NHANES studyPopulation characteristics of the study subjects

In the current study, 32,308 individuals aged 20 to 60 were included (30,086 with RA and 2,222 without RA, Table 1). Compared to healthy individuals, patients with a history of RA were more likely to be older, predominantly female, and of non-Hispanic Black origin (P < 0.05). Additionally, they had lower educational attainment, a lower poverty-to-income ratio, and a higher prevalence of smoking (P < 0.05). Individuals who were widowed, divorced, or separated had a higher incidence of RA compared to those who were married, cohabiting or single and never married (P < 0.05). As expected, RA patients had significantly higher BMI and alcohol intake (P < 0.05).

Table 1 Baseline characteristics of participants with and without RA history

We arranged the RA patients’ alcohol intake values from smallest to largest and divided them into four equal quartiles. Based on these quartile, the individuals were subsequently separated equally into 4 groups, and their baseline characteristics were compared according to the quartile group (Q1-Q4). As shown in Table 2, significant differences were observed between the highest and lowest quartiles of alcohol intake in terms of sex, race, marital status, education attainment, and BMI distributions (P < 0.05). Patients who consumed more alcohol were predominantly male and had a higher incidence of smoking (P < 0.05). No significant differences were observed between the highest and lowest quartiles in terms of citizenship or poverty-to-income ratio.

Table 2 Baseline characteristics of participants with RA according to alcohol intake quartilesRelationship between alcohol intake and RA based on observations

The odds ratios (ORs) for Model 1, Model 2, and Model 3 were 1.027 [95% confidence interval (CI) 1.023–1.031], 1.030 (95% CI 1.026–1.034), and 1.030 (95% CI 1.025–1.034), respectively, indicating a positive correlation between the incidence of RA and alcohol intake (Table 3). Furthermore, alcohol consumption quartiles were compared, using the first quartile used as a reference (Q1). In Model 1, which was unadjusted for covariates, patients with increasing quartiles of alcohol intake (Q2, OR: 0.070, 95% CI: 0.047–0.103; Q3, OR: 1.266, 95% CI: 0.926–1.732; Q4, OR: 3.885, 95% CI: 3.222–4.684) exhibited a higher risk of RA compared to Q1 (all P values for trend < 0.05). This increased risk remained significant in Model 2 (P < 0.05), adjusted for age, sex, and race (Q2, OR: 0.092, 95% CI: 0.063–0.136; Q3, OR: 1.557, 95% CI: 1.121–2.218; Q4, OR: 6.045, 95% CI: 4.884–7.483) and in Model 3, which was further adjusted for education level, smoking status, marital status, and BMI (Q2, OR: 0.091, 95% CI: 0.062–0.134; Q3, OR: 1.909, 95% CI: 1.332–2.736; Q4, OR: 6.930, 95% CI: 5.570–8.623).

Table 3 Associations between alcohol intake and RA risk

To gain further insight into the relationship between alcohol intake and the risk of RA, we conducted subgroup analyses stratified by age, sex, race, martial status, smoking status, education level, and BMI. In each subgroup, a significant positive correlation was observed between alcohol intake and RA risk, both before and after adjusting for possible variables (all OR > 1, P < 0.05, Table 4).

Table 4 Subgroup analysis for the association between alcohol intake and RA riskMendelian randomization studyCausal relationship between spirits intake and the risk of RA

An MR analysis was conducted to evaluate the possible causal relationship between spirits intake and RA risk, as the multivariate linear regression analysis above indicated a substantial positive correlation between the two variables. In the majority of the studies, we considered 17 IVs (genetic variations) associated with spirits intake. As shown in Fig. 2, the estimated effect of spirits intake on RA by each SNP is displayed in a forest plot. The direction of each SNP’s OR indicates either a protective or risk effect. Five SNPs (rs11612594, rs140255666, rs9826278, rs34300267, and rs79668485) to the left of the invalid line exhibited a protective effect against RA, while 12 SNPs (rs138046610, rs144504331, rs28603365, rs73568958, rs117933182, rs77094623, rs111610561, rs138235027, rs79181129, rs190371197, rs10257147, and rs73340194) to the right of the invalid line were associated with an increased RA risk. According to the IVW findings, there was a significant correlation between genetically predicted spirits intake and an elevated risk of RA (OR = 1.043, P < 0.05, Table 5).

Fig. 2

Forest plot of SNPs associated with RA risk. SNPs are ordered by chromosome numbers. The dot and the bar indicate the causal effect of spirits intake level on the risk of RA. Protective SNPs against RA risk are shown on the left of the invalid line; SNPs with predisposition risk to RA are shown on the right of the invalid line. Note A shot of spirits (whisky or Baijiu), about 30 mL, with an alcohol content of 40% [29]

Table 5 MR estimates for the causal effect alcohol intake on RA risk

However, MR-Egger (OR = 1.022, P = 0.686), the weighted median (OR = 1.044, P = 0.095), the simple mode (OR = 1.042, P = 0.384), and the weighted mode (OR = 1.055, P = 0.165) did not strongly support a causal association between spirits intake and RA risk (Table 5). The effect estimates of exposure to RA, as determined by several MR techniques, are displayed in Fig. 3. Each point in the scatter plot represents a SNP, and the 95% CI is indicated by the line at each point. The horizontal coordinate shows the impact of SNPs on exposure factors (spirits intake), and the vertical coordinate represents the impact of SNPs on outcomes (RA). The colored lines show the MR fit results. The results show that spirits intake had a suggestive risk effect on RA, with an OR of 1.043 (P < 0.05), meaning that for every unit increase in log-odds of spirits intake, the risk of RA increased by 1.043.

Fig. 3

Scatter plot for of the associations between spirits intake and RA risk. Each point in the scatter plot represents an IV (genetic variant). The line on each point reflects the 95% CI, and the horizontal coordinate is the effect of SNPs on spirits intake. The vertical coordinate is the effect of SNPs on RA. SNP effects were plotted as lines for the IVW (red line), MR-Egger (blue line), simple mode (green line), weighted median (purple line), and weighted mode (orange line) methods. The slope of each line represents the causal estimation. A positive slope indicates that spirits intake had a positive effect on RA risk. Note A shot of spirits (whisky or Baijiu), about 30 mL, with an alcohol content of 40% [29]. MR, Mendelian randomization; SNP, single nucleotide polymorphism

Sensitivity analysis

Several sensitivity analyses, including Cochrane’s Q test, the MR-Egger intercept test, and the leave-one-out analysis for spirits intake with 17 genetic variations, were conducted as assess the stability of the results. Cochrane’s Q test revealed no heterogeneity in the causal effect of spirits intake on RA (all P > 0.05, Table 5). Additionally, no horizontal pleiotropy was detected during spirits exposure according to the MR-Egger intercept test (P = 0.685, Table 5). As shown in Fig. 4A, the funnel plot depicting the causal relationship between spirits intake and RA was nearly symmetrically distributed. The reliability and dependability of the MR results were supported by the leave-one-out graph, which showed that all lines were positioned to the right of the “0 line”, indicating that the exclusion of any single SNPs did not significantly impact the results (Fig. 4B).

Fig. 4

Sensitivity analyses for the effect of spirits intake on RA risk. (A) The funnel plot of the causality of spirits intake and RA was almost symmetrically distributed, suggesting that no significant heterogeneity existed. (B) The leave-one-out graph showed that all lines were located to the right of the “0 line”, suggesting that the removal of any SNPs did not fundamentally affect the results, supporting the reliability of the MR results. Note A shot of spirits (whisky or Baijiu), about 30 mL, with an alcohol content of 40% [29]. MR, Mendelian randomization