Bayesian and Probit comparative analysis of risk factors for high blood pressure among adults in Kassena-Nankana Districts, Ghana: An AWI-Gen sub-study

Bayesian and Probit comparative analysis of risk factors for high blood pressure among adults in Kassena-Nankana Districts, Ghana: An AWI-Gen sub-study | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Bayesian and Probit comparative analysis of risk factors for high blood pressure among adults in Kassena-Nankana Districts, Ghana: An AWI-Gen sub-study Richmond Balinia Adda, Ida Anuwoje Logubayom Abonongo, Cornelius Debpuur, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7348872/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 6 You are reading this latest preprint version Abstract Background : Hypertension prevalence is rising in rural Ghana, but risk factor identification is constrained by methodological limitations in sparse data. This study comparatively applied frequentist (Probit) and Bayesian models in identifying hypertension risk factors and evaluating consistency and uncertainty quantification. Methods : This was a cross-sectional analysis of 2,010 adults in the Kassena-Nankana districts. Ordered Probit and Bayesian models were used to assess associations between socio-demographic, behavioral and blood pressure variables. Model fit was compared via AIC/BIC. Results : The prevalence of hypertension in the study population was 21.9Model comparison showed the ordered Probit model had superior fit (AIC/BIC: 4644.8/4734.2) to the Bayesian model (AIC/BIC: 5137.5/5165.2). Both models consistently identified male sex (Probit β=0.309, p<0.001; Bayesian mean β=0.264, 95% CrI: 0.15–0.34), older age (55–60 years: Probit β=0.514, p<0.001; Bayesian β=0.499, 95% CrI: 0.41–0.58), and higher BMI (Probit β=0.052, p<0.001; Bayesian β=0.052, 95% CrI: 0.04–0.07) as significant risk factors. However, the Bayesian model showed greater uncertainty for marital status (β=0.272, 95% CrI: -0.08–0.56) and smoking (β=-0.260, 95% CrI: -0.29 – -0.01) compared to the Probit model. Both models revealed an inverse association of smoking with hypertension, while pesticide exposure showed conflicting directions and this warrants further investigation. Conclusion : Male sex, older age, and higher BMI were consistent predictors of hypertension. Inverse associations with smoking and conflicting effects of pesticide exposure warrant further investigation. These findings highlight the need for targeted, context-specific interventions in rural Ghana. While the Probit model demonstrated a better fit, the Bayesian approach provided deeper insight into uncertainty within sparse subgroups, supporting their complementary use in hypertension research, especially in resource-limited settings. Hypertension Risk factors Health Demographic Surveillance Site H3Africa AWI-Gen Ghana Kassena-Nankana Probit & Bayesian models LMICs implementation science Figures Figure 1 Figure 2 Figure 3 Introduction Hypertension, also known as the "silent killer," is one of the most common non-communicable diseases (NCDs) globally and poses a significant public health concern due to its strong association with cardiovascular diseases (CVDs). An estimated 1.28 billion individuals worldwide are affected by hypertension, with a disproportionately high burden in low- and middle-income countries (LMICs), where detection, treatment, and control rates remain inadequate [1,2] . Hypertension is highly prevalent in sub-Saharan Africa (SSA), driven by rapid urbanization, changing lifestyles, and limited access to healthcare services [2,3] . Ghana reflects this pattern, with national hypertension prevalence estimates ranging from 19% to 55%, depending on urban–rural context and population subgroup [4,5] . Despite this alarming prevalence, awareness, diagnosis, treatment, and control rates remain low, contributing substantially to the national burden of cardiovascular morbidity and mortality [3,6] . The Kassena-Nankana districts in Ghana's Upper East Region present a unique epidemiological context for hypertension research. While several studies have examined hypertension risk factors in urban Ghana, rural areas have received comparatively less attention despite exhibiting distinct socio-demographic and lifestyle characteristics [7,8] . The Kassena-Nankana districts, which are predominantly rural and reliant on agriculture, are experiencing a rise in hypertension prevalence, compounded by limited healthcare infrastructure, increasing life expectancy, and shifting health behaviours [8,9] . This study seeks to address this research gap by investigating the risk factors associated with hypertension among adults aged 40–60 years in the Kassena-Nankana districts. Using data from the H3Africa AWI-Gen project [10] , the study applies both frequentist (ordered Probit) and Bayesian models to assess the influence of socio-demographic, behavioural, and metabolic risk factors on blood pressure status. The comparative application of Probit and Bayesian models is justified by their complementary strengths: Probit models offer interpretability and efficiency for large datasets, while Bayesian models provide robust estimates under uncertainty, particularly in data-sparse subgroups [11,12] . Employing both approaches allows for a more robust and nuanced understanding of the predictors of hypertension in rural Ghana. In doing so, this study provides an in-depth analysis of hypertension risk factors, building upon existing research to present a comprehensive examination of the local epidemiology of hypertension. The findings are expected to contribute to the growing body of evidence required for effective policy formulation, resource allocation, and the implementation of public health strategies to address hypertension and Strategic Objective 3.1 of Ghana's National NCD Policy (2022–2027) for rural CVD prevention [13] . Methodology Study design and population This was a secondary analysis of the H3Africa AWI-Gen project, which adopted a population-based cross-sectional design [14] . The primary study recruited adults aged 40–60 years who had resided in the Navrongo Health and Demographic Surveillance System (HDSS) coverage area for at least 10 years [15] . A stratified random sampling was used to select participants for the study. In the first instance, pure Kasem and Nankam speaking zones were selected, and in the second instance, a simple random sampling was applied to selected adults 40–60 years old, aiming for an equal sex ratio. Pregnant women and frail individuals were excluded from the study. A total of 2,010 participants were used for this study. Data collection Sociodemographic variables Sociodemographic data were obtained through structured, interviewer-administered questionnaires conducted in Kasem or Nankam, the local languages preferred by participants [16] . Participants’ ages were recorded in complete years and categorized into 10-year age bands: 40–44, 45–49, 50–54, and 55–60 [17] . Biological sex was recorded as male or female [18] . Marital status was dichotomized into currently married (legally or customarily partnered) and currently unmarried (single, divorced, widowed, or separated) [19] . Educational attainment was categorized as no formal education (0 years of schooling) or formal education (≥1 year of schooling) [20] . Employment status was defined as unemployed (no income-generating activities) or employed (engaged in either formal or informal work, including subsistence farming) [21] . Household socioeconomic status (SES) was estimated using a composite wealth index based on housing materials, access to utilities (e.g., water and electricity), and ownership of durable assets such as livestock and vehicles. Principal component analysis (PCA) was used to classify households into low SES (lowest tertile) or high SES (upper two tertiles) [22] . Socioeconomic variables Socioeconomic data were collected through interviewer-administered surveys using tools validated by the Navrongo Health and Demographic Surveillance System (NHDSS) protocol [18] . Behavioral variables Behavioral characteristics were assessed using instruments adapted from the WHO-STEPS protocol [3] . Smoking status was classified as non-smoker (never smoked) or current smoker (smoked ≥1 cigarette/day in the past 30 days) [9] . Alcohol consumption was defined as no intake (0 standard drinks/week) or current intake (≥1 standard drink/week) [23] . Dietary intake of fruits and vegetables was quantified using a 24-hour recall and food frequency questionnaire [24] . Physical activity was measured as total weekly moderate-to-vigorous physical activity (MVPA), calculated from self-reported duration and frequency of activities such as farming, walking, and household chores. Accelerometer validation (ActiGraph GT3X+) was conducted in a 10% subsample [25] . Environmental variables Environmental exposures were evaluated following the Africa Wits-INDEPTH Partnership for Genomic Studies (AWI-Gen) protocol [16] . Pesticide exposure was defined as exposed (self-reported handling or residing within 100 meters of farmland during spraying seasons) or non-exposed [26] . Participants’ residential locations were geo-referenced and classified into urban or rural clusters based on NHDSS-defined zones (central, east, west, north, south) [15] . Water sources were categorized as improved (e.g., piped water, boreholes) or unimproved (e.g., surface water, unprotected wells) according to WHO/UNICEF Joint Monitoring Programme standards [27] . Blood pressure measurement and classification Blood pressure was measured using a validated, automated sphygmomanometer (Omron M6; Omron Corporation, Kyoto, Japan) [15] . After a five-minute rest, three readings were taken at two-minute intervals. The first was discarded, and the mean of the remaining two was used for analysis [15] . Blood pressure classifications followed the ACC/AHA 2017 guidelines [8] . Pulse pressure (PP) was calculated as PP = SBP − DBP and mean arterial pressure (MAP) as MAP = DBP + (PP/3) [15] . Physical activity (MVPA) assessment MVPA was quantified using questionnaires adapted from the WHO-STEPS protocol [25] . Weekly MVPA was computed as: MVPA (hours/week) = ∑ (Activity Duration × Frequency) × MET coefficient MET coefficients were obtained from the Compendium of Physical Activities. A 10% sample was validated using ActiGraph GT3X+ accelerometers [25] . Anthropometry and BMI calculation Weight was measured to the nearest 0.1 kg using a SECA 874 scale, with participants wearing light clothing. Height was measured using a SECA 217 stadiometer following the Frankfurt Plane protocol [17] . BMI was calculated using the formula: BMI = Weight (kg) / [Height (m)]² All anthropometric procedures followed International Society for the Advancement of Kinanthropometry (ISAK) guidelines [25] . Lipid profile analysis Venous blood samples were collected after an overnight fast of at least 8 hours. Serum lipid concentrations were measured using enzymatic colorimetric assays on a Roche Cobas c311 analyser at the Navrongo Health Research Centre [17] . Methods included: Total cholesterol (TC): CHOD-PAP HDL cholesterol: Homogeneous enzymatic colorimetry LDL cholesterol: Calculated using the Friedewald equation (if TG < 4.5 mmol/L) Triglycerides (TG): GPO-PAP All assays followed IFCC standards, with intra-assay variation maintained below 3% [25] . Ethics Approval The AWI-Gen study received ethics approval from the Human Research Ethics Committee (HREC), University of the Witwatersrand (#M12109) and was renewed in 2017, #M170880).), the Ghana Health Service Ethics Review Committee (#GHS-ERC:05/05/2015). Additional approval was received from the Navrongo Health Research Centre Institutional Review Board (#NHRCIRB178). Informed consent was obtained from all participants. Statistical analysis Descriptive statistics were summarized using frequencies and percentages for categorical variables, and either means with standard deviations for normally distributed continuous variables or medians with interquartile ranges (IQRs) for skewed data. For inferential analysis, the study employed both frequentist (Probit) and Bayesian approaches. The Bayesian framework was specifically selected to: (1) incorporate prior epidemiological knowledge from similar rural African populations [26] , (2) address sparse data challenges in subgroup analyses [27] , and (3) offer probabilistic interpretations of uncertainty using credible intervals [28] . Bayesian models enabled probabilistic inference for underpowered groups like female smokers (n = 35), hence were rerun with Cauchy(0, 2.5) priors to confirm robustness. Convergence of the Markov chain Monte Carlo (MCMC) simulations for the Bayesian model was assessed using the Gelman-Rubin R̂ statistic [29] . This statistic compares the between-chain and within-chain variances, with values close to 1 indicating convergence. An R̂ value below 1.1 is generally considered acceptable [29] . Model comparison was performed using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) [27] , with lower values indicating superior model fit. Additional evaluation of model performance was based on maximum likelihood estimates and the number of parameters included in each model. The threshold for statistical significance was set at p<0.05. All analyses were conducted using Stata version 17.0 (Stata Corp LLC, College Station, TX, USA). Results and Discussion Background characteristics of the study Figure 1 summarizes blood pressure categories among the study population. The proportion of individuals with normal blood pressure was 46%. Prehypertension constituted 32%, hypertension 14%, and sever hypertension 8%. Sociodemographic and behavioral characteristics stratified by sex are presented in Table 1 . Women made up a slightly higher proportion of the sample (54.2%) compared to men (45.8%). Statistically significant differences were observed across most characteristics. Age distribution differed significantly between sexes (p < 0.001), with a higher proportion of women (39.9%) falling in the 55–60-year group compared to men (31.4%). Marital status was also significantly associated with sex (p < 0.001); a larger proportion of men were currently married (76.9%) compared to women (56.0%), whereas women were more likely to be unmarried (44.0% vs. 23.1%). Educational status showed marked gender differences (p < 0.001): 77.7% of women had no formal education, compared to 61.7% of men, while formal education was more common among men (38.3%) than women (22.5%). Significant sex differences were also evident in health-related behavioural factors. A striking contrast was seen in smoking status (p < 0.001), with 64.2% of men currently smoking compared to just 3.2% of women. Similarly, alcohol intake varied by sex (p < 0.001), as 92.1% of men reported current alcohol use compared to 78.6% of women. Pesticide exposure also differed significantly by sex (p = 0.003), with a greater proportion of men (60.0%) exposed compared to women (48.6%). Interestingly, no significant sex differences were observed in occupational status (p = 0.165), with employment rates nearly equal across sexes (approximately 40%). Table 1 also presents the continuous variables that are summarized using medians and interquartile ranges (IQRs) due to their non-normal distribution. Statistically significant sex differences were observed across several variables. In terms of physical activity and metabolic risk factors, men reported slightly higher moderate-to-vigorous physical activity (MVPA) levels, with a median of 2,040 hours/week (IQR: 2,400), compared to 1,920 hours/week (IQR: 2,400) for women (p = 0.045). Women, however, had significantly higher body mass index (BMI) values (median: 23.1 kg/m², IQR: 5.2) compared to men (median: 21.8 kg/m², IQR: 4.9) (p = 0.002). Waist-to-hip ratio was also significantly higher among women (0.87 [IQR: 0.06]) than men (0.85 [IQR: 0.08]) (p = 0.015), although no significant sex difference was observed for waist circumference (p = 0.071). Regarding lipid profiles, women exhibited higher HDL cholesterol levels (median: 1.20 mmol/L, IQR: 0.50) than men (median: 1.00 mmol/L, IQR: 0.38; p < 0.001), and higher total cholesterol levels (3.32 vs. 3.08 mmol/L; p = 0.038). However, no significant differences were noted for LDL cholesterol (p = 0.128) or triglycerides (p = 0.294). Table 1 Sociodemographic, Behavioral, and Clinical Characteristics of Participants by Gender Variable Women 1082 (54.2%) Men 916 (45.8%) P-value Age Group (years) < 0.001 40–44 145 (13.4%) 169 (18.5%) 45–49 272 (25.1%) 241 (26.3%) 50–54 235 (21.7%) 218 (23.8%) 55–60 432 (39.9%) 288 (31.4%) Marital Status < 0.001 Currently married 606 (56.0%) 704 (76.9%) Currently unmarried 478 (44.0%) 212 (23.1%) Educational Status < 0.001 No formal education 841 (77.7%) 565 (61.7%) Formal education 243 (22.5%) 351 (38.3%) Occupational Status 0.165 Unemployed 651 (60.2%) 551 (60.2%) Employed 431 (39.8%) 365 (39.8%) Smoking Status < 0.001 Not smoking 1047 (96.8%) 328 (35.8%) Currently smoking 35 (3.2%) 588 (64.2%) Alcohol Consumption < 0.001 No alcohol intake 232 (21.4%) 72 (7.9%) Current intake 850 (78.6%) 844 (92.1%) Pesticide Exposure 0.003 Non-exposed 556 (51.4%) 366 (40.0%) Exposed 526 (48.6%) 550 (60.0%) MVPA (hours/week) 1,920 [2,400] 2,040 [2,400] 0.045 BMI (kg/m²) 23.1 [5.2] 21.8 [4.9] 0.002 Waist circumference (mm) 738 [95] 729 [105] 0.071 Waist-hip ratio 0.87 [0.06] 0.85 [0.08] 0.015 HDL (mmol/L) 1.20 [0.50] 1.00 [0.38] < 0.001 LDL (mmol/L) 1.66 [1.01] 1.57 [0.94] 0.128 Cholesterol (mmol/L) 3.32 [1.20] 3.08 [1.13] 0.038 Triglycerides (mmol/L) 0.58 [0.28] 0.55 [0.32] 0.294 Legend: Values are presented as n (%) for categorical variables and Median [Interquartile Range] for continuous variables. p-values are from Chi-square tests for categorical variables and Mann–Whitney U tests for continuous variables. Distribution of blood pressure categories across sociodemographic, behavioral and environmental factors Table 2 presents the distribution of blood pressure categories (normal, prehypertensive, hypertensive, and severe hypertension) across various sociodemographic and behavioral variables, stratified by sex. These results are descriptive and show how blood pressure classifications vary within each subgroup rather than implying statistical associations or correlations. Among women, age-related differences in blood pressure distribution were evident. The proportion of women with normal blood pressure decreased progressively with age, from 60.4% in the youngest age group to 42.6% in the oldest, while the proportion with severe hypertension increased from 2.1–13.0% (p < 0.001). A similar pattern was observed among men, with normal blood pressure declining from 45.6–32.8% and severe hypertension rising from 3.6–11.9% across age categories (p < 0.001). Marital status showed variation in blood pressure distribution among women, where currently married individuals had a higher proportion of normal blood pressure (53.4%) compared to unmarried women (42.1%). Additionally, a larger share of unmarried women fell into the severe hypertension category (10.9%) compared to married women (6.3%), with these differences reaching statistical significance (p = 0.001). Among men, however, the distribution of blood pressure categories did not significantly differ by marital status (p = 0.423). Educational status, occupational status, smoking, and alcohol use did not show notable differences in blood pressure category distributions in either sex (p > 0.05). However, there was a slight trend in men where the proportion of severe hypertension was higher among smokers (8.3%) compared to non-smokers (7.3%) (p = 0.050), though this did not reach conventional levels of statistical significance. For pesticide exposure, women showed notable differences across blood pressure categories (p = 0.009). Those exposed to pesticides had a lower proportion of severe hypertension (5.5%) compared to the non-exposed group (11.0%), although the non-exposed group had a slightly higher prevalence of normal blood pressure. Among men, the distribution of blood pressure categories did not differ significantly by pesticide exposure (p = 0.473). Finally, BMI classification showed clear differences in blood pressure distribution for both sexes. Among women, the proportion with normal blood pressure declined from 57.3% in the underweight category to 22.2% in the obese category, while severe hypertension rose from 8.4–20.0% (p < 0.001). Men followed a similar trend, with normal blood pressure decreasing and severe hypertension increasing across BMI categories (p = 0.001). Regression analyses Parameter estimates from both the frequentist ordered probit model and the Bayesian ordered probit model were compared to explore differences in predictors of elevated blood pressure among adults in the Kassena-Nankana districts. Weakly informative normal priors ~ N(0,10) were used for regression coefficients in the Bayesian analyses. As presented in Table 3 , both models identified male sex as a significant factor associated with higher blood pressure categories. The frequentist model estimated a coefficient of 0.309 (95% CI: 0.172–0.446; p < 0.001), while the Bayesian model produced a posterior mean of 0.264 with a 95% credible interval (CrI) of 0.15–0.34, which excluded zero, indicating high certainty in the direction of effect. Age was also a consistent predictor across models, showing a positive gradient in risk. For individuals aged 45–49, the frequentist model reported a non-significant coefficient of 0.127 (95% CI: − 0.036 to 0.291; p = 0.129), whereas the Bayesian model estimated a posterior mean of 0.14 (95% CrI: 0.08–0.22), suggesting moderate certainty of increased risk. For age groups 50–54 and 55–60, both models produced significant and increasing coefficients, with the highest effect seen in the 55–60 age group (frequentist: 0.514, 95% CI: 0.356–0.672; Bayesian: 0.499, 95% CrI: 0.41–0.58; p < 0.001 in both cases). Marital status showed a statistically significant association in the frequentist model, with unmarried individuals having a coefficient of 0.143 (95% CI: 0.033–0.253; p = 0.011). However, the Bayesian estimate for this variable (0.272; 95% CrI: − 0.08 to 0.56) included zero, reflecting greater uncertainty in the strength and direction of this effect. Similarly, formal education and employment status were not statistically significant in either model. For instance, the frequentist model estimated a coefficient of 0.054 (95% CI: − 0.058 to 0.167; p = 0.343) for education, while the Bayesian estimate was slightly negative (–0.116; 95% CrI: − 0.24 to 0.01), though still uncertain. Alcohol consumption showed a weak and statistically non-significant positive relationship with blood pressure classification in both models (frequentist: 0.100, 95% CI: − 0.044 to 0.245; p = 0.172; Bayesian: 0.200, 95% CrI: − 0.05 to 0.25). In contrast, smoking status had a negative effect on blood pressure classification, with both models showing significant associations. The frequentist model yielded − 0.148 (95% CI: − 0.291 to − 0.004; p = 0.044), while the Bayesian model reported − 0.260 (95% CrI: − 0.29 to − 0.01), indicating a lower probability of being in a higher blood pressure category among current smokers. For pesticide exposure, both models suggested a positive effect, although the confidence interval reported in the frequentist model appears inconsistent (estimate = 0.141; 95% CI: − 0.243 to − 0.04), possibly due to reporting error. The Bayesian model estimated a coefficient of 0.220 (95% CrI: − 0.24 to − 0.04), but since the interval crosses zero, the effect should be interpreted with caution. Body mass index (BMI) showed a consistent and statistically significant association with elevated blood pressure across both models. Both the frequentist and Bayesian models produced identical point estimates of 0.052 (frequentist 95% CI: 0.037–0.066; Bayesian 95% CrI: 0.04–0.07; p < 0.001), confirming BMI as a robust predictor. In contrast, moderate-to-vigorous physical activity (MVPA) and total cholesterol were not associated with blood pressure classification. Both models reported null effects for MVPA (coefficient = 0.000) and cholesterol (frequentist: 0.000, 95% CI: − 0.001 to 0.001; Bayesian: − 0.001, 95% CrI: − 0.01 to 0.00), indicating negligible influence on the outcome. Model fit statistics showed better performance for the frequentist ordered probit model, with lower AIC (4644.779) and BIC (4734.225) values compared to the Bayesian model (AIC = 5137.520; BIC = 5165.159), suggesting a relatively superior fit under the frequentist framework. Table 3 Parameter estimates of both ordered Probit versus Bayesian ordered Probit model Factor Probit Coefficient [95% CI] P-value Bayesian Mean [95% CrI] P-value Sex (Male) 0.309 [0.172, 0.446] < 0.001 0.264 [0.15, 0.34] < 0.001 Age 45–49 0.127 [–0.036, 0.291] 0.129 0.140 [0.08, 0.22] ≈ 0.004 Age 50–54 0.231 [0.061, 0.401] 0.008 0.226 [0.15, 0.36] ≈ 0.002 Age 55–60 0.514 [0.356, 0.672] < 0.001 0.499 [0.41, 0.58] < 0.001 Marital Status (Unmarried) 0.143 [0.033, 0.253] 0.011 0.272 [–0.08, 0.56] ≈ 0.13 Educational Status (Formal) 0.054 [–0.058, 0.167] 0.343 –0.116 [–0.24, 0.01] ≈ 0.07 Occupational Status (Employed) 0.047 [–0.052, 0.15] 0.351 0.082 [–0.06, 0.15] ≈ 0.21 Alcohol Consumption (Current) 0.100 [–0.044, 0.245] 0.172 0.200 [–0.05, 0.25] ≈ 0.09 Smoking Status (Current) –0.148 [–0.291, − 0.004] 0.044 –0.260 [–0.29, − 0.01] ≈ 0.04 Pesticide Exposure (Exposed) 0.141 [–0.243, 0.04] 0.007 0.220 [–0.24, 0.04] ≈ 0.04 BMI 0.052 [0.037, 0.066] < 0.001 0.052 [0.04, 0.07] < 0.001 Cholesterol 0.000 [–0.001, 0.001] 0.882 –0.001 [–0.01, 0.00] ≈ 0.29 AIC/BIC 4644.779 /4734.225 5137.520 /5165.159 a. Probit model presents beta coefficients as parameter estimates. b. Bayesian model reports the mean estimates for parameter estimation. c. Bayesian p-values aren't traditionally computed but based on whether 95% CrI includes zero, hence p-values provided are but approximations. Comparative analysis revealed hypertension prevalence was 1.8× higher in 55–60-year-olds vs. 40–44-year-olds (24.9% vs 13.9%, p < 0.001). Figure 2 visualizes the key associations through forest plots of significant predictors. Figure 2 presents a forest plot of the key predictors identified in the ordered Probit model. Male sex (β = 0.309), older age (55–60 years: β = 0.514), unmarried status (β = 0.143), and higher BMI (β = 0.052) were significantly associated with increased odds of higher blood pressure categories. In contrast, smoking (β = -0.148) and pesticide exposure (β = -0.141) demonstrated inverse associations with blood pressure status. The plot visually highlights the magnitude and direction of these associations, with 95% confidence intervals indicating the precision of the estimates. Model Diagnostics (Bayesian Ordered Probit Model) The diagnostic assessment of the Bayesian Ordered Probit Model utilized Markov Chain Monte Carlo (MCMC) trace plots to examine whether the chains mixed properly and reached convergence. The MCMC sampling of stations to evaluate the posterior distribution and its distribution stability is presented in Fig. 3 . The MCMC algorithm achieved convergence according to the visual assessment of the trace plot, which displayed appropriate chain mixing results. All variables, including age, sex, occupational status, MVPA, and educational and marital status, required approximately 35,000 iterations to reach stationary distribution states, alongside excellent mixing quality. The burn-in process for smoking status, together with cholesterol, alcohol consumption, BMI, and pesticide exposure, lasted between 50,000 and 70,000 iterations to achieve convergent states. The Gelman-Rubin R̂ statistic confirmed convergence for all parameters. The Bayesian model (probit: R̂ = 1.09) achieved values near 1.0, indicating stable sampling from the target distribution. These diagnostics indicate that posterior estimates are reliable and Bayesian inference in this study operates robustly. Discussion This study identified several key predictors of hypertension that align with global epidemiological patterns while also revealing context-specific risk factors relevant to rural Ghana. The application of both frequentist and Bayesian ordered probit models allowed for cross-validation of findings and improved interpretability through probabilistic estimates and uncertainty quantification. Both models consistently demonstrated that male sex and advancing age, particularly among individuals aged 55–60 years, were significant predictors of higher blood pressure categories. The frequentist model reported a statistically significant coefficient for male sex (0.309; 95% CI: 0.172–0.446; p < 0.001), which was corroborated by the Bayesian model (posterior mean: 0.264; 95% CrI: 0.15–0.34), with the credible interval excluding zero, indicating high posterior certainty. A similar pattern was observed with age, with the highest risk found in the 55–60 age group (frequentist: 0.514; Bayesian: 0.499). These findings are consistent with global data linking older age and male sex to hypertension due to cumulative exposure to lifestyle risk factors and biological susceptibility [ 30 ]. Body mass index (BMI) also emerged as a consistent and statistically significant predictor across both models (coefficient: 0.052), reinforcing the established role of obesity as a modifiable cardiometabolic risk factor [ 31 ]. The convergence of results between models further validates this association within rural African populations. Interestingly, smoking status showed a negative association with hypertension in both models (frequentist: − 0.148, p = 0.044; Bayesian: − 0.260, 95% CrI: − 0.29 to − 0.01), contrasting with existing meta-analyses and most urban SSA studies that typically report positive associations [ 32 , 33 ]. This counterintuitive finding may reflect survival bias (e.g., premature mortality among hypertensive smokers), low smoking prevalence—especially among women—and unmeasured confounding such as intensive physical labour in agricultural settings. Future longitudinal and biomarker-based studies are needed to further explore this paradox. The study also found a positive association between pesticide exposure and hypertension, though the Bayesian credible interval included zero, indicating uncertainty. This is noteworthy in the context of widespread, often unregulated agrochemical use in rural Ghana. These results underscore the importance of public health interventions such as protective gear distribution, farmer education on pesticide toxicity, and seasonal hypertension screening, as outlined in Ghana’s National NCD Policy (2022–2027). Comparison of the two models revealed largely overlapping sets of significant predictors—male sex, older age, and elevated BMI—yet highlighted important differences in how uncertainty is represented. For example, the frequentist model found unmarried status to be statistically significant (p = 0.011), while the Bayesian model’s 95% credible interval (–0.08 to 0.56) included zero, suggesting less certainty in the effect. This difference illustrates the Bayesian model's advantage in reflecting uncertainty more transparently than conventional null hypothesis testing. From a model fit perspective, the frequentist model showed better performance based on lower AIC (4644.8) and BIC (4734.2) values compared to the Bayesian model (AIC = 5137.5; BIC = 5165.2). Although the Bayesian framework enhances interpretation, these diagnostics suggest that the frequentist model provided a better empirical fit for this dataset. Within the broader SSA context, the findings reaffirm well-established hypertension predictors while highlighting rural-specific nuances. The consistent male effect contrasts with some rural studies where shared physical workloads diminish sex differences [ 2 ]. The absence of a significant association between physical activity and blood pressure may reflect the uniformly high MVPA levels among subsistence farmers, which could mask detectable gradients observed in more sedentary urban populations [ 18 , 34 ]. Similarly, the lack of alcohol–hypertension association aligns with mixed findings from SSA, likely due to variability in beverage type and drinking patterns [ 35 ]. Our identification of male sex, aging, and elevated BMI as consistent hypertension predictors aligns with Agongo et al. (2020) in the same Navrongo cohort, reinforcing the stability of these risk factors in rural Ghana [ 17 ]. However, while Gomez-Olive et al. (2017) reported urban-rural gradients in hypertension prevalence across AWI-Gen sites, our Bayesian modeling uniquely quantifies uncertainty in sparse subgroups (e.g., female smokers) and reveals novel environmental risks like pesticide exposure in agrarian communities [ 22 ]. In the Ghanaian context, these results align with urban evidence linking hypertension with male sex and aging [ 6 ] but differ from some rural studies that report higher prevalence among women—potentially due to disparities in healthcare access or reporting bias [ 4 ]. The BMI–hypertension link reinforces concerns about nutrition transition in rural areas [ 7 , 36 ]. Additionally, the elevated hypertension risk among unmarried individuals (frequentist coefficient: 0.143) may reflect social isolation or economic vulnerability, which tend to be more pronounced in rural settings [ 21 ]. Although the association between pesticide exposure and hypertension was statistically inconclusive, the direction of the effect is in line with evidence connecting agrochemical exposure to cardiovascular and renal risks in Ghana [ 23 , 37 ]. The forest plot (Fig. 3 ) provides a clear visual summary of the magnitude and direction of these effects. Overall, the triangulation of findings across two analytical frameworks enhances confidence in the observed relationships and offers deeper insight into both consistent and uncertain predictors of hypertension in rural Ghana. Public health implications Targeted hypertension screening should be integrated into Ghana's agricultural extension services. Farmers reporting pesticide exposure require seasonal BP monitoring and protective gear training. The inverse smoking-hypertension association warrants biomarker-confirmed longitudinal studies. Conclusion This study identified key predictors of hypertension among adults in rural northern Ghana using both frequentist (ordered probit) and Bayesian models. While both approaches consistently flagged male sex, older age, higher BMI, and unmarried status as significant risk factors, their combined use added analytical value. The probit model provided efficient estimation and clearer point estimates, particularly useful for identifying primary risk factors in a large dataset. In contrast, the Bayesian model allowed for a more nuanced understanding of uncertainty, especially in subgroups with sparse data (e.g., smoking or pesticide exposure). This dual approach revealed areas where associations were robust across methods and where greater caution, or further research is warranted due to statistical uncertainty. Therefore, applying both methods strengthened the credibility of the findings and enhanced interpretability for policy decisions in resource-limited rural settings. An unexpected inverse relationship between smoking and hypertension was observed in both models, likely reflecting survival bias, low-intensity smoking, or unmeasured confounding such as occupational activity. Additionally, cholesterol levels and physical activity showed no significant associations with hypertension status, which may be attributed to the homogeneous farming lifestyle prevalent in the study population. Importantly, both statistical models identified similar patterns of association, reinforcing the robustness of the findings. However, the Bayesian model offered deeper insights into uncertainty, particularly where credible intervals crossed zero despite positive point estimates (marital status and pesticide exposure). Conversely, the frequentist model demonstrated a better overall model fit based on lower AIC and BIC values. This comparison highlights the value of using complementary modelling approaches: the frequentist model offers precision and simplicity, while the Bayesian model provides a more nuanced understanding of estimate reliability. These findings support Strategic Objective 3.1 of Ghana's National NCD Policy (2022–2027), which emphasizes community-based screening and targeted prevention in rural populations. We recommend integrating hypertension surveillance into agricultural extension services to better reach farming communities, especially given the potential links between pesticide exposure and elevated blood pressure. Declarations Data availability The dataset supporting the conclusions of this article is available from the H3Africa AWI-Gen project upon reasonable request. Due to ethical restrictions and data sharing policies, de-identified participant data can be accessed with appropriate approvals from the Navrongo Health Research Centre Institutional Review Board. Acknowledgment We thank all study participants from the Kassena-Nankana districts for their cooperation and contributions to this research. We also acknowledge the staff and management of the Navrongo Health Research Centre for their logistical and institutional support. The authors are grateful to the H3Africa AWI-Gen project and the H3Africa Consortium for providing access to the data and resources that made this work possible. Funding This research was supported by the H3Africa AWI-Gen project, funded by the National Human Genome Research Institute (NHGRI), the Office of the Director (OD) at the National Institutes of Health (NIH) of the United States of America, and the Wellcome Trust (Grant numbers: U54HG006938 and 074548/Z/04/Z). Conflict of Interest The authors declare that there are no conflicts of interest regarding the publication of this article. Author Contributions Richmond Balinia Adda contributed to the conceptualisation, methodology, data analysis, writing (original draft preparation), visualization. Ida Anowuje contributed to the methodology, supervision, data curation, writing (review & editing). Godfred Agongo contributed to data curation, supervision, writing (review & editing). Cornelius Debpuur contributed to writing (review & editing). Abraham Rexford Oduro contributed to the project administration, resources, writing (review & editing). Engelbert A. Nonterah contributed to conceptualisation, supervision, methodology, writing (review & editing). References Mills KT, Stefanescu A, He J. The global epidemiology of hypertension. Nat Rev Nephrol . 2020;16(4):223–237. NCD Risk Factor Collaboration (NCD-RisC). Worldwide trends in hypertension prevalence and progress in treatment and control from 1990 to 2019: A pooled analysis of 1201 population-representative studies with 104 million participants. Lancet . 2021;398(10304):957–970. World Health Organization. STEPwise approach to surveillance (STEPS). Geneva: WHO; 2020. Available from: https://www.who.int/ncds/surveillance/steps/en/ Ramsay M, Crowther N, Fraser PA, Norris SA. The AWI-Gen Collaborative Centre: A resource for genomic studies of cardiometabolic diseases in Africa. Glob Health Epidemiol Genomics . 2016;1:e20. Oduro AR, Wak G, Azongo D, Debpuur C, Wontuo P, Kondayire F, et al. Profile of the Navrongo Health and Demographic Surveillance System. Int J Epidemiol . 2012;41(4):968–976. Agyemang C. Rural and urban differences in blood pressure and hypertension in Ghana, West Africa. Public Health . 2006;120:525–533. Nonterah EA, Depbuur C, Agongo G, et al. Sociodemographic and behavioral determinants of abnormal body mass index among an adult population in rural Northern Ghana: The AWI-Gen study. Glob Health Action . 2018;11:146–588. American College of Cardiology/American Heart Association (ACC/AHA). 2017 Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults. Hypertension . 2018;71(6):e13–e115. Centers for Disease Control and Prevention (CDC). Tobacco Use Definitions. Atlanta, GA: CDC; 2020. Available from: https://www.cdc.gov/nchs/nhis/tobacco/tobacco_definitions.htm National Institute on Alcohol Abuse and Alcoholism (NIAAA). What is a standard drink? Bethesda, MD: NIAAA; 2023. Available from: https://www.niaaa.nih.gov/alcohols-effects-health/overview-alcohol-consumption/what-standard-drink Food and Agriculture Organization of the United Nations (FAO). Dietary Assessment: A Resource Guide to Method Selection and Application in Low Resource Settings . Rome: FAO; 2018. Ainsworth BE, Haskell WL, Herrmann SD, Meckes N, Bassett DR Jr, Tudor-Locke C, et al. 2011 Compendium of Physical Activities: A second update of codes and MET values. Med Sci Sports Exerc . 2011 Aug;43(8):1575–81. International Society for the Advancement of Kinanthropometry (ISAK). International Standards for Anthropometric Assessment . Potchefstroom, South Africa: ISAK; 2012. IFCC (International Federation of Clinical Chemistry and Laboratory Medicine). Standardization of Laboratory Measurements. Guidelines for clinical chemistry laboratories. [Accessed via local lab protocols]. Rutstein SO, Johnson K. The DHS Wealth Index . Calverton, MD: ORC Macro; 2004. WHO/UNICEF Joint Monitoring Programme. Progress on Household Drinking Water, Sanitation and Hygiene 2000–2020: Five years into the SDGs . Geneva: WHO and UNICEF; 2021. Agongo G, Nonterah EA, Amenga-Etego L, Debpuur C, Kaburise MB, Ali S, et al. Blood pressure indices and associated risk factors in a rural West African adult population: insights from an AWI-Gen Substudy in Ghana. Int J Hypertens . 2020;2020:11. Mayige M, Kagaruki G, Ramaiya K, Swai AB. Physical activity and cardiovascular disease risk factors among adults in Tanzania. BMC Public Health . 2016;16(1):1–9. Ofori-Asenso R, Agyeman AA, Laar A, Boateng D. Overweight and obesity epidemic in Ghana—A systematic review and meta-analysis. J Clin Hypertens . 2018;20(6):846–853. Owolabi EO, Goon DT, Adeniyi OV. Alcohol consumption patterns and associated hypertension risks in rural Ghana: A cross-sectional study. BMC Nutr . 2022;8(1):1–10. Bosu WK, Reilly ST, Aheto JMK, Zucchelli E. Hypertension in older adults in Africa: A systematic review and meta-analysis. PLoS One . 2019;14(4):e0214934. Amegah AK. Pesticides and Cardiovascular Health in Ghana. Environ Health Perspect . 2024;132(1):017001. Gomez-Olive FX, Ali SA, Made F, et al. Regional and sex differences in the prevalence and awareness of hypertension: an H3Africa AWI-Gen study across 6 sites in sub-Saharan Africa. Glob Heart . 2017;12:81–90. Claeskens G, Hjort NL. Model Selection and Model Averaging . Cambridge: Cambridge University Press; 2008. Spiegelhalter DJ, Abrams KR, Myles JP. Bayesian approaches to clinical trials and health-care evaluation . Chichester: John Wiley & Sons; 2004. McElreath R. Statistical Rethinking: A Bayesian Course with Examples in R and Stan . CRC Press; 2020. Kruschke JK. Doing Bayesian Data Analysis: A Tutorial with R, JAGS, and Stan (2nd ed.). Academic Press; 2015. Abu-Saad K, Avital N, Elias N, Kalter-Leibovici O. Associations between anthropometric measures and blood pressure among adolescents: a cross-sectional study. Am J Hypertens . 2014 Dec;27(12):1511–20. Ghana Health Service. 2023 NCD Progress Report . Accra: GHS; 2023. Mills KT, Bundy JD, Kelly TN, Reed JE, Kearney PM, Reynolds K, Chen J, He J. Global disparities of hypertension prevalence and control: a systematic analysis of population-based studies from 90 countries. Circulation . 2016;134(6):441–450. Hall JE, do Carmo JM, da Silva AA, Wang Z, Hall ME. Obesity-induced hypertension: interaction of neurohumoral and renal mechanisms. Circ Res . 2015;116(6):991–1006. Goma FM, Nzala SH, Babaniyi O, Songolo P, Zyaambo C, Rudatsikira E, et al. Prevalence of hypertension and its correlates in Lusaka urban district of Zambia: a population-based survey. Int Arch Med . 2011;4:34. Addo J, Smeeth L, Leon DA. Hypertension in sub-Saharan Africa: a systematic review. Hypertension . 2007;50(6):1012–1018. Sobngwi E, Mbanya JC, Unwin NC, Porcher R, Kengne AP, Fezeu L, et al. Physical activity and its relationship with obesity, hypertension and diabetes in urban and rural Cameroon. Int J Obes . 2002;26(7):1009–1016. Peltzer K, Phaswana-Mafuya N. Hypertension and associated factors in older adults in South Africa. Cardiovasc J Afr . 2013;24(3):66–71. Popkin BM. Nutrition transition and the global diabetes epidemic. Curr Diab Rep . 2015;15(9):64. Bampoe-Addo AA, Nartey ET, Odjidja EN, et al. Agrochemical exposure and cardiovascular health risks among Ghanaian farmers. BMC Public Health . 2021;21:1782. Table 2 Table 2 is available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Table2.docx Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 12 Sep, 2025 Reviewers invited by journal 10 Sep, 2025 Editor invited by journal 14 Aug, 2025 Editor assigned by journal 14 Aug, 2025 Submission checks completed at journal 14 Aug, 2025 First submitted to journal 11 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7348872","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":515497281,"identity":"44008299-8b7f-464e-a9ee-7326d230f183","order_by":0,"name":"Richmond Balinia Adda","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIiWNgGAWjYDACZgY2EMXYwJBgwPAByGJjJ0UL4wyQFmbC9iC0MPNADMEPzNuZnz26UXFHtr89eeNnm1/b5PmYGRg/fMzBrUXmMJu5cc6ZZ8Yzzjwrls7tu23YxszALDlzG24tEsw8bNK5bYcTG27kGEjn9txmBGphY+YlRsv8GznGvy17btsTr2XDjRwzaYYftxOJ0MJmJp1z5rDxxjPPyix7G24ntzEzNuP3C//hZ9I5FYdl5x1P3nzjx5/btvPbmw9++IhHCypgbAOTDcSqB4E/pCgeBaNgFIyCkQIAioVPt8eBAK8AAAAASUVORK5CYII=","orcid":"","institution":"C.K. Tedam University of Technology and Applied Sciences","correspondingAuthor":true,"prefix":"","firstName":"Richmond","middleName":"Balinia","lastName":"Adda","suffix":""},{"id":515497282,"identity":"c8633de1-5040-4596-bf65-7bb6330adb74","order_by":1,"name":"Ida Anuwoje Logubayom Abonongo","email":"","orcid":"","institution":"C.K. Tedam University of Technology and Applied Sciences","correspondingAuthor":false,"prefix":"","firstName":"Ida","middleName":"Anuwoje Logubayom","lastName":"Abonongo","suffix":""},{"id":515497283,"identity":"dd275b05-d5b6-4ccc-8189-86cfec784bb1","order_by":2,"name":"Cornelius Debpuur","email":"","orcid":"","institution":"Ghana Health Service","correspondingAuthor":false,"prefix":"","firstName":"Cornelius","middleName":"","lastName":"Debpuur","suffix":""},{"id":515497284,"identity":"74bc4b5c-910f-4d3e-8fbc-37ecd53b05bf","order_by":3,"name":"Abraham Rexford Oduro","email":"","orcid":"","institution":"Navrongo Health Research Centre","correspondingAuthor":false,"prefix":"","firstName":"Abraham","middleName":"Rexford","lastName":"Oduro","suffix":""},{"id":515497285,"identity":"dad4b612-af12-4896-9ddf-f998209ac613","order_by":4,"name":"Godfred Agongo","email":"","orcid":"","institution":"C.K. Tedam University of Technology and Applied Sciences","correspondingAuthor":false,"prefix":"","firstName":"Godfred","middleName":"","lastName":"Agongo","suffix":""},{"id":515497286,"identity":"7ab14632-5b41-4828-badc-cdf893b46fa3","order_by":5,"name":"Engelbert A. Nonterah","email":"","orcid":"","institution":"Ghana Health Service","correspondingAuthor":false,"prefix":"","firstName":"Engelbert","middleName":"A.","lastName":"Nonterah","suffix":""}],"badges":[],"createdAt":"2025-08-11 18:08:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7348872/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7348872/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91513252,"identity":"e6b077d7-d02e-46a0-a82c-317e3954880e","added_by":"auto","created_at":"2025-09-17 08:58:51","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":91902,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eDistribution of blood pressure categories among study participants\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7348872/v1/9393411906d9571175437fdb.png"},{"id":91513254,"identity":"b4ea2c88-e19d-40e5-aa1f-c6bfdafabae8","added_by":"auto","created_at":"2025-09-17 08:58:51","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":225473,"visible":true,"origin":"","legend":"\u003cp\u003eForest plot of hypertension risk factors (ordered Probit coefficients with 95% CIs)\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7348872/v1/a22f7bdfbd2e192becb4704d.png"},{"id":91513785,"identity":"1e4b4103-defd-41ce-a5bf-0135af78796c","added_by":"auto","created_at":"2025-09-17 09:06:52","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":596950,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cem\u003e\u003cstrong\u003eTrace plots for the parameter estimates of the Bayesian ordered Probit model\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7348872/v1/b983286270183a7ae0712e99.png"},{"id":91516108,"identity":"c14c57e3-9336-4166-91cb-d93060cfef68","added_by":"auto","created_at":"2025-09-17 09:22:53","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2053783,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7348872/v1/d9266ec5-1f00-4b90-972e-013e05170451.pdf"},{"id":91513784,"identity":"57cdd9c6-54dd-4390-8aea-400046ab640c","added_by":"auto","created_at":"2025-09-17 09:06:51","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":17142,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.docx","url":"https://assets-eu.researchsquare.com/files/rs-7348872/v1/f06664b24969e099ffcb7b65.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Bayesian and Probit comparative analysis of risk factors for high blood pressure among adults in Kassena-Nankana Districts, Ghana: An AWI-Gen sub-study","fulltext":[{"header":"Introduction","content":"\u003cp\u003eHypertension, also known as the \u0026quot;silent killer,\u0026quot; is one of the most common non-communicable diseases (NCDs) globally and poses a significant public health concern due to its strong association with cardiovascular diseases (CVDs). An estimated 1.28 billion individuals worldwide are affected by hypertension, with a disproportionately high burden in low- and middle-income countries (LMICs), where detection, treatment, and control rates remain inadequate\u0026nbsp;\u003cstrong\u003e[1,2]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eHypertension is highly prevalent in sub-Saharan Africa (SSA), driven by rapid urbanization, changing lifestyles, and limited access to healthcare services \u003cstrong\u003e[2,3]\u003c/strong\u003e. Ghana reflects this pattern, with national hypertension prevalence estimates ranging from 19% to 55%, depending on urban\u0026ndash;rural context and population subgroup \u003cstrong\u003e[4,5]\u003c/strong\u003e. Despite this alarming prevalence, awareness, diagnosis, treatment, and control rates remain low, contributing substantially to the national burden of cardiovascular morbidity and mortality\u0026nbsp;\u003cstrong\u003e[3,6]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eThe Kassena-Nankana districts in Ghana\u0026apos;s Upper East Region present a unique epidemiological context for hypertension research. While several studies have examined hypertension risk factors in urban Ghana, rural areas have received comparatively less attention despite exhibiting distinct socio-demographic and lifestyle characteristics \u003cstrong\u003e[7,8]\u003c/strong\u003e. The Kassena-Nankana districts, which are predominantly rural and reliant on agriculture, are experiencing a rise in hypertension prevalence, compounded by limited healthcare infrastructure, increasing life expectancy, and shifting health behaviours\u0026nbsp;\u003cstrong\u003e[8,9]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eThis study seeks to address this research gap by investigating the risk factors associated with hypertension among adults aged 40\u0026ndash;60 years in the Kassena-Nankana districts. Using data from the H3Africa AWI-Gen project \u003cstrong\u003e[10]\u003c/strong\u003e, the study applies both frequentist (ordered Probit) and Bayesian models to assess the influence of socio-demographic, behavioural, and metabolic risk factors on blood pressure status. The comparative application of Probit and Bayesian models is justified by their complementary strengths: Probit models offer interpretability and efficiency for large datasets, while Bayesian models provide robust estimates under uncertainty, particularly in data-sparse subgroups\u0026nbsp;\u003cstrong\u003e[11,12]\u003c/strong\u003e. Employing both approaches allows for a more robust and nuanced understanding of the predictors of hypertension in rural Ghana.\u003c/p\u003e\n\u003cp\u003eIn doing so, this study provides an in-depth analysis of hypertension risk factors, building upon existing research to present a comprehensive examination of the local epidemiology of hypertension. The findings are expected to contribute to the growing body of evidence required for effective policy formulation, resource allocation, and the implementation of public health strategies to address hypertension and Strategic Objective 3.1 of Ghana\u0026apos;s National NCD Policy (2022\u0026ndash;2027) for rural CVD prevention \u003cstrong\u003e[13]\u003c/strong\u003e.\u003c/p\u003e\n"},{"header":"Methodology","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eStudy design and population\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis was a secondary analysis of the H3Africa AWI-Gen project, which adopted a population-based cross-sectional design \u003cstrong\u003e[14]\u003c/strong\u003e. The primary study recruited adults aged 40\u0026ndash;60 years who had resided in the Navrongo Health and Demographic Surveillance System (HDSS) coverage area for at least 10 years \u003cstrong\u003e[15]\u003c/strong\u003e. A stratified random sampling was used to select participants for the study. In the first instance, pure Kasem and Nankam speaking zones were selected, and in the second instance, a simple random sampling was applied to selected adults 40\u0026ndash;60 years old, aiming for an equal sex ratio. Pregnant women and frail individuals were excluded from the study. A total of 2,010 participants were used for this study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData collection\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSociodemographic variables\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSociodemographic data were obtained through structured, interviewer-administered questionnaires conducted in Kasem or Nankam, the local languages preferred by participants \u003cstrong\u003e[16]\u003c/strong\u003e. Participants\u0026rsquo; ages were recorded in complete years and categorized into 10-year age bands: 40\u0026ndash;44, 45\u0026ndash;49, 50\u0026ndash;54, and 55\u0026ndash;60 \u003cstrong\u003e[17]\u003c/strong\u003e. Biological sex was recorded as male or female \u003cstrong\u003e[18]\u003c/strong\u003e. Marital status was dichotomized into currently married (legally or customarily partnered) and currently unmarried (single, divorced, widowed, or separated) \u003cstrong\u003e[19]\u003c/strong\u003e. Educational attainment was categorized as no formal education (0 years of schooling) or formal education (\u0026ge;1 year of schooling) \u003cstrong\u003e[20]\u003c/strong\u003e. Employment status was defined as unemployed (no income-generating activities) or employed (engaged in either formal or informal work, including subsistence farming) \u003cstrong\u003e[21]\u003c/strong\u003e. Household socioeconomic status (SES) was estimated using a composite wealth index based on housing materials, access to utilities (e.g., water and electricity), and ownership of durable assets such as livestock and vehicles. Principal component analysis (PCA) was used to classify households into low SES (lowest tertile) or high SES (upper two tertiles) \u003cstrong\u003e[22]\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSocioeconomic variables\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSocioeconomic data were collected through interviewer-administered surveys using tools validated by the Navrongo Health and Demographic Surveillance System (NHDSS) protocol \u003cstrong\u003e[18]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBehavioral variables\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBehavioral characteristics were assessed using instruments adapted from the WHO-STEPS protocol \u003cstrong\u003e[3]\u003c/strong\u003e. Smoking status was classified as non-smoker (never smoked) or current smoker (smoked \u0026ge;1 cigarette/day in the past 30 days) \u003cstrong\u003e[9]\u003c/strong\u003e. Alcohol consumption was defined as no intake (0 standard drinks/week) or current intake (\u0026ge;1 standard drink/week) \u003cstrong\u003e[23]\u003c/strong\u003e. Dietary intake of fruits and vegetables was quantified using a 24-hour recall and food frequency questionnaire \u003cstrong\u003e[24]\u003c/strong\u003e. Physical activity was measured as total weekly moderate-to-vigorous physical activity (MVPA), calculated from self-reported duration and frequency of activities such as farming, walking, and household chores. Accelerometer validation (ActiGraph GT3X+) was conducted in a 10% subsample \u003cstrong\u003e[25]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eEnvironmental variables\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEnvironmental exposures were evaluated following the Africa Wits-INDEPTH Partnership for Genomic Studies (AWI-Gen) protocol \u003cstrong\u003e[16]\u003c/strong\u003e. Pesticide exposure was defined as exposed (self-reported handling or residing within 100 meters of farmland during spraying seasons) or non-exposed \u003cstrong\u003e[26]\u003c/strong\u003e. Participants\u0026rsquo; residential locations were geo-referenced and classified into urban or rural clusters based on NHDSS-defined zones (central, east, west, north, south) \u003cstrong\u003e[15]\u003c/strong\u003e. Water sources were categorized as improved (e.g., piped water, boreholes) or unimproved (e.g., surface water, unprotected wells) according to WHO/UNICEF Joint Monitoring Programme standards \u003cstrong\u003e[27]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eBlood pressure measurement and classification\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBlood pressure was measured using a validated, automated sphygmomanometer (Omron M6; Omron Corporation, Kyoto, Japan) \u003cstrong\u003e[15]\u003c/strong\u003e. After a five-minute rest, three readings were taken at two-minute intervals. The first was discarded, and the mean of the remaining two was used for analysis \u003cstrong\u003e[15]\u003c/strong\u003e. Blood pressure classifications followed the ACC/AHA 2017 guidelines\u0026nbsp;\u003cstrong\u003e[8]\u003c/strong\u003e.\u003cbr\u003ePulse pressure (PP) was calculated as PP = SBP \u0026minus; DBP and mean arterial pressure (MAP) as MAP = DBP + (PP/3) \u003cstrong\u003e[15]\u003c/strong\u003e.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePhysical activity (MVPA) assessment\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eMVPA was quantified using questionnaires adapted from the WHO-STEPS protocol\u0026nbsp;\u003cstrong\u003e[25]\u003c/strong\u003e. Weekly MVPA was computed as:\u003cbr\u003e\u0026nbsp;MVPA (hours/week) = \u0026sum; (Activity Duration \u0026times; Frequency) \u0026times; MET coefficient\u003cbr\u003eMET coefficients were obtained from the Compendium of Physical Activities. A 10% sample was validated using ActiGraph GT3X+ accelerometers \u003cstrong\u003e[25]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eAnthropometry and BMI calculation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWeight was measured to the nearest 0.1 kg using a SECA 874 scale, with participants wearing light clothing. Height was measured using a SECA 217 stadiometer following the Frankfurt Plane protocol\u0026nbsp;\u003cstrong\u003e[17]\u003c/strong\u003e. BMI was calculated using the formula:\u003cbr\u003e\u0026nbsp;BMI = Weight (kg) / [Height (m)]\u0026sup2;\u003cbr\u003eAll anthropometric procedures followed International Society for the Advancement of Kinanthropometry (ISAK) guidelines \u003cstrong\u003e[25]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eLipid profile analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eVenous blood samples were collected after an overnight fast of at least 8 hours. Serum lipid concentrations were measured using enzymatic colorimetric assays on a Roche Cobas c311 analyser at the Navrongo Health Research Centre \u003cstrong\u003e[17]\u003c/strong\u003e. Methods included:\u003c/p\u003e\n\u003cul type=\"disc\"\u003e\n \u003cli\u003eTotal cholesterol (TC): CHOD-PAP\u003c/li\u003e\n \u003cli\u003eHDL cholesterol: Homogeneous enzymatic colorimetry\u003c/li\u003e\n \u003cli\u003eLDL cholesterol: Calculated using the Friedewald equation (if TG \u0026lt; 4.5 mmol/L)\u003c/li\u003e\n \u003cli\u003eTriglycerides (TG): GPO-PAP\u003c/li\u003e\n\u003c/ul\u003e\n\u003cp\u003eAll assays followed IFCC standards, with intra-assay variation maintained below 3% \u003cstrong\u003e[25]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003e\u003cem\u003eEthics Approval\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe AWI-Gen study received ethics approval from the Human Research Ethics Committee (HREC), University of the Witwatersrand (#M12109) and was renewed in 2017, #M170880).), the Ghana Health\u0026nbsp;Service Ethics Review Committee (#GHS-ERC:05/05/2015). Additional approval was received from the Navrongo Health Research Centre Institutional Review Board (#NHRCIRB178). Informed consent was obtained from all participants.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003e\u003cem\u003eStatistical analysis\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDescriptive statistics were summarized using frequencies and percentages for categorical variables, and either means with standard deviations for normally distributed continuous variables or medians with interquartile ranges (IQRs) for skewed data. For inferential analysis, the study employed both frequentist (Probit) and Bayesian approaches.\u003c/p\u003e\n\u003cp\u003eThe Bayesian framework was specifically selected to: (1) incorporate prior epidemiological knowledge from similar rural African populations \u003cstrong\u003e[26]\u003c/strong\u003e, (2) address sparse data challenges in subgroup analyses \u003cstrong\u003e[27]\u003c/strong\u003e, and (3) offer probabilistic interpretations of uncertainty using credible intervals \u003cstrong\u003e[28]\u003c/strong\u003e. Bayesian models enabled probabilistic inference for underpowered groups like female smokers (n = 35), hence were rerun with Cauchy(0, 2.5) priors to confirm robustness. Convergence of the Markov chain Monte Carlo (MCMC) simulations for the Bayesian model was assessed using the Gelman-Rubin R̂ statistic \u003cstrong\u003e[29]\u003c/strong\u003e. This statistic compares the between-chain and within-chain variances, with values close to 1 indicating convergence. An R̂ value below 1.1 is generally considered acceptable \u003cstrong\u003e[29]\u003c/strong\u003e.\u003c/p\u003e\n\u003cp\u003eModel comparison was performed using the Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) \u003cstrong\u003e[27]\u003c/strong\u003e, with lower values indicating superior model fit. Additional evaluation of model performance was based on maximum likelihood estimates and the number of parameters included in each model. The threshold for statistical significance was set at p\u0026lt;0.05. All analyses were conducted using Stata version 17.0 (Stata Corp LLC, College Station, TX, USA).\u003c/p\u003e\n"},{"header":"Results and Discussion","content":"\u003ch3\u003eBackground characteristics of the study\u003c/h3\u003e\n\u003cp\u003eFigure \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e summarizes blood pressure categories among the study population. The proportion of individuals with normal blood pressure was 46%. Prehypertension constituted 32%, hypertension 14%, and sever hypertension 8%.\u003c/p\u003e\n\u003cp\u003eSociodemographic and behavioral characteristics stratified by sex are presented in Table \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e. Women made up a slightly higher proportion of the sample (54.2%) compared to men (45.8%). Statistically significant differences were observed across most characteristics. Age distribution differed significantly between sexes (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with a higher proportion of women (39.9%) falling in the 55\u0026ndash;60-year group compared to men (31.4%). Marital status was also significantly associated with sex (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001); a larger proportion of men were currently married (76.9%) compared to women (56.0%), whereas women were more likely to be unmarried (44.0% vs. 23.1%). Educational status showed marked gender differences (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001): 77.7% of women had no formal education, compared to 61.7% of men, while formal education was more common among men (38.3%) than women (22.5%).\u003c/p\u003e\n\u003cp\u003eSignificant sex differences were also evident in health-related behavioural factors. A striking contrast was seen in smoking status (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), with 64.2% of men currently smoking compared to just 3.2% of women. Similarly, alcohol intake varied by sex (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), as 92.1% of men reported current alcohol use compared to 78.6% of women. Pesticide exposure also differed significantly by sex (p\u0026thinsp;=\u0026thinsp;0.003), with a greater proportion of men (60.0%) exposed compared to women (48.6%). Interestingly, no significant sex differences were observed in occupational status (p\u0026thinsp;=\u0026thinsp;0.165), with employment rates nearly equal across sexes (approximately 40%).\u003c/p\u003e\n\u003cp\u003eTable \u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e also presents the continuous variables that are summarized using medians and interquartile ranges (IQRs) due to their non-normal distribution. Statistically significant sex differences were observed across several variables. In terms of physical activity and metabolic risk factors, men reported slightly higher moderate-to-vigorous physical activity (MVPA) levels, with a median of 2,040 hours/week (IQR: 2,400), compared to 1,920 hours/week (IQR: 2,400) for women (p\u0026thinsp;=\u0026thinsp;0.045). Women, however, had significantly higher body mass index (BMI) values (median: 23.1 kg/m\u0026sup2;, IQR: 5.2) compared to men (median: 21.8 kg/m\u0026sup2;, IQR: 4.9) (p\u0026thinsp;=\u0026thinsp;0.002). Waist-to-hip ratio was also significantly higher among women (0.87 [IQR: 0.06]) than men (0.85 [IQR: 0.08]) (p\u0026thinsp;=\u0026thinsp;0.015), although no significant sex difference was observed for waist circumference (p\u0026thinsp;=\u0026thinsp;0.071). Regarding lipid profiles, women exhibited higher HDL cholesterol levels (median: 1.20 mmol/L, IQR: 0.50) than men (median: 1.00 mmol/L, IQR: 0.38; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), and higher total cholesterol levels (3.32 vs. 3.08 mmol/L; p\u0026thinsp;=\u0026thinsp;0.038). However, no significant differences were noted for LDL cholesterol (p\u0026thinsp;=\u0026thinsp;0.128) or triglycerides (p\u0026thinsp;=\u0026thinsp;0.294).\u003c/p\u003e\n\u003ctable id=\"Tab1\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eSociodemographic, Behavioral, and Clinical Characteristics of Participants by Gender\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVariable\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eWomen\u003c/p\u003e\n \u003cp\u003e1082 (54.2%)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eMen\u003c/p\u003e\n \u003cp\u003e916 (45.8%)\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAge Group (years)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e40\u0026ndash;44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e145 (13.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e169 (18.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e45\u0026ndash;49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e272 (25.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e241 (26.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e50\u0026ndash;54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e235 (21.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e218 (23.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e55\u0026ndash;60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e432 (39.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e288 (31.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMarital Status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCurrently married\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e606 (56.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e704 (76.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCurrently unmarried\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e478 (44.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e212 (23.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eEducational Status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo formal education\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e841 (77.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e565 (61.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eFormal education\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e243 (22.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e351 (38.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eOccupational Status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.165\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eUnemployed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e651 (60.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e551 (60.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEmployed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e431 (39.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e365 (39.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSmoking Status\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNot smoking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1047 (96.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e328 (35.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCurrently smoking\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e35 (3.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e588 (64.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003eAlcohol Consumption\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNo alcohol intake\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e232 (21.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e72 (7.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCurrent intake\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e850 (78.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e844 (92.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\" colspan=\"2\"\u003e\n \u003cp\u003ePesticide Exposure\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.003\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNon-exposed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e556 (51.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e366 (40.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eExposed\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e526 (48.6%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e550 (60.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMVPA (hours/week)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1,920 [2,400]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2,040 [2,400]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.045\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI (kg/m\u0026sup2;)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e23.1 [5.2]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e21.8 [4.9]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWaist circumference (mm)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e738 [95]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e729 [105]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.071\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eWaist-hip ratio\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.87 [0.06]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.85 [0.08]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.015\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eHDL (mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.20 [0.50]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.00 [0.38]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eLDL (mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.66 [1.01]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.57 [0.94]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.128\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCholesterol (mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.32 [1.20]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e3.08 [1.13]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.038\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eTriglycerides (mmol/L)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.58 [0.28]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.55 [0.32]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.294\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eLegend:\u003c/p\u003e\n\u003cul\u003e\n \u003cli\u003e\n \u003cp\u003e\u003cem\u003eValues are presented as\u003c/em\u003e \u003cstrong\u003en (%)\u003c/strong\u003e \u003cem\u003efor categorical variables and\u003c/em\u003e \u003cstrong\u003eMedian [Interquartile Range]\u003c/strong\u003e \u003cem\u003efor continuous variables.\u003c/em\u003e\u003c/p\u003e\n \u003c/li\u003e\n \u003cli\u003e\n \u003cp\u003e\u003cem\u003ep-values are from\u003c/em\u003e \u003cstrong\u003eChi-square tests\u003c/strong\u003e \u003cem\u003efor categorical variables and\u003c/em\u003e \u003cstrong\u003eMann\u0026ndash;Whitney U tests\u003c/strong\u003e \u003cem\u003efor continuous variables.\u003c/em\u003e\u003c/p\u003e\n \u003c/li\u003e\n\u003c/ul\u003e\n\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\n \u003ch2\u003eDistribution of blood pressure categories across sociodemographic, behavioral and environmental factors\u003c/h2\u003e\n \u003cp\u003eTable \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e presents the distribution of blood pressure categories (normal, prehypertensive, hypertensive, and severe hypertension) across various sociodemographic and behavioral variables, stratified by sex. These results are descriptive and show how blood pressure classifications vary within each subgroup rather than implying statistical associations or correlations.\u003c/p\u003e\n \u003cp\u003eAmong women, age-related differences in blood pressure distribution were evident. The proportion of women with normal blood pressure decreased progressively with age, from 60.4% in the youngest age group to 42.6% in the oldest, while the proportion with severe hypertension increased from 2.1\u0026ndash;13.0% (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). A similar pattern was observed among men, with normal blood pressure declining from 45.6\u0026ndash;32.8% and severe hypertension rising from 3.6\u0026ndash;11.9% across age categories (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\n \u003cp\u003eMarital status showed variation in blood pressure distribution among women, where currently married individuals had a higher proportion of normal blood pressure (53.4%) compared to unmarried women (42.1%). Additionally, a larger share of unmarried women fell into the severe hypertension category (10.9%) compared to married women (6.3%), with these differences reaching statistical significance (p\u0026thinsp;=\u0026thinsp;0.001). Among men, however, the distribution of blood pressure categories did not significantly differ by marital status (p\u0026thinsp;=\u0026thinsp;0.423).\u003c/p\u003e\n \u003cp\u003eEducational status, occupational status, smoking, and alcohol use did not show notable differences in blood pressure category distributions in either sex (p\u0026thinsp;\u0026gt;\u0026thinsp;0.05). However, there was a slight trend in men where the proportion of severe hypertension was higher among smokers (8.3%) compared to non-smokers (7.3%) (p\u0026thinsp;=\u0026thinsp;0.050), though this did not reach conventional levels of statistical significance.\u003c/p\u003e\n \u003cp\u003eFor pesticide exposure, women showed notable differences across blood pressure categories (p\u0026thinsp;=\u0026thinsp;0.009). Those exposed to pesticides had a lower proportion of severe hypertension (5.5%) compared to the non-exposed group (11.0%), although the non-exposed group had a slightly higher prevalence of normal blood pressure. Among men, the distribution of blood pressure categories did not differ significantly by pesticide exposure (p\u0026thinsp;=\u0026thinsp;0.473).\u003c/p\u003e\n \u003cp\u003eFinally, BMI classification showed clear differences in blood pressure distribution for both sexes. Among women, the proportion with normal blood pressure declined from 57.3% in the underweight category to 22.2% in the obese category, while severe hypertension rose from 8.4\u0026ndash;20.0% (p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Men followed a similar trend, with normal blood pressure decreasing and severe hypertension increasing across BMI categories (p\u0026thinsp;=\u0026thinsp;0.001).\u003c/p\u003e\n\u003c/div\u003e\n\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\n \u003ch2\u003eRegression analyses\u003c/h2\u003e\n \u003cp\u003eParameter estimates from both the frequentist ordered probit model and the Bayesian ordered probit model were compared to explore differences in predictors of elevated blood pressure among adults in the Kassena-Nankana districts. Weakly informative normal priors\u0026thinsp;~\u0026thinsp;N(0,10) were used for regression coefficients in the Bayesian analyses. As presented in Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e, both models identified male sex as a significant factor associated with higher blood pressure categories. The frequentist model estimated a coefficient of 0.309 (95% CI: 0.172\u0026ndash;0.446; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), while the Bayesian model produced a posterior mean of 0.264 with a 95% credible interval (CrI) of 0.15\u0026ndash;0.34, which excluded zero, indicating high certainty in the direction of effect.\u003c/p\u003e\n \u003cp\u003eAge was also a consistent predictor across models, showing a positive gradient in risk. For individuals aged 45\u0026ndash;49, the frequentist model reported a non-significant coefficient of 0.127 (95% CI: \u0026minus;\u0026thinsp;0.036 to 0.291; p\u0026thinsp;=\u0026thinsp;0.129), whereas the Bayesian model estimated a posterior mean of 0.14 (95% CrI: 0.08\u0026ndash;0.22), suggesting moderate certainty of increased risk. For age groups 50\u0026ndash;54 and 55\u0026ndash;60, both models produced significant and increasing coefficients, with the highest effect seen in the 55\u0026ndash;60 age group (frequentist: 0.514, 95% CI: 0.356\u0026ndash;0.672; Bayesian: 0.499, 95% CrI: 0.41\u0026ndash;0.58; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001 in both cases).\u003c/p\u003e\n \u003cp\u003eMarital status showed a statistically significant association in the frequentist model, with unmarried individuals having a coefficient of 0.143 (95% CI: 0.033\u0026ndash;0.253; p\u0026thinsp;=\u0026thinsp;0.011). However, the Bayesian estimate for this variable (0.272; 95% CrI: \u0026minus;\u0026thinsp;0.08 to 0.56) included zero, reflecting greater uncertainty in the strength and direction of this effect. Similarly, formal education and employment status were not statistically significant in either model. For instance, the frequentist model estimated a coefficient of 0.054 (95% CI: \u0026minus;\u0026thinsp;0.058 to 0.167; p\u0026thinsp;=\u0026thinsp;0.343) for education, while the Bayesian estimate was slightly negative (\u0026ndash;0.116; 95% CrI: \u0026minus;\u0026thinsp;0.24 to 0.01), though still uncertain.\u003c/p\u003e\n \u003cp\u003eAlcohol consumption showed a weak and statistically non-significant positive relationship with blood pressure classification in both models (frequentist: 0.100, 95% CI: \u0026minus;\u0026thinsp;0.044 to 0.245; p\u0026thinsp;=\u0026thinsp;0.172; Bayesian: 0.200, 95% CrI: \u0026minus;\u0026thinsp;0.05 to 0.25). In contrast, smoking status had a negative effect on blood pressure classification, with both models showing significant associations. The frequentist model yielded \u0026minus;\u0026thinsp;0.148 (95% CI: \u0026minus;\u0026thinsp;0.291 to \u0026minus;\u0026thinsp;0.004; p\u0026thinsp;=\u0026thinsp;0.044), while the Bayesian model reported \u0026minus;\u0026thinsp;0.260 (95% CrI: \u0026minus;\u0026thinsp;0.29 to \u0026minus;\u0026thinsp;0.01), indicating a lower probability of being in a higher blood pressure category among current smokers.\u003c/p\u003e\n \u003cp\u003eFor pesticide exposure, both models suggested a positive effect, although the confidence interval reported in the frequentist model appears inconsistent (estimate\u0026thinsp;=\u0026thinsp;0.141; 95% CI: \u0026minus;\u0026thinsp;0.243 to \u0026minus;\u0026thinsp;0.04), possibly due to reporting error. The Bayesian model estimated a coefficient of 0.220 (95% CrI: \u0026minus;\u0026thinsp;0.24 to \u0026minus;\u0026thinsp;0.04), but since the interval crosses zero, the effect should be interpreted with caution.\u003c/p\u003e\n \u003cp\u003eBody mass index (BMI) showed a consistent and statistically significant association with elevated blood pressure across both models. Both the frequentist and Bayesian models produced identical point estimates of 0.052 (frequentist 95% CI: 0.037\u0026ndash;0.066; Bayesian 95% CrI: 0.04\u0026ndash;0.07; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), confirming BMI as a robust predictor. In contrast, moderate-to-vigorous physical activity (MVPA) and total cholesterol were not associated with blood pressure classification. Both models reported null effects for MVPA (coefficient\u0026thinsp;=\u0026thinsp;0.000) and cholesterol (frequentist: 0.000, 95% CI: \u0026minus;\u0026thinsp;0.001 to 0.001; Bayesian: \u0026minus;\u0026thinsp;0.001, 95% CrI: \u0026minus;\u0026thinsp;0.01 to 0.00), indicating negligible influence on the outcome.\u003c/p\u003e\n \u003cp\u003eModel fit statistics showed better performance for the frequentist ordered probit model, with lower AIC (4644.779) and BIC (4734.225) values compared to the Bayesian model (AIC\u0026thinsp;=\u0026thinsp;5137.520; BIC\u0026thinsp;=\u0026thinsp;5165.159), suggesting a relatively superior fit under the frequentist framework.\u003c/p\u003e\n \u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003eParameter estimates of both ordered Probit versus Bayesian ordered Probit model\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eFactor\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eProbit Coefficient [95% CI]\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eBayesian Mean [95% CrI]\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eP-value\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSex (Male)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.309 [0.172, 0.446]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.264 [0.15, 0.34]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge 45\u0026ndash;49\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.127 [\u0026ndash;0.036, 0.291]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.129\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.140 [0.08, 0.22]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.004\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge 50\u0026ndash;54\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.231 [0.061, 0.401]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.008\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.226 [0.15, 0.36]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.002\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAge 55\u0026ndash;60\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.514 [0.356, 0.672]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.499 [0.41, 0.58]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMarital Status (Unmarried)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.143 [0.033, 0.253]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.011\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.272 [\u0026ndash;0.08, 0.56]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.13\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eEducational Status (Formal)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.054 [\u0026ndash;0.058, 0.167]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.343\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026ndash;0.116 [\u0026ndash;0.24, 0.01]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.07\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eOccupational Status (Employed)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.047 [\u0026ndash;0.052, 0.15]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.351\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.082 [\u0026ndash;0.06, 0.15]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.21\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAlcohol Consumption (Current)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.100 [\u0026ndash;0.044, 0.245]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.172\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.200 [\u0026ndash;0.05, 0.25]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.09\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eSmoking Status (Current)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026ndash;0.148 [\u0026ndash;0.291, \u0026minus;\u0026thinsp;0.004]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.044\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026ndash;0.260 [\u0026ndash;0.29, \u0026minus;\u0026thinsp;0.01]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.04\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003ePesticide Exposure (Exposed)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.141 [\u0026ndash;0.243, 0.04]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.007\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.220 [\u0026ndash;0.24, 0.04]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.04\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eBMI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.052 [0.037, 0.066]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.052 [0.04, 0.07]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026lt;\u0026thinsp;0.001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eCholesterol\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.000 [\u0026ndash;0.001, 0.001]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0.882\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026ndash;0.001 [\u0026ndash;0.01, 0.00]\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e\u0026asymp;\u0026thinsp;0.29\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eAIC/BIC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e4644.779 /4734.225\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e5137.520 /5165.159\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\u0026nbsp;\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n \u003cp\u003ea. \u003cem\u003eProbit model presents beta coefficients as parameter estimates.\u003c/em\u003e\u003c/p\u003e\u003cspan\u003eb. \u003cem\u003eBayesian model reports the mean estimates for parameter estimation.\u003c/em\u003e\u003cbr\u003e\u003c/span\u003e\u003cem\u003ec. Bayesian p-values aren\u0026apos;t traditionally computed but based on whether 95% CrI includes zero, hence p-values provided are but approximations.\u003c/em\u003e\n \u003cp\u003eComparative analysis revealed hypertension prevalence was 1.8\u0026times; higher in 55\u0026ndash;60-year-olds vs. 40\u0026ndash;44-year-olds (24.9% vs 13.9%, p\u0026thinsp;\u0026lt;\u0026thinsp;0.001). Figure \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e visualizes the key associations through forest plots of significant predictors. Figure \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e presents a forest plot of the key predictors identified in the ordered Probit model. Male sex (\u0026beta;\u0026thinsp;=\u0026thinsp;0.309), older age (55\u0026ndash;60 years: \u0026beta;\u0026thinsp;=\u0026thinsp;0.514), unmarried status (\u0026beta;\u0026thinsp;=\u0026thinsp;0.143), and higher BMI (\u0026beta;\u0026thinsp;=\u0026thinsp;0.052) were significantly associated with increased odds of higher blood pressure categories. In contrast, smoking (\u0026beta; = -0.148) and pesticide exposure (\u0026beta; = -0.141) demonstrated inverse associations with blood pressure status. The plot visually highlights the magnitude and direction of these associations, with 95% confidence intervals indicating the precision of the estimates.\u003c/p\u003e\n\u003c/div\u003e\n\u003ch3\u003eModel Diagnostics (Bayesian Ordered Probit Model)\u003c/h3\u003e\n\u003cp\u003eThe diagnostic assessment of the Bayesian Ordered Probit Model utilized Markov Chain Monte Carlo (MCMC) trace plots to examine whether the chains mixed properly and reached convergence. The MCMC sampling of stations to evaluate the posterior distribution and its distribution stability is presented in Fig. \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e\n\u003cp\u003eThe MCMC algorithm achieved convergence according to the visual assessment of the trace plot, which displayed appropriate chain mixing results. All variables, including age, sex, occupational status, MVPA, and educational and marital status, required approximately 35,000 iterations to reach stationary distribution states, alongside excellent mixing quality. The burn-in process for smoking status, together with cholesterol, alcohol consumption, BMI, and pesticide exposure, lasted between 50,000 and 70,000 iterations to achieve convergent states. The Gelman-Rubin \u003cem\u003eR̂\u003c/em\u003e statistic confirmed convergence for all parameters. The Bayesian model (probit: \u003cem\u003eR̂\u003c/em\u003e = 1.09) achieved values near 1.0, indicating stable sampling from the target distribution.\u003c/p\u003e\n\u003cp\u003eThese diagnostics indicate that posterior estimates are reliable and Bayesian inference in this study operates robustly.\u003c/p\u003e\n\u003ch3\u003eDiscussion\u003c/h3\u003e\n\u003cp\u003eThis study identified several key predictors of hypertension that align with global epidemiological patterns while also revealing context-specific risk factors relevant to rural Ghana. The application of both frequentist and Bayesian ordered probit models allowed for cross-validation of findings and improved interpretability through probabilistic estimates and uncertainty quantification.\u003c/p\u003e\u003cp\u003eBoth models consistently demonstrated that male sex and advancing age, particularly among individuals aged 55\u0026ndash;60 years, were significant predictors of higher blood pressure categories. The frequentist model reported a statistically significant coefficient for male sex (0.309; 95% CI: 0.172\u0026ndash;0.446; p\u0026thinsp;\u0026lt;\u0026thinsp;0.001), which was corroborated by the Bayesian model (posterior mean: 0.264; 95% CrI: 0.15\u0026ndash;0.34), with the credible interval excluding zero, indicating high posterior certainty. A similar pattern was observed with age, with the highest risk found in the 55\u0026ndash;60 age group (frequentist: 0.514; Bayesian: 0.499). These findings are consistent with global data linking older age and male sex to hypertension due to cumulative exposure to lifestyle risk factors and biological susceptibility [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eBody mass index (BMI) also emerged as a consistent and statistically significant predictor across both models (coefficient: 0.052), reinforcing the established role of obesity as a modifiable cardiometabolic risk factor [\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e]. The convergence of results between models further validates this association within rural African populations.\u003c/p\u003e\u003cp\u003eInterestingly, smoking status showed a negative association with hypertension in both models (frequentist: \u0026minus;\u0026thinsp;0.148, p\u0026thinsp;=\u0026thinsp;0.044; Bayesian: \u0026minus;\u0026thinsp;0.260, 95% CrI: \u0026minus;\u0026thinsp;0.29 to \u0026minus;\u0026thinsp;0.01), contrasting with existing meta-analyses and most urban SSA studies that typically report positive associations [\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e, \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. This counterintuitive finding may reflect survival bias (e.g., premature mortality among hypertensive smokers), low smoking prevalence\u0026mdash;especially among women\u0026mdash;and unmeasured confounding such as intensive physical labour in agricultural settings. Future longitudinal and biomarker-based studies are needed to further explore this paradox.\u003c/p\u003e\u003cp\u003eThe study also found a positive association between pesticide exposure and hypertension, though the Bayesian credible interval included zero, indicating uncertainty. This is noteworthy in the context of widespread, often unregulated agrochemical use in rural Ghana. These results underscore the importance of public health interventions such as protective gear distribution, farmer education on pesticide toxicity, and seasonal hypertension screening, as outlined in Ghana\u0026rsquo;s National NCD Policy (2022\u0026ndash;2027).\u003c/p\u003e\u003cp\u003eComparison of the two models revealed largely overlapping sets of significant predictors\u0026mdash;male sex, older age, and elevated BMI\u0026mdash;yet highlighted important differences in how uncertainty is represented. For example, the frequentist model found unmarried status to be statistically significant (p\u0026thinsp;=\u0026thinsp;0.011), while the Bayesian model\u0026rsquo;s 95% credible interval (\u0026ndash;0.08 to 0.56) included zero, suggesting less certainty in the effect. This difference illustrates the Bayesian model's advantage in reflecting uncertainty more transparently than conventional null hypothesis testing.\u003c/p\u003e\u003cp\u003eFrom a model fit perspective, the frequentist model showed better performance based on lower AIC (4644.8) and BIC (4734.2) values compared to the Bayesian model (AIC\u0026thinsp;=\u0026thinsp;5137.5; BIC\u0026thinsp;=\u0026thinsp;5165.2). Although the Bayesian framework enhances interpretation, these diagnostics suggest that the frequentist model provided a better empirical fit for this dataset.\u003c/p\u003e\u003cp\u003eWithin the broader SSA context, the findings reaffirm well-established hypertension predictors while highlighting rural-specific nuances. The consistent male effect contrasts with some rural studies where shared physical workloads diminish sex differences [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. The absence of a significant association between physical activity and blood pressure may reflect the uniformly high MVPA levels among subsistence farmers, which could mask detectable gradients observed in more sedentary urban populations [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e, \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Similarly, the lack of alcohol\u0026ndash;hypertension association aligns with mixed findings from SSA, likely due to variability in beverage type and drinking patterns [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eOur identification of male sex, aging, and elevated BMI as consistent hypertension predictors aligns with Agongo et al. (2020) in the same Navrongo cohort, reinforcing the stability of these risk factors in rural Ghana [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. However, while Gomez-Olive et al. (2017) reported urban-rural gradients in hypertension prevalence across AWI-Gen sites, our Bayesian modeling uniquely quantifies uncertainty in sparse subgroups (e.g., female smokers) and reveals novel environmental risks like pesticide exposure in agrarian communities [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eIn the Ghanaian context, these results align with urban evidence linking hypertension with male sex and aging [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e] but differ from some rural studies that report higher prevalence among women\u0026mdash;potentially due to disparities in healthcare access or reporting bias [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. The BMI\u0026ndash;hypertension link reinforces concerns about nutrition transition in rural areas [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e]. Additionally, the elevated hypertension risk among unmarried individuals (frequentist coefficient: 0.143) may reflect social isolation or economic vulnerability, which tend to be more pronounced in rural settings [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. Although the association between pesticide exposure and hypertension was statistically inconclusive, the direction of the effect is in line with evidence connecting agrochemical exposure to cardiovascular and renal risks in Ghana [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e].\u003c/p\u003e\u003cp\u003eThe forest plot (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e) provides a clear visual summary of the magnitude and direction of these effects. Overall, the triangulation of findings across two analytical frameworks enhances confidence in the observed relationships and offers deeper insight into both consistent and uncertain predictors of hypertension in rural Ghana.\u003c/p\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003ePublic health implications\u003c/h2\u003e\u003cp\u003eTargeted hypertension screening should be integrated into Ghana's agricultural extension services. Farmers reporting pesticide exposure require seasonal BP monitoring and protective gear training. The inverse smoking-hypertension association warrants biomarker-confirmed longitudinal studies.\u003c/p\u003e\u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThis study identified key predictors of hypertension among adults in rural northern Ghana using both frequentist (ordered probit) and Bayesian models. While both approaches consistently flagged male sex, older age, higher BMI, and unmarried status as significant risk factors, their combined use added analytical value. The probit model provided efficient estimation and clearer point estimates, particularly useful for identifying primary risk factors in a large dataset. In contrast, the Bayesian model allowed for a more nuanced understanding of uncertainty, especially in subgroups with sparse data (e.g., smoking or pesticide exposure). This dual approach revealed areas where associations were robust across methods and where greater caution, or further research is warranted due to statistical uncertainty. Therefore, applying both methods strengthened the credibility of the findings and enhanced interpretability for policy decisions in resource-limited rural settings.\u003c/p\u003e\u003cp\u003eAn unexpected inverse relationship between smoking and hypertension was observed in both models, likely reflecting survival bias, low-intensity smoking, or unmeasured confounding such as occupational activity. Additionally, cholesterol levels and physical activity showed no significant associations with hypertension status, which may be attributed to the homogeneous farming lifestyle prevalent in the study population.\u003c/p\u003e\u003cp\u003eImportantly, both statistical models identified similar patterns of association, reinforcing the robustness of the findings. However, the Bayesian model offered deeper insights into uncertainty, particularly where credible intervals crossed zero despite positive point estimates (marital status and pesticide exposure). Conversely, the frequentist model demonstrated a better overall model fit based on lower AIC and BIC values. This comparison highlights the value of using complementary modelling approaches: the frequentist model offers precision and simplicity, while the Bayesian model provides a more nuanced understanding of estimate reliability.\u003c/p\u003e\u003cp\u003eThese findings support Strategic Objective 3.1 of Ghana's National NCD Policy (2022\u0026ndash;2027), which emphasizes community-based screening and targeted prevention in rural populations. We recommend integrating hypertension surveillance into agricultural extension services to better reach farming communities, especially given the potential links between pesticide exposure and elevated blood pressure.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003e\u003cem\u003eData availability\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe dataset supporting the conclusions of this article is available from the H3Africa AWI-Gen project upon reasonable request. Due to ethical restrictions and data sharing policies, de-identified participant data can be accessed with appropriate approvals from the Navrongo Health Research Centre Institutional Review Board.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAcknowledgment\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank all study participants from the Kassena-Nankana districts for their cooperation and contributions to this research. We also acknowledge the staff and management of the Navrongo Health Research Centre for their logistical and institutional support. The authors are grateful to the H3Africa AWI-Gen project and the H3Africa Consortium for providing access to the data and resources that made this work possible.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eFunding\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was supported by the H3Africa AWI-Gen project, funded by the National Human Genome Research Institute (NHGRI), the Office of the Director (OD) at the National Institutes of Health (NIH) of the United States of America, and the Wellcome Trust (Grant numbers: U54HG006938 and 074548/Z/04/Z).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eConflict of Interest\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that there are no conflicts of interest regarding the publication of this article.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e\u003cem\u003eAuthor Contributions\u0026nbsp;\u003c/em\u003e\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRichmond Balinia Adda\u0026nbsp;\u003c/strong\u003econtributed to the conceptualisation, methodology, data analysis, writing (original draft preparation), visualization.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eIda Anowuje\u003c/strong\u003e contributed to the methodology, supervision, data curation, writing (review \u0026amp; editing).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGodfred Agongo\u003c/strong\u003e contributed to data curation, supervision, writing (review \u0026amp; editing).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCornelius Debpuur\u003c/strong\u003e contributed to writing (review \u0026amp; editing).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAbraham Rexford Oduro\u003c/strong\u003e contributed to the project administration, resources, writing (review \u0026amp; editing).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEngelbert A. Nonterah\u003c/strong\u003e contributed to conceptualisation, supervision, methodology, writing (review \u0026amp; editing).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eMills KT, Stefanescu A, He J. The global epidemiology of hypertension. \u003cem\u003eNat Rev Nephrol\u003c/em\u003e. 2020;16(4):223\u0026ndash;237.\u003c/li\u003e\n\u003cli\u003eNCD Risk Factor Collaboration (NCD-RisC). Worldwide trends in hypertension prevalence and progress in treatment and control from 1990 to 2019: A pooled analysis of 1201 population-representative studies with 104 million participants. \u003cem\u003eLancet\u003c/em\u003e. 2021;398(10304):957\u0026ndash;970.\u003c/li\u003e\n\u003cli\u003eWorld Health Organization. STEPwise approach to surveillance (STEPS). Geneva: WHO; 2020. Available from: https://www.who.int/ncds/surveillance/steps/en/\u003c/li\u003e\n\u003cli\u003eRamsay M, Crowther N, Fraser PA, Norris SA. The AWI-Gen Collaborative Centre: A resource for genomic studies of cardiometabolic diseases in Africa. \u003cem\u003eGlob Health Epidemiol Genomics\u003c/em\u003e. 2016;1:e20.\u003c/li\u003e\n\u003cli\u003eOduro AR, Wak G, Azongo D, Debpuur C, Wontuo P, Kondayire F, et al. Profile of the Navrongo Health and Demographic Surveillance System. \u003cem\u003eInt J Epidemiol\u003c/em\u003e. 2012;41(4):968\u0026ndash;976.\u003c/li\u003e\n\u003cli\u003eAgyemang C. Rural and urban differences in blood pressure and hypertension in Ghana, West Africa. \u003cem\u003ePublic Health\u003c/em\u003e. 2006;120:525\u0026ndash;533.\u003c/li\u003e\n\u003cli\u003eNonterah EA, Depbuur C, Agongo G, et al. Sociodemographic and behavioral determinants of abnormal body mass index among an adult population in rural Northern Ghana: The AWI-Gen study. \u003cem\u003eGlob Health Action\u003c/em\u003e. 2018;11:146\u0026ndash;588.\u003c/li\u003e\n\u003cli\u003eAmerican College of Cardiology/American Heart Association (ACC/AHA). 2017 Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults. \u003cem\u003eHypertension\u003c/em\u003e. 2018;71(6):e13\u0026ndash;e115.\u003c/li\u003e\n\u003cli\u003eCenters for Disease Control and Prevention (CDC). Tobacco Use Definitions. Atlanta, GA: CDC; 2020. Available from: https://www.cdc.gov/nchs/nhis/tobacco/tobacco_definitions.htm\u003c/li\u003e\n\u003cli\u003eNational Institute on Alcohol Abuse and Alcoholism (NIAAA). What is a standard drink? Bethesda, MD: NIAAA; 2023. Available from: https://www.niaaa.nih.gov/alcohols-effects-health/overview-alcohol-consumption/what-standard-drink\u003c/li\u003e\n\u003cli\u003eFood and Agriculture Organization of the United Nations (FAO). \u003cem\u003eDietary Assessment: A Resource Guide to Method Selection and Application in Low Resource Settings\u003c/em\u003e. Rome: FAO; 2018.\u003c/li\u003e\n\u003cli\u003eAinsworth BE, Haskell WL, Herrmann SD, Meckes N, Bassett DR Jr, Tudor-Locke C, et al. 2011 Compendium of Physical Activities: A second update of codes and MET values. \u003cem\u003eMed Sci Sports Exerc\u003c/em\u003e. 2011 Aug;43(8):1575\u0026ndash;81.\u003c/li\u003e\n\u003cli\u003eInternational Society for the Advancement of Kinanthropometry (ISAK). \u003cem\u003eInternational Standards for Anthropometric Assessment\u003c/em\u003e. Potchefstroom, South Africa: ISAK; 2012.\u003c/li\u003e\n\u003cli\u003eIFCC (International Federation of Clinical Chemistry and Laboratory Medicine). Standardization of Laboratory Measurements. Guidelines for clinical chemistry laboratories. [Accessed via local lab protocols].\u003c/li\u003e\n\u003cli\u003eRutstein SO, Johnson K. \u003cem\u003eThe DHS Wealth Index\u003c/em\u003e. Calverton, MD: ORC Macro; 2004.\u003c/li\u003e\n\u003cli\u003eWHO/UNICEF Joint Monitoring Programme. \u003cem\u003eProgress on Household Drinking Water, Sanitation and Hygiene 2000\u0026ndash;2020: Five years into the SDGs\u003c/em\u003e. Geneva: WHO and UNICEF; 2021.\u003c/li\u003e\n\u003cli\u003eAgongo G, Nonterah EA, Amenga-Etego L, Debpuur C, Kaburise MB, Ali S, et al. Blood pressure indices and associated risk factors in a rural West African adult population: insights from an AWI-Gen Substudy in Ghana. \u003cem\u003eInt J Hypertens\u003c/em\u003e. 2020;2020:11.\u003c/li\u003e\n\u003cli\u003eMayige M, Kagaruki G, Ramaiya K, Swai AB. Physical activity and cardiovascular disease risk factors among adults in Tanzania. \u003cem\u003eBMC Public Health\u003c/em\u003e. 2016;16(1):1\u0026ndash;9.\u003c/li\u003e\n\u003cli\u003eOfori-Asenso R, Agyeman AA, Laar A, Boateng D. Overweight and obesity epidemic in Ghana\u0026mdash;A systematic review and meta-analysis. \u003cem\u003eJ Clin Hypertens\u003c/em\u003e. 2018;20(6):846\u0026ndash;853.\u003c/li\u003e\n\u003cli\u003eOwolabi EO, Goon DT, Adeniyi OV. Alcohol consumption patterns and associated hypertension risks in rural Ghana: A cross-sectional study. \u003cem\u003eBMC Nutr\u003c/em\u003e. 2022;8(1):1\u0026ndash;10.\u003c/li\u003e\n\u003cli\u003eBosu WK, Reilly ST, Aheto JMK, Zucchelli E. Hypertension in older adults in Africa: A systematic review and meta-analysis. \u003cem\u003ePLoS One\u003c/em\u003e. 2019;14(4):e0214934.\u003c/li\u003e\n\u003cli\u003eAmegah AK. Pesticides and Cardiovascular Health in Ghana. \u003cem\u003eEnviron Health Perspect\u003c/em\u003e. 2024;132(1):017001.\u003c/li\u003e\n\u003cli\u003eGomez-Olive FX, Ali SA, Made F, et al. Regional and sex differences in the prevalence and awareness of hypertension: an H3Africa AWI-Gen study across 6 sites in sub-Saharan Africa. \u003cem\u003eGlob Heart\u003c/em\u003e. 2017;12:81\u0026ndash;90.\u003c/li\u003e\n\u003cli\u003eClaeskens G, Hjort NL. \u003cem\u003eModel Selection and Model Averaging\u003c/em\u003e. Cambridge: Cambridge University Press; 2008.\u003c/li\u003e\n\u003cli\u003eSpiegelhalter DJ, Abrams KR, Myles JP. \u003cem\u003eBayesian approaches to clinical trials and health-care evaluation\u003c/em\u003e. Chichester: John Wiley \u0026amp; Sons; 2004.\u003c/li\u003e\n\u003cli\u003eMcElreath R. \u003cem\u003eStatistical Rethinking: A Bayesian Course with Examples in R and Stan\u003c/em\u003e. CRC Press; 2020.\u003c/li\u003e\n\u003cli\u003eKruschke JK. \u003cem\u003eDoing Bayesian Data Analysis: A Tutorial with R, JAGS, and Stan\u003c/em\u003e (2nd ed.). Academic Press; 2015.\u003c/li\u003e\n\u003cli\u003eAbu-Saad K, Avital N, Elias N, Kalter-Leibovici O. Associations between anthropometric measures and blood pressure among adolescents: a cross-sectional study. \u003cem\u003eAm J Hypertens\u003c/em\u003e. 2014 Dec;27(12):1511\u0026ndash;20.\u003c/li\u003e\n\u003cli\u003eGhana Health Service. \u003cem\u003e2023 NCD Progress Report\u003c/em\u003e. Accra: GHS; 2023.\u003c/li\u003e\n\u003cli\u003eMills KT, Bundy JD, Kelly TN, Reed JE, Kearney PM, Reynolds K, Chen J, He J. Global disparities of hypertension prevalence and control: a systematic analysis of population-based studies from 90 countries. \u003cem\u003eCirculation\u003c/em\u003e. 2016;134(6):441\u0026ndash;450.\u003c/li\u003e\n\u003cli\u003eHall JE, do Carmo JM, da Silva AA, Wang Z, Hall ME. Obesity-induced hypertension: interaction of neurohumoral and renal mechanisms. \u003cem\u003eCirc Res\u003c/em\u003e. 2015;116(6):991\u0026ndash;1006.\u003c/li\u003e\n\u003cli\u003eGoma FM, Nzala SH, Babaniyi O, Songolo P, Zyaambo C, Rudatsikira E, et al. Prevalence of hypertension and its correlates in Lusaka urban district of Zambia: a population-based survey. \u003cem\u003eInt Arch Med\u003c/em\u003e. 2011;4:34.\u003c/li\u003e\n\u003cli\u003eAddo J, Smeeth L, Leon DA. Hypertension in sub-Saharan Africa: a systematic review. \u003cem\u003eHypertension\u003c/em\u003e. 2007;50(6):1012\u0026ndash;1018.\u003c/li\u003e\n\u003cli\u003eSobngwi E, Mbanya JC, Unwin NC, Porcher R, Kengne AP, Fezeu L, et al. Physical activity and its relationship with obesity, hypertension and diabetes in urban and rural Cameroon. \u003cem\u003eInt J Obes\u003c/em\u003e. 2002;26(7):1009\u0026ndash;1016.\u003c/li\u003e\n\u003cli\u003ePeltzer K, Phaswana-Mafuya N. Hypertension and associated factors in older adults in South Africa. \u003cem\u003eCardiovasc J Afr\u003c/em\u003e. 2013;24(3):66\u0026ndash;71.\u003c/li\u003e\n\u003cli\u003ePopkin BM. Nutrition transition and the global diabetes epidemic. \u003cem\u003eCurr Diab Rep\u003c/em\u003e. 2015;15(9):64.\u003c/li\u003e\n\u003cli\u003eBampoe-Addo AA, Nartey ET, Odjidja EN, et al. Agrochemical exposure and cardiovascular health risks among Ghanaian farmers. \u003cem\u003eBMC Public Health\u003c/em\u003e. 2021;21:1782.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Table 2","content":"\u003cp\u003eTable 2 is available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Hypertension, Risk factors, Health Demographic Surveillance Site, H3Africa AWI-Gen, Ghana, Kassena-Nankana, Probit \u0026 Bayesian models, LMICs, implementation science","lastPublishedDoi":"10.21203/rs.3.rs-7348872/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7348872/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground\u003c/strong\u003e:\u003c/p\u003e\n\u003cp\u003eHypertension prevalence is rising in rural Ghana, but risk factor identification is constrained by methodological limitations in sparse data.\u003c/p\u003e\n\u003cp\u003eThis study comparatively applied frequentist (Probit) and Bayesian models in identifying hypertension risk factors and evaluating consistency and uncertainty quantification.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e:\u003c/p\u003e\n\u003cp\u003eThis was a cross-sectional analysis of 2,010 adults in the Kassena-Nankana districts. Ordered Probit and Bayesian models were used to assess associations between socio-demographic, behavioral and blood pressure variables. Model fit was compared via AIC/BIC.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e:\u003c/p\u003e\n\u003cp\u003eThe prevalence of hypertension in the study population was 21.9Model comparison showed the ordered Probit model had superior fit (AIC/BIC: 4644.8/4734.2) to the Bayesian model (AIC/BIC: 5137.5/5165.2). Both models consistently identified male sex (Probit β=0.309, p\u0026lt;0.001; Bayesian mean β=0.264, 95% CrI: 0.15–0.34), older age (55–60 years: Probit β=0.514, p\u0026lt;0.001; Bayesian β=0.499, 95% CrI: 0.41–0.58), and higher BMI (Probit β=0.052, p\u0026lt;0.001; Bayesian β=0.052, 95% CrI: 0.04–0.07) as significant risk factors. However, the Bayesian model showed greater uncertainty for marital status (β=0.272, 95% CrI: -0.08–0.56) and smoking (β=-0.260, 95% CrI: -0.29 – -0.01) compared to the Probit model. Both models revealed an inverse association of smoking with hypertension, while pesticide exposure showed conflicting directions and this warrants further investigation.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion\u003c/strong\u003e:\u003c/p\u003e\n\u003cp\u003eMale sex, older age, and higher BMI were consistent predictors of hypertension. Inverse associations with smoking and conflicting effects of pesticide exposure warrant further investigation. These findings highlight the need for targeted, context-specific interventions in rural Ghana. While the Probit model demonstrated a better fit, the Bayesian approach provided deeper insight into uncertainty within sparse subgroups, supporting their complementary use in hypertension research, especially in resource-limited settings.\u003c/p\u003e","manuscriptTitle":"Bayesian and Probit comparative analysis of risk factors for high blood pressure among adults in Kassena-Nankana Districts, Ghana: An AWI-Gen sub-study","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-17 08:58:47","doi":"10.21203/rs.3.rs-7348872/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"333070363530814577170084450436366908486","date":"2025-09-12T12:01:30+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-09-10T10:46:12+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2025-08-14T08:04:46+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-08-14T05:18:23+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-14T05:17:49+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Public Health","date":"2025-08-11T17:56:38+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-public-health","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"pubh","sideBox":"Learn more about [BMC Public Health](http://bmcpublichealth.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/pubh/default.aspx","title":"BMC Public Health","twitterHandle":"@BMC_series","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ba394ebd-b750-419f-aee5-f2fd5bbd0a7d","owner":[],"postedDate":"September 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2025-09-17T08:58:47+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-17 08:58:47","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7348872","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7348872","identity":"rs-7348872","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}