High Estimated 24-Hour Urinary Sodium Excretion Is Related to Symptomatic Knee Osteoarthritis: A Nationwide Cross-Sectional Population-Based Study

Abstract

Objectives

High salt intake results in various harmful effects on human health including hypertension, cardiovascular disease, and reduced bone density. Despite this, there are very few studies in the literature that have investigated the association between sodium intake and osteoarthritis (OA). Therefore, we aimed to explore these associations in a Korean population.

Methods

This study used cross-sectional data from adult subjects aged 50–75 years from two consecutive periods of the Korean National Health and Nutrition Examination Survey V–VII (2010–2011 and 2014–2016). The estimated 24-hour urinary sodium excretion (24HUNa) was used as a surrogate marker of salt intake. In the 2010–2011 dataset, knee OA (KOA) was defined as the presence of the radiographic features of OA and knee pain. The association between KOA and salt intake was analysed using univariable and multivariable logistic regression methods. For the sensitivity analysis, the same procedures were conducted on subjects with self-reported OA (SR-OA) with knee pain in the 2010–2011 dataset and any site SR-OA in the 2014–2016 dataset.

Results

Subjects with KOA had significantly lower energy intake, but higher 24HUNa than those without KOA. The restricted cubic spline plots demonstrated a J-shaped distribution between 24HUNa and prevalent KOA. When 24HUNa was stratified into five groups (<2, 2–3, 3–4, 4–5 and ≥5 g/day), subjects with high sodium intake (≥5 g/day) had a higher risk of KOA (odds ratio [OR] = 1.64, 95% confidence interval [CI] 1.03–2.62) compared to the reference group (3–4 g/day) after adjusting for covariates. The sensitivity analysis based on SR-OA with knee pain showed that high sodium intake was also significantly associated with increased prevalence of OA (OR = 1.84, 95% CI 1.10–3.10) compared with the reference group. Regarding SR-OA at any site in the 2014–2016 dataset, estimated 24HUNa showed a significantly positive association with the presence of SR-OA after adjusting for potential confounders.

Conclusions

This nationwide Korean representative study showed a significant association between symptomatic KOA and high sodium intake (≥5 g/day). Avoidance of a diet high in salt might be beneficial as a non-pharmacologic therapy for OA.

Introduction

Osteoarthritis (OA), the most common joint disease, is characterised by progressive cartilage loss, subchondral bone sclerosis, osteophyte formation, and joint inflammation (1). Although OA has been traditionally classified as a non-inflammatory disease, the current understanding of OA pathogenesis is that chronic low-grade inflammation and oxidative stress play a pivotal role in the development or progression of OA (2). Its prevalence and socioeconomic burden have been increasing worldwide in accordance with a growing global elderly and/or obese population (3). Data from the Global Burden of Diseases, Injuries, and Risk Factors Study 2017 shows that the age-standardised prevalence, annual incidence, and years lived with disability (YLD) rate of OA were estimated to be 3,754 (95% uncertainty intervals [UI] 3389–4188), 181 (95% UI 163–202) and 119 (95% UI 60–236) per 100,000, respectively; moreover, these values increased by nearly 9% between 1990 and 2017 (4).

Although the aetiology of OA is complex and has not been fully elucidated, age, female sex, obesity, physical activity, and previous joint injury are considered risk factors. Additionally, dietary or nutritional components have been reported as factors contributing to the development or progression of OA (5), and these include vitamin C, D, or K, β-carotene, and polyunsaturated fatty acid (6, 7, 8, 9). This may directly affect pain or joint function and the structure of the joint or have an indirect effect through the diet-intestinal microbiota axis (10). Although a single dietary or nutritional factor alone could not be sufficient for OA development, dietary or nutritional supplements have been constantly drawing the attention of OA patients and health care providers because these factors are adjustable.

Sodium intake has been reported to be associated with many health issues, including cardiovascular diseases (CVDs) and bone mineral density. A study estimated that globally, 1.65 million annual deaths from cardiovascular causes were attributed to a sodium intake >2.0 g/day in 2010 (11). Accordingly, the World Health Organization (WHO) suggested a reduction in daily sodium intake of <2 g/day (5 g/day salt) for adults (12). Sodium has been proposed as an inducer of the inflammatory response through various mechanisms including induction of pathogenic CD4+ T helper cells that produce interleukin-17 and inhibition of regulatory T cells (13). Additionally, high salt intake has also been reported to induce excessive oxidative stress (14). However, there have been only few studies on the relationship between OA and dietary sodium intake.

In this study, we aimed to analyse the cross-sectional association between sodium intake and OA, especially focusing on knee OA (KOA) in a Korean population using two consecutive datasets from the Korean National Health and Nutrition Examination Survey (KNHANES) V-VII.

Methods

Study design and population

This is a cross-sectional study based on data obtained during two different time points; KNHANES 2010–2011 and 2014–2016. KNHANES is an ongoing, multicomponent, nationally representative survey of the general Korean population administered by the Korea Centers for Disease Control and Prevention. Its target population comprises nationally representative non-institutionalised Korean citizens and each survey year includes a new sample of approximately 10,000 individuals aged ≥ 1 year. The sampling plan follows a multistage clustered probability design, as previously described (15). In the KNHANES data, inclusion errors owing to differences in the number of households and populations between the sampling design and the survey time, the unequal extraction rate, and the error from non-respondents of the survey were corrected to increase the representativeness and accuracy of the estimates of survey items of Koreans. Accordingly, individual weights were constructed so that participating individuals could represent the entire Korean population. Annual assessment consists of a health interview, a medical examination, and a nutrition survey. Health interview and medical examination were performed at the mobile check-up centre by trained staff members, and the nutrition survey was conducted by dieticians visiting the homes of participants. Investigated variables are slightly different in the each year's survey, as described in Figure 1. Knee radiographs were performed on those aged ≥50 years in the KNHANES V (2010–2012), while urine sodium concentration was not measured in a random urine sample in 2012 and 2013. We chose two consecutive analysis groups; 1) 2010–2011 dataset available for both knee radiographs and random urine sodium levels (Period A), 2) 2014–2016 dataset available for random urine sodium levels but not knee radiographs (Period B). In both groups, participants aged from 50 to 75 years were included. Subjects with implausible data on daily energy intake (<500 kcal or > 5,000 kcal) were excluded. We also excluded patients with liver cirrhosis, malignancies, renal insufficiency (serum creatinine >2.0 mg/dL) and missing sodium values. After applying the criteria mentioned above, 4,588 participants from the Period A dataset and 5,987 participants from the Period B dataset were finally included in the present analyses.

Figure 1.

Availability of the main variables in each year's survey and definitions of osteoarthritis

KNHANES = Korean National Health and Nutrition Examination Survey; OA = osteoarthritis; IPAQ = International Physical Activity Questionnaire; SR-OA = self-reported physician diagnosed osteoarthritis; NA = not available; *, performed on subjected aged □ 50 years; **, knee joint pain lasting more than 30 days during the 3 months prior to the interview; ○, available in the dataset; ●, variables used to define OA; †, presence of knee pain and a knee radiograph of Kellgren—Lawrence (KL) grade ≥2 (radiographic symptomatic knee OA); ‡, participants who answered yes to the question “Have you ever been diagnosed with osteoarthritis by a doctor?”; ¶, SR-OA with the presence of knee pain.

This study was approved by the Institutional Review Board of Seoul National University Bundang Hospital (IRB No. X-1902/524-901). Informed consent was waived because subject consent had already been given for the KNHANES and the dataset was provided after anonymisation.

Definition of OA

Self-reported physician-diagnosed OA (SR-OA) was defined based on the response to the question from the KNHANES Period A and Period B: “Have you ever been diagnosed with osteoarthritis by a doctor?”. SR-OA includes current and past physician-diagnosed OA at any site. Knee pain was defined as positive when the subject had knee joint pain lasting for ≥ 30 days during the 3 months prior to the interview in the KNHANES Period A. Plain knee radiographs were taken using digital X-ray machines (SD3000 Synchro Stand; SYFM, Namyangju, South Korea) in the KNHANES 2010–2012. Bilateral anterior-posterior, lateral (30° flexion), and weight-bearing anterior-posterior radiographs of the knees were obtained. Reading of X-rays was performed by two radiologists, referencing the Kellgren-Lawrence (KL) grading system 0–4 (16). If there was any discrepancy greater than two KL grades between radiologist scores, a third radiologist was consulted.

Because the association between clinical symptoms of OA and radiographic evidence of the disease is relatively weak, various definitions have been used in OA studies (17). In the present study, symptomatic radiographic KOA was defined as the presence of knee pain and a knee radiograph of KL grade of ≥2, using the items included in KNHANES Period A dataset. To check for robustness of the primary findings in symptomatic radiographic KOA, we performed sensitivity analyses using differently defined populations such as SR-OA with knee pain in KNHANES Period A and SR-OA at any site in KNHANES Period B. The control groups included subjects without radiographic or self-reported KOA.

Extraction and processing of variables

Sex, age, body mass index (BMI), educational level, household income, smoking status, hypertension (HTN) medication, diabetes mellitus (DM) and depression were considered as potential confounding variables affecting symptomatic radiographic KOA or SR-OA with knee pain and sodium intake or excretion. Physical activity was estimated by metabolic equivalent minutes (METs) scores, based on subjects' answers on the short-form International Physical Activity Questionnaire (IPAQ) (18). IPAQ data were not available in the KNHANES 2014–2016 dataset (i.e., Period B — subjects with SR-OA).

Sodium-related variables

In the KNHANES, daily sodium intakes were estimated from the 24-hour dietary recall questionnaire using the Food Composition Table developed by the National Rural Resources Development Institute. The 24-hour dietary recall method is reportedly an inaccurate reflection of 24-hour urinary sodium excretion (24HUNa) and tends to underestimate it (19). Although the gold standard for assessing sodium intake is 24HUNa (20), 24-hour urine collection has not been considered feasible in a large-scale population-based study. Therefore, we used Tanaka's formula to calculate 24HUNa (estimated 24HUNa), as a surrogate marker of daily sodium intake, using morning fasting urine Na concentration (21). Tanaka's equation was developed to estimate 24HUNa using Japanese data from the INTERSALT study, which demonstrated a significant correlation between estimated 24HUNa and measured sodium excretion using 24-hour urine collection. The method has been subsequently validated in other populations (22, 23). This is a valid and simple method to estimate sodium intake in a large population study and has been suggested as the best alternative to the 24-hour urine collection method (24). The equation for estimated 24HUNa (mg/day) is based on the urinary sodium (UNa) and creatinine (UCr), as follows:

Statistical analysis

The KNHANES was designed using a complex, stratified, multistage probability-sampling model. We used the weighted values of the health examination survey (wt_itvex) as the weight variable, the variance estimate layer (kstrata) as the stratification variable, and the number of enumeration districts (primary sampling unit) as the cluster information variable. Weighted univariable comparisons between subjects with KOA and controls were performed using Rao-Scott chi-square tests for categorical variables and Student's t-test for continuous variables. Restricted cubic spline (RCS) plots were used to explore the shape of the association between estimated 24HUNa and symptomatic radiographic KOA or SR-OA with knee pain. Estimated 24HUNa was categorised by 1 g levels, as did previous epidemiological studies on sodium intake (25, 26). Subjects were divided into five groups according to 24HUNa as follows: very low (estimated 24HUNa <2.000 g/ day), low (2.000 to 2.999 g/day), reference (3.000 to 3.999 g/ day), moderate (4.000 to 4.999 g/day) and high (≥5.0 g/day). A multivariable logistic regression analysis was conducted to estimate the odds ratio (OR) and 95% confidence interval (CI) to identify factors associated with symptomatic radiographic KOA or SR-OA with knee pain by adjusting covariates showing significant differences in previous univariable analyses. We included the 24HUNa variable in two ways: 1) the continuous value of 24HUNa and 2) five 24HUNa groups as a categorical variable. We considered a p value of <0.05 as statistically significant. Statistical analysis was conducted using R, version 4.1.0 (R Foundations for Statistical Computing, Vienna, Austria) and SAS 9.4 (SAS Institute Inc., Cary, NC, USA)

Results

Characteristics of subjects with symptomatic radiographic KOA

From the Period A dataset, a total of 4,588 subjects, which corresponds to the weighted number of 10,395,909 subjects, were finally analysed, as summarised in Supplementary Figure 1. Values of random spot urine sodium and creatinine levels were missing in 431 participants (weighted n = 570,192) and calculation of the estimated 24HUNa was impossible in 440 subjects because items necessary for the calculation were omitted from the analysis (weighted n = 598,964). The prevalence of symptomatic radiographic KOA was 10.5% (weighted n = 1,214,665) among the included subjects. Comparisons of the baseline characteristics and nutrient variables between subjects with and without symptomatic radiographic KOA are presented in Table 1. Subjects with symptomatic radiographic KOA were older (p<0.001) and predominantly female (p<0.001) and had higher BMI (p<0.001) and lower socioeconomic status (p=0.049). They had more comorbidities, including higher rates of HTN and depression (both p<0.001), but not DM (p=0.199). The proportion of current- or ex-smokers was lower in symptomatic radiographic KOA subjects than in those without symptomatic radiographic KOA (p<0.001). However, subjects with symptomatic radiographic KOA had a lower intake of total energy (p<0.001) and daily dietary sodium (p<0.001), and estimated 24HUNa levels were higher in subjects with symptomatic radiographic KOA than in those without symptomatic radiographic KOA (p=0.037).

Table 1.

Baseline characteristics of the study population and comparisons according to symptomatic radiographic knee osteoarthritis (KOA) from the KNHANES 2010–2011 (Period A) dataset

	Total (n = 11,610,574)	KOA (n = 1,214,665)	Non-KOA (n = 10,395,909)	P-value
Age	60.0 ± 0.2	65.1 ± 0.4	59.4 ± 0.2	<0.001
Sex				<0.001
Men	5,593,559 (48.2)	240,187 (19.8)	5,353,372 (51.5)
Women	6,017,015 (51.8)	974,478 (80.2)	5,042,537 (48.5)
BMI (kg/m2)	24.1 ± 0.1	25.5 ± 0.2	23.9 ± 0.1	<0.001
Physical activity (METs)	3052 ±144	3354 ± 329	3019±139	0.248
Education				<0.001
< Elementary	4,840,737 (41.8)	940,330 (77.7)	3,900,408 (37.6)
Middle-High	5,318,654 (45.9)	246,036 (20.3)	5,072,618 (48.9)
> College	1,424,640 (12.3)	24,039 (2.0)	1,400,602 (13.5)
Income				0.049
Low	2,942,494 (25.6)	368,244 (30.7)	2,574,250 (25.0)
Mid-low	2,994,446 (26.1)	330,620 (27.5)	2,663,826 (25.9)
Mid-High	2,943,101 (25.6)	285,288 (23.8)	2,657,814 (25.8)
High	2611580 (22.7)	216,344 (18.0)	2,395,236 (23.3)
DM diagnosis	1,560,815 (13.4)	189,016 (15.6)	1,371,799 (13.2)	0.199
HTN medication	3,805,387 (32.8)	616,309 (50.7)	3,189,078 (30.7)	<0.001
Depression	1,562,505 (13.5)	309,875 (25.6)	1,252,630 (12.1)	<0.001
Smoking				<0.001
Never-smoker	6,340,137 (54.7)	949,131 (78.2)	5,391,006 (51.9)
Ex-smoker	2,920,945 (25.2)	143,975 (11.9)	2,776,971 (26.8)
Current smoker	2,332,985 (20.1)	120,153 (9.9)	2,212,832 (21.3)
Total energy intake (kcal)	1,963 ± 18	1,670 ± 35	1,998 ± 20	<0.001
Dietary sodium intake (g/day)	4.936 ± 0.069	4.226 ± 0.172	5.027 ± 0.071	<0.001
Estimated 24-hour urinary sodium excretion* (g/day)	3.437 ± 0.018	3.535 ± 0.051	3.425 ± 0.018	0.037

Values are described as mean ± SE or frequency (%); BMI, body mass index; METs, metabolic equivalent minutes; DM, diabetes mellitus; HTN, hypertension; KNHANES, Korean National Health and Nutrition Examination Survey; SE, standard error of the mean; *, 24-hour urinary sodium excretion estimated by Tanaka's method. The equation is described in the text.

Factors associated with symptomatic radiographic KOA

In unadjusted regression analyses, older age, female sex, high BMI and low socioeconomic status showed significant associations with symptomatic radiographic KOA (Table 2). However, physical activity did not show significant associations with symptomatic radiographic KOA. Comorbidities, such as HTN and depression, were significantly associated with the presence of symptomatic radiographic KOA (both p<0.001). The estimated 24HUNa showed a positive association with symptomatic radiographic KOA. After adjusting for all significant variables, the association with the continuous value of estimated 24HUNa did not remain statistically significant.

Table 2.

Univariable and multivariable logistic regression analyses to determine associated factors for symptomatic radiographic knee osteoarthritis from the KNHANES 2010–2011 (Period A)

Variables	Unadjusted OR	95% confidence interval	P value	Adjusted OR†	95% confidence interval	P value
Age	1.109	1.093–1.125	<0.001
Sex
Men	Reference
Women	4.293	3.296–5.592	<0.001
BMI	1.169	1.128–1.213	<0.001
Education
Elementary	Reference
Middle-High	0.200	0.152–0.263	0.127
College	0.072	0.039–0.131	<0.001
Income
Low	Reference
Mid-low	0.859	0.635–1.163	0.469
Mid-high	0.738	0.543–1.001	0.488
High	0.630	0.447–0.889	0.033
Diabetes Mellitus	1.221	0.908–1.643	0.187
Hypertension	2.326	1.864–2.902	<0.001
Depression	2.522	1.964–3.237	<0.001
Smoking
Never	Reference
Ever	0.290	0.216–0.389	<0.001
Current	0.305	0.196–0.475	0.017
Physical activity (METs)	1.000	1.000–1.000	0.177
Total energy intake (kcal)	0.999	0.999–1.000	<0.001
Dietary sodium intake (g/day)	0.991	0.985–0.996	<0.001	1.003	0.999–1.007	0.181
Estimated 24-hour urinary sodium excretion* (g/d)	1.015	1.001–1.029	0.040	1.005	0.991–1.019	0.513

BMI, body mass index; KNHANES, Korean National Health and Nutrition Examination Survey; METs, metabolic equivalent minutes; OR, odds ratio; *, 24-hour urinaiy sodium excretion estimated by Tanaka's method. The equation is described in the text; †, Adjusted for total energy intake, age, sex, body mass index, education level, income, use of hypertension medication, and smoking

Distribution of estimated 24HUNa and its relationship with symptomatic radiographic KOA

To explore the relevant pattern of estimated 24HUNa with symptomatic radiographic KOA, we drew RCS plots, as shown in Figure 2. The OR for symptomatic radiographic KOA showed a J-shaped association with estimated 24HUNa and gradually increased at approximately 3.5 g/day of estimated 24HUNa. Based on the distribution, we set the 3–3.999 g/day group as the reference group in the subsequent analyses.

Figure 2.

Restricted cubic spline plot for the association between estimated 24-hour urinary sodium excretion and symptomatic radiographic knee osteoarthritis (KOA) from the KNHANES 2010–2011 (Period A)

To determine the relationship between estimated 24HUNa levels and symptomatic radiographic KOA, we performed multiple regression analyses using their categorical variables with adjustment for all significant covariates (Table 3). Subjects with the highest estimated 24HUNa (≥5.0 g/day) level were more likely to have symptomatic radiographic KOA than the reference group (OR 1.64, 95% CI 1.03–2.62, p<0.05). However, in the correlation analysis, knee pain intensity and 24HUNa in the Period A dataset were not significantly associated with each other.

Table 3.

Association of estimated 24-hour urinary sodium excretion with symptomatic radiographic knee osteoarthritis (KOA) from the KNHANES 2010–2011 (Period A) dataset

24HUNa Group	Total	KOA, n (%)	Non-KOA, n (%)	Univariable†	Multivariable‡
<2	418,434	64,622 (15.9)	353,812 (84.1)	1.61 (0.85–3.06)	1.47 (0.74–2.91)
2 – 2.999	3,209,243	278,855 (8.8)	2,930,388 (91.2)	0.91 (0.68–1.23)	0.89 (0.65–1.22)
3 – 3.999	5,249,601	502,473 (9.7)	4,747,128 (90.3)	Reference	Reference
4 – 4.999	2,392,179	273,133 (11.6)	2,119,046 (88.4)	1.20 (0.91–1.58)	1.02 (0.77–1.35)
≥ 5	472,867	95,582 (20.5)	377,285 (79.5)	2.17 (1.39–3.38)*	1.64 (1.03–2.62)*

Values are presented as odds ratios (95% confidence intervals); 24HUNa, estimated 24-hour urinary sodium excretion (g/day); KNHANES, Korean National Health and Nutrition Examination Survey; KOA, knee osteoarthritis; *P<0.05, p values from univariable and multivariable logistic regression with KOA as the dependent variable; † Univariable analysis adjusted for age and sex; ‡ Multivariable analysis adjusted for total energy intake, age, sex, body mass index, education level, use of hypertension medication, depression, and smoking

Association between m24UHNa and SR-OA with knee pain

To confirm whether the above results were obtained for other definitions of OA, we performed similar analyses using SR-OA with knee pain in the same dataset. Among the study population, SR-OA with knee pain accounted for 11% of the total subjects (weighted n = 10,874,419). The characteristics of SR-OA with knee pain were not different from those of symptomatic radiographic KOA (Supplementary Table 1). Subjects with SR-OA and knee pain had higher estimated 24HUNa despite a lower energy intake. However, after adjusting for potential confounders, the association between estimated 24HUNa and the presence of SR-OA with knee pain was insignificant (Supplementary Table 2). When estimated 24HUNa levels were stratified into five groups, the highest level (≥5 g/day) was significantly associated with SR-OA with knee pain (OR 1.84, 95% CI 1.10–3.10, p<0.05) compared with the reference level (3–4 g/day, Table 4). This finding was consistent with the results for symptomatic radiographic KOA.

Table 4.

Association of estimated 24-hour urinary sodium excretion with self-reported osteoarthritis with knee pain from the KNHANES 2010–2011 (Period A) dataset

24HUNa Group	Total	KOA, n (%)	Non-KOA, n (%)	Univariable†	Multivariable‡
<2	391,758	52,089 (13.2)	339,669 (86.8)	1.24 (0.60–2.57)	1.02 (0.46–2.26)
2 – 2.999	2,960,875	286,766 (9.7)	2,674,109 (90.3)	0.97 (0.70–1.34)	0.96 (0.70–1.31)
3 – 3.999	4,899,516	502,399 (10.3)	4,397,117 (89.7)	Reference	Reference
4 – 4.999	2,180,902	257,632 (11.8)	1,923,270 (88.2)	1.17 (0.89–1.55)	1.01 (0.75–1.36)
≥ 5	431,368	98,486 (22.8)	332,882 (77.2)	2.33 (1.44–3.77)*	1.84 (1.10–3.10)*

Values are presented as odds ratios (95% confidence intervals); 24HUNa, estimated 24-hour urinary sodium excretion (g/day); KNHANES, Korean National Health and Nutrition Examination Survey; KOA, knee osteoarthritis; *P<0.05, p values from univariable and multivariable logistic regression with self-reported osteoarthritis with knee pain as the dependent variable; † Univariable analysis adjusted for age and sex; ‡ Multivariable analysis adjusted for total energy intake, age, sex, body mass index, education level, use of hypertension medication, depression, and smoking

Association between m24UHNa and SR-OA

Next, we conducted analyses on SR-OA to investigate the relationship between OA at any site and m24UHNa. Among the study population, SR-OA accounted for 17.8% (weighted n = 2,075,275) and 17% (weighted n = 2,193,783) of the total subjects in the KNHANES Period A and Period B datasets, respectively. The characteristics of SR-OA subjects in the KNHANES Period A and Period B datasets were almost similar to those of symptomatic radiographic KOA in the Period A dataset (Supplementary Table 3 and 4). Subjects with SR-OA in the KNHANES Period A did not show any significant association with estimated 24HUNa level after adjusting for covariates (Supplementary Table 5). However, the OR for SR-OA also showed a J-shaped association with estimated 24HUNa (Supplementary Figure 2A). The SR-OA group in the KNHANES Period B had lower dietary sodium intake and higher estimated 24HUNa. The RCS plot of the OR for SR-OA revealed a positive linear association with estimated 24HUNa (Supplementary Figure 2B). Additionally, multiple logistic regression analysis showed that estimated 24HUNa was positively associated with the presence of SR-OA after adjusting for potential confounders (OR 1.01, p=0.046, Supplementary Table 6).

Discussion

This large cross-sectional study provides the first evidence of an association between the highest estimated 24HUNa level (≥5g /day) and symptomatic KOA. Moreover, our results suggest that a higher estimated 24HUNa level may be associated with SR-OA at any site, regardless of the presence of knee pain.

OA is a highly prevalent and disabling disease, and globally, KOA affects 16% of people aged ≥15 years and 23% of people aged ≥40 years (27). Although the prevalence and incidence of this high impact disease are expected to rise globally, the pharmacologic treatment options for OA are currently quite limited. Joint replacement or fusion are indicated as final treatments for patients with advanced OA. Therefore, research is ongoing to identify an effective tool for preventing OA development, modifying its course or relieving symptoms.

High sodium intake has been a serious and important public health issue because it is associated with various diseases including HTN and CVDs, obesity, and osteoporosis (28, 29, 30). Despite the obvious deleterious effects of high salt intake, there have been few studies on the association between salt intake and OA to date. In a recent cross-sectional study from the US NHANES 2011–2016, Matsunaga et al. reported no significant association between OA and high sodium intake (31). On the contrary, in our study, estimated 24HUNa, as a surrogate marker of daily sodium intake, was positively associated with symptomatic radiographic KOA and SR-OA at any site. The discrepancy between our results and that of Matsunaga et al. may have been the result of different research methods. They defined OA subjects as those who self-reported physician-diagnosed OA, they calculated salt intake by using 24-hour dietary recalls and simply stratified salt intake levels into three tertiles. Additionally, the prevalence of SR-OA differs between the two studies: 7.0% in the US NHANES study and 10.5–17.0% in our KNHANES study. The prevalence of OA can depend on the definition of OA, the age or sex distribution of the study population, race/ethnicity of the study population or the accessibility of medical services. The previous Korean community-based cohort studies reported that the prevalence of symptomatic radiographic KOA was 5.4–24.2% in subjects aged ≥40–50 years (32, 33). These figures were in line with our results in the present study.

As shown in Figure 2, the RCS plot for the association between KOA and 24HUNa demonstrated a J-shaped curve, with a gradual increase of 24HUNa at approximately 3.5 g/ day. This pattern was similar to that reported in previous studies regarding the relationship between sodium intake and cardiovascular events or mortality, showing a J- or U-shaped relationship (34). In addition to high salt intake, very low salt intake could cause a harmful cardiovascular effect by affecting neurohumoral, lipid, and insulin abnormality (34). Although the very low 24HUNa group (<2 g/day) did not show any significantly increased risk of KOA or SR-OA with knee pain in our results (Tables 3 and 4), the relationship between extreme salt restriction and OA needs to be further studied.

In the present study, the association between salt intake and OA appeared more clearly in subjects with KOA than in those with SR-OA at any site. Although the joint-specific prevalence of OA reportedly varies across different studies, the prevalence of symptomatic radiographic KOA is known to be higher than symptomatic hand or hip OA (35, 36). Szoeke et al. reported that patient-reported physician-diagnosed OA reflects the presence of radiological (Cohen's kappa 0.64) or symptomatic KOA (Cohen's kappa 0.73) more closely than self-perceived or self-reported OA (37). However, nonaxial OA involves the joints of the hands and hips as well as the knees, and SR-OA in KNHANES literally includes current and past physician-diagnosed OA at any site. OA is a complex disease with a multifactorial interaction of genetic and environmental factors. Twin and sibling studies show that the heritable contribution to OA is approximately 50% and that overall heritability is estimated to be stronger for hand and hip OA than for knee OA (38). The heritability of radiographic KOA and hand distal interphalangeal OA in the TwinsUK Registry was 37% and 65%, respectively (39). Moreover, a recent large genome-wide association study estimated the proportion of the total heritability explained by OA loci to be 14.7% for KOA, 51.9% for hip OA and 22.5% for OA at any site (40). These findings indicate that non-axial OA could have different pathophysiologic mechanisms and that there is a stronger non-genetic influence in KOA than in other nonaxial OA. Based on the assumption that KOA is more strongly associated with poor health-related physical function or quality of life and employment reduction than other non-axial OA (41, 42), the association between dietary salt intake and KOA in the present study may provide important clinical implications for the management or secondary prevention of KOA.

In the present study, the association between salt intake and OA could have been mediated by the association between salt intake and OA pain, because the association was more pronounced in the analysis of symptomatic OA (radiographic KOA or SR-OA with current knee pain) than in the analysis of SR-OA with or without knee pain. Interestingly, when mice are fed a high-salt diet, they demonstrate reduced pain thresholds to mechanical stimuli and CCR2-mediated inflammatory response underlies such salt-induced modulation of pain sensitivity (43). Moreover, Ccr2-null mice are reported to be protected against KOA (44). Furthermore, salt increases the expression of CCR2 in monocytes and circulation of the ligand MCP-1 in human beings (45). It is also of note that, in human OA synovium, CCR2+ cells are abundant and associated with OA cartilage erosion (44). Knee pain has been considered an important predictor of disease progression or incident KOA (46). In fact, there have been several studies showing an association between dietary sodium intake and other painful conditions such as headache (47, 48). Therefore, a prospective study is warranted to evaluate whether dietary salt restriction can decrease knee pain or slow the progression of KOA.

Although OA has been traditionally classified as a noninflammatory disease caused by cartilage wear and tear, it is now understood to involve multiple inflammatory pathways leading to chronic low-grade inflammation. Inflammation in the joint triggers cell senescence, increases sensitivity of nociceptive afferent neurons, and aggravates osteochondral pathology (49). Furthermore, systemic inflammatory conditions including obesity might play a role in OA pathogenesis through proinflammatory adipokines and cytokines (50). A meta-analysis showed that serum C-reactive protein (CRP) levels are significantly elevated and are associated with pain and decreased physical function in OA patients (51). Yilmaz et al. reported a significant association between urinary sodium excretion and serum CRP levels (52). There is growing evidence that high salt drives inflammation through direct and indirect effects on immune cells. High salt increases proinflammatory cytokines and decreases anti-inflammatory cytokines in macrophages and T cells (13), and induces inflammatory adipokines in adipocytes (53). Thus, the results of the current study could be related to proinflammatory effects driven by high salt.

It is well known that the radiographic damage in OA patients does not correlate with the presence and severity of joint pain and disability (54). Particularly, a higher pain intensity is associated with functional deterioration in patients with KOA (55). Although Western Ontario and McMaster Universities Arthritis Index is a validated tool for assessing pain and physical function in hip and knee OA (56), the KNHANES has evaluated the intensity of knee pain using a numerical rating scale (0-10-point scale) during 2010–2014. We did not find any association between knee pain intensity and 24HUNa in the Period A dataset. Pain medications, the information of which was not available in the KNHANES dataset, could have affected the results.

Many large cohort studies and meta-analyses have shown a significant association between OA and CVDs (57, 58, 59). Kim et al. also reported that the prevalence of radiographic KOA was positively correlated with the Framingham risk score using data from the KNHANES (60). Additionally, OA can be associated with CVD-related mortality (61). High salt intake is a well-known risk factor for CVD and accounted for 3 million deaths and 70 million disability adjusted life years globally in 2017 (62). Plausible explanations for the association between OA and CVD include physical inactivity and the use of non-steroidal anti-inflammatory drugs (63). Our results might add high salt intake as another common link between OA and CVDs. Considering the interconnection among high salt intake, OA, CVDs and CVD-related mortality, dietary salt restriction should be recommended to OA patients with an excessive salt intake.

There were several limitations to the current study. First, because of the cross-sectional study design, our results do not support a causal relationship between sodium intake and OA. Second, there may have been some yet-to-be identified or uncaptured confounding factors that could have affected the association between urinary sodium excretion and OA. Many medications such as glucocorticoids, diuretics, and metformin may have affected urinary sodium excretion (64, 65, 66), but data on specific medications were unavailable. Third, we could not analyse the association between salt intake and non-knee OA, and the association may be limited to KOA only. Fourth, although OA is a common arthropathy worldwide, there are racial or national differences in OA prevalence, incidence, and phenotypes. The present study is a single-country study. Further research is therefore needed in other races or nations to confirm our findings. Despite these limitations, this study is the first to demonstrate a significant association between high salt intake and KOA or SR-OA from a large population-representative dataset. Further large-scaled studies are required to reproduce our findings in multiple nations. Finally, a large-scaled prospective cohort study is warranted to investigate the causal relationship between salt intake and OA development or progression in various races/ethnics.

Conclusion

In conclusion, this cross-sectional study revealed a significantly higher prevalence of KOA in subjects with high sodium intake (≥ 5 g/day). The results suggest that the incorporation of dietary salt reduction should be considered as a non-pharmacological treatment for OA.

Funding

This study was supported by the Seoul National University Bundang Hospital Research Fund (grant no. 02-2018-047 to YJH and 21-2021-0013 to YJL).

Conflicts of Interest

The authors declare that they have no competing interests.

Ethical approval

This study conformed to the ethical guidelines of the 2008 Declaration of Helsinki and was approved by the Institutional Review Board of Seoul National University Bundang Hospital. All participants provided written informed consent prior to being administered the survey.

Electronic Supplementary Material

Supplementary material is available for this article at https://doi.org/10.1007/s12603-022-1804-x and is accessible for authorized users.

Supplementary material, approximately 182 KB.