Clinical characteristics and lifestyle behaviors among individuals with arthritis: an analysis of 2017 Behavioral Risk Factor Surveillance System data

Context: The combination of osteoarthritis and rheumatoid arthritis affects nearly one-fourth of Americans. Many of the risk factors for arthritis are lifestyle related, such as obesity, physical activity, dietary behaviors, and alcohol and tobacco use. Objective: To analyze lifestyle behaviors of patients with vs. those without arthritis. Methods: A cross-sectional analysis of 2017 Behavioral Risk Factor Surveillance System (BRFSS) data was conducted on patients with and without a history of arthritis. Physical activity, smoking status, alcohol use, and dietary behaviors were extracted. Logistic regression models were constructed to calculate adjusted risk ratios (ARRs). All confidence intervals (CIs) were reported at 95%. Results: The median response rate for the 2017 BRFSS survey was 45.1% (sample n=292,808; population N=118,751,156). The overall prevalence of arthritis was 44% (n=128,850). Respondents with arthritis were signi ﬁ cantly more likely to be physically inactive than nonarthritic respondents, with a signi ﬁ cant gender interaction (male absolute risk reduction [ARR],1.2[CI,1.15 – 1.25];femaleARR,1.29[CI,1.24 – 1.33]).Those with arthritis were also more likely to be current smokers (ARR 1.1; CI, 1.05 – 1.15) and less likely to have attempted to quit smoking in the last 12 months (ARR, 0.91; CI, 0.87 – 0.96). Finally, men with arthritis were more likely to drink alcohol heavily (ARR, 1.14; CI, 1.02 – 1.27]). Conclusion: Patients with arthritis were more likely to be women, White, current smokers, and physically inactive, and to have poor mental health, lower education levels, and difficulty paying for healthcare and medications than those without arthritis. Further efforts should be made to address these findings through resource allocation in the care of patients with arthritis. Practitioners should also seek increased understanding of the psychological, social, and economic impacts of physical activity and smoking in patients with arthritis.

Although the use of cigarettes by Americans has decreased in recent years, physical activity levels and diet quality remain suboptimal. [8][9][10] Because these modifiable lifestyle factors play a large role in the development of arthritis, furthering the knowledge about lifestyle factors contributing to and worsening these debilitating conditions is very important for making advances in treatment and management. A clinical model should incorporate behavioral health, preventive medicine, and restoration of health, which is embedded within the osteopathic models.
Many studies have evaluated risk factors contributing to the disease process of arthritis. However, to our knowledge, no recent studies have analyzed the lifestyle behaviors of those diagnosed vs. those not diagnosed with arthritis using a nationally representative sample, such as the Behavioral Risk Factor Surveillance System (BRFSS), which is a CDC-managed, self-reported, previously validated, nationally representative database collecting lifestyle behavior and chronic disease information. Therefore, the primary objective of this study was to describe and analyze the sociodemographics and lifestyle behaviors of those diagnosed with vs. without arthritis using the BRFSS. The secondary objective was to determine which sociodemographics and lifestyle behaviors need further attention after a diagnosis of arthritis.

Methods
We conducted a cross-sectional analysis of lifestyle behaviors and clinical characteristics in Americans with and without a history of arthritis. Data were extracted from the 2017 BRFSS, which incorporated our variables of interest. In fulfillment of the appropriate reporting of recommended observational studies, this study was based on the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. All data included in this study were publicly available and contained no protected health information with patient identifiers; therefore, this study did not require IRB submission or approval.
The BRFSS collects data from all 50 states, the District of Columbia, Puerto Rico, and Guam and includes all noninstitutionalized individuals or those in the military over age 18 years. Random sampling collapsed all available telephone numbers into two strata based on area code and density of population in the area code, followed by computer automated sampling. The data we extracted included sociodemographics, policy-level variables, and lifestyle behaviors including smoking status, alcohol use, physical activity, and diet. Respondents were classified as having arthritis if they responded "yes" to the question: "Has a doctor, nurse, or healthcare professional ever diagnosed you as having some form of arthritis?" All respondents answering "no" were classified as having no history of arthritis. Participants aged 18-44 years were excluded from analysis due to the low prevalence of autoimmune and nonautoimmune arthritis in this age group.
Statistical analysis was performed on March 3, 2020 using Stata 16.1. Sampling weights were obtained from BRFSS and used to calculate the estimated population size and 95% confidence interval (CI). BRFSS utilizes an iterative proportional fitting method in determining the appropriate weights. We then computed descriptive statistics to describe sociodemographics, lifestyle behaviors, and policy-level variables. To determine independent risk factors for arthritis, we constructed a multivariate logistic regression model to calculate adjusted risk ratios (ARRs) by incorporating lifestyle behavior variables including smoking status, attempts to quit smoking, binge and heavy alcohol use, physical activity, and fruit/ vegetable consumption. ARRs were adjusted using age, gender, race, education, and healthcare coverage.

Sociodemographics
The median response rate for the entire 2017 BRFSS survey was 45.1% (sample n=447,222; population N=254,205,400). For the remainder of the manuscript, n indicates sample size and N indicates estimated population size. After excluding respondents who did not provide a response to the question regarding the presence of arthritis as well as respondents under age 45, the remaining sample size was 292,808 (N=118,751,156; a population estimate based on the sample meeting inclusion criteria). Of all patients who reported a history of arthritis (n=147,288), the prevalence of respondents over 45 years of age was 87% (128,850 of 147,288). Those reporting a history of arthritis were predominantly elderly, with the majority being age 55 to 64 (39,620; 26.9%), female (88,019; 59.76%), and White, non-Hispanic (107,490; 72.98%; Table 1). Respondents with arthritis were more likely to have healthcare coverage (95.27% of those with arthritis vs. 92.33% of those without), to have a BMI greater than 25 kg/m 2 (74.83% of those with arthritis vs. 66.84% of those without), and to be high school than college graduates (30.21% of those with arthritis vs. 21.56% of those without). Finally, participants reporting greater than 14 poor mental health days per month were significantly more likely to have a history of arthritis (15.90%; CI, 15.42-16.40) compared to those with no arthritis history (6.85%; CI, 6.58-7.14).

Work and activity limitations
Considering respondents with a history of arthritis, the following work and activity data were obtained. Approximately 79% of respondents (102,268) reported that arthritis prevents them from doing everything they want, and 51% (66,022) reported that arthritis prevents them from doing their usual activities. Over 33% (43,138) reported that arthritis limits the type or amount of work they can perform. Finally, 48% (61,332) reported that arthritis limits their social activities (Table 2). On a scale of 0-10, the mean amount of joint pain over the past 30 days was 4.74 (SD, 2.84).

Policy-level factors
Of respondents with a history of arthritis, only 12.76% (16,441) reported having taken a class on managing arthritis symptoms (Table 2). Among participants with a history of arthritis and a BMI over 25 kg/m 2 , 47.28% (60,920) reported having a doctor suggest weight loss as a viable option for managing arthritis. Among respondents with arthritis and physical inactivity, 55.25% (71,189) reported having a doctor suggest the use of physical activity. In contrast, among those reporting a history of arthritis and  physical activity, only 59.58% (76,768) stated that they achieve 150 min of physical activity weekly. Participants with vs. without a history of arthritis were significantly more likely to be unable to afford doctor visits and medications (15,

Lifestyle behaviors
The prevalence of respondents with arthritis who were current smokers was significantly higher than those without arthritis (ARR, 1.10; CI, 1.05-1.15), as shown in Tables 3 and 4. Those with arthritis were also less likely to report an attempt to quit smoking in the previous 12 months (ARR, 0.91; CI, 0.87-0.96). In the analysis of binge drinking, respondents with arthritis were significantly less likely to binge drink (10,865, 8.43%; CI, 8.06-8.82) than the non-arthritis sample (17,379, 10.6%; CI, 10.24-10.97). After adjustments were applied, a gender interaction existed in the calculated ARR. The ARR in women who binge drink was 0.88 (CI, 0.8-0.97; p<0.05), although statistical significance was not achieved in men (ARR, 1.032; CI, 0.96-1.11). The difference of prevalence in arthritic patients reporting heavy alcohol use was statistically significant compared to the nonarthritic cohort. The ARR in men who heavily drink was 1.14 (CI, 1.02-1.27) and in women was 0.86 (CI, 0.78-0.95).
Finally, physical activity and diet behaviors were analyzed. Respondents with arthritis were more likely to be physically inactive (n=47,300, 36.71%; CI, 36.07-37.34) than those without arthritis (n=43,268, 26.39%; CI, 25.88-26.92). Considering gender interactions, physically inactive men had an ARR of 1.2 (CI, 1.15-1.25), whereas physically inactive women had an ARR of 1.29 (CI, 1.24-1.33). Respondents with arthritis were more likely to not meet the recommended intake guidelines for fruits and vegetables (5+ servings per  day) compared to those without arthritis (ARR, 1.15; CI, 1. 10-1.20). No statistically significant difference existed between arthritic and nonarthritic respondents regarding the consumption of sugar-sweetened beverages.

Discussion
Our study found that patients with arthritis were significantly less likely to be physically active than those without arthritis. Furthermore, our data also demonstrated that those living with arthritis were less likely to be able to do the preferred activities, type and amount of work, and social activities that could be performed. These findings are concerning because participating in joint-friendly physical activity can improve pain, function, quality of life, and even mental health in those living with arthritis. 11,12 Likewise, when a patient's work and social life are inhibited, his or her morale may also be affected. Additionally, the present study identified that arthritic patients were more likely to report an increased number of poor mental health days, which has been shown to worsen the self-management of arthritis. 13,14 Other contributors to the association between poor mental health days and arthritis include biochemical and genetic factors, which should also be considered in the management of arthritis. 15,16 Finally, because our study found that those diagnosed with arthritis were more likely to have unpaid medical bills and were less likely to be able to afford medications than those without a diagnosis, financial strain is likely contributing to the poor mental health days reported by this patient population. Thus, it is important to bring the mental health aspects of these conditions to light to provide patient-centered care in the management of arthritis.
A further concern elucidated in our study is that those diagnosed with arthritis were more likely to be current smokers than those who were not diagnosed. Although this is predictable because smoking has been identified as a risk factor for arthritis, 3-7 the prevalence of smoking continuation should be of concern to those providing care to these patients. Unfortunately, in addition to the additive effects of smoking and arthritis, current evidence also suggests that the effects of biologics used in the treatment of some arthritis may be diminished in heavy smokers. 17 Thus, the importance of quitting smoking should be emphasized to manage arthritis.
The findings of our research were supported by several other studies. First, physical inactivity was widespread   18,19 Furthermore, in prospective studies and clinical trials, physical inactivity has been confirmed as a risk factor for arthritis. 20,21 Additionally, a meta-analysis of observational studies identified smoking as a strong risk factor for the development of seropositive rheumatoid arthritis, 22 and another report has also noted that patients with arthritis are more likely to be depressed or to report poor mental health days, both of which are supported by this study. 23 These studies, among other current research, indicate that the findings of the present study are in alignment and directly supported by the current knowledge of arthritis and its associated risk factors.
Our study identified alcohol abuse patterns that differed from other studies. The 2011-2014 National Health and Nutrition Examination Survey (NHANES) study found that, among 4,792 patients with arthritis, 15.6% (749) reported recent binge drinking behaviors, compared to 8% (10,865) in our sample. 24 Another study found that 6% of binge drinkers reported a history of arthritis. 25 However, variations in arthritis survey questions between these studies may be an explanation for the differences. These variations may have been due to sampling differences or response bias, although the lack of current research into how binge and heavy drinking impact arthritis is an area in which clarity is needed. The lack of research is particularly concerning due to our findings that men with arthritis were significantly more likely to report heavy drinking compared to their nonarthritic cohorts.
This study had several limitations and some strengths. First, the study design was cross-sectional; thus, causation claims could not be made. Second, measured variables were self-reported, which may have imputed response bias into our analyses because respondents may have reported false statements. Finally, the median response rate for the entire 2017 BRFSS survey was 45.1%, which may have limited the generalizability of the findings. The strengths of this study included its large sample size and the thoroughness of the sampling techniques.

Conclusion
The present study identified several behaviors and policylevel factors associated with patients diagnosed with arthritis. Notably, those diagnosed with arthritis were more likely to be women, White, above the ideal body weight, and to have the following characteristics: physical inactivity, tobacco use, poor mental health, lower education levels, difficulty paying for healthcare and medications, and men were more likely to drink heavily. This population was less likely to have attempts to quit smoking. These findings offer new areas on which clinicians should focus and better allocate resources in the care of patients with arthritis. Further research should be made to address these findings and should be applied to finding methods for improving physical activity and smoking cessation, as well as improving the understanding of the psychological, social, and economic impacts of physical activity and smoking, in patients with arthritis.