Association Between Urine Uranium and Asthma Prevalence: NHANES 2007-2016

Background: Previous studies showed that urine uranium (U) is associated with asthma prevalence in adults. However, the association between them among the general population is unclear. Therefore, this study aimed at exploring this unclear association. Methods: The data of the participants were collected from the 2007-2016 National Health and Nutrition Examination Survey (NHANES) performed in the United States. Continuous variables with skewed distribution were analyzed using Ln-transformation. The association between urine U and asthma prevalence was analyzed by multiple regression analysis, and the linear association between them was evaluated by the smooth curve tting. The subgroup analysis was performed by the hierarchical multivariate regression analysis. Results: A total of 13,581 participants were included in our analysis. The multivariate regression analysis showed that LnU was independently and positively correlated with asthma prevalence in the general population (OR=1.12; 95% CI:1.04,1.20; P=0.002). The subgroup analysis revealed that the College Graduate or above showed the stronger association between LnU and asthma prevalence (<9 th Grade: OR=0.84; 95% CI: 0.61, 1.14; 9-11 th Grade: OR=1.23; 95% CI: 0.99,1.52; High School Grade: OR=1.00; 95% CI: 0.84, 1.19; College: OR=1.04; 95% CI: 0.91,1.19; ≥ College Graduate: OR=1.32; 95% CI: 1.11, 1.57; P for interaction=0.0389). Conclusions: Our research suggested that urinary U levels are positively correlated with asthma prevalence in the general population of the United States, and the association is especially strong among people with high level of education.


Introduction
Asthma is a chronic in ammatory disease of the airways affecting approximately 300 million people worldwide, and has become a serious global health problem affecting all mankind, as revealed by the Global Initiative for Asthma (GINA) report in 2021 [1]. The control of asthma is not optimistic. The results of a survey on patients diagnosed with asthma indicate that this disease is still poorly controlled in the United States (U.S.), Canada, Europe [2], and Asia [3,4]. In addition, Yaghoubi et al. estimated that the total economic burden of uncontrolled asthma among adolescents and adults in the U.S. will exceed $900 billion over the next 20 years [5].
The pathogenesis of asthma is relatively complicated. It is related to an early abnormal immune maturation, genomics, and environmental factors and it is a heterogeneous disease [6]. Environmental factors play a very important role in asthma prevalence. Indeed, an association between uranium (U) and asthma prevalence in adults was found in China and U.S. [7][8][9].
Exposure to harmful heavy metals such as lead (Pb), mercury (Hg) and arsenic (As) can cause a series of human diseases [10,11], and among them, also U poisoning can cause serious health problems such as respiratory diseases and lung cancer [7][8][9][10]12], renal toxicity [13], as well as reproductive system toxicity [14] and immune system damage [15]. The sources of U pollution include mining, military activities, nuclear facilities, groundwater, and phosphate fertilizers [16]. At present, few studies reported the association between U exposure and the prevalence of asthma in adults [7][8][9], and the epidemiological association between U and asthma prevalence in the general population is not well known. In view of the enormous harm of asthma to human health and the economic burden, a large sample of epidemiological data on the impact of the environmental factors on asthma is necessary to carry out the primary prevention of asthma. Therefore, this study was performed using data from the National Health and Nutrition Examination Survey (NHANES), which is a program of studies performed to assess the health and nutritional status of adults and children in the U.S, with the aim of investigating the association between urine U and asthma prevalence in the general population. The association between them in terms of age, gender, body mass index (BMI), economic level, race, education level, marital status, diabetes, hypertension, liver disease, smoking, alcohol consumption, and kidney function was also explored to evaluate the in uence of these factors.

Study population
Our analysis was performed on NHANES data from 2007 to 2016. These data were obtained through a multi-level probability sampling of the nutrition and health status of the U.S. population representative of the general population in the U.S. We included participants aged 6-150 years in the 2007-2016 NHANES database. A total of 50588 participants were considered in this analysis. However, participants who met the following exclusion criteria were removed: participants who were unsure whether they had asthma (n = 2159); participants without urine U values (n = 3,4739); pregnant women (n=109).

Exposure variable
Urine U was the exposure variable in this study, detected by inductively coupled plasma mass spectrometry (ICP-MS). ICP-MS detects the U ion intensity and from that, it elaborates the concentration.
The urine samples were collected, processed, and transported to the National Environmental Health Center, at the Department of Environmental Health Laboratory Science, and the Centers for Disease Control and Prevention to be analyzed. A detailed description of the method used for the urine uranium detection is available on the following website: https://wwwn.cdc.gov/Nchs/Nhanes/.

Outcome variable
The outcome variable in this study was the self-reported prevalence of asthma, and the participants were divided into asthma and non-asthma group according to the self-reported prevalence.
The age, gender, race, PIR, education level, and marital status were obtained from the demographic data of NHANES. The race was divided into the following ve categories: Mexican American, Non-Hispanic White, Non-Hispanic Black, Other Hispanic, Other Races. The BMI of the study population was calculated using the following formula: weight (Kg)/height (m 2 ). The education level included the following levels: <9 th Grade, 9-11 th Grade, High School Grade, College, ≥College Graduate. The marital status was classi ed as follows: married, divorced, widowed, separated, living with the partner, never married. The information regarding the presence of diabetes, hypertension, liver disease and whether they have smoking and drinking habits was obtained from the questionnaire lled by the participants. If the participant smoked at least 100 cigarettes during their life, this behavior was de ned as smoking behavior. Alcohol use was de ned as drinking at least 12 glasses of alcoholic beverages in the past 12 months. Moreover, the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation was used to convert the serum creatinine of the participants into the glomerular ltration rate (GFR) in order to assess and adjust the in uence of renal function on the association between urine U and asthma prevalence [17,18]. A GFR <60 ml/min/1.73 m 2 was de ned as renal insu ciency [19][20][21][22].

Statistical analysis
Statistical analysis was performed using the R software (The R Foundation; https://www.r-project.org).
The baseline characteristics of the study population was evaluated by the descriptive analysis. Since the levels of Ba, Cd, Co, Cs, Mo, Pb, Sb, Tl, Tu, Hg, and U in urine were characterized by a skewed distribution, the Ln-conversion was performed to improve the normality. Age was presented as median (Q1-Q3).
Continuous variables were indicated as mean ± standard deviation (SD), while the categorical variables were indicated by numbers (percentages). The comparison between the asthma and non-asthma group was performed using the unpaired t-test or Mann-Whitney U test, Pearson chi-squared tests or the Fisher's exact, as appropriate. The association between urine U and asthma prevalence was assessed by multiple regression model. Two adjustment models were used to explore the association between them in our study. The covariates that contribute to the odds ratio change greater than 10% such as age, gender, race, and BMI were included as the adjustment variables for the adjusted model I. The adjusted model 2 included the variables in the model I and the ones in which the P-value of the covariate to the dependent variable was less than 0.1, such as age, gender, race, BMI, education level, marital status, PIR, diabetes, hypertension, alcohol use, BUN, TB, TP, globulin, LnBa, LnCd, LnCo, LnCs, LnPb, LnSb, LnTl, LnTu, and LnHg. LnU was transformed into a quartile categorical variable and the P-value for the trend was calculated to verify the possibility of non-linearity between LnU as a continuous variable and asthma prevalence. Furthermore, the correlation graph between urine U and asthma prevalence was shown using the smooth curve tting (penalty spline method). Multiple imputation analysis was used to assess whether the missing values of the covariates were the cause of bias in the results [23]. Hierarchical multivariate regression analysis was used for subgroup analysis. A P-value < 0.05 was considered statistically signi cant.

Results
Baseline characteristics of the subjects A total of 13,581 participants from the NHANES data from 2007 to 2016 were included in our analysis according to the inclusion and exclusion criteria of this study. The baseline characteristics of the study population are listed in Table 1. The participants in the asthma and non-asthma group were 2081 and 11500, respectively. Their Median age was 36.0 (15-57) and 29.0 (14-52) years, respectively (P<0.001). The mean LnU of the asthma and non-asthma group was -5.1 ± 1.0 and -5.0 ± 1.0 ug/L, respectively (P<0.001). In addition, the asthma and non-asthma group showed a statistically signi cant difference in BMI, PIR, race, education level, marital status, diabetes, hypertension, liver disease, smoking, alcohol consumption, TP, albumin, globulin, BUN, TB, GFR, LnBa, LnCd, LnCo, LnPb, LnSb, LnTl, LnTu, and LnHg (all P<0.05).

Association between urinary U and asthma prevalence
The association between urine U and asthma prevalence assessed by the logistic regression model is described in Table 2. The continuous variable urine U was associated with asthma prevalence in the nonadjusted model (OR=1.09; 95% CI: 1.04, 1.14; P=0.0003). Urine U was still associated with asthma after the adjustment of the model I (OR=1.10; 95% CI: 1.05, 1.15; P<0.0001) and model II (OR=1.12; 95% CI: 1.04,1.20; P=0.002) for different confounding factors. Moreover, participants in the second, third, and highest LnU quartile showed a statistically signi cantly higher asthma prevalence in the non-adjusted model (P for trend=0.0007), adjustment model I (P for trend=0.0002), and model II (P for trend=0.0104) compared with the lowest-quantile LnU after converting LnU from a continuous variable to a categorical variable (quartile). P for trends for all models was signi cant and consistent with the P-value of LnU as a continuous variable, thus suggesting a linear association between LnU and asthma prevalence. Further analysis with the smoothing curve tting con rmed the linear association between urine LnU and asthma prevalence after the adjustment for confounding factors (adjustment model II) (Figure 1).
The effect of other covariates on the relationship between LnU and asthma prevalence was also evaluated in the subgroup analysis ( Figure 2). The subgroup analysis revealed that College Graduate or above had the stronger association between LnU and asthma prevalence (<9 th Grade: OR=0.84; 95% CI: 0.61, 1.14 9-11 th Grade: OR=1.23; 95% CI: 0.99, 1.52; High School Grade: OR=1.00; 95% CI: 0.84, 1.19; College: OR=1.04; 95% CI: 0.91, 1.19; ≥College Graduate: OR=1.32; 95% CI: 1.11, 1.57; P for interaction=0.0389). The association between urine LnU and asthma prevalence was consistent in the following subgroups: age, gender, BMI, PIR, race, marital status, diabetes, hypertension, liver disease, smoking, alcohol use, and GFR (P for interaction of all covariates > 0.05). The results of the imputed data set and the complete data were consistent.

Discussion
Our results demonstrated that urine uranium level was positively correlated with asthma prevalence in the whole population of the U.S., and this association was particularly signi cant in people with an education level ≥ college graduate.
The bioavailability of U compounds the human gastrointestinal tract is estimated between <0.1% and 6% [24]. The kidney is the main excretory organ of U compounds [25]. Since the urine U level is correlated with U exposure [26,27], in this study urine U is used as a sign of U exposure.
A case-control study in a Chinese adult population (sample size 1:1 matched, 551 patients in both the asthma group and the control group) as well as an analysis considering 1857 American adults in the NHANES 2007-2008 survey suggested that the prevalence of asthma was positively correlated with the level of U in the urine [7] , [8,9]. Since then, this conclusion has been con rmed by Li, X et al, who obtained consistent results after considering 3425 American adults aged 20-59 (2011-2014) [8,9]. In this study, a sample size of 13,581 of American general population was considered as the study population (NHANES 2007(NHANES -2016, and their age was ≥6 years old, which basically represents the general population in the United States. The results of this study not only con rmed the association between urinary U levels and the incidence of asthma in adults, but it was the rst revealing the prevalence of asthma in adolescents and children (<18 years) positively correlated with urinary U levels. Our research has good clinical signi cance and represents a good reference value considering the high prevalence of asthma in adolescents and children, further revealing that this association cannot be ignored. In addition, our study also discovered that the education level of the general population had an interactive effect on the relationship between urinary U level and asthma prevalence, since this association was stronger in people with higher level of education, suggesting that their urinary U level (uranium exposure) needs a more stringent management in order to reduce the prevalence of asthma in this population.
The pathogenesis of asthma is a complex and incompletely clear process involving multiple cellular components [28][29][30]. Asthma caused by heavy metals is currently explained with the involvement of oxidative stress or airway in ammation [7]. For example, immunocompetent cells are induced by heavy metals to produce oxidants [31], which can cause airway hyper-responsiveness [32], airway spasm and contraction [33], and increase airway mucus secretion [34]. Additionally, Cr and Cd can induce airway in ammation, cause airway obstruction and airway hyperresponsiveness [35,36]. Despite the above evidence, no research on the mechanism regulating the association between U and asthma prevalence is available. Thus, further longitudinal studies between uranium and asthma are necessary to discover the potential mechanism.
O'Conor R et al. suggested that people with higher level of education have more awareness for asthma than people with lower level of education, and this aspect is more helpful in the control of asthma [37]. Our study revealed that the association between urine U and asthma prevalence in college graduates is more evident. This may be explained with the fact that people with high education level pay more attention to their own health, thus, allowing doctors or investigators to obtain information about their asthma prevalence. Hence, it is necessary to monitor urinary U levels in the general population, especially in the population with a high level of education.
The data from NHANES of a large sample survey with a random sampling of the general population in the U.S. was used in this work, since it has a good population representation. The multiple regression analysis was used to explore the independent relationship between urine U and asthma prevalence after adjusting for potential confounding factors. Despite this strength, this study has also some limitations. The simultaneous occurrence of U exposure and asthma cannot be considered a causal relationship between them since the design of this study is cross-sectional. A urine sample per participant was used to detect the urine U concentration, potentially causing measurement errors due to individual variability in short-term uranium excretion. Therefore, multiple or 24-hour urine samples should be used to reduce measurement errors. Self-reported asthma history may have recall bias.

Conclusions
Our research suggested that urinary U levels are positively correlated with asthma prevalence in the U.S. general population and the association is especially strong among people with high level of education. Therefore, the exposure should be prevented, the concentration of U in the urine should be monitored and U pollution should be removed from the environment.