RCS further demonstrated a negative dose–response relationship between the weighted score and infertility prevalence, with a significant overall trend (p < 0.001) and no evidence of nonlinearity (p = 0.435) (Figure 1A).

After adjusting for all covariates, subjects with 5–6 healthy lifestyle factors had significantly lower serum uric acid measurements as compared to those with 0–1 healthy lifestyle factors (β: –0.35, 95% CI: −0.55 to −0.15) (Supplementary Table S11). The multivariable-adjusted OR (95% CI) for infertility comparing extreme uric acid quartiles was 1.98 (1.39–2.85) (Supplementary Table S12). Per SD increment in serum uric acid was associated with a 31% (OR: 1.31; 95% CI: 1.16–1.49) elevated risk of infertility (Supplementary Table S12). RCS analysis consistently indicated a strong positive and linear association between serum uric acid and infertility (P-overall<0.001, P-nonlinearity = 0.618) (Figure 1B). As further depicted in Figure 2, uric acid was found to mediate 9.61% (95% CI: 3.93 to 21.88%) of the inverse healthy lifestyle score-infertility association.

To further validate the association between serum uric acid concentration and female infertility, a real-world data analysis was performed at the Center for Reproductive Medicine, Jiangxi Maternal and Child Health Hospital with ethical approval (No. 2024–11). A total of 41,289 women were enrolled from January 2014 to December 2022, of whom 92 were above 45 years old, 707 lacked infertility duration data, 2,645 had infertility due to male factor only, 3,701 had no serum uric acid measurements, and 263 had absent covariates. Finally, 33,881 participants were included in the analysis, and 32,027 (94.5%) were diagnosed with infertility, confirmed as ≥12 months of unsuccessful attempts to conceive without contraception (Supplementary Table S13).

Consistently, women with higher serum uric acid levels were found to have higher odds of infertility (p = 0.002) (Supplementary Table S14). After adjusting for covariates such as age, gravidity, parity, smoking status, systolic and diastolic BP, and year of treatment, the positive association remained statistically significant (p = 0.007). When categorized into quartiles, the risk of infertility was ~1.2 times greater among women with serum uric acid levels >328 μmol/L (Q4) than those ≤ 248 μmol/L (Q1) in both crude (OR: 1.21, 95% CI: 1.06–1.39) and adjusted (OR: 1.19, 95% CI: 1.04–1.37) models (Supplementary Table S14).