Our study analyzed data from the National Health and Nutrition Survey (NHANES), which collected information representative of the civilian noninstitutionalized population in the United States to investigate the prevalence of major diseases and identify risk factors for diseases. The NHANES study was approved by the National Center for Health Statistics Institutional Review Board, and all participants provided written informed consent. The data used in this study were limited to three consecutive cycles from 2007 to 2012 because we needed complete exposure and outcome data. We included individuals with complete information on the needed variables in these cycles.

A total of 30,442 participants were included in the NHANES during 2007–2012. For this cross-sectional analysis, we excluded the following participants: (1) missing data on Se intake (n = 6,966), (2) unqualified data on lung function (specific standards can be found in Section 2.3) (n = 10,392), (3) incomplete data on asthma (n = 0), and (4) incomplete data on covariates (n = 8,543). As a result, 4,541 participants were eventually enrolled in the present study (Figure 1).

In each NHANES cycle, participants provided detailed dietary intake information for two 24-h periods, which were then used to estimate intakes of energy, nutrients, and other food components. The first dietary recall was collected in person during the NHANES visit, while the second was collected by telephone 3 to 10 days later. For these analyses, total estimated Se intake (micrograms, mcg) was averaged over the two recall periods. Se levels were categorized into high (Se+) and low (Se−) according to the medians of the distribution (105.1 mcg).

Survey participants aged 6 to 79 years were eligible for the spirometry component. Specific exclusion criteria that were employed for baseline spirometry included criteria employed in earlier NHANES spirometry surveys. Excluded from testing were examinees who had current chest pain or a physical problem with forceful expiration, were taking supplemental oxygen, had recent surgery of the eye, chest or abdomen, had a recent heart attack, stroke, tuberculosis exposure or had recently coughed up blood. Adults with a personal history of detached retina or a collapsed lung and children with painful ear infections were also excluded. Lung function was measured by Ohio 822/827 dry-rolling seal volume spirometers. Spirometry was performed in the standing position unless the participant was physically impaired. The participant was asked to elevate the chin and extend the neck slightly and place a nose clip on his or her nose during testing, the latter to prevent air leaks. He/she was then instructed to perform a series of maximal forced expiratory maneuvers. In each maneuver, the participant took the deepest breath possible to fill the lungs with air, then put the mouthpiece into his/her mouth making a tight seal and then blew the air out as hard and fast as possible. Participants aged 6–10 years were asked to blow out the air for a minimum of 3 s of exhalation, whereas those aged 11–79 years were asked to blow out the air for a minimum of 6 s of exhalation. We used forced vital capacity (FVC), forced expiratory volume 1st second (FEV1) and the ratio of FEV1 to FVC to analyze the effects of Se intake and asthma on pulmonary function.

Current physician-diagnosed asthma (hereafter referred to as “current asthma”) was defined by a positive answer to the following questions: “Do you still have asthma?.” If the respondent answered “yes” to this question, she/he was classified as a participant with current asthma. If not, s/he was classified as a participant without current asthma. The asthma-Se patterns were divided into four categories: without current asthma and low Se intake (Asthma− Se−), without current asthma and high Se intake (Asthma− Se+), with current asthma and low Se intake (Asthma+ Se−) and with current asthma and high Se intake (Asthma+ Se+).

Based on previous studies related to lung function, we used the following variables as covariates (25). Sociodemographic characteristics were collected using the sample person and family demographics questionnaires, including age, sex, race, ratio of family income to poverty (PIR) and education level. We dichotomized educational attainment into three levels, including less than a high school degree, high school grade/GED or some college/AA degree and college graduate or above. Body mass index (BMI) was calculated as weight (kg) divided by the square of height in (m). Triglycerides in blood, hemoglobin, high-density cholesterol (HDL) and urinary creatinine were measured using the Roche Modular P chemistry analyzer. Smoking status was defined as never smokers (<100 cigarettes in a lifetime), ever smokers (>100 cigarettes in a lifetime, now not smoking at all), and current smokers (>100 cigarettes in a lifetime, some days or every day).

All statistical analyses were performed in R (version 4.3.1) software. Participants were characterized according to the quartiles of Se intake, with means (standard deviations) for continuous variables and percentages for categorical variables. Baseline characteristics were compared across Se intake quartiles using chi-square tests, ANOVA and Wilcoxon rank-sum tests depending on the nature of the data. To assess the association among asthma, Se intake and pulmonary function, we conducted univariate and multivariate linear regression models, including the unadjusted model (model 1) and fully adjusted model (model 2, age, race, PIR, education levels, triglycerides, smoking status, BMI, ASBP, ADBP, urinary creatinine, high-density cholesterol (HDL), and hemoglobin), to reduce the impact of confounding factors. The variance inflation factor (VIF) was used to test for multicollinearity between variables. To explore the role of Se intake in the relationship between asthma and pulmonary function, multivariable linear regressions and restricted cubic splines (from the R package “rms”) with four default knots based on sample size were performed across Se intake. We also used quantile regression to assess the relationship among asthma, Se intake and pulmonary function. Finally, we used linear regression to investigate whether the difference in lung function between participants with current asthma and participants without current asthma disappeared under different selenium intake levels.

Table 1 shows the baseline characteristics of the total population and participants by asthma status. Participants in the asthma+ group were more likely to be non-Hispanic but had high levels of hemoglobin. These participants also tended to have higher levels of Se intake. Most importantly, we observed that asthmatic participants had poor FEV1, FVC and FEV1/FVC. We found that all participants were older than 20 years. No other significant differences were observed in the results.

The differences in FEV1, FVC and FEV1/FVC in the patterns of asthma-Se are shown in Figure 2. The asthma-Se+ group had the highest FEV1, FVC and FEV1/FVC, whereas the lowest FEV1 and FVC was observed in the asthma+ Se− group. Furthermore, we found that the FEV1, FVC and FEV1/FVC in the asthma+ Se+ group were higher than those in the asthma+ Se-group. Asthma− Se + and asthma− Se − groups showed similar outcomes. All results are significant (p < 0.05).

In the unadjusted model (model 1), Se was positively associated with FEV1 (β = 6.30, 95% CI: 5.50 to 7.10) and FVC (β = 7.80, 95% CI: 6.70 to 8.70), while asthma was negatively associated with FE1 (β = −255, 95% CI: −421 to −89) and FEV1/FVC (β = −0.04, 95% CI: −0.06 to −0.03). In the adjusted model (model 2), we found that the positive association of Se with FEV1 and FVC was still significant, and an inverse association among asthma, FEV1 and FEV1/FVC also existed. The adjusted model indicated that every 1-unit increase in Se intake, FEV1 and FVC increased 3.30 mL (95% CI: 2.60 to 4.00) and 4.30 mL (95% CI: 3.50 to 5.10), respectively. Furthermore, participants with current asthma had decreased lung function compared to those without current asthma (FEV1/FVC: β = −0.04, 95% CI: −0.06 to −0.02; FEV1: β = −215, 95% CI: −340 to −90). Multicollinearity was not present for all variables (variance inflation factor, VIF < 5) (Table 2).

The restricted cubic spline visualized the associations among asthma, Se intake and lung function (Figure 3). In the asthma group (Figures 3A–C), the intake of selenium was correlated with FEV1, FVC, and FEV1/FVC (P for Se < 0.05), and this correlation was nonlinear (P-nonlinear <0.05). Se intake was positively correlated with FEV1 (Figure 3B) and FVC (Figure 3C) when it was approximately higher than 100 mcg. Moreover, this positive effect increased with increasing intake of Se. In contrast, when the intake of Se was approximately lower than 100 mcg, Se intake was negatively correlated with FEV1 (Figure 3B) and FVC (Figure 3C). This positive effect decreased with increasing intake of Se. In the asthma+ group, the intake of selenium was associated with FEV1 (Figure 3E) and FVC (Figure 3F) (P for Se < 0.05), but this association was linear (P-nonlinear >0.05). Similarly, Se intake was positively correlated with FEV1 (Figure 3E) and FVC (Figure 3F), when the intake of Se was approximately higher than 100 mcg.

Decimal quantile regression was used to assess the effects of different levels of Se intake on lung function (Figure 4A). We found that in the asthma+ group, the positive effects of Se intake on FEV1 (Figure 4B) were significant for all percentiles and were rarely affected by the intake of Se. (The β of Se intake did not vary significantly with Se intake). However, the positive effects of Se intake on FVC (Figure 4C) were enhanced with the increase in Se intake and were significant for all percentiles except the 30th percentile and 70th percentile (the β of Se intake increased with increasing Se intake). The greatest positive effects were found in the 90th percentile of Se intake. We did not find any significant association between FEV1/FVC and Se intake for all percentiles. Next, to investigate the effects of different intakes of selenium on lung function in the asthma+ group, we classified selenium intake by quartile and performed multiple linear regression in the asthma+ group (Figure 5). Q1 was used as a reference. The FVC and FEV1 of participants with Q4 intake increased 353 mL (95% CI: 80 to 625) and 543 mL (95% CI: 118 to 969), respectively, relative to Q1 intake. Other results were not significant.

We used multiple linear regression to investigate whether the associations between asthma and lung function were still significant at different Se levels (Figure 6). The asthma group was used as a reference. We found that at the Q2 and Q4 levels, the correlation between FEV1 and asthma disappeared. In addition, asthma had the greatest adverse effect on FEV1/FVC at the Q2 level. In summary, maintaining Se intake at Q4 levels had the best beneficial impact on FEV1 and FVC.

To verify whether the results of this study were robust, we conducted subgroup analysis based on demographic stratification (Table 3). Q4 levels of Se intake improved the FEV1 and FVC in participants older than 40 years. Similar results were observed in the male and non-Hispanic groups. This effect was not significant in the 20- to 40-year-old, female, Mexican American and other Hispanic groups. In other race groups, the data were not enough to analyze the relationship. No significant interaction was observed in any of the groups.