Prevalence of Hypertension in Rural Areas of China: A Meta-Analysis of Published Studies

Background Hypertension is one of the leading causes of disease burden across the world. In China, the latest nationwide survey of prevalence of hypertension was ten year ago, and data in rural areas is little known. More information about hypertension prevalence could help to improve overall antihypertensive health care. We aimed to estimate the pooled prevalence of hypertension in rural areas of China. Methods Comprehensive electronic searches of PubMed, Web of Knowledge, Chinese Web of Knowledge, Wangfang, Weipu and SinoMed databases were conducted to identify any study in each database published from January 1, 2004 to December 31, 2013, reporting the prevalence of hypertension in Chinese rural areas. Prevalence estimates were stratified by age, area, sex, publication year, and sample size. All statistical calculations were made using the Stata Version 11.0 (College Station, Texas) and Statsdirect Version 2.7.9. Results We identified 124 studies with a total population of 3,735,534 in the present meta-analysis. Among people aged 18 years old in Chinese rural areas, the summarized prevalence is 22.81% (19.41%–26.41%). Subgroup analysis shows the following results: for male 24.46% (21.19%–27.89%, for female 22.17% (18.25%–26.35%). For 2004–2006: 18.94% (14.41%–23.94%), for 2007–2009, 21.24% (15.98%–27.01%) for 2010–2013: 26.68%, (20.79%–33.02%). For Northern region 25.76% (22.36%–29.32%), for Southern region 19.30%, (15.48%–24.08%). Conclusions The last decade witnessed the growth in prevalence of hypertension in rural areas of China compared with the fourth national investigation, which has climbed the same level as the urban area. Guidelines for screening and treatment of hypertension in rural areas need to be given enough attention.


Introduction
Throughout the world, cardiovascular diseases have become a major public health problem and have been recognized as a leading cause of death and disability in most developed and some developing countries [1,2]. Hypertension is the most important risk factor for cardiovascular diseases such as stroke [3], coronary artery disease [4,5], end-stage renal disease [6,7] and heart failure [8]. It is estimated that about 25% of the world's adult population have hypertension, and it will be likely to increase 29% by 2025 [9]. In Europe, an estimated 37%-55% of the adult population are affected by hypertension [10,11]. The prevalence of hypertension is even higher in some developing countries. These observations indicate the high burden of hypertension in the general population.
Since the geographic, demographic and socioeconomic characteristics are hugely different throughout China, the prevalence and awareness rates, treatments, and control of hypertension may differ widely [12]. The 1991 National Investigation Data suggested that the prevalence among Chinese adult population was 11.2%, with 16.3% in urban and 11.1% in rural areas [13]. A national epidemiology study in 2002 showed that almost 18% of Chinese people aged 15 years and older were hypertensive. The prevalence of hypertension is 19.3% in urban areas, and 18.6% in rural areas [14]. Over the decade, the prevalence of hypertension across China has increased significantly, and the rural areas had witnessed faster growth compared with the urban areas, but the gap is rapidly closing. After more than ten years since 2002, the recent meta-analysis indicated that the prevalence of hypertension was 21.5% in urban area [15], and few data are available in rural areas.
In China, about 650 million rural people are spreading in 34 regions with very diverse socioeconomic and cultural characteristics. However, nationally representative studies are scarce, and the latest nationwide survey about prevalence of hypertension was ten year ago. More relevant information can help to improve the overall antihypertensive health care. Therefore, the aims of the present study are to systematically review the studies on hypertension and to estimate the pooled prevalence of hypertension in rural areas of China.

Literature Search Strategy
Comprehensive electronic searches of PubMed, Web of Knowledge, Chinese Web of Knowledge (CNKI), Wangfang, Weipu (VIP) and SinoMed databases were conducted to identify any study in each database published from January 1, 2004 to December 31, 2013, reporting the prevalence of hypertension in Chinese rural areas. Related articles were identified with the following search strategy ('hypertension' OR 'high blood pressure') AND ('prevalence' OR 'epidemiologic studies') AND 'rural' in all databases. The relevant reference lists retrieved from databases also was searched to obtain full-scale studies. Two authors (XFC and ZZL) were entrusted to screen the titles and abstracts and reviewed the full-text of the eligible articles. When they disagree with each other, the third (LZL) makes the final decision. The search language is limited in English and Chinese. All objects included studies were approved by the Medical Ethics Committee.

Criteria for Inclusion
In the present meta-analysis, the included studies must meet the following criteria: 1) an original epidemiological study conducted among Chinese rural population over 15 years old; 2) A cross-sectional study or data, or first phase of longitudinal study; 3) Provide information about sample size and prevalence estimation. 4) Have defined diagnostic criteria for hypertension: systolic pressure §140 mmHg and/or diastolic pressure §90 mmHg, or have been diagnosed with hypertension or have taken antihypertensive drug within two weeks, and secondary hypertension was excluded. 5) sample size §300. We did not excluded these studies that were limited to a specific age group, but studies limited to a specific occupation or particular population were excluded.

Data Extraction and Quality Assessment
Two investigators independently extracted necessary information from all included publications. Disagreements would be discussed and resolved by a third investigator if these two investigators could not reach a consensus. For all included articles, we extracted the following information: first author, publication year, screening year, province, study design, sample selection method, sample source, diagnostic criteria, method of measurement, area (northern vs southern), prevalence of overweight and obesity, onset age of study, sex ratio, case, sample, prevalence estimation, and age-specific prevalence if possible.
The quality of eligible literature was assessed according to the criteria of observational studies in recommended by Agency of Healthcare Research and Quality [16]. The tool includes 11 items: 1) Define the source of information (survey, record review); 2) List inclusion and exclusion criteria for exposed and unexposed subjects (cases and controls) or refer to previous publications; 3) Indicate time period used for identifying patients; 4) Indicate whether or not subjects were consecutive if not population-based; 5) Indicate if evaluators of subjective components of study were masked to other aspects of the status of the participants; 6) Describe any assessments undertaken for quality assurance purposes (e.g., test/retest of primary outcome measurements); 7) Explain any patient exclusions from analysis; 8) Describe how confounding was assessed and/ or controlled; 9) If applicable, explain how missing data were handled in the analysis; 10) Summarize patient response rates and completeness of data collection; 11) Clarify what follow-up, if any, was expected and the percentage of patients for which incomplete data or follow-up was obtained. A maximum point of 11 were assigned to each of the items equally.

Statistical Analysis
Because the weight of inverse variance is not optimum for fixed effects model when deal with binary data with low proportion. For the purpose of metaanalysis, the prevalence estimates were firstly transformed into a quantity (the Freeman-Tukey variant of the arcsine square root transformed proportion) suitable for usual fixed and random effects models [17]. Moreover, the transformed prevalence is weighted very slightly towards 50% and some studies with zero prevalence can still be included in the meta-analysis. The pooled proportion is calculated as the back-transform of the weighted mean of the transformed proportions, using inverse arcsine variance weights for the fixed effects model and DerSimonian-Laird weights for the random effects model [18]. Heterogeneity between studies were evaluated with Cochran chi-square (x 2 ) and quantified with the I 2 statistic, which shows the percentage of variation between studies that is due to heterogeneity rather than chance; I 2 ,25% is considered low, 25%-50% is moderate, and .50% is treated as high-level heterogeneity [19,20]. Considering the high heterogeneity between studies, we used random effects models for the pooled estimation of prevalence. Subgroup analysis was also conducted in order to deal with heterogeneity. Subgroups were defined as in difference in study year (2004-2006, 2007-2009, 2010-2013), sample size(,3000 vs §3000), area (southern vs northern), sex (male vs female), onset age of study (15-17, 18-19, 20-29, 30-39, 40-49, 50-60), age-specific group (15-, 30-, 40-, 40-, 50-, 60-, 70-), rate of overweight and obesity (20%#, 230%, 240%,.40%.). In order to explore the potential sources of heterogeneity, meta-regression was also conducted by arranging groups of studies according to potentially relevant characteristics. Covariates examined both univariate and multivariable models were publication year, screening year, sex ratio, area, response rate, sample size (continuous), sample size (,3000 vs §3000), rate of overweight and obesity, quality score. Publication bias was evaluated by testing for funnel plot asymmetry, Begg's Test and Egger's Test. Significance was set at a P value of less than 0.05. All statistical calculations were made using the Stata Version 11.0 (College Station, Texas) and Statsdirect Version 2.7.9 (http://www.statsdirect.com).

Results
Our searches returned a total of 1619 records. After removal of duplicates and initial screening, we identified 931 records for further screening by title and abstract. After exclusion of ineligible studies, we reviewed 253 records in full. Finally, we identified 124 studies (S1 References) with a total population of 3735534 in the present meta-analysis. The 124 records were divided into 6 groups depending on the onset age of study: 15-(N519, n52551480), 18-(N528, n5136249), 20-(N519, n5272549), 30-(N542, n5701659), 40-(N57, n548129), 50-60(N59, n525468). The flow diagram of the search process is exhibited in Fig. 1. 107 studies reported data on men (n51694586) and women (n51868799). Two studies investigated women (n52306) and men (n5747) respectively. In the surveys with samples, more than 60% of the study population was women. 54 reports were from south of China (n5431222), 70 reports were from north of China (n53304312), S1 Table and S2 Table shows detailed information of the 124 studies selected.
The present meta-analysis focused on these studies with onset age at 18 years old , and other studies added to the age-specific groups. The point prevalence of hypertension with 28 study populations ranged between 8.15% and 40.67%, with an overall prevalence of 22.81% (95%CI: 19.41%-26.41%, Fig. 2) and high level heterogeneity between-study heterogeneity (I 2 599.6%, P,0.0001). Table 1 shows the pooled prevalence of all subgroups depending on study year, sample size, area, sex, age-specific group, and rate of overweight and obesity. Among people aged 18 years old in rural China, the summarized prevalence in male (24.46%, 95%CI: 21.19%-27.89%, Fig. 3 We noticed high level heterogeneity between studies and subgroups (P,0.001, I 2 586.0%-100.0%). In univariate meta-regression analyses (Table 2), we used variables including year of publication, sample size, sex ratio (male/female), sample size (,3000 vs. $3000), overweight and obesity rate, and quality score to define hypertension, but did not change the estimation of prevalence. We noted that the prevalence of hypertension increased 0.39% among the participants from north China compared with the participants from the south China (metaregression P50.046). However, in multivariable analysis, none of the factors was significantly associated with heterogeneity on meta-regression (Table 2). Egger's linear regression test (P50.093) and Begg's test (P50.204) shown no significant publication bias.

Discussion
There are no nationwide data regarding the prevalence of hypertension in rural areas of China. This is the first report attempting to synthesize the prevalence  Prevalence of Hypertension in Rural Areas estimations of hypertension among Chinese rural population by using metaanalysis. This systematic review with meta-analysis of observational studies done in China included 124 reports done in China in the last decade. Therefore, it is possible to provide a reliable estimate of prevalence. A national epidemiology study in 2002 found that almost 18% of Chinese aged 15 years and older were hypertensive [14]. This meta-analysis indicates that the prevalence of hypertension in rural areas of China is 22.8% ($18 years old) in the last decade, which is well-above the level in 2002 (18.0%) of national investigation and even 1.3% higher than the prevalence (21.5%) of urban. The present level of Chinese rural areas has reached an epidemic proportion, and prevalence is still increasing on the time. Several reasons account for the current situation. For such a widely prevalent disease, even modest improvement in interventions can profoundly impact the population level. Therefore, prevention measures aimed at high-risk populations should be taken immediately. Our prevalence estimation of hypertension for rural residents was not different from previous reports in other countries. In India, the prevalence of hypertension was 19.0% in a rural community [49]. In 2008, a cross-sectional study of rural community health status among 1078 adults (aged .or 518 years) shown that the crude prevalence of hypertension was 18.3% in Nigeria while a total of 30 peer-reviewed publications were identified that reported the prevalence of hypertension in 33 143 patients was 32.6% from rural Ibero-America [50]. A Prevalence of Hypertension in Rural Areas study suggested that the prevalence of hypertension in young men is higher than in young women [51], but the situation was reverse before the age of 40, and after that the prevalence of hypertension did not differ between males and females in rural areas. The prevalence of hypertension were substantially different in regions.
The prevalence was higher in north China than in south China (25.7% vs 19.3%), and difference still exists between urban and rural population [14] (25.8% vs 20.4%), which may be attributed to discrepancy in eating habits. The proportion of salt intake among the local residents in Hainan and the long-term living from the north is 7.2% and 8.4%, while the short-term residents from the north is more than16.9% [52]. Such findings should be confirmed by investigation with a larger sample size. Overweight and obesity are also important risk factors of hypertension. The pooled results indicated that the prevalence increases with the growing rate of overweight and obesity. The proportion of overweight and obesity among Chinese population is 38.5% according to the latest data [53]. However, a challenge prevailing in Chinese rural areas is the lower levels of awareness rate, treatment and control of hypertension. The hypertension awareness rates in rural and urban areas were 13.5% vs. 35 Over the ten years, the hypertension treatment rate in rural areas was rising, but not significant. There is still an obvious gap between China and the developed countries. For instance, the United States was 59.0% in 2000 [57]. The rate of blood pressure control is at very low level. The rates in urban and rural areas were 4.1% vs. 2.8% in 1991 [54], 4.4% vs. 2.6% in 1998 [55], and 9.7% vs. 3.5% in 2002 [56]. The disparities in hypertension prevalence and control among population indicate a need for interventions that span the population and focus on vulnerable groups.
Although this meta-analysis includes more studies encompassing larger sample sizes than individual studies, some limitations need to be illustrated clearly. First, the whole analysis shows higher heterogeneity, but most of the studies had large sample sizes that can produce very precise estimation. In addition, metaregression analysis did not show that any factors that may be associated with the prevalence of hypertension. We still considered that other factors may influence heterogeneity and partly contribute to the increased prevalence of hypertension, such as dietary fats, high intake of sodium, stressful life, sedentary lifestyle urbanization and industrialization. Unfortunately, we did not obtain enough information about these aspects for further analysis. Besides, specific statistical comparisons with the previous results in all subgroups have not been done because of the data limitation, different grouping methods and unspecific analysis from the previous. Considering the huge rural population base, a little bit of growth in prevalence may have important implication for public health. Second, the included studies the meta-analysis are almost cross-sectional design. We did not identify any longitudinal studies, which may be difficult to carry out but quite important for further research because it could provide information on the development of hypertension. Third, although the same standard diagnosis of hypertension was adopted, the time point and time of blood pressure measurement may have an influence on the identification of hypertension. Another problem is the impossibility of identifying white-coat hypertension, which affects the pooled results because it occurs in 15% to 30% of subjects with an elevated office blood pressure [58,59]. Finally, there are some disparities in the distribution of healthcare resources in China. Some low-income residents did not have resources for examination, and we have restricted sample size in inclusion criteria. The potential information bias may have an effect on the pooled rates.
In conclusion, the last decade witnessed the growth in prevalence of hypertension in rural areas of China compared with the fourth national investigation, which has climbed the same level as the urban area. Guidelines for screening and treatment of hypertension in rural are need to be given enough attention.
Supporting Information S1