Rural-urban variation in hypertension among women in Ghana: insights from a national survey

In this study, we used data from the 2014 GDHS. Since the inception of the GDHS, it is only the 2014 edition that assessed hypertension status of Ghanaian women. The GDHS is a five-year interval nationally representative survey mostly carried out by the Demographic and Health Surveys (DHS) Program, Ghana Statistical Service and Ghana Health Service [25]. The survey seeks to collect, analyse, and circulate representative and reliable data on core health indicators in over 90 countries including Ghana. These core healthcare indicators comprise adult health and lifestyle including hypertensive status, nutrition as well as maternal and child health. In 2014, the survey recruited 9396 women within the 15–49 age group. The survey made use of an updated frame prepared for the 2010 Population and Housing Census (PHC) and had a response rate of 97%. We included 9333 women in the present study because this sample had complete information for the analysis. The sample was derived through a two-stage sampling approach aimed at permitting estimation of core indicators throughout the then 10 administrative regions. The first stage involved the selection of sample points or clusters made up of enumeration areas (EAs) whereas the second stage constituted a sampling of households following systematic sampling. Between January and March 2014, household listing was conducted for this purpose. Approximately 30 households were identified per cluster. This resulted in 427 clusters (with 216 from urban and 211 from rural settings) and 12,831 households throughout the country [25]. The sample excluded institutional and nomadic persons such as those in hotels and prisons. The data was deemed suitable for the study because it is nationally representative and the first of its kind to investigate the hypertensive status of women in their reproductive age at the national level. We had access to the dataset from the website of Measure DHS and is freely available through https://​dhsprogram.​com/​data/​available-datasets.​cfm.

Hypertensive status (measured by blood pressure) of Ghanaian women aged 15–49 was the dependent variable for the study. During the 2014 DHS, blood pressure was monitored and measured on three occasions following the UA-767F/FAC (A&D Medical) blood pressure with at least 10 min interval [25]. In determining hypertensive status, an average of the second and third measurements were computed, and this conforms to calibration by other studies on hypertension that are underpinned by the DHS datasets [16, 26, 27]. Following the Joint National Committee Seven (JNC7) guideline, hypertension was conceptualised as an average systolic blood pressure of ≥140 mmHg and/or an average diastolic blood pressure of ≥90 mmHg. Hypertensive women were coded as 1 whilst non-hypertensive women were coded otherwise ‘0’.

The main explanatory variable was place of residence (rural or urban), in line with the categorisation of the DHS survey. The choice of this explanatory variable emerged from its statistically significant association with hypertension with dominance among the urban population [18, 28, 29] whilst some evidence also documents high inclination toward rural residents [9]. There was, therefore, the need to interrogate and identify the situation in Ghana. We included some socio-demographic characteristics of the women; age, wealth quintile, marital status, occupation, salted fish consumption, region and tobacco use (comprising cigarette, tobacco, and snuff). We included salted fish consumption because some evidence indicates an association between hypertension and salt intake [30, 31]. Behavioural factors such tobacco use and some related lifestyles have been documented as precursors to hypertension [32‐34]. The following variables were recoded to suit the analysis; marital status was recoded as “single = 0”, “married = 1”, “cohabiting = 2”, “widowed = 3”, “divorced = 4” and “separated=5; occupation recoded into “not working=0” “agriculture = 1″ “manual = 2″ and “service = 1″; salted fish consumption into “No = 0″ and “Yes = 1.”

In our analysis, we calculated the proportion of women with hypertension by place of residence (rural or urban) as shown in Fig. 1. We also computed the proportion of hypertension by the socio-demographic variables as shown in Table 1, and also explored which of them had a significant association with hypertension. Out of the ten variables tested (see Table 1), six were significant and were used in our inferential analysis (residence, age, wealth, marital status, occupation, and region). To ensure that there is no multicollinearity between the explanatory variables, tests for multicollinearity was conducted and it was revealed that the socio-demographic variables are not highly correlated [mean VIF = 1.42, maximum = 2.31, minimum = 1.02]. Due to the dichotomous nature of our dependent variable, binary logistic regression analysis was conducted where odds ratios (ORs) and adjusted odds ratios (aORs) with their respective 95% confidence intervals (95% CI) were reported (Table 2). Model 1 focused on the bivariate analysis between residence and hypertension whilst Model 2 presents a multivariable model adjusting for the effect of the significant socio-demographic variables. The results were weighted in order to achieve proportionality at the national level and the entire analysis was conducted using Stata version 13.

DHS reports that informed consent was sought from all the women prior to their participation in the survey. The DHS sought ethical approval from the Ethics Committee of ORC Macro Inc. and that of Ghana Health Service. Authors of this manuscript were not directly involved in the data collection processes but rather obtained access by applying to the DHS (via https://​dhsprogram.​com/​data/​available-datasets.​cfm) in order to obtain access.

The findings indicate that 14.6% of the participants had hypertension nationwide. Of these, 9.5% emerged from urban locations whilst the remaining 5.1% occurred among women in rural residential areas of Ghana.

As indicated in Table 1, 71.2% of urban women aged 30–34 had hypertension with 55.5% of those aged 15–19 women also having it. Nearly all the richest urban residents had hypertension (97.9%), whilst only 13.8% poorest women in urban Ghana had hypertension. Single women in urban Ghana had hypertension (72.9%) and most women who had secondary education and resided in urban settings had hypertension (72.8%). With respect to occupation, 77.5% of those in service profession and living in urban Ghana had hypertension, however, only 18.7% of those in urban settings in agriculture had hypertension. A significant proportion of women in urban locations who were not consuming salted fish were hypertensive (66.3%). Across the regions, at least nine out of ten urban residents in the Greater Accra region had hypertension (92.2%). Hypertension is quite phenomenal among urban women who use tobacco (71.4%). With the exception of tobacco use and salted fish consumption, the rest of the socio demographic characteristics were significantly associated with hypertension at 95% significance level.

Outcome of the binary logistic regression is presented in Table 2. In the crude model (model I), rural women had lower odds of hypertension [OR = 0.59; 95% CI = 0.52, 0.67] compared to urban women, however, this was insignificant in the adjusted/final model. The propensity to be hypertensive was lower for women aged 15–19 [aOR = 0.074; 95% CI = 0.051, 0.11] relative to those aged 45–49. Compared to the richest women, poorest women were less likely to be hypertensive [aOR = 0.63; 95% CI = 0.45, 0.89]. Single women were less probable to experience hypertension as compared with the divorced [aOR = 0.66; 95% CI = 0.46, 0.97]. We also realised that, compared to women in the Upper West region, women of all other regions were more probable to be hypertensive particularly those in the Greater Accra region [aOR = 2.17; 95% CI = 1.50, 3.13].

The conclusions drawn for the study are based on the larger sample size derived by probabilistic method used, hence, having a true representation of the population studied. A weakness of the study is that due to the cross-sectional nature of the survey, causality could not be established. Also, the study only reflected women’s hypertension situation, hence the conclusions drawn may not be applicable for men. Finally, the study methodology, which is cross-sectional in nature, limited the effort to explore reasons behind some of our observations.