Prevalence and income-related equity in hypertension in rural China from 1991 to 2011: differences between self-reported and tested measures

We used a national representative database from China Health and Nutrition Surveys (CHNS). CHNS is a longitudinal survey from the late 1980s conducted by the University of North Carolina Center for Population Studies, the National Institute of Nutrition and Food Safety and the China’s Center for Disease Control. The CHNS data contains new household formation, replacement communities and households, and all household members [27]. The questionnaire contains 12 dimensions: population density, economic activity, traditional markets, modern markets, transportation infrastructure, sanitation, communications, housing, education, diversity, health infrastructure and social services.

CHNS survey covers nine provinces that vary substantially in geography, economic level, public resources and health indicators. The samples in each province were selected with a multistage, random cluster process. The counties were stratified by income (low, middle and high) in each province and a weighted sampling scheme was used to select four counties randomly. Villages and towns within the counties and urban and suburban neighborhoods within the cities were chosen randomly. Approximately there were 4400 households in the whole survey covering 19,000 individuals [28].

The CHNS contains a weakness; the follow-ups were missing some chunks of data every year. There were three major reasons: 1) missing population that couldn’t be found because of travel, hours of work or play, 2) school children who were in boarding schools, 3) migrants work for working population. But the CHNS considered loss follow-ups into their design and recruited new participants as replenishment population if there were no more than 20 households or if respondents had constituted a new family [29]. This design of replenishment sample made up the weakness caused by the loss of enrolled subjects. The cross sectional data of each wave was regarded as national representative in many other researches [30‐32].

The feasibility and reliability for Concentration Index(C) and decomposition of C to measure health equity have been well documented [36, 37]. The concentration index can expose the relationship between health outcomes, such as self-reported health status and living standards like income level and wealth index. More widely, the concentration index can examine inequality not only in health outcomes but also in any health sector variable of interest [38], such as hypertension prevalence in this article. In our study, the two key variables underlying the concentration index are hypertension prevalence, the distribution of which is the subject of interest, and income level against which the distribution is to be assessed. We can see the degree of inequality of the hypertension prevalence distributes among different living standards. Further more, we decomposed the C to figure out how such inequality can be explained. The following specifically shows how we computed C and sub-section 2.4 decomposition of C.

Excluding respondents under 18 and singular values we have a sample of 122,945 observations. The sample values contain 11,119 in 1991, 10,828 in 1993, 11,891 in 1997, 13,324 in 2000, 13,194 in 2004, 15,922 in 2006, 16,313 in 2009 and 19,722 in 2011 respectively. The descriptive results of our sample are presented in Table 2. A table about baseline subjects involved in 1991 and new subjects of each wave is shown in the (Additional file 1: Table S1).

Figure 1 displays the prevalence rate of self-reported hypertension and tested hypertension, which suggests that the prevalence rate of self-reported hypertension in rural China has been increasing from 1991 (2.72%) to 2011 (13.2%). The prevalence of tested hypertension in rural China also has increased from 11.01% in 1991 to 25.05% in 2011. The increasing trend of self-reported prevalence and tested prevalence appears consistent. Figure 1 also indicates that the morbidity rate of self-reported hypertension increased more rapidly after 2000.

Considering the age of follow-ups would increase across the time, this study also evaluated the age-adjusted prevalence of both self-reported and tested hypertension. Both the prevalence’s, self-reported and tested hypertension, was adjusted to age distribution of the corresponding year in order correct the prevalence deviation caused by the follow-up getting older across time. The results of age-adjusted prevalence are placed in the [Additional file 2: Figure S1]. Additionally, to guarantee the hypertension screening in this survey would not affect the self-reported prevalence, identified baseline subjects and new subjects of each year and conducted Chi-Squared test. The null hypothesis was that the respondent’s self-reported hypertension status was independent of the respondent being in baseline population or new population. The results of Chi-Squared test suggested that the prevalence of two kinds of respondents varied in some years, but had no significant difference in most years, which is shown in the [Additional file 3: Table S2]. The results confirmed that hypertension screening across time has little effect on self-reported hypertension prevalence.

To build the confidence in our concentration index results, we excluded the results whose income level ranks the first 1% and the last 1% in our sample and calculated the concentration indexes of 8 years again. Table 6 shows the results of our sensitivity analysis. The table suggests that after exclusion of extreme data, the concentration indexes were consistent with the former results. Beyond that, the trend of 8 years in altered sample was also identical with the trend in the original sample.

In addition, to find out whether the new subjects of each wave would affect signs of concentration index of tested hypertension prevalence in total sample, we also conducted the Cs of baseline subjects in each wave and compared them with Cs of total population. The results are shown in the [Additional file 5: Table S5]. It is apparent that the signs of Cs of baseline subjects in each wave were generally consistent with signs of total population and the 95% CIs overlap in each year.

We computed separately self-reported prevalence and tested prevalence from 1991 to 2011 in this article. Consistent with those studies, we find that the prevalence of both self-reported hypertension and tested hypertension have rapidly increased from 1991 to 2011 in rural China. This rapid increase may be because of the change of health behaviors in rural China. Overweight rose from 15.56 to 32.49%, and obesity raised from 3.04 to 12.05%. The role that obesity plays in hypertension prevalence has already been discussed in other literatures [41]. Our study emphasizes on this opinion and verifies that obesity in China has a rapid grow in the past decades. The population with tested hypertension was always significantly larger than that with self-reported hypertension [40]. The rise of self-reported prevalence is from 2.72 to 13.2% from 1991 to 2011, while the tested is from 11.01 to 25.05%. However, compared with prior studies, we found a lower prevalence of both self-reported hypertension and tested hypertension in rural China and a possible reason is that the new subjects of each year may drag the prevalence rate down. One study found that the prevalence of self-reported hypertension and tested hypertension in 2009 is 12.6 and 29.6% respectively [42], while the prevalence in our study is 9.46 and 21.07% respectively. Another study on rural resident aged 35–74 indicates that hypertension prevalence increased by 20% from 1991 to 2011 [40], however, in our study self-reported prevalence and tested prevalence increased by 6.52 and 4.25% respectively. This difference suggests that the growth of hypertension prevalence is lower in younger people than in elder people.

Our study also indicates that the prevalence rate of tested hypertension is nearly twice as that of self-reported hypertension; the ratio is lower than previous findings using a national survey [42]. A potential reason is that our results are from rural areas where basic public health services are less developed compare with urban areas and thus the ratio between tested hypertension and self-reported hypertension is a bit less than nationwide area. In other literatures, some researchers did not adopt the definitions of self-reported hypertension and tested hypertension but conducted research on hypertension prevalence and awareness. In a sense, the awareness of hypertension can describe self-reported prevalence.

Based on the uneven access and utilization of health resources, both our findings and prior findings indicate that different measurements of hypertension prevalence vary significantly [43]. Using self-reported hypertension measures implies substantial bias against the real prevalence of hypertension, and thus, the findings based on self-reported measures can be expected to mislead the government’s policy. The deviation caused by self-reported measures exists in many countries and is supposed to be higher in low-income and middle-income countries such as China [44].

Our study indicates that age, BMI, region and marital status are all risk factors for hypertension, which is consistent with prior studies. For instance, in 2011, aging 60 and above shares more than 150% in the contribution to the concentration indexes of both self-reported prevalence and tested prevalence. BMI and region have significant impacts on hypertension prevalence. Some studies also suggested that income level could have an impact on hypertension prevalence [45], but in our study the impact of income level on the self-reported hypertension and the tested hypertension is respectively insignificant and negative. While many studies suggested that higher education level could reduce the probability of hypertension prevalence [8], our study shows a conflicting result that education level can strikingly affect tested hypertension prevalence, but not self-reported hypertension. The disparity may reflect education can help to improve individuals’ health consciousness, which in turn affects the actual control of blood pressure. Additionally, we find people having physical examinations in the past 4 weeks are more likely to get self-reported hypertension but unexpectedly, this result is not so significant for tested hypertension. A possible explanation is that those people who take physical examinations usually have more chances to be diagnosed by doctors and therefore it appears that they have more probability to have self-reported hypertension. More efforts should be put into equalizing basic public health services especially popularizing physical examination as it plays an important role in hypertension awareness and control.

There are both similarities and differences between the findings of our study and prior studies. Some of them have proved that not only hypertension, but also some other chronic diseases are inequitable and pro-poor, such as diabetes and heart diseases [8]. We find that this is also true for tested hypertension. It is generally considered that the poor possess less health resources than the rich and thus suffer a worse health status. However, in our findings, the concentration index of self-reported prevalence is positive, for example, 0.118 [95%CI (0.023,0.213)] in 1991 and 0.065 [95%CI (0.026,0.104)] in 2009, which means that self-reported prevalence concentrates on the rich from 1991 to 2011, while the Cs of tested prevalence were negative in some years. This conflicting result might be due to the ignorance of differences in two measurements of hypertension in prior studies. The significant disparities between the rich and the poor, in the access to and utilization of basic health services are taken into consideration for our study. Some researchers shared the same reasoning – it is evident that people in states that provide more education and better medical and health facilities are in a better position to be diagnosed and aware of their own particular illnesses than the people in states providing less education and worse medical and health facilities, where there is less awareness of treatable conditions [46].

Combining the equity with the prevalence of the two hypertensions, we can find that the income-related inequality of self-reported prevalence is pro-rich as its concentration indexes and horizontal-inequity indexes are always positive in 8 years. In addition, although the tested prevalence has rapidly increased from 1991 to 2011, the concentration indexes are always close to 0. This indicates that the prevalence of tested hypertension is less related to income level. This reveals the access to health resources and services in rural China are pro-rich, even though China is strongly pushing the equalization of basic public health services. There is no doubt that China’s basic public health services are getting increasingly equal to everyone and the quantity of health funds has been devoted into basic services such as hypertension screening ever since the equalization policy of basic public health services was carried out. Nevertheless, there still is a gap which cannot be ignored between the poor and rich in health accessibility and utilization. The imbalanced accessibility and utilization of health services might be the cause of the contradictory result in this study and this finding should raise our attention to put more effort into health service equity. Hence, if we focus on self-reported hypertension solely, a biased conclusion or policy will probably come out. Additionally, for the realization of the right to maintain health, China stressed the importance of universal health coverage (UHC) [47]. As a result, in 2011, about 95.7% of the Chinese population is covered by three main health insurances. In spite of the achievement of UHC, the access to health services and resources is not sufficient yet. Researchers conducted a study to assess the effective coverage of health insurance to explore whether the expansion of health insurance can improve health status [48], while the insufficient access to health services and resources indicated in our study may reveal that the effective coverage of health insurance in rural China is still low. The government may implement relevant policies to promote effective coverage of health services, not only crude coverage.

China has implemented a series of policies to improve the access to health care in rural areas. But current policies cannot sufficiently meet the challenges of promoting effective coverage of health care. Developing social capital in rural area can be a potential solution to promote the management of chronic diseases [49], but systematic measurements have not been well documented yet. It is advisable to establish free health management model in rural China to solve pro-rich access problem. Currently, free physical examination can be received by rural residents. But for those who have been monitored for chronic diseases, free management measures are not provided except outpatient reimbursement for certain kinds of diseases [50]. The feasibility of providing national essential medicines for free in chronic diseases among the elderly has been studied and researchers found that it could be financially guaranteed but a further systematic study is needed [51].