Study design
We conducted a stratified cluster survey from September to December 2017.
Study setting
Three different altitude areas were purposively selected: Bomi county of Nyingchi city, Dagze district of Lhasa city, and Nagarze county of Lhokha city. These three areas had an average altitude of 2,500 meters, 4,100 meters, and 4,500 meters above sea level, respectively [22]. Bomi county is the farthest from the capital city (Lhasa), followed by Nagarze county and Dagze District, with average distances from the capital city of 630, 127, and 50 kilometers, respectively. In Bomi county, the primary sources of income are agriculture, forestry, and tourism. Being close to the capital city of Lhasa, Dageze district is more urbanized, although, in some parts, agriculture is the primary source of income. Nagarze county is situated in south-eastern Tibet, mostly surrounded by hills where animal husbandry has become the primary source of income for residents.
Sample selection
Two townships located within 50 kilometers from the center of each county were selected using simple random sampling. Thus, we chose a total of 6 townships as the primary sampling units. The most recent prevalence of hypertension in Lhasa city was 51.2% [23]. With a two-sided, 95% confidence interval, an error of 0.05, a design effect of 1.3, and including 10% non-respondents, we required a total of 550 participants in each county.
Inclusion and exclusion criteria
Eligibility criteria for participants included age ≥18 years, Tibetan ethnicity, and residents in the village for at least one year. Excluded from this survey were those who had severe mental dysfunction, pregnancy, or severe complications of hypertension.
Data collection
The researchers invited eligible participants to the nearest local primary health center or village committee offices. After giving informed consent, all study participants were physically examined by trained investigators following standard protocols. Physical measurements included weight, height, and blood pressure. The body weight and height of participants wearing no shoes or overcoat were measured using the Suhong RGZ-120 height and weight scale. Before the first examination of their blood pressure, we allowed participants to relax for at least five minutes in a quiet room. Investigators advised all participants to avoid drinking tea and alcohol, cigarette smoking, over-exercising, and to void urine half an hour before their examination. In this survey, we used an electronic sphygmomanometer with high reliability and validity (Omron HEM-7201automatic blood pressure monitor) at a high altitude area [24]. Blood pressure was measured twice at 60 seconds intervals for all participants. Those who had a discordant blood pressure of greater than ten mmHg on the previous two measurements had a third measurement taken. The final result was the arithmetic mean of all BP measurements. To explore the participants previously diagnosed with hypertension, we administered self-completed questionnaires and confirmed the results by checking the record books of the participants.
Variable definitions
Independent variables included socio-demographic characteristics, biological, and behavioral determinants, and history of diseases. Behavioral determinants included the consumption of tobacco and alcohol, and biological factors included body mass index (BMI) defined as the weight (kg) divided by the square of height (m2) and waist circumference. According to Chinese BMI classification, the BMI range for overweight is from 24.0-27.9 kg/m2, and the cut point for obesity is ≥28.0 kg/m2[25].
We identified those who had regular measurements by community health workers in the hypertension intervention program to determine the screening coverage of hypertension screening. We recorded the results of the routine hypertension screening as hypertension or no hypertension detected by the routine screening by community health workers.
The effectiveness of hypertension screening was the proportion of those previously diagnosed as hypertension by local health workers whose blood pressure was under control, defined as systolic blood pressure (SBP) < 140 mmHg and diastolic blood pressure (DBP) < 90 mmHg.
Individuals previously informed by a doctor or local health worker about their hypertension status, despite their current hypertension status, were considered to have hypertension awareness. Individuals who responded "no" to the question: “In the past, have you received measurement of hypertension by a local doctor or healthcare provider?” and had hypertension during their physical examination, were categorized as unaware hypertension.
The presence of hypertension was defined as SBP ≥140mm Hg and/or DBP ≥ 90mmHg, and/or self-reported treatment for hypertension with antihypertensive medications taken in the past two weeks[26].
Participants who answered "yes" to the question “In the past, have you received a diagnosis of hypertension by a local doctor or healthcare provider?” were categorized as previously diagnosed hypertension.
Statistical Analysis
We described the characteristics of the participants in our study by a weighted analysis. We used the survey design to adjust the estimates, and the iterative proportional fitting (raking) to reduce the sampling bias by fitting the data using known demographic characteristics from the 2010 census. R version 3.5.1 (https://cran.r-project.org) was used to analyze the data. We used the survey-weighted logistic regression models to find factors associated with the screening of hypertension in the past.
A Venn diagram visualizes the overlapping sets of newly discovered hypertension cases by this study, previously diagnosed hypertension by screening program, and those who received antihypertensive treatment in the past, both controlled and uncontrolled states.
Figure 1 schematizes the three groups of people. Those who were covered by this survey are shown in the dotted-lined circle. The solid-lined circle represents those who were previously approached by the screening program. The small dash-dot-lined circle represents those who were screened and received antihypertensive treatment. The segment marked "a" represents those who diagnosed with hypertension but never tested or unscreened hypertension in the past and denoted as unaware hypertension. Uncontrolled hypertension (or diagnosed hypertension in both the survey and screening program) is the summation of segments “b” and “d." Segment “c” represents those who previously had a history of diagnosed hypertension but normal blood pressure in our study. Segments “d” and “e” represent those uncontrolled and controlled hypertension after medical treatment, respectively. Finally, anyone falling outside the circles marked as "f” is normotensive or non-hypertensive subjects.
The effective coverage was defined by Shengelia et al (2005)[27] and later by Ng et al (2014) [28], as the fraction of potential health gain that was actually delivered to the population through the health system. The calculation of the effective coverage follows the following formula:
ECij = Uij*Qij/Nij
Where subscriptions i and j represent individual and intervention. ECij is the effective coverage of individual i with intervention j. In our case, where the intervention is only the HT screening program, thus, i remains in the formula, and j is omitted. Then,
ECi = Ui*Qi/Ni, or EC = U*Q/N, in a simple term.
Where Q is the quality or health gain ratio. U is the utilization of health service and refers to the probability that the individual with a need will receive the intervention. N is the need indicator, which refers to individuals who will gain actual benefits from receiving or true need; if N=1 is the true need for receiving the healthcare services and N=0 for individual no need for coverage. In this survey, we calculated the effective coverage of HT screening by replacing U with the coverage of HT screening program, and Q/N is the effectiveness of HT treatment, both by medical and lifestyle modification, among the screened population.
Variables excluded from the final multivariate model were those with a univariate p-value greater than 0.2. Factors assisting control of hypertension programs were identified using survey-weighted logistic regression modeling and determined based on a backward step-wise process and included only variables with a p-value less than 0.05.