The Association between Depression and Diabetes and Associated Risk Factors by Racial/Ethnic Status among Adults in Arizona: Arizona Behavioral Risk Factor Surveillance System, 2014-2017

Objective: The purpose of our study was to examine the association between depression with diabetes, current smokers, and overweight or obesity among adults in four racial/ethnic groups: non-Hispanic whites (NHWs), non-Hispanic African Americans/blacks (NHAA/blacks), Hispanics, and American Indians/Alaskan Natives (AI/ANs), in Arizona. Methods: Data from the 2014-2017 Arizona Behavioral Risk Factor Surveillance System was used to examine the association between self-reported provider-diagnosed depression and self-reported provider-diagnosed diabetes and associated risk factors for each racial/ethnic group (n=31,671). The analysis was extended to test whether current smoking status or overweight or having obesity status modified the association between depression and diabetes among all Arizona adults. Weighted adjusted prevalence ratios (APRs) accounting for potential confounders were estimated using Cox’s proportional hazards regression analysis. Results: For all Arizona adults after adjusting for potential confounders, the prevalence of diabetes (APR =1.60: 95% confidence interval=1.43-1.72), current smoking status (APR=1.04: 1.02-1.07), and overweight or obesity status (APR=1.07:1.03-1.11) was greater among adults with depression versus without depression. For NHWs (APR=1.63:1.47-1.80), Hispanics (APR=1.71:1.39-2.12), and AI/ANs (APR=1.44:1.11-2.05) the prevalence of diabetes was greater among adults with depression versus without depression. In addition, the association between depression and diabetes was greater for current smokers and those overweight or with obesity than their counterparts in all Arizona adults (p<0.05 for both). Conclusion: Depression was associated with diabetes among all Arizona adults and NHWs, Hispanics, and AI/ANs. Current smoking status and overweight/obese status modified the association between depression and diabetes for all Arizona


Introduction
More than 30 million people (9.4% of United States [U.S]. population) in the U.S. have diabetes with 1.5 million new cases of diabetes in 2015 [1]. In 2014-2015, diabetes was the 7 th leading cause of death in the U.S. and in Arizona [2].
However, in Arizona, diabetes was the 3 rd leading cause of death for non-Hispanic African Americans/blacks (NHAA/ blacks), and Hispanics and the 5 th leading cause of death for American Indians/Alaska Natives (AI/ANs) in 2014-2015 [3]. In Arizona, 42.8% of the population is comprised of racial and

KEY POINTS
• Although research has shown that depression symptoms are common among individuals with diabetes, limited research has been conducted exploring the relationship between diabetes and depression among racial/ethnic groups.
• As the nation's lead public health agency, the Centers for Disease Control and Prevention support health departments on innovative projects that help raise awareness of type 2 diabetes and its comorbidities. While African Americans are at greatest risk for type 2 diabetes, the Hispanic community being the fastest and largest growing ethnic group in the country, and Arizona being the home to the largest Native American/Alaska Native population, this study advances the body of science on the association between diabetes and depression pertaining to three groups with the highest prevalence of diabetes.
• Culturally tailored interventions need to be developed to address depression and diabetes among minority populations who are disproportionally impacted by diabetes.
Depressive symptoms are common among patients with diabetes and may have a large impact on disease self-management and health outcomes [5]. With depression affecting more people than any other mental health disorder and a leading cause of disability, it is unclear if depression increases the risk of developing diabetes or if individuals with diabetes may be at an increased risk for depression or depression-like symptoms [5][6][7][8]. Diabetes compounded with depression has been associated with a higher risk of diabetes-specific complications, such as decreased treatment adherence, unhealthy eating, smoking, weight gain, decrease quality of life, increased healthcare expenditures, and loss of productivity and early disability [9,10]. Furthermore, people with diabetes experience greater levels of depression and the depressive symptoms play a greater role in mortality among people with diabetes than among people without diabetes [11].
Greater prevalence of comorbid diabetes and depression have been found among NHAA/blacks and AI/ANs compared to NHWs [12,13]. Also, compared to non-Hispanic whites (NHWs), major depression and factors associated with depression were more frequently identified among individuals of minority groups [14]. However, these findings were limited to national or local settings and not state level, especially one like Arizona with a greater number of minority populations. Relatively little information is known about the relationship between diabetes and depression among racial/ethnic communities in the state of Arizona. To address these gaps, we examined the association between depression and diabetes among adults in four racial/ ethnic groups: NHWs, NHAA/blacks, Hispanics, and AI/ANs, in Arizona. We hypothesized that having diabetes would be associated with having depression and that current smoking and overweight or obesity status will modify this relationship. In addition, this study aimed at exploring the associations between depression and risk factors associated with diabetes, including overweight or obesity and current smoking.

Methods and Materials
The Behavioral Risk Factor Surveillance System (BRFSS) is a population-based telephone survey conducted annually in all 50 states, the District of Columbia and U.S. territories to collect information on health-related behavioral risk factors, preventable health practices, health care access, and chronic conditions among noninstitutionalized U.S. adults aged 18 years or older [15]. Additional detailed information about the BRFSS survey design, sampling methods, data collection, and weights is available at https://www.cdc.gov/brfss/index.html.
Data from the 2014-2017 Arizona BRFSS (N=31,671), gathered by the Arizona Department of Health Services (ADHS), was used to examine the association between self-reported provider diagnosed depression (hereafter depression) and self-reported provider diagnosed diabetes (hereafter diabetes).
In the BRFSS, race/ethnicity was self-reported based on which one or more subgroups a respondent self-identified  15) was greater among those with depression than adults without depression. For Hispanics, the positive association between depression and diabetes (APR=1.71:1.39-2.12) remained significant but not for the association between depression and current smokers, and overweight or having obesity in adjusted models. In adjusted models, for AI/ANs, the association between depression and diabetes (APR=1.44:1.01-2.05) and between depression and current smokers (APR=1.11: 1.03-1.20) remained significant but not the association between depression and overweight or having obesity. There were no significant associations among NHAA/blacks (Table 4). racial/ethnic groups, the prevalence of depression varied across gender, age, lower socioeconomic status, those who were overweight or obese and current smokers. In our study, NHWs, AI/ANs and Hispanics had a greater prevalence of depression among those with diabetes when adjusting for potential confounders. Racial/ethnic differences among individuals with depression has been previously reported with major depression and factors associated with depression more commonly found among individuals of minority groups compared to NHWs [14]. Greater health burdens and lack of insurance have been largely associated with elevated depression rates among minority groups [14]. In addition, studies have found that patients with depression and diabetes, compared with those with diabetes alone, have been found to have poor adherence to diet, smoking cessation, and physical activity recommendations [17].
While one in 10 people reported diabetes in 2014-2016, there was significant variation in the prevalence of diabetes by race/ ethnicity (9.8-18.5%).

Discussion
Our findings highlight a significant association between depression and diabetes among all Arizona adults, NHWs, Hispanics and AI/ANs, and found a stronger association for current smokers and those overweight or who had obesity in the overall population.
Almost one in five adults in Arizona reported depression with variation across racial/ethnic groups (15.4-20.0%). Across Abbreviation: CI = Confidence Interval * Weighted percentages might not sum to 100% because of round †Chi-square tests were performed to examine differences in study characteristics across race/ethnicity subgroups.
higher prevalence of diabetes among AI/ANs, NHAA/black, and Hispanics [8,12,18]. There are several factors that could be making diabetes more common among certain racial/ethnic groups. Factors contributing to the high rates of type 2 diabetes among AI/ANs have been identified as genetic, environmental, and behavioral issues [18]. The high obesity rates, having a sedentary lifestyle and living in a stressful environment due to poverty, historical trauma and violence, can impact mental and physical health, contributing to the high rates of diabetes and complications due to diabetes [18]. In addition to having diabetes and having greater prevalence of overweight or obesity, AI/ANs and NH AA/blacks had the highest prevalence of current smokers in our study. Smoking is a major cardiovascular risk factor for patients with diabetes as smoking can lead to secondary cardiovascular complications, such as high blood pressure and stroke [19,20].
Depression prevalence was slightly higher among current smokers than in non-smokers and smoking status was an effect modifier between depression and diabetes with the association being stronger among current smokers. Previous studies have found a strong association between depression and smoking in aspects of quantity and frequency of smoking. Adults with depression are more likely to smoke more a pack a day and smoke their first cigarette within 5 minutes of waking up than adults without depression [21,22]. The sooner a person smokes upon waking is a measure of the level of nicotine addiction [21,22]. Also, adults with depression were more likely to be heavier smokers and smoke at greater rates than those without depression [22,23]. Furthermore, adults with depression have been found to have a higher rate of smoking initiation as well as a lower quit rate, compared to those without depression [22,23]. Although smoking rates continue to decline among the general population, smoking continues to increase among individuals with depression and other mental health illnesses [24]. Similar to the findings in this study, smoking has been reported to be a potential effect modifier between depression and diabetes [25]. Individuals who smoke have a 30-40% greater risk of developing type 2 diabetes and are more likely to have a difficult time controlling their diabetes, causing serious health problems resulting from diabetes [19].
Several cross-sectional studies have found an association between depression and overweight or having obesity [26].
In addition, studies have found a reciprocal association where obesity was found to increase the risk of depression and  Abbreviation: CIs = Confidence Intervals; BRFSS= Behavioral Risk Factor Surveillance System * Weighted percentages might not sum to 100% because of round †Chi-square tests were performed to examine differences in study characteristics across race/ethnicity subgroups.
depression was also found to increase the risk of becoming obesity [27]. Individuals with obesity have a 55% increased risk of developing depression and those with depression have a 58% increased risk of becoming obesity [26]. One potential explanation for the association between depression and obesity could be due to the side effects of antidepressive medication, such as weight gain and fatigue [26]. Physical inactivity resulting from fatigue due to antidepressive medication and depression can contribute to weight gain, and increase the risk for developing type 2 diabetes [26]. Although obesity has been found to be an individual risk factor for depression and type 2 diabetes [27], it has also been found to be an effect modifier in the association between depression and diabetes as demonstrated in this study and in others [28].
This study has several limitations which require thoughtful consideration when interpreting our findings. One significant limitation relates to self-reported diagnosed diabetes and depression status. This study did not verify depression and diabetes; many participants may not have been diagnosed with clinically present depression or diabetes and many may misreport condition status due to the stigma associated with depression or diabetes. Thus, diabetes and depression status may be subject to recall and social desirability biases. Self-reported tobacco use, BMI, income, and education are also subject to social desirability bias [29]. Additionally, it is possible that participants with diagnosed diabetes or depression are more likely to be diagnosed with the condition based on their access to or seeking of health care. Secondly, this is a cross-sectional designed study and exploration of causality cannot be determined. Therefore, we do not know the temporal association between depression, diabetes, and associated risk factors. Finally, due to a limitation of statistical power resulting from a small sample size in certain groups, there is potential for a type II error, which tells you the probability of falsely accepting the null hypothesis. Despite these limitations, our findings provide important information about the role of depression on diabetes and related risk factors among racial/ethnic groups supporting further investigation.  Cox's proportional hazards regression models were used to estimate the APR for the association between depression and diabetes among non current smokers. Adjusted for age, gender, education, income, and overweight/obesity status. ** Cox's proportional hazards regression models were used to estimate the APR for the association between depression and current smoking status. Adjusted for age, gender, education, income, and overweight/obesity status. † † Cox's proportional hazards regression models were used to estimate the APR for the association between depression and overweigh/obesity status. Adjusted for age, gender, education, income, and current smoking status.

Dependent Variables
Depression and diabetes are important health issues that need to be addressed in a multifaceted approach considering effect modifying factors such as overweight or having obesity and smoking status. Similar to our findings, several studies have identified the association between diabetes and depression and associated risk factors such as overweight/obesity and smoking [30][31][32][33][34]. In a study looking at the association between depression and diabetes, authors found that patients who reported high levels of depression symptoms, reported higher levels of diabetes nonadherence regiments [34]. In addition, authors found depressive symptoms associated with poorer diet and lack of medication regimen adherence [35]. Despite the availability of mental health screening tools, depression among individuals with diabetes and chronic diseases can be overlooked [30]. Individuals with depression and diabetes are more likely to rate their health as poor in comparison to individuals with depression only [36]. The negative health outcomes associated with diabetes and depression call for both conditions to be treated at the same time [37]. Health care providers, including diabetes educators, nurses, physicians, and dietitians caring for individuals with diabetes and other chronic diseases should consider screening, referring to selfmanagement programs, and treating depression to reduce the effects and the racial/ethnic disparities in diabetes outcomes and risk factors associated with diabetes (i.e., tobacco use, nutrition, and physical activity) and medication compliance [33]. In addition, involving a multidisciplinary team in healthcare systems through a team-based care model [31] and providing culturally tailored services and traditional healing practices can optimize treatment outcomes and increase referral capacity. Funding received from the CDC allowed for the implementation of the program "State Public Health Actions to Prevent Obesity, Diabetes and Heart Disease and Stroke", commonly known as the 1305 cooperative agreement. The funding afforded the ADHS, as well as other state health departments across the nation, integration of a team-based care approach to promote the prevention and control of chronic diseases and their associated risk factors [34]. Although the 1305 cooperative agreement did not award funding to address depression or mental health, it provided the opportunity for health authorities to design statewide public health interventions and collaborate with diabetes educators, allied health care professionals, academic institutions, and the greater Arizona diabetes stakeholder community to reduce comorbidities associated with depression and diabetes. Therefore, prioritizing and embedding surveillance efforts into statewide health assessments and improvement plans can potentially increase awareness of this comorbid condition and identify public health strategies to alleviate diabetes and depression.