Are Employees in Oil and Gas Companies More Vulnerable to Hypertension Than the General Population of Working Age? Evidence From Men and Women in Northwest of China

Objective To examine whether the employees of an oil and gas company in Shaanxi Province are more vulnerable to hypertension than the general residents, aged 18-60, living in the same geographical region. Design Cross-sectional analysis of the hypertension prevalence of two different population by using propensity score matching (PSM). Participants Employees of Changqing Oil Field Filiale took part in Health Risk Factors Survey (2013) and residents took part in the National Health Service Survey (2013) in Shaanxi Province, China. Main outcome measures The primary outcome was hypertention which was defined as systolic blood pressure of at least 140 mm Hg, or diastolic blood pressure of at least 90 mm Hg, or self-reported antihypertensive medication use in the previous 2 weeks, or self-reported history of hypertention. Results Changqing employees were much younger, with a higher proportion of men, and their lifestyle was less healthy, with higher BMI, more drinkers and smokers. Before PSM, hypertension prevalence of Changqing employees was slightly lower than Shaanxi residents (5.8% vs 7.6%). After PSM, the results of the adjusted logistic model showed that Changqing employees were more likely to be hypertensive ( OR =1.10, 95% CI =1.02-1.19,


Introduction
Hypertension is an important health challenge around the world primarily due to its high prevalence but also because it is one of the leading risk factors of cardiovascular disease, premature death and disability. [1][2][3] For low-and middle-income countries, like China, the prevention and control of hypertension are essential, as the prevalence of hypertension in these countries has been increasing since 2000. [4][5][6] In order to effectively prevent and control hypertension, more health resources should be invested in the working population (employees) because the majority of employees are adults aged 18-60 and this age range is the key period for hypertension prevention. 7 For employees in oil and gas industries, health risk factors include not only unhealthy lifestyles but also occupational hazards coming from exposure to various industrial pollutions. 8 Employees in petrochemical industries are regularly engaged in long work shifts (12 hours) that often occur at night and some live in remote regions far away from their families. These working conditions are often associated with high calorific intake, fatty diets with poor fiber, physical inactivity, worse sleep quality, and high level of job strain. [8][9] Some studies have shown that the unhealthy lifestyles listed above can contribute to increased cardiovascular disease risk and this, in turn, may also affect productivity. [10][11][12] Employees in oil and gas industries are more likely to be exposed to gas flaring, oil-polluted surfaces and underground water than the general population. [13][14] Evidence suggests that chronic high-level and prolonged low-level exposure to these contaminants increases cardiovascular risk including hypertension. [15][16][17][18] In China most oil and gas companies are large state-owned enterprises. Compared to the private sector labor force, employees in state-owned companies have unique advantages in coping with the damage caused by chronic diseases. For example, they have access to a better health care system, jointly established by companies and the government. This provides almost free medical treatment, free periodic physical examinations and a wider scope of health management services. 19 In addition, the income of employees in oil and gas companies are much higher on average, which allows them to better respond to health risks.
Following on from the previous discussion this study asks, are the employees of oil and gas industries more vulnerable to hypertension than the general population of China? We also ask, do the protective factors (e.g. better health care and higher incomes) effectively reduce the prevalence of hypertension? Furthermore, some studies have shown that there are significant gender differences of the exposures to risk factors of hypertension. [20][21][22] For example, women in oil and gas companies are more likely to work in offices or logistics, and their occupational exposure is lower than that of men.
Therefore, an additional question that can be asked is whether a gender difference exists in the vulnerability to hypertension? The answers to the questions raised above are essential for the health management of oil and gas industry. More generally, they are also directly relevant to health resource allocation and policy making.
In order to address these questions, this study uses the employees of Changqing Oil Field Company, one of the largest oil and gas companies in China, and residents aged 18-60 in Shaanxi Province, where the company is located, to conduct a quantitative comparative study. However, it is difficult to compare the prevalence of hypertension between the two samples directly because many hypertension-associated risk factors (for example age, gender, and health behavior) are distributed differently between the two groups. Therefore, before comparing the prevalence of hypertension, the distributions of possible hypertension-associated factors should be balanced between the two groups. The Propensity Score Matching (PSM) method, 23 which is widely used in observational studies to balance covariate distributions between study groups, is used in this study. The primary aim of this current study is to ascertain whether the employees of an oil and gas company in Shaanxi Province are more vulnerable to hypertension than the general residents aged 18-60 living in Shaanxi Province. Also, we examine whether a gender difference exists in the vulnerability.

Health Risk Factors Survey (HRFS) in Changqing Oil Field Filiale 24
Changqing Oil Field has organized annual health examinations for employees The interviewers were recruited from health administrators at all levels of the company and were trained before the survey was implemented. All the 25 secondary units of the company participated in the survey and 52,000 questionnaires were distributed. In total, 50,013 valid questionnaires were collected. The validity of the survey questionnaires was checked by health administrators before data entry.

National Health Service Survey (NHSS) in Shaanxi Province, China
In Shaanxi's 2013 NHSS, a representative sample including 32 districts/counties, 160 sub-districts/townships, 320 residential committees/villages and 20,702 households was randomly selected by using a 3-stage cluster random sampling method.

Definition of the study sample
The employees in Changqing Oil Field were adult, in-service individuals, most between 19 to 60 years old. However, the participants of the NHSS cover a wider age range. In order to improve comparability, only 19-60 year old participants were selected in this study. The analysis was conducted using samples containing 49,698 Changqing employees and 34,820 residents from Shaanxi province (after excluding cases with missing values in key analysis variables).

Definition of hypertension
For the employees of Changqing, we defined hypertension as systolic blood pressure of at least 140 mm Hg, or diastolic blood pressure of at least 90 mm Hg, or self-reported antihypertensive medication use in the previous 2 weeks. For the residents of Shaanxi, chronic conditions including hypertension were self-reported and the interviewers ensured that the chronic conditions were clinically diagnosed by medical organizations to minimize any mistakes due to self-reporting.
Health behavior included smoking (no smoking, smoking cessation and smoking) and drinking (drinking more than one time a week and no drinking).

Statistical analysis
The propensity score was calculated by the logistic regression model, in which Changqing employees were taken as the treatment group and the residents of Shaanxi Province as the control group. Before PSM, z-tests were used to compare the means difference of continuous variables and Chi-Square tests were used to compare the proportions difference of categorical variables. P-values smaller than 0.05 were statistically significant. Health risk factors including gender, drinking and smoking as categorical variables along with age and BMI as continuous variables, were incorporated in the logistic model as covariates. The nearest neighbor matching method was used with a matching ratio of 1:1 and a caliper value of 0.038. This resulted in 26,178 matched cases. Absolute standardized difference in mean (ASDM) was adopted to evaluate the matching results and a confounding factor was considered to be balanced between treatment and control group when ASDM was smaller than 0.1. 25 After PSM, two types of logistic models were used to test whether Changqing staff were more vulnerable to hypertension than Shaanxi residents. In the unadjusted logistic model, group of population (Changqin or Shaanxi) was the only covariate, while in the adjusted logistic model, those baseline variables with P-values smaller than 0.1 in the balance tests were added as covariates to adjust the bias further. The PSM was conducted by using the SPSS's plug-in "psmatching 3.02", and the remaining statistical analysis and drawing were done by SPSS 24 and R 3.2.5.

Patients and public involvement
Patients were not involved in this study.

The prevalence of hypertension before PSM
The prevalence of hypertension of the overall sample and different strata, before PSM, is shown in Table 2. The overall hypertension prevalence of the treatment group was much lower than the control group (5.8% vs 7.6%, P<0.01). As the stratified results show, some stratum's prevalence was much lower in the treatment group -such as those who were female, non-drinking, non-smoking and those in any BMI stratum -while the prevalence in other strata was much lower in the control group. All comparisons are statistically significant (P<0.01).

Evaluation of the balance of covariates after PSM
After PSM, the distributions of the baseline factors between treatment and control groups are shown in Table 3. Most of the baseline factors were balanced between the two groups (P>0.1) except for weight, BMI (both continuous and categorical variables) and age (both continuous and categorical variables). As shown in Table 4 and Figue 1, the ASDMs of most covariate variables were larger than 0.1 before PSM, except for status of smoking. After PSM, all of them became smaller than 0.1, which indicated that all the covariate variables were balanced between the two groups.

The prevalence of hypertension and logistic model results after PSM
As shown in Table 5, hypertension prevalence in the treatment group was slightly higher than that of the control group with no statistical significance in differences after PSM (6.2% vs 5.8%, P=0.09). This pattern is repeated in the male stratum but with a statistically significant difference (treatment group vs control group: 9.0% vs 6.2%, P<0.01). The opposite pattern was found in the female stratum -the prevalence of hypertension of female employees in the treatment group was significantly lower than that of the control group (treatment group vs control group: 2.9% vs 5.4%, P<0.01).
Similar results are found in both male and female groups for BMI and age which are both significantly associated with hypertension. The ORs in these two variables increase as the levels become higher, which indicates the gradient effects on hypertension.

Discussion
Changqing Oil Field is a large state-owned oil and gas company, with nearly 80,000 employees, headquartered in Shaanxi Province. Employees in oil and gas enterprises are usually exposed to more occupational health risk factors than the general workforce, which may lead to a higher risk of chronic disease, including hypertension.
However, the average incomes of employees in this company are higher and medical care is much better, which can have certain protective effects on chronic diseases that may, in turn, offset the negative effects of exposure to occupational health risk factors.
Furthermore, chronic diseases are affected not only by the above external factors but also by internal individual factors, including for example, demographic and health behavior factors. For these reasons, we argue that when evaluating the efficacy of health management measures, companny's internal health management departments and more generally health policy makers, should increase the comparability and reliability of results by balancing the influence of these factors, before comparing the prevalence of chronic diseases between their employees and the general population.
In the current study, employees in Changqing Oil Field are younger on average but have unhealthier lifestyles with higher BMI and a greater incidence of drinking and smoking. Also, for the company examined, there is a higher proportion of male employees. Related and as shown in previous studies, age, gender and lifestyle are strongly associated with hypertension. 21 Therefore, if we directly compare the prevalence of hypertension without adjusting for the above confounding factors, this may potentially bias the results. The results of the direct comparison of the prevalence of hypertension showed that the direction of the difference in different strata between the two groups was not consistent. For example, in Changqing Oil Field, the hypertension prevalence of those strata who were female, aged 19-30, non-drinker and non-smoker were much lower. This imbalance in covariates between the two groups examined in this study, meant that it was necessary to balance the multiple confounding factors using an appropriate technique i.e. PSM.
After using the PSM method to balance the confounders measured in the two surveys (such as age, gender, BMI, smoking and drinking), the results for both the unadjusted and adjusted (by BMI, age and gender) models showed that Changqing employees were more vulnerable to hypertension than Shaanxi residents. One possible reason is that employees in oil and gas enterprises are more likely to be exposed to occupational hazards such as physical factors (dust, high temperature and pressure, noise and power frequency electromagnetic fields) and chemical factors (including H2S, aromatic hydrocarbons, alkanes, etc.). [8][9] If the enterprise's protective facilities were defective, or the staff's self-protection consciousness was not strong, this may increase the risk of hypertension. A second possible explanation is that Changqing employees' workplaces are mostly located in remote areas, far from the city. The costs associated with transportation and time taken for workers to go to their health organization are therefore realtively high, which may result in reduced use of available medical and health services. Lastly, many oil and gas enterprises implement a 24- Another strength of this study is that the two groups are in the same region. Therefore, apart from occupational characteristics, the two groups are similar, which in turn increases comparability. Furthermore, use of the PSM method to balance the distribution of age, gender, BMI, smoking and drinking within the two groups, improves the comparability of hypertension prevalence between the two groups. A limitation of this study is that Changqing employees were not randomly selected, which may lead to selection bias. However, after correcting for the main confounding factors of hypertension using PSM and the logistic model, we believe that potential biases have been effectively reduced. Another limitation is that although several confounding factors (such as age, gender, BMI, smoking and alcohol consumption) are corrected, some unobserved confounding factors were not adjusted. Hence, bias due to this potential omission has not been accounted for.
In conclusion, we find that Changqing employees were more vulnerable to hypertension than the general residents in Shaanxi Province. This study suggests that, with regards to policies concerning the prevention, treatment and control of hypertension, the department of health management in the company should formulate different policies for male and female workers respectively and should be intensely focused male and front-line workers.

Strengths and limitations of this study
• This study compares the hypertention prevalence of employees in oil and gas industries with general population of working age for the first time. • Data is derived from large-scale, high-quality health surveys performed in the in the same geographical region. • Results provide implication that the company's health management department should formulating different policies for male and female workers respectively and should be intensely focused male and front-line workers. • Cross-sectional study design precludes the demonstration of a causal relationship between the vulnerability of hypertention and the characteristic of the employees. • Unobserverd important confoundings cann't be adjusted by using PSM method.