Association between secondhand smoke exposure and rheumatoid arthritis in US never-smoking adults: a cross-sectional study from NHANES

While smoking is widely acknowledged as a risk factor for rheumatoid arthritis (RA), the connection between secondhand smoke (SHS) exposure and RA in never-smoking adults remains limited and inconsistent. This study aims to explore and quantify this association using serum cotinine levels. We conducted a cross-sectional study with 14,940 adults who self-report as never smokers, using National Health and Nutrition Examination Survey data from 1999 to 2018. Based on previous literature, SHS exposure was categorized into four groups according to serum cotinine levels. Compared to individuals in the unexposed group (serum cotinine < 0.05 ng/mL), the adjusted odds ratio (OR) for RA was 1.37 (95% CI 1.14–1.64, p = 0.001) in the low exposure group (serum cotinine at 0.05 to 0.99 ng/mL) after adjusting for covariates. However, no significant association was found in the moderate exposure group (serum cotinine at 1 to 10 ng/mL) or the heavy exposure group (serum cotinine ≥ 10 ng/mL). Furthermore, we detected a non-linear, positively saturated correlation between the cotinine levels after log2 transformation and RA, with a turning point at approximately − 2.756 ng/mL (OR = 1.163, 95% CI 1.073–1.261, p = 0.0002). The stability of the results was confirmed by subgroup analysis.


Smokers and RA assessment
In this study, never smokers were defined as participants who reported consuming fewer than 100 cigarettes in their lifetime and did not use any products containing nicotine in the past 5 days.Former smokers were participants who had a history of smoking over 100 cigarettes but were not presently engaged in smoking.Current smokers were participants who had smoked over 100 cigarettes in their lifetime and were actively smoking during the study.Participants were considered to have RA if they responded "RA" to the NHANES questionnaire questions, which asked whether a doctor had diagnosed them with arthritis and, if so, specified the type of arthritis.

Covariates
NHANES data collection involved administering a standardized participant questionnaire through household interviews, complemented by a comprehensive medical evaluation for each individual.In accordance with prior literature, the covariates taken into account in this study encompassed age, gender, marital status, race/ethnicity, educational level, family income, hypertension, diabetes, coronary heart disease, body mass index, physical activity, alcohol user, fish intake, asthma.Marital status was categorized as married, living with a partner, or living alone.Race/ethnicity included classifications such as non-Hispanic white, non-Hispanic black, Mexican American, or other races.Educational attainment was classified as less than 9 years, 9-12 years, and more than 12 years.Family income, based on the poverty income ratio (PIR), was grouped as low (PIR ≤ 1.3), medium (1.3 < PIR ≤ 3.5), and high (PIR > 3.5) according to a US government report 26 .Physical activity was classified into sedentary, moderate, and vigorous.Moderate activity involved at least 10 min of movement within the last 30 days, resulting in only light sweating or a mild to moderate increase in breathing or heart rate.Vigorous activity also required at least 10 min of movement within the last 30 days but resulted in profuse sweating or an increase in breathing or heart rate.Alcohol user was categorized as never, former, mild, moderate, or heavy drinker.The definitions of hypertension, diabetes, coronary heart disease, and asthma were based on questionnaire inquiries about whether participants had been informed by their doctor of these conditions in the past.The definition of fish intake is determined through dietary interview questionnaires, asking whether fish-containing foods have been consumed in the past 30 days.BMI was calculated through a standardized method of measuring height and weight.

Statistical analysis
A secondary analysis was performed on publicly accessible datasets, categorical variables were represented as percentages (%), while continuous variables were expressed as means (SD) or median (IQR), according to the data distribution.To assess differences between groups, we employed analysis of variance, Kruskal-Wallis, and chi-squared tests.Additionally, to explore the relationship between SHS exposure and the occurrence of RA, we calculated odds ratios (OR) with 95 percent confidence intervals (95% CI) using statistical methods, such as multivariate logistic regression models.Model 1 underwent adjustments for sociodemographic characteristics, comprising age, gender, race/ethnicity, marital status, education level, and family income.Model 2 incorporated additional adjustments by including factors from Model 1, along with hypertension, diabetes, coronary heart disease, body mass index, physical activity, alcohol user, fish intake.Model 3 achieved full adjustment, encompassing factors from Model 2 and incorporating asthma.
After adjusting for covariates as specified in Model 3, a multivariate-adjusted restricted cubic spline analysis with four knots was employed to explore the potential nonlinear dose-response relationship between log2transformed serum cotinine levels and RA.
In addition, after adjusting for covariates as specified in Model 3, a two-piecewise logistic regression model was used to explore the saturated relationship between log2-transformed serum cotinine levels and RA.
Possible alterations in this association were further explored based on age, gender, and BMI categories through multivariate logistic regression analysis.Likelihood ratio tests for interactions were incorporated.Consistent findings were observed even with the model adjustment through multivariate logistic and restricted cubic spline analyses.
All analyses were performed using the statistical software packages R (http:// www.R-proje ct.org, The R Foundation) and Free Statistics software version 1.9.2.Descriptive statistics were applied to characterize all participants, and two-tailed testing was utilized with a significance level of p < 0.05.

Study population
From 1999 to 2018, a total of 14,940 individuals were successfully enrolled in this study (Fig. 1).Exclusions were applied to participants aged under 20 years (n = 46,235).We also excluded individuals who were pregnant (n = 689) and those with missing data on RA (n = 1,923), serum cotinine (n = 2,874), and other covariates (n = 4,067).

Baseline characteristics
The baseline characteristics of participants, classified according to the presence or absence of RA, are depicted in Table 1.Among the 14,940 individuals, 714 (4.8%) had RA.The study participants had an average age of 46.7 ± 17.6 years, with 8,992 (60.2%) being female and 5,937 (39.7%) identifying as non-Hispanic White.In contrast to participants without RA, those with RA tended to be older, living alone, have a lower educational level, exhibit a lower family income, and experience a higher incidence of hypertension and diabetes.

Association of SHS exposure with RA
Table 2 exhibits the results of multivariate logistic regression.Following adjustments for age, gender, marital status, race/ethnicity, educational level, and family income (model 1), an OR of 1.43 (95% CI 1.19-1.71,p < 0.001) was observed for the association between the low exposure group and RA.Following adjustments for the factors encompassed in model 1 and hypertension, diabetes, coronary heart disease, body mass index, physical activity, alcohol user, fish intake (model 2), consistency with model 1 was observed, with an OR of 1.36 (95% CI 1.13-1.63,p = 0.001).After adjusting for the factors included in model 2 and asthma, the association remained statistically significant, with an OR of 1.37 (95% CI:1.14-1.64)and the p for trend value was 0.001.Notably, a significant association with RA was observed exclusively in the low exposure group across all SHS exposure categories.www.nature.com/scientificreports/Moreover, a non-linear and positively saturated association was observed between log2-transformed serum cotinine levels and RA through restricted cubic spline analysis (p = 0.028 for non-linearity)(Fig.2).In the threshold analysis, for individuals with log2-transformed serum cotinine levels below − 2.756 ng/mL, an OR of 1.163 (95% CI 1.073-1.261,p = 0.0002) was observed for each 1-unit increase in log2-transformed serum cotinine levels (ng/mL).However, no significant association was found between log2-transformed serum cotinine levels and RA in subjects with levels surpassing − 2.756 ng/mL, with an OR of 1.003 (95% CI 0.933-1.077)and a p-value of 0.943 (Table 3).www.nature.com/scientificreports/

Stratified analyses
No significant interactions were observed in any subgroup when stratified by age, gender, or BMI in the stratified analyses of multiple subgroups (Fig. 3).Similar results were obtained even after the model was adjusted through multivariate logistic and restricted cubic spline analyses.

Discussion
In this large cross-sectional study, we focused on non-smokers in the United States and observed a non-linear, positively saturated relationship between SHS exposure and RA.When the log2-transformed serum cotinine levels were below − 2.756 ng/mL, each 1 ng/mL increase in log2-transformed serum cotinine levels corresponded to a 16.3% increase in the risk of developing RA.However, no significant correlation was found between SHS exposure and RA risk when the log2-transformed serum cotinine levels was exceeded − 2.756 ng/mL.Our study indicates that exposure to low levels of SHS increases the risk of developing RA, with a saturation effect observed between the two variables.Furthermore, the robustness of this relationship was also confirmed through subgroup analyses.
Previous investigations on the association between smoke exposure and RA are limited, and the results have been inconsistent.It was reported by Nguyen in a substantial prospective cohort study involving French women that an elevated risk of RA was associated with passive smoking exposure during childhood and/or adulthood, particularly in individuals who had never been active smokers 27 .Karen discovered a significant association between past and present tobacco use and the onset of RA in a prospective analysis involving 103,818 women, with a particular emphasis on seropositive RA 1 .While some investigations support a link between smoke exposure and RA, others contradict such a connection.No apparent relationship was observed between exposure to passive smoking and RA in a Swedish population-based case-control study 28 .Vanessa provides evidence contradicting the notion that passive smoking or an earlier age of smoking makes individuals more susceptible to RA 29 .However, no further study has been conducted to explore the relationship between SHS exposure and RA in never-smoking adults, utilizing serum cotinine to quantify the extent of SHS exposure.Serum cotinine levels can be employed to quantify exposure to smoking, enhancing the broad applicability of research findings.The NHANES presents a unique opportunity to investigate the potential association between SHS exposure and RA, as well as the dose-response relationship between the two.In contrast to previous studies, our research took a dual approach.We used self-reported criteria to identify non-smokers and employed serum cotinine levels to quantify SHS exposure.In our study, we observed that when log2-transformed serum cotinine levels surpassed a certain threshold, there was no significant association between SHS exposure and RA.However, below this threshold, an increase in serum cotinine levels correlated with a higher prevalence of RA.To validate our findings, further multicenter randomized controlled trials are imperative.Furthermore, research indicates that thirdhand smoke exposure, which refers to the inhalation of residual nicotine substances in the environment, can affect serum cotinine levels 30 .However, there is currently no universally accepted biomarker to effectively distinguish between thirdhand smoke exposure and SHS exposure.This indicates that there may be biases present when using serum cotinine levels to assess SHS exposure.
While the precise pathogenic impact of SHS exposure on RA remains unclear, various mechanisms have been suggested to enhance comprehension of the role of cigarette smoking in RA.The persistence of inflammation in RA may be linked to autoimmunity, characterized by oxidative stress, a proinflammatory condition, the production of autoantibodies, and epigenetic effects 7 .Researchers believed that tobacco exposure and its potential mechanisms in relation to RA involve various aspects as follows: Firstly, both the cellular and humoral components of the immune system are affected by cigarette smoking, resulting in a systemic proinflammatory condition 31,32 .Fibroblast-like synoviocytes affected by RA exhibit an induction of proinflammatory cytokinessuch as IL-1a, IL-1B, IL-6, and IL-8, at both mRNA and protein levels-due to exposure to cigarette smoke condensate 33,34 .Secondly, cigarette smoke can trigger the release of endogenous free radicals 35 , compromising antioxidant systems and increasing oxidative stress, thereby contributing to the pathogenesis of RA 36 .Additionally, evidence presented by researchers from the Swedish Epidemiological Investigation of Rheumatoid Arthritis (EIRA) suggests that smoking could activate autoimmune responses specific to RA and citrullinated proteins 4 .Last but not least, smoking elevates the levels of Fas (CD95) and CD4 T-cells, making cells more prone to apoptosis.This heightened susceptibility to apoptosis results in increased levels of cellular debris, which might not be sufficiently eliminated in autoimmune disorders 37 .Smoking influences the immune system, resulting in diminished natural killer cells, suppressed hormonal cells, and cell-mediated immunity, along with the induction of dysfunction in T lymphocytes [38][39][40][41][42] .In a meta-analysis involving six cohorts, it was reported by De Rooy et al. that the impact of smoking on joint damage was mediated through ACPA 43 .In summary, tobacco exposure may contribute to or exacerbate rheumatoid arthritis through multiple pathways.Nevertheless, further prospective inquiries are required to authenticate the impact of SHS exposure on RA.   www.nature.com/scientificreports/Coupled with its nationwide and multi-ethnic scope, our findings hold broad applicability and generalizability.Additionally, the dual approach-combining self-reported criteria for non-smoker identification and serum cotinine levels for quantifying SHS exposure-minimizes potential classification errors and subjectivity, significantly enhancing the study's overall reliability.However, our study does have limitations to consider.Firstly, being a cross-sectional study, it falls short of establishing causal relationships among dependent, independent, and covariate variables.Addressing this limitation requires the initiation of future prospective studies.Moreover, the relatively short half-life of serum cotinine restricts its ability to reflect long-term SHS exposure, instead emphasizing recent exposure.Additionally, exposure to thirdhand smoke can influence serum cotinine levels, potentially introducing bias when defining SHS exposure based on cotinine levels.Furthermore, the wide timeline of this study could potentially impact the obtained results, introducing the possibility of temporal bias.Despite the utilization of regression models and stratified analyses, it is crucial to acknowledge that complete elimination of residual confounding effects from unmeasured or unknown factors may not be attainable.

Conclusion
In conclusion, our study centered on the never-smoking individuals in the United States and revealed a positively saturated relationship between SHS exposure and RA, indicating an inflection point of serum cotinine level after log2 conversion at approximately − 2.756 ng/mL.These findings aim to capture the attention of non-smokers.Moreover, further large-scale prospective investigations are necessary to confirm and extend our findings.

Table 1 .
Baseline characteristics of 14,940 participants in the NHANES 1999-2018 cycles.NHANES, National Health and Nutrition Examination Survey.RA, rheumatoid arthritis.BMI, body mass index.SHS, secondhand smoke.

Table 2 .
Association between SHS exposure and RA among never-smoking adults in the NHANES 1999-2018 cycles.SHS, secondhand smoke; RA, rheumatoid arthritis; NHANES, National Health and Nutrition Examination Survey; OR,odds radio; CI, confidence interval; Ref, reference.*Model1 adjusted for age, gender, marital status, race/ethnicity, educational level, family income.Model 2 was adjusted for model 1 + hypertension, diabetes, coronary heart disease, body mass index, physical activity, alcohol user, fish intake.Model 3 was adjusted for model 2 + asthma.

Table 3 .
Association between log2-transformed serum cotinine levels and RA using two-piecewise regression models.RA, rheumatoid arthritis; OR, odds ratio; CI, confidence interval.*Adjusted for age, gender, marital status, race/ethnicity, educational level, family income, hypertension, diabetes, coronary heart disease, body mass index, physical activity, alcohol user, fish intake, asthma.
Our study's strengths lie in the utilization of data from multiple survey cycles, which ensures a robust sample size.
Stratified analysis of the relationship between SHS exposure and RA.Except for the stratification component itself, each stratification factor was adjusted for all other variables (age, gender, marital status, race/ ethnicity, educational level, family income, hypertension, diabetes, coronary heart disease, body mass index, physical activity, alcohol user, fish intake, asthma).OR, odds ratio; CI, confidence interval.