Association of prognostic nutritional index with the presence and all-cause mortality of rheumatoid arthritis: the National Health and Nutrition Examination Survey 2003–2018 Authors: Junjie Wang, Rui Zhu, Haonan Fang, Xing Xing, Liru Ge, Guoqi Cai Published: 2024 ABSTRACT BackgroundThe prognostic nutritional index (PNI) is a comprehensive measure of individual immune and nutritional status. This study aimed to evaluate the role of PNI in the presence and mortality of rheumatoid arthritis (RA).MethodsThis study used data of participants aged ≥ 40 years from the National Health and Nutrition Examination Survey (NHANES) 2003–2018. PNI was calculated using serum albumin and lymphocyte count. The relationship between PNI and the prevalence of RA and mortality among RA patients was assessed using logistic and Cox regression models. Nonlinear associations were explored using restricted cubic splines (RCS).ResultsOf 18,245 participants (mean 55.4 years, 49% female), 1901 had RA, among whom (480/1899, 25%) died during a median follow-up period of 84 months. PNI was inversely associated with the likelihood of having RA (odds ratio = 0.97, 95% confidence interval [CI]: 0.95–0.98). Compared to participants whose PNI was in the lowest quartile, those in other quartiles had a reduced likelihood of having RA by 21-38% (Ptrend <0.01). Cox regression analysis revealed an inverse association between PNI and all-cause mortality (hazard ratio = 0.95, 95%CI: 0.91–0.99). An L-shaped association was observed between PNI and the presence and all-cause mortality of RA, with turning points occurring around the mean value of PNI. The presence and all-cause mortality of RA was significantly reduced before the turning points of PNI and plateaued afterwards.ConclusionIn middle-aged and older adults, there is an inverse association between PNI and the presence and all-cause mortality of RA.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12889-024-20795-0. Background Rheumatoid arthritis (RA) is a systemic inflammatory autoimmune disease, affecting about 1% of adults worldwide [1]. The incidence of RA increases in people aged ≥ 40 years [2]. According to the latest Global Burden of Disease study, about 31.7 million people worldwide will suffer from RA by 2050 [3]. As a global public health problem, the burden of RA on patients and society is enormous. Pain, swelling, and joint damage are common manifestations of RA [4], which can also lead to depression [5], increase the risk of cardiovascular disease [6], and reduce quality of life [7]. In addition to these, RA cannot be completely cured, disability and early death is the primary outcome for RA, which impose a huge socioeconomic burden [8]. The pathological mechanism of RA is complex and not fully understood. Growing evidence has shown that immunity and nutrition are closely related to the development and progression of RA [9, 10]. In RA patients, a variety of immune cells induce adaptive immune system activation, which is contributed to the development of RA, such as inducing joint erosions, promoting synovial angiogenesis and tissue remodeling and repair [11]. Besides, nutritional status also has an impact on the prognosis of RA. RA patients have poor nutritional status compared to normal individuals [12]. Previous study showed that malnutrition was strongly associated with increased all-cause mortality in RA patients [13]. Thus, both immunity and nutritional status need to be considered in the diagnosis and prognosis of RA. At present, there is no unified diagnostic standard for RA, and the diagnosis is mainly based on joint symptoms and serum markers [14]. However, autoantibodies (serum markers used to diagnose RA) are negative in one-third of RA patients [15], and the symptoms of RA are insidious and tend to worse over time [16]. Thus, there is unmet need in identifying markers for the diagnosis and prognosis of RA. The prognostic nutritional index (PNI) is an indicator to reflect the immune and nutritional status of an individual, which is calculated by serum albumin and lymphocyte count [17]. PNI is often utilized as a prognostic marker in cancer patients, based on the inversely correlation between PNI and the risk of death [18, 19]. In a small study of patients with RA (n = 138), PNI was closely related to RA disease activity [20]. However, it is unclear whether PNI plays a role in the presence and all-cause mortality of RA. Therefore, this study aimed to evaluate the association between PNI and the presence and all-cause mortality of RA in a national representative cohort in the US. Materials and methods Data source and study participants This study was reported following the guidelines of the Strengthening the Reporting of Observational studies in Epidemiology (STROBE) statement [21]. The National Health and Nutrition Examination Survey (NHANES) is an ongoing survey conducted by the National Center for Health Statistics (NCHS) to describe the health and nutritional status of American, and the Research Ethics Review Board of NCHS approved the protocols of NHANES [22]. Our study used data on demographic, laboratory, questionnaire, and death databases of NHANES from 2003 to 2018. We excluded some ineligible participants based on the following criteria: (1) age less than 40 years; (2) insufficient data for the calculation of PNI; (3) the data for diagnosing arthritis was missing; (4) patients diagnosed with other arthritis. Eventually, a total of 18,245 participants were selected in our study (16,344 non-RA individuals and 1,901 RA patients). In survival analysis conducted in patients with RA, we excluded two participants without data on death. The sample selection flowchart is presented in Fig. 1. Fig. 1Flowchart of the sample selection from NHANES 2003–2018 Flowchart of the sample selection from NHANES 2003–2018 Exposure: PNI PNI was firstly proposed by Buzby et al., which was calculated using serum albumin and lymphocyte count [17]. The bromocresol violet dye method was performed to measure serum albumin levels, and the Beckman Coulter method of counting and sizing was utilized to determine lymphocyte count [23]. PNI = 10 * serum ALB (g/dL) + 5 * lymphocyte count (109/L). Outcome: RA RA was assessed through interview data in health questionnaires. Participants were asked two arthritis-related questions (i.e. “Doctor ever said you had arthritis?” and “What type of arthritis?”). Those who responded to the interviewer and without arthritis were defined as non-RA, and those who self-reported RA were defined as having RA. Although the definition of RA was self-reported rather than based on clinical examination, the consistency between them was considered acceptable for the diagnosis accuracy of arthritis [24]. Outcome: all-cause mortality Date of death was determined using the NHANES Public Use-Related Mortality data as of December 31, 2019, which linked the NCHS with the National Death Index (NDI) through a probabilistic matching algorithm. Covariates We selected covariates related to RA based on previously publish research [25–27]. In addition, a Directed Acyclic Graph was used to identify potential confounders for the association between PNI and RA (Figure S1). The identified confounders included age (year), sex, race, educational levels, poverty-income ratio (PIR), body mass index (BMI, kg/m2), diabetes, hypertension, physically active, smoking status, and drinking status. Educational levels were divided into lower than high school, high school or equivalent, and college or above. PIR was categorized as low income (< 1.3), middle income (1.3–3.5) and high income (> 3.5). Diabetes and hypertension were self-reported. Physically active was described as any vigorous or moderate activity of at least 10 min in the past 30 days or work involving high-intensity or moderate-intensity activity that cause a significant increase in breathing or heart rate. Smoking status was classified as never smoking (smoking fewer than 100 cigarettes in their entire life), former smoking (smoking more than 100 cigarettes but currently not smoking), and current smoking (smoking more than 100 cigarettes and currently smoking). Drinking status was specified as no (drinking lower than 12 alcohol drinks in entire life), moderate (males: ≤ 2 drinks per day, female: ≤ 1 drink per day), and vigorous (male: > 2 drinks per day, female: > 1 drink per day). Statistical analysis We considered masked variance and used the weighting methodology in analyses [28]. Characteristics of participants were described and compared according to the presence or absence of RA. Baseline characteristics were compared between RA and non-RA participants by performing chi-square tests and t-tests. Weighted multivariable logistic regression analysis and weighted multivariable Cox regression were employed to explore the association between PNI (as continuous variable or quartiles) and the presence and all-cause mortality of RA, respectively. The nonlinear association between PNI (as continuous variable) and the presence and all-cause mortality of RA was explored using restricted cubic splines (RCS), with the 5th, 35th, 65th and 95th percentiles determined as four knots of the splines. Subgroup analysis was conducted splitting by the turning point identified by RCS. The interaction between PNI and each of the baseline characteristics for the presence and all-cause mortality of RA was assessed. In sensitivity analysis, multiple imputation with chained equations (MICE) was performed to handle missing data on covariates, with five imputations conducted assuming missing at random [29]. All analyses were performed by using R software version 4.3.2 (http://www.R-project.org) and P < 0.05 (two-sided) was considered statistically significant. Results The baseline characteristics of participants Among 18,245 participants included in this study, 1901 had RA. There were significant differences in baseline characteristics between RA and non-RA participants (Table 1). For example, RA patients were older and physically more active, had higher BMI and lower PNI and PIR, more likely to be women, smoker, and to comorbid with diabetes and hypertension. Table 1Baseline characteristics of the participantsCharacteristicTotalNon-RARAP value18,24516,3441901Age (mean (SD)), years55.44 (11.40)54.96 (11.24)60.98 (11.73)< 0.01 Age (%) < 0.01Age < 6010,308 (67.8)9636 (69.5)672 (47.4)Age ≥ 607937 (32.2)6708 (30.5)1229 (52.6) Sex (%) < 0.01male9627 (51.3)8822 (52.0)85 (42.5)female8618 (48.7)7522 (48.0)1096 (57.5) Race (%) < 0.01Mexican American3086 (7.4)2786 (7.4)300 (7.1)Other Hispanic1725 (4.9)1555 (4.9)170 (4.7)Non-Hispanic White7539 (69.8)6784 (70.2)755 (65.6)Non-Hispanic Black3859 (10.4)3296 (9.9)563 (16.7)Other Race2036 (7.5)1923 (7.6)113 (5.8) Educational levels (%) < 0.01Lower than high school5062 (16.5)4393 (15.8)669 (24.4)High school or equivalent4108 (23.7)3649 (23.3)459 (28.0)College or above9050 (59.9)8279 (60.9)771 (47.6) PIR (%) < 0.01Low income2995 (10.6)2554 (9.9)441 (18.9)Moderate income6957 (33.1)6133 (32.3)824 (43.0)High income6696 (56.3)6220 (57.9)476 (38.1)BMI (mean (SD)), kg/m228.73 (6.16)28.59 (6.04)30.33 (7.29)< 0.01 BMI (%) < 0.01< 254930 (28.2)4526 (28.7)404 (23.2)25–306618 (36.8)6058 (37.4)560 (30.4)≥ 306420 (35.0)5524 (34.0)896 (46.4) Diabetes (%) < 0.01No14,883 (86.2)13,570 (87.1)1313 (75.4)Borderline455 (2.2)385 (2.1)70 (3.1)Yes2894 (11.6)2378 (10.7)516 (21.5) Hypertension (%) < 0.01No10,606 (63.2)9912 (65.0)694 (42.6)Yes7603 (36.8)6398 (35.0)1205 (57.4) Physically active (%) 0.02No10,595 (52.2)9473 (51.9)1122 (56.0)Yes7555 (47.8)6808 (48.1)747 (44.0) Smoking status (%) < 0.01Never9729 (53.5)8865 (54.5)864 (42.3)Former5054 (28.0)4430 (27.5)624 (34.3)Current3448 (18.5)3036 (18.1)412 (23.4) Drinking status (%) < 0.01No2458 (13.4)2163 (13.1)295 (17.6)Moderate5835 (48.2)5291 (48.4)544 (45.8)Vigorous4678 (38.3)4252 (38.5)426 (36.6) PNI (mean (SD)) 52.74 (7.41)52.85 (7.46)51.54 (6.65)< 0.01RA, rheumatoid arthritis; PIR, poverty income ratio; BMI, body mass index; PNI, prognostic nutritional index Table 1Baseline characteristics of the participantsCharacteristicTotalNon-RARAP value18,24516,3441901Age (mean (SD)), years55.44 (11.40)54.96 (11.24)60.98 (11.73)< 0.01 Age (%) < 0.01Age < 6010,308 (67.8)9636 (69.5)672 (47.4)Age ≥ 607937 (32.2)6708 (30.5)1229 (52.6) Sex (%) < 0.01male9627 (51.3)8822 (52.0)85 (42.5)female8618 (48.7)7522 (48.0)1096 (57.5) Race (%) < 0.01Mexican American3086 (7.4)2786 (7.4)300 (7.1)Other Hispanic1725 (4.9)1555 (4.9)170 (4.7)Non-Hispanic White7539 (69.8)6784 (70.2)755 (65.6)Non-Hispanic Black3859 (10.4)3296 (9.9)563 (16.7)Other Race2036 (7.5)1923 (7.6)113 (5.8) Educational levels (%) < 0.01Lower than high school5062 (16.5)4393 (15.8)669 (24.4)High school or equivalent4108 (23.7)3649 (23.3)459 (28.0)College or above9050 (59.9)8279 (60.9)771 (47.6) PIR (%) < 0.01Low income2995 (10.6)2554 (9.9)441 (18.9)Moderate income6957 (33.1)6133 (32.3)824 (43.0)High income6696 (56.3)6220 (57.9)476 (38.1)BMI (mean (SD)), kg/m228.73 (6.16)28.59 (6.04)30.33 (7.29)< 0.01 BMI (%) < 0.01< 254930 (28.2)4526 (28.7)404 (23.2)25–306618 (36.8)6058 (37.4)560 (30.4)≥ 306420 (35.0)5524 (34.0)896 (46.4) Diabetes (%) < 0.01No14,883 (86.2)13,570 (87.1)1313 (75.4)Borderline455 (2.2)385 (2.1)70 (3.1)Yes2894 (11.6)2378 (10.7)516 (21.5) Hypertension (%) < 0.01No10,606 (63.2)9912 (65.0)694 (42.6)Yes7603 (36.8)6398 (35.0)1205 (57.4) Physically active (%) 0.02No10,595 (52.2)9473 (51.9)1122 (56.0)Yes7555 (47.8)6808 (48.1)747 (44.0) Smoking status (%) < 0.01Never9729 (53.5)8865 (54.5)864 (42.3)Former5054 (28.0)4430 (27.5)624 (34.3)Current3448 (18.5)3036 (18.1)412 (23.4) Drinking status (%) < 0.01No2458 (13.4)2163 (13.1)295 (17.6)Moderate5835 (48.2)5291 (48.4)544 (45.8)Vigorous4678 (38.3)4252 (38.5)426 (36.6) PNI (mean (SD)) 52.74 (7.41)52.85 (7.46)51.54 (6.65)< 0.01RA, rheumatoid arthritis; PIR, poverty income ratio; BMI, body mass index; PNI, prognostic nutritional index Baseline characteristics of the participants RA, rheumatoid arthritis; PIR, poverty income ratio; BMI, body mass index; PNI, prognostic nutritional index Linear and nonlinear associations between PNI and the presence of RA As shown in Table 2, there was an inverse association between PNI and RA before and after adjusting for confounders (odds ratio [OR] = 0.97, 95% confidence interval [CI]: 0.95–0.98). Compared to the participants whose PNI was in the lowest quartile (Q1) group, the likelihood of having RA was reduced by 21-38% in higher quartile groups (P for trend < 0.01). Table 2Cross-sectional association between PNI and the presence of RARA/TotalUnadjustedAdjustedOR (95%CI)P valueOR (95%CI)P valuePNI (continuous)1901/18,2450.95 [0.93,0.96]< 0.010.97 [0.95,0.98]< 0.01PNI (quartile)Q1 (< 49.5)666/4548RefRefQ2 (49.5–52.5)492/46060.68 [0.56,0.83]< 0.010.79 [0.59,1.06]0.11Q3 (52.5–55.5)370/43700.52 [0.43,0.63]< 0.010.66 [0.52,0.85]< 0.01Q4 (≥ 55.5)373/47210.49 [0.40,0.59]< 0.010.62 [0.48,0.80]< 0.01P for trend< 0.01< 0.01Adjusted: adjusted for age, sex, race, educational levels, PIR, BMI, diabetes, hypertension, physically active, smoking status, drinking statusQ1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile; OR, odds ratio; CI, confidence intervalRA, rheumatoid arthritis; PNI, prognostic nutritional index Table 2Cross-sectional association between PNI and the presence of RARA/TotalUnadjustedAdjustedOR (95%CI)P valueOR (95%CI)P valuePNI (continuous)1901/18,2450.95 [0.93,0.96]< 0.010.97 [0.95,0.98]< 0.01PNI (quartile)Q1 (< 49.5)666/4548RefRefQ2 (49.5–52.5)492/46060.68 [0.56,0.83]< 0.010.79 [0.59,1.06]0.11Q3 (52.5–55.5)370/43700.52 [0.43,0.63]< 0.010.66 [0.52,0.85]< 0.01Q4 (≥ 55.5)373/47210.49 [0.40,0.59]< 0.010.62 [0.48,0.80]< 0.01P for trend< 0.01< 0.01Adjusted: adjusted for age, sex, race, educational levels, PIR, BMI, diabetes, hypertension, physically active, smoking status, drinking statusQ1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile; OR, odds ratio; CI, confidence intervalRA, rheumatoid arthritis; PNI, prognostic nutritional index Cross-sectional association between PNI and the presence of RA Adjusted: adjusted for age, sex, race, educational levels, PIR, BMI, diabetes, hypertension, physically active, smoking status, drinking status Q1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile; OR, odds ratio; CI, confidence interval RA, rheumatoid arthritis; PNI, prognostic nutritional index The RCS plot (Fig. 2A) indicated a nonlinear relationship between PNI and RA (P for nonlinear < 0.01). In subgroup analysis, we fitted two weighted logistic regressions before and after the turning point (Table 3). Before the turning point (PNI = 52.1), the likelihood of having RA was reduced by 5% for each unit increase in PNI (OR = 0.95, 95%CI: 0.91–0.99, P = 0.02), whereas the relation was not statistically significant after the turning point (OR = 0.99, 95%CI: 0.98–1.01, P = 0.47). Fig. 2The nonlinear association between PNI and (A) the presence of RA, (B) mortality among RA patients The nonlinear association between PNI and (A) the presence of RA, (B) mortality among RA patients Table 3Cross-sectional associations between PNI and the presence of RA by the turning point of PNIPNIOR (95%CI)P value< 52.10.95 [0.91,0.99]0.02≥ 52.10.99 [0.98,1.01]0.47RA, rheumatoid arthritis; PNI, prognostic nutritional index; OR, odds ratio; CI, confidence interval Table 3Cross-sectional associations between PNI and the presence of RA by the turning point of PNIPNIOR (95%CI)P value< 52.10.95 [0.91,0.99]0.02≥ 52.10.99 [0.98,1.01]0.47RA, rheumatoid arthritis; PNI, prognostic nutritional index; OR, odds ratio; CI, confidence interval Cross-sectional associations between PNI and the presence of RA by the turning point of PNI RA, rheumatoid arthritis; PNI, prognostic nutritional index; OR, odds ratio; CI, confidence interval There were no significant interactions between PNI and any categorical covariates for the presence of RA (all P for interaction > 0.05). Linear and nonlinear associations between PNI and all-cause mortality of RA During a median follow-up duration of 84 months (IQR: 41.5-132.5 months), a total of 480 deaths were recorded among 1899 RA patients (25.28%). Table S1 shows the baseline characteristics of these RA patients. After adjusting for all confounders, each unit increase in PNI was associated with a 5% reduction in all-cause mortality among RA patients (adjusted hazard ratio [HR] = 0.95, 95%CI: 0.91–0.99). Compared to patients whose PNI was the lowest quartile (Q1) group, those who had higher PNI showed a decreased risk of all-cause mortality by 34-51% (P for trend = 0.03, Table 4). Table 4Association between PNI and all-cause mortality among RA patientsDeath/TotalUnadjustedAdjustedHR (95%CI)P valueHR (95%CI)P valuePNI (continuous)480/18990.93 [0.90, 0.96]< 0.010.95 [0.91, 0.99]0.02PNI (quartile)Q1 (< 48.0)180/461RefRefQ2 (48.0-51.5)123/5090.62 [0.46,0.84]< 0.010.49 [0.31,0.79]< 0.01Q3 (51.5–54.5)92/4420.48 [0.33,0.70]< 0.010.66 [0.40,1.09]0.10Q4 (≥ 54.5)85/4870.37 [0.27,0.52]< 0.010.54 [0.34,0.84]0.01P for trend< 0.010.03Adjusted: adjusted for age, sex, race, educational levels, PIR, BMI, diabetes, hypertension, physically active, smoking status, drinking statusQ1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile; HR, hazard ratio; CI, confidence intervalRA, rheumatoid arthritis; PNI, prognostic nutritional index Table 4Association between PNI and all-cause mortality among RA patientsDeath/TotalUnadjustedAdjustedHR (95%CI)P valueHR (95%CI)P valuePNI (continuous)480/18990.93 [0.90, 0.96]< 0.010.95 [0.91, 0.99]0.02PNI (quartile)Q1 (< 48.0)180/461RefRefQ2 (48.0-51.5)123/5090.62 [0.46,0.84]< 0.010.49 [0.31,0.79]< 0.01Q3 (51.5–54.5)92/4420.48 [0.33,0.70]< 0.010.66 [0.40,1.09]0.10Q4 (≥ 54.5)85/4870.37 [0.27,0.52]< 0.010.54 [0.34,0.84]0.01P for trend< 0.010.03Adjusted: adjusted for age, sex, race, educational levels, PIR, BMI, diabetes, hypertension, physically active, smoking status, drinking statusQ1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile; HR, hazard ratio; CI, confidence intervalRA, rheumatoid arthritis; PNI, prognostic nutritional index Association between PNI and all-cause mortality among RA patients Adjusted: adjusted for age, sex, race, educational levels, PIR, BMI, diabetes, hypertension, physically active, smoking status, drinking status Q1, first quartile; Q2, second quartile; Q3, third quartile; Q4, fourth quartile; HR, hazard ratio; CI, confidence interval RA, rheumatoid arthritis; PNI, prognostic nutritional index RCS results showed that there was an L-shaped correlation between PNI and all-cause mortality among RA patients (Fig. 2B). Before the turning point (PNI = 51.1), the risk of mortality was decreased by 13% for each unit increase in PNI (HR = 0.87, 95%CI: 0.81–0.94), whereas this relationship was not significant after the turning point (HR = 0.97, 95%CI: 0.88–1.07). There were no significant interactions between PNI and any categorical covariates for all-cause mortality of RA (all P for interaction > 0.05). Sensitivity analysis After multiple imputations for missing covariates, the relationship between PNI and the presence of RA as well as all-cause mortality among RA patients remained largely consistent (Table S2, Table S3). Discussion Using data from a national representative study in the US, this study showed an inverse correlation between PNI and the risk and all-cause mortality of RA in middle-aged and older adults after adjusting for common risk factors (including age, sex, race, educational levels, PIR, BMI, diabetes, hypertension, smoking status, and drinking status). Furthermore, we found an L-shaped relationship between PNI and the presence and all-cause mortality of RA, which indicated that the protective effect of PNI has a certain threshold. The turning point was around the mean level of PNI, which was consistent with other studies that evaluated the association between PNI and other diseases [30, 31]. Specifically, a lower PNI increased the likelihood of having RA as well as the risk of all-cause mortality in RA patients only if PNI was below the turning point (close to the mean level of PNI). These findings suggested that PNI, as a simple combination of serum albumin and lymphocyte count, might provide insight into the diagnosis and prognosis of RA. Importantly, immune and nutritional factors play an important role in the pathogenesis of RA, and the management of RA should take these factors into account. There are several potential mechanisms underlying the association between PNI and RA. Existing evidence has shown that serum albumin levels are significantly lower in RA patients compared to the normal population [32]. Serum albumin is an important indicator to evaluate the nutritional status of individuals, which is mainly synthesized in the liver and has anti-inflammatory and antioxidant effects [33]. An increase in reactive oxygen species will destroy proteins, nucleic acids and other substances in the body, resulting in intracellular signal transduction disorders and inducing inflammation [34]. Lymphocytes are mainly produced by lymphoid organs, which play an important role in the immune response and are a key component in the pathogenesis of autoimmune disease [35]. This is important because RA is essentially a progressive systemic autoimmune disease characterized by ongoing immune dysregulation [36]. Immunity and nutritional status should be considered comprehensively for the diagnosis and prognosis of RA. Firstly, RA is a chronic inflammatory disease [1], immunity permeates the entire course of the disease. Secondly, RA patients are more susceptible to malnutrition as the disease progressed [37]. Moreover, the relationship between inflammation and nutritional status is bidirectional, with inflammation causing insulin resistance and stimulating catabolism, which can lead to malnutrition [38]. Malnutrition contributes to impair mitochondrial function, which enhances the production of reactive oxygen species and activates inflammatory pathways [39]. At present, the mechanism of therapeutic drugs for RA is mainly to block or delay the immune and inflammatory response [14]. For nutrition, most studies only consider the role of dietary factors in prevention and management of RA [40–42]. Our findings suggest that both immune and nutritional factors should be considered when targeting RA. Our study has the following advantages. Firstly, the data were derived from a large sample of the general population, with complex sampling methods applied; therefore, the results have good generalizability. Secondly, potential confounders were identified through epidemiological evidence and a Directed Acyclic Graph, making the association between PNI and RA more robust. Lastly, this study has a long follow-up time (median 7 years), which better reflected the mortality risk in RA patients. There are also some limitations in the present study. Firstly, the diagnosis of RA came from the questionnaire survey, therefore misclassification bias cannot be ruled out. Secondly, the chronological order between PNI and the presence of RA cannot be certain in our cross-sectional analysis. Future longitudinal studies are warranted to determine whether PNI predicts the onset and progression of RA. Conclusion In middle-aged and older adults, there is an inverse association between PNI and the presence and all-cause mortality of RA. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1: Table S1. Baseline characteristics of rheumatoid arthritis. Table S2. Logistic regression analysis on the association between PNI and RA in sensitivity analysis. Table S3. Cox regression analysis on the association between PNI and rheumatoid arthritis in sensitivity analysis. Supplementary Material 1: Table S1. Baseline characteristics of rheumatoid arthritis. Table S2. Logistic regression analysis on the association between PNI and RA in sensitivity analysis. Table S3. Cox regression analysis on the association between PNI and rheumatoid arthritis in sensitivity analysis. Supplementary Material 2: Figure S1. Directed acyclic plot (DAG) between prognostic nutritional index and rheumatoid arthritis (RA). Supplementary Material 2: Figure S1. Directed acyclic plot (DAG) between prognostic nutritional index and rheumatoid arthritis (RA).