Hypertension is the leading risk factor for cardiovascular disease (CVD), which remains an important public health problem as one of the top causes of death in China, causing heavy social, familial, and economic dysfunctions [
For hypertensive patients, antihypertension treatment is fundamental to improve health outcomes, minimize their impact, prevent further disability, and reduce health care costs [
Northwest China is economically less developed, with large land area and sparse population density. One-fifth of local residents are leading nomadic or seminomadic lives, making the penetration rate of medical resources low, health awareness poor, and the lifestyle unhealthy [
Considering so many associated risk factors, accurate prediction nontreatment tools and early intervention may be the most effective actions toward unsatisfactory treatment. Therefore, this study aimed to identify the risk factors associated with untreated hypertension in the management of hypertension in primary health care of Northwest China and to develop a predictive nomogram to estimate the probability of nontreatment in a given visit, according to five dimensions related to the demographic factors, socioeconomic status, living environment, health-related behaviors, and anthropometric value.
The cross-sectional study is reported according to the STROBE checklist standards. This study design and methods have been described previously [ Residents who are willing to participate in the investigation and sign an informed consent form Local inhabitants aged ≥15 years Residing at the current address for ≥6 months Women who are not pregnant
For the current analysis, we excluded participants without response (
The flow chart of inclusion and screening of surveyed subjects.
Trained study staff used a standardized questionnaire to collect data on demographic characteristics (such as sex, age, and ethnicity), socioeconomic status (occupation, education attainment status, marital status, and family income per member), health-related behaviors (alcohol consumption and cigarette smoking) [
Trained observers measured the body height, weight, waist circumference (WC), and blood pressure (BP) of each participant according to the standardized equipment and procedures. In order to protect the privacy of participants during the anthropometric measurements, we arranged the measurement site in a room with a suitable temperature, and only one participant can enter at a time. In addition, participants wore light clothing, while their weight and waist circumference were measured. Each participant’s BP records were measured by using the automatic sphygmomanometer OMRON HBP-1300 Professional Portable Blood Pressure Monitor (OMRON, Kyoto, Japan) three times on the right upper arm after the participant rested 5 min in a seated position, with 30 seconds between each measurement with an observer present. The mean value of the three measurements was used for analysis. Body mass index (BMI) was calculated as weight divided by the square of height (kg/m2).
All subjects fasted for ≥8 h, and a 5 mL fasting blood sample was collected. Next, fasting plasma glucose (FPG), triglyceride (TG), total cholesterol, low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) were tested using standard methods.
Descriptive data were conducted for 895 subjects at baseline using SPSS 20.0 for Windows (SPSS Inc., Chicago, IL). Continuous variables were presented as means ± standard deviations (M ± SD) and categorical variables were expressed as frequency (
Steps of the formulation and assessment of the nomogram were carried out using the package of “rms” in R version 3.5.1 (
In total, 895 patients with mean age of 52.65 ± 17.49 years were enrolled, with women accounting for 52.7%. The baseline characteristics between training and validation sets are displayed in Table
Baseline characteristics of the study population by training set and validation set.
Variables | Training set ( | Validation set ( | Total ( | |
---|---|---|---|---|
Age (years) | 52.63 ± 17.61 | 52.69 ± 17.26 | 52.65 ± 17.49 | 0.963 |
<45 y | 191 (30.5) | 77 (28.6) | 268 (29.9) | 0.820‡ |
45–60 y | 201 (32.1) | 91 (33.8) | 292 (32.6) | |
>60 y | 234 (37.4) | 101 (37.5) | 335 (37.5) | |
Gender, women ( | 337 (53.8) | 135 (50.2) | 472 (52.7) | 0.316# |
Herdsman ( | 260 (41.5) | 116 (43.1) | 376 (42.0) | 0.659# |
Education levels ( | ||||
Primary and lower | 361 (57.7) | 159 (59.1) | 520 (58.1) | 0.923‡ |
Junior high | 145 (23.2) | 60 ( 22.3) | 205 (22.9) | |
Senior high and higher | 120 (19.2) | 50 (18.6) | 170(19.0) | |
Ethnicity ( | ||||
Han | 269 (43.0) | 111 (41.3) | 380 (42.5) | 0.198# |
Kazakh | 90 (14.4) | 54 (20.1) | 144 (16.1) | |
Tajik | 74 (11.8) | 28 (10.4) | 102 (11.4) | |
Others | 193 (30.8) | 76 (28.3) | 269 (30.1) | |
Number of family members ( | ||||
1 | 33 (5.3) | 10 (3.7) | 43 (4.8) | 0.551‡ |
2–4 | 403 (64.4) | 172 (63.9) | 575 (64.3) | |
≥5 | 190 (30.4) | 87 (32.3) | 277 (30.9) | |
Marital status ( | ||||
Single | 64 ( 10.2) | 21 (7.8) | 85 (9.5) | 0.508# |
Married | 511 (81.6) | 224 (83.3) | 735 (82.1) | |
Separated | 51 (8.1) | 24 (8.9) | 75 (8.4) | |
Family income per member | ||||
<¥500/month | 142 (22.7) | 65 (24.2) | 207 (23.1) | 0.004‡ |
¥500–1000/month | 73 (11.7) | 30 (11.2) | 103 (11.5) | |
¥1001–3000/month | 286 (45.7) | 146 (54.3) | 432 (48.3) | |
>¥3000/month | 125 (20.0) | 28 (10.4) | 153 (17.1) | |
Altitude of habitation (m) | ||||
<1000 | 387 (61.8) | 164 (61.0) | 551 (61.6) | 0.717‡ |
1000–3000 | 163 (26.0) | 76 (28.3) | 239 (26.7) | |
>3000 | 76 (12.1) | 29 (10.8) | 105 (11.7) | |
Current smokers ( | 150 (24.0) | 50 (18.6) | 200 (22.3) | 0.077# |
Current drinkers ( | 138 (22.0) | 46 (17.1) | 184 (20.6) | 0.093# |
Body mass index | 27.29 ± 4.30 | 27.25 ± 4.48 | 27.28 ± 4.35 | 0.917 |
BMI: <23.9 kg/m2 | 124 (19.8) | 53 (19.7) | 177 (19.8) | 0.669‡ |
BMI: 24.0–27.9 kg/m2 | 242 (38.7) | 112 (41.6) | 354 (39.6) | |
BMI: ≥28.0 kg/m2 | 260 (41.5) | 104 (38.7) | 364 (40.7) | |
Abdominal obesity (n, %) | 424 (67.7) | 183 (68.0) | 607 (67.8%) | 0.930# |
CVD ( | 21 (3.4) | 2 (0.7) | 23 (2.6) | 0.195# |
Diabetes ( | 91 (14.5) | 31 (11.5) | 122 (13.6) | 0.228# |
Dyslipidemia ( | 113 (18.1) | 41 (15.2) | 154 (17.2) | 0.307# |
Comorbidity ( | 193 (30.8) | 66 (24.5) | 259 (28.9) | 0.057# |
Blood pressure (mmHg) | ||||
Systolic blood pressure | 149.37 ± 20.58 | 152.09 ± 21.34 | 150.19 ± 20.83 | 0.073 |
Diastolic blood pressure | 85.67 ± 12.31 | 85.97 ± 13.68 | 85.76 ± 12.73 | 0.749 |
CVD, cardiovascular disease.
Table S1 showed that the untreated rate among patients with hypertension was 20.9% (187/895). As compared with patients under antihypertensive treatment, those who were untreated were more likely to be older, be herdsmen, experience lower education status, and have less family income per member and less likely to have comorbidity.
The samples of the training set were used for building models. Of demographic features, socioeconomic status, living region, health-related behaviors, and anthropometric values, 17 factors were reduced to five potential predictors in the study (∼3 : 1 ratio; Figures
Demographic features, socioeconomic status, live setting, health-related behaviors, and anthropometric value selection using the LASSO binary logistic regression model. (a) Optimal parameter (lambda) selection in the LASSO model used fivefold cross-validation via minimum criteria. The partial likelihood deviance (binomial deviance) curve was plotted versus log (lambda). Dotted vertical lines were drawn at the optimal values by using the minimum criteria and the 1-SE of the minimum criteria (the 1-SE criteria). (b) LASSO coefficient profiles of the 17 features. A coefficient profile plot was produced against the log (lambda) sequence. Vertical line was drawn at the value selected using fivefold cross-validation, where optimal lambda resulted in five features with nonzero coefficients. LASSO: least absolute shrinkage and selection operator; SE: standard error.
Prediction factors for nontreatment in hypertension from study population by multiple logistic regression model.
Stratification | OR (95% CI) | ||
---|---|---|---|
Age | 0.517 | 1.68 (1.27–2.21) | <0.001 |
Herdsman | 1.054 | 2.87 (1.77–4.64) | <0.001 |
Family income per member | −1.644 | 0.19 (0.15–0.25) | <0.001 |
Altitude of habitation | 1.134 | 3.11 (2.17–4.44) | <0.001 |
Comorbidity | −0.776 | 0.46 (0.27–0.78) | 0.004 |
Constant | −1.920 | 0.147 | <0.001 |
OR: odds ratio; CI: confidence interval.
The results of multivariable logistic regression analysis were shown in Table
Developed medication nontreatment nomogram.
The AUC for the prediction nomogram was 0.859 (95% CI: 0.812–0.906) (Figure
Receiver operating characteristic curve of the untreated nomogram prediction in the study.
Calibration curves of the nomogram in the training set (a) and validation set (b). The
The decision curve analysis for the untreated nomogram showed that if the threshold probability of a hypertensive patient ranges from 7% to 91%, using this untreated nomogram to predict untreated risk adds more benefit than the scheme (Figure
Decision curve analysis for the untreated nomogram. The
Herein, this is the first study in relatively representative patients with hypertension among primary health care of less developed northwest China to develop a predictive nomogram to evaluate the factors influencing the access to treatment for hypertension. Untreated rate of hypertension is still very high among patients with hypertension, despite the existence of universal access and the availability of effective treatments. The nomogram developed is simple (consisting of only five factors; during selection of variables for each block, many were eliminated because they were not associated with nontreatment or because they were strongly collinear with other variables) and shows good standardization and ability to discriminate. Its high sensitivity (85%) is worth mentioning, indicating that the factors included are able, as a whole, to predict properly hypertensive patients who have no access to treatment.
Herdsmen are those who are leading a nomadic or seminomadic lifestyle. Common antihypertension agents are not readily available in many stock-raising regions due to their nomadic lifestyles. In addition, the proportions of individuals who could not afford these drugs are much higher among them, based on their household income [
With the increase in geographical altitude, the population will become sparse, the style of living and production will become more singular, and health needs and access to health care become considerable challenges in high-altitude regions [
The elderly and hypertensive patients without comorbidities were associated with nontreatment. This may be explained by the fact that elderly people in rural and stock-raising regions have traffic obstacles when they go to town to buy antihypertensive drugs and most elderly farmers or herdsmen generally lose a large part of their economic resources compared with their younger counterparts, which may have led them to take antihypertensive drugs intermittently [
Our study has several strengths. First, the nontreatment risk prediction tools may provide important insight to clinicians in delivering optimal health care services. Using accessible metrics such as age, occupation, number of family members, altitude of habitation, and information about comorbidity, this tool can help the clinicians, especially primary health providers, to better identify patients who are at high risk of not taking medications as prescribed. Finding such patients is an important step before intervening to improve their adherence. Second, our model showed good accuracy and excellent agreement in training set and validation set, which suggests that it has good transportability and generalizability.
Some limitations in the current report should be kept in mind when interpreting the results. First, the study sample may not stand for all Chinese with hypertension. Patients who are not aware of their hypertensive status were also excluded from the analysis. Second, risk factor analysis may have missed some potential variables that could affect the commence of antihypertensive treatment such as the therapy-related factors (e.g., type of medicine, side effects, and medicine-related questions) and social support. Third, although the robustness of our nomogram was examined with internal validation, we failed to perform external validation, and therefore the generalizability of current data may be limited in other populations and regions/countries and relevant external validation is needed.
There is a considerable nontreatment rate among hypertensive patients in primary health care of less developed Northwest China. This study developed a novel nomogram with a relatively good accuracy to help primary health care access the risk of nontreatment in hypertensive patients when they manage these patients. Herdsmen living in high-altitude areas with low family income may be the key population for enhancing antihypertensive treatment.
Materials included in the manuscript, excluding the relevant raw data, will be made freely available to any researchers who wish to use them for noncommercial purposes, while preserving any necessary confidentiality and anonymity.
Ethics approval was obtained from Ethics Review Committee of People’s Hospital of Xinjiang Uygur Autonomous Region.
The authors declare that they have no conflicts of interest.
Nanfang Li and Lin Wang contributed to the study design. Nanfang Li, Lin Wang, Mulalibieke Heizhati, Mei Li, Zhikang Yang, Zhongrong Wang, and Reyila Abudereyimu participated in the data collection. Lin Wang and Xintian Cai performed the statistical analysis. Lin Wang drafted the manuscript. Mulalibieke Heizhati critically revised the manuscript. Nanfang Li, Mei Li, Zhikang Yang, Zhongrong Wang, and Reyila Abudereyimu gave important suggestions and did significant changes. All authors reviewed and approved the final version of the paper.
This work was supported by a grant from National Key Research and Development Plan Projects (2018YFC1311503). The authors thank all individuals who participated in the present study and also thank all participants of the survey including the population. They acknowledge the Department of Science and Technology of Xinjiang Uygur Autonomous Region of China for funding the project.
Table S1: Baseline characteristics of the treatment and nontreatment groups of the study population.