The use of electronic health records to examine the association between obesity and chronic conditions: Results from a population-based sample in Saudi Arabia

is a problem worldwide. Over the prevalence of has increased signicantly in Saudi putting population health at an increased risk of mortality and morbidity. Because of the wide variation in previous local estimates, this study used electronic records of a population-based sample to estimate the prevalence of obesity and its association with diabetes and hypertension.


Background
Obesity is a growing public health problem worldwide and has become an epidemic in many countries. Statistics of the World Health Organization (WHO) suggest that obesity rates have nearly tripled since 1975 [1]. In 2016, it was estimated that there were more than 1.9 billion overweight adults worldwide (39%). Of these adults, over 650 million were obese, representing 13% of all adults worldwide. [1,2] Over the past few decades, the prevalence of obesity has increased signi cantly in the Gulf Cooperation Council (GCC) countries (Kuwait, Saudi Arabia, Oman, United Arab Emirates, Qatar, and Bahrain). This increase paralleled the rapid economic growth and prosperity following oil discovery in the region [2].
Factors such as urbanization, limited access to walkable areas, sedentary lifestyle, hot weather, changes in dietary habits, and car-dependent cities have shifted the leading causes of death from infectious to non-communicable diseases (NCDs) in the GCC countries, including Saudi Arabia (SA) [3]. These factors also constitute signi cant drivers of obesity in SA. Previously published nationwide surveys showed an increase of about 74% in the prevalence of obesity in SA [4]. This progressive increase led to obesity, rising from 22% in 1990 to 36% in 2005 [5,6]. However, according to the WHO, adult obesity decreased slightly to 33.7% in 2016 [7]. This minimal decrease does not change the ongoing war against the obesity epidemic in SA, especially with the existent gender disparity in obesity prevalence; 39.5% vs. 29.5% among women than men, respectively [7].
High body mass index (BMI) can put someone at an increased risk of mortality and multiple morbidities, including diabetes mellitus, hypertension, dyslipidemia, cardiovascular diseases (CVD), and cancer [8][9][10]. Having a higher BMI is associated with an increased risk of death from all-cause mortality and CVD compared to people in the healthy weight range [11]. Notably, high BMI tops the list of the risk factors associated with disease burden in SA [12]. Nowadays, NCDs account for about 73% of SA's deaths.
CVD, which hypertension is a major risk factor for, account for about 37% of deaths, making CVD the leading cause of death in SA [13]. Diabetes, on the other hand, places high among the top 10 causes of death (5% of total deaths) and disability (years lived with disability or YLDs). This makes it a signi cant contributor to the disease burden in SA [7,14]. The increasing burden of NCDs is not limited to population health; it also creates an enormous economic burden on the health system [10].
In 2017, SA had adopted a uni ed excise tax on sugar-sweetened beverages [15]. Moreover, to create greater public awareness of food choices, the government recently mandated that all food establishments label menus show meals' calorie content [16]. Some studies showed an increasing trend of obesity prevalence in SA [6,17]. But the WHO estimates tell a different story [7]. This discrepancy emphasizes the need for more updated and comprehensive data than what is already published. Furthermore, in light of the recent obesity preventive interventions, there is a need for an effective tracking system to measure the impact of these interventions and monitor local progress. Most of the previous studies were based on convenience samples using traditional data sources, such as community surveys.
An emerging, more comprehensive surveillance source of chronic diseases and other population healthrelated data is electronic medical records (EMR). The use of EMR data for chronic disease surveillance has been encouraged as it can provide timely, geographically speci c, and high-quality information [18,19]. Additionally, it has been reported that once the EMR system is in place, the time and cost commitment for data extraction is minimal [20]. Therefore, the EMR system is a potentially cost-effective tool that can be used to bridge the current gap in knowledge and represent a promising tool in population health improvement.
A good understanding of the obesity epidemic in SA and its association with cardiometabolic diseases, including diabetes and hypertension, is needed. Awareness of the magnitude of the target population's problem and characteristics is critical in targeting at-risk groups and designing effective populationbased interventions. SA covers a wide geographic area, and its residents come from various cultural backgrounds. This study aims to estimate the prevalence of overweight and obesity and their association with sociodemographic factors and morbidities in the central, eastern, and western provinces of the Kingdom.

Methods
This population-based study used the EMR system from the National Guard Health Affairs (NGHA) in SA. NGHA is a government entity that serves all employees of the national guard and their dependents. In addition, individuals with health insurance may also be seen privately through the business center. This network of ve hospitals is located in three regions of the Kingdom: central, western, and eastern regions. The network is estimated to serve around one million bene ciaries. Care is coordinated via a single EMR system known as BestCare. This system was implemented in January of 2016. The main medical city is located in the capital of Riyadh, which is also the home for a large university for health sciences and a research center with branches in the eastern and western regions. In addition, there are several outpatient clinics distributed around the Kingdom that serves patients.
In this study, we included individuals ages 17 years or older who visited any outpatient clinic in the past BMI was calculated automatically in the system using weight (in kg) divided by height (in meters squared). Subjects were then categorized as underweight (BMI <18.5), normal (BMI=18.5-24.9), or overweight (BMI=25-29.9) or obese (=>30) [4]. The following variables were obtained from the BestCare system; age, gender, nationality, BMI, diabetes, hypertension, and cancer. If the patient attended an outpatient appointment or was hospitalized for any reason, a primary diagnosis is documented. Patients were classi ed as diabetic if the discharge diagnosis was diabetes. The same was true for hypertensive patients. This study was reviewed and approved by the Institutional Review Board (IRB) of King Abdulah International Medical Research Center (KAIMRC).
Statistical analysis STATA 15 and Excel for Mac were used in all the analyses. Descriptive statistics by BMI category were calculated for various variables, and differences by demographic characteristics were evaluated using Chi-2 tests. A p-value of <0.05 was declared as statistically signi cant. In addition, differences in BMI categories were depicted across genders, regions, and nationalities.
To evaluate the association between obesity and diabetes or hypertension controlling for multiple factors, we constructed logistic regression models to calculate the odds ratios (ORs) and associated 95% con dence intervals. Normal weight individuals (BMI=18.5-24.9) were used as the reference category. Age was categorized into (17-25 as the reference, 26-45, 46-64, and 65 and older). Females and the central region were used as the reference group for gender and region, respectively.

Results
The study initially identi ed 876,602 individuals. After excluding those younger than 17 years old, the remaining were 616,092 individuals. Most of the study population were Saudi nationals 573,698 (93.1%), and 338,724 (55%) of the overall population were females. The distribution of individuals in each BMI category was underweight=33,332 (5.41%), normal weight=162,100 (26.32%), overweight=180,431(29.30%) and obese=239,913 (38.96%). Approximately 68% of the study population were either obese or overweight. Comorbidities like diabetes and hypertension were found in 18.42% and 16.23% of individuals, respectively (Table 1).
An independent association also observed between obesity and hypertension; obese patients were 2.1 times more likely to have hypertension (p-value <0.01). Gender was also a signi cant predictor of hypertension. Males were 18% more likely to be diagnosed with hypertension than females (OR=1.18; 95% CI=1. 16

Discussion
This study found a signi cant association between higher BMI and being diagnosed with diabetes or hypertension. Our results revealed that around 45% of women in our sample were obese, and 26% were overweight. Obesity prevalence was projected to rapidly increase between 1992 -2022 from 12% to 41% among men, and 21% to 78% among women [17]. Our nding underlines the major impact that obesity plays on NCDs and healthcare utilization and on population health. NCDs are currently responsible for around 73% of all death in the Kingdom [12,21]. If this burden continues, it will likely play a devastating impact on population health in the next decade in the Kingdom. Using electronic records to examine the impact of interventions to reduce obesity and chronic conditions may help monitor and improve population health. Other countries, such as the United States, have explored using EMR to evaluate conditions like diabetes, hyperlipidemia, hypertension, smoking, obesity, and depression to better understand population health In addition, several previous regional cross-sectional studies have indicated variations in obesity prevalence in the Kingdom [21, 24, 25]. Although these rates varied between regions due to limited sample size and differences in age groups, their ndings consistently align with our ndings on the higher obesity prevalence among women. Furthermore, our result on the obesity prevalence being higher among women was also similar to ndings from a recently published study (PURE-Saudi) [26]. While our ndings indicate that 16% of our study participants have hypertension and 18% have diabetes, the PURE-Saudi study showed a prevalence of 30% hypertension and 25% diabetes among participants. This might be explained by the older cohort in the PURE study or its limited representativeness.
Lifestyle has become more westernized and sedentary in the Kingdom during the last three decades, leading to an increased obesity prevalence among both men and women [27]. In particular, women have been shown in our study and other previous research to have a higher prevalence of obesity than men. [12,21,24]. A combination of social and policy factors may be leading to this inequality. These factors include that women are more prone to stay home, have limited access to culturally acceptable exercise activities, and the high cost of female gyms relative to those for men [4,12,21].
Our ndings have implications for both healthcare policies and population health initiatives and research funding. On the national level, these ndings call for strengthening preventive care to reduce obesity in the Kingdom and to address inequality between men and women in terms of obesity burden and chronic disease management. Our ndings can also be used to inform the modelling of future obesity burden and inform targeted-awareness initiatives in the Kingdom. Finally, this study adds to the growing evidence that obesity and NCDs are increasing threats in the Kingdom.
The Saudi Vision 2030 is a strategic plan to effectively transform numerous sectors in the Kingdom, including healthcare [28]. Multiple initiatives, under the Vision 2030, have been recently implemented to reduce the burden of NCDs in the Kingdom and its risk factors, including obesity. For example, the Kingdom recently introduced a tax on carbonated drinks (50%), which has been shown to be effective in lowering the consumption of carbonated beverages [15]. In addition, there is a new model of care being developed for the Saudi healthcare system as part of the Saudi Vision 2030. This model prioritizes NCDs prevention and emphasizes the public health role in healthcare [29]. Addressing the inequalities between women and men is a critical indicator in Vision 2030. This comes alongside other public health initiatives to improve the quality of life in SA and promote women's access to exercise facilities that are safe, affordable, and culturally acceptable [28]. Further studies need to assess the trend in women's physical activity in the light of the recent policies aimed to promote physical activity among women and evaluate the acceptability and e cacy of promoting home gyms in the Kingdom.
Since 2016, the Kingdom has experienced rapid growth in food home delivery via smartphone applications. Moreover, the Kingdom has also experienced complete and partial lockdown between March and June 2020 to mitigate the coronavirus disease 2019 (COVID-19) pandemic. Consequently, this inevitably limited physical activity. These two factors are expected to contribute to the pre-existing obesity endemic in Saudi Arabia. Our study has several strengths. First, the study included a large sample of diverse populations. To our knowledge, this is the largest cohort that aimed to determine the extent of the burden of obesity in Saudi Arabia. In fact, we do not know of any other study that used a population-based sample capturing over half a million individuals in the Kingdom. Previous similar research was limited either to the Saudi population, small sample size, or to a speci c region. In addition, BMI measurements were recorded during hospital visits by trained nurses, which improves the reliability of BMI measurements in our data. Finally, the use of a uni ed electronic system captured the latest data measured in terms of BMI or disease diagnosis for all patients. This will help future studies in terms of identifying targeted groups for prevention or intervention.
However, our present study has a few limitations. The data are based on visits to the healthcare facility, raising the possibility that the prevalence of obesity is overestimated compared with the general population. This is because those who did not visit the hospital in a period of four years are not represented in our sample and likely to be healthier than those who visited the hospital or clinic. Still, even if that had occurred, it is expected that the magnitude of the bias is minimal as NGHA provides healthcare to all military personnel, staff, and students who may have shown up at the clinic for a regular check-up. Second, some patients may have had a change in their BMI since their last visit due to nutritional programs, exercise, or other means of weight reduction. However, it is unlikely that a drastic change had occurred without any visit to our facility in which BMI would have been captured. Therefore, it is doubtful that this would affect our ndings. Finally, the study did not discriminate between type 1 and type 2 diabetes. Because the latter is the one known to be associated with obesity, the potential underestimation of association is possible.

Conclusion
In summary, this large sample population-based study showed a signi cant association between BMI, diabetes, and hypertension. The extent of obesity prevalence in the study population was high, and more pronounced among women. These ndings support the ongoing efforts to increase preventive measures and population health research. Future work is needed to continuously monitor the obesity trend and evaluate the e cacy of Vision 2030 obesity-related policies and initiatives. Using electronic records to examine the impact of interventions to reduce obesity and chronic conditions may help monitor and improve population health.   prevalence of obesity and overweight by gender: Non-Saudis