Descriptive epidemiology of blood pressure in a rural adult population in Northern Ghana

Introduction: There are limited data on blood pressure (BP) levels in rural populations in sub-Saharan Africa. This is a cause for concern considering high BP contributes to a substantial public health burden in this population. Methods: Blood pressure levels were measured in a representative rural sample from Kassena-Nankana District of Northern Ghana, West Africa, and associations with anthropometric indices, age, sex, and time of BP measurement were assessed. This cross-sectional survey was made on a random sample drawn from a population register and included 207 males and 367 females aged between 18 and 65 years. Data collection consisted of anthropometric and blood pressure measurements after questionnaire interviews. Household salt consumption was also measured in 20 randomly selected households. Results: Overall prevalence of casual high BP was 19.3% (111/574). Age-adjusted mean (standard deviation) systolic BPs (SBPs) and diastolic BPs (DBPs) were 124.25 (18.67) and 69.92 (12.09) mmHg in males and 122.07 (22.01) and 72.11 (12.35) mmHg in females. Mean body mass index (BMI) was 21.78 (3.09) kg/m 2 ; 21.13 (2.27) for males and 22.15 (3.41) kg/m 2 for females (p >0.05). Blood pressures increased with age, 4 and 3 mmHg per decade for SBP and DBP, respectively (p < 0.000). Regression analyses showed that SBP was significantly associated with age, BMI, waist circumference and time of BP measurement (p <0.000). In multiple regression analyses, age and BMI were independently associated with SBP and DBP in women (p <0.001). Time of BP measurement was independently associated with SBP but only in men (p <0.001). There was no increase in BMI with age. Repeat measurements in 89 subjects yielded a regression dilution factor of 0.57 for SBP. Salt consumption per individual was estimated as equivalent to 12.5 g of pure salt/day.


Introduction
Less than a century ago, rural populations living traditional lives had no problems related to high blood pressure (BP) and it has been erroneously assumed by many that the situation is unchanged.This is largely attributable to a lack of reliable data on BP levels in these settings.Recent reports have shown that high BP is no longer uncommon in rural settings in sub-Saharan African countries 1 .Cooper et al 2 reported a high BP prevalence of 15.4% in rural Cameroon.A mean prevalence of approximately 16% has been reported in rural West African countries 3 .For those over 65 years, recent studies have shown a prevalence as high as between 30% and 40% in rural West Africa 4 .
There are limited data on BP profiles in Ghana, especially for the rural population.Studies undertaken in rural populations in the 1970s and 1980s generally reported a low prevalence of high blood pressure.In 1973, Pobee et al 5 reported high BP prevalences of 2% to 5% for those in 20 rural Ghanaian villages.The study concluded that high BP was not a significant health problem in rural Ghanaians and that large scale programs for the control of high BP should therefore be limited to the urban populations.In the 1980s and 1990s, the prevalence of high BP in rural Ghanaian populations was reported as 4.1% 5 .However, in recent times a 2002 BP survey of two rural communities in the Volta Region of Ghana reported the prevalence of high BP to be 32.8% 6 ; and a 2005 cross-sectional study of four rural communities in the Ga District of Ghana reported a high BP prevalence of 25.4% 7 .
The aims of this article were to measure BP levels, to determine the proportion of individuals with high blood pressure, to estimate the regression dilution factor in order to guide judgments about the likely distribution of usual blood pressures, and to assess associations with anthropometric and other variables putatively related to BP in a rural adult population in Northern Ghana.

Study area
The study was conducted in the Kassena-Nankana District (KND) of Northern Ghana.Ghana is a tropical country located on West Africa's Gulf of Guinea, a few degrees north of the Equator.Ghana borders Cote d'Ivoire to the west, Burkina Faso (formerly Upper Volta) to the north, Togo to the east and the Gulf of Guinea to the south, and has a total area of 238 540 km 2 .Of its estimated population of 19.7 million, almost one-third live in towns or cities.The literacy rate is 74.8%.
The KND is one of the eight districts in the northernmost part of Ghana with Navrongo as its commercial and political capital.It is a relatively flat area and covers 1675 km 2 .The rainy season extends from April to September and the prolonged and intensely hot dry season from October to March.Rainfall averages from 850 to 950 mm per year with the peak occurring in August.Daily mean temperatures range from 20ºC to 40ºC.
The mode of life is traditional with 90% of the inhabitants being small-scale subsistence farmers who depend mostly on the weather for their farming activities.In addition, women work at minor trading activities to supplement agricultural production.The population of the KND is largely dispersed but is also clustered in a few concentrated settlements.In 2006 the population was 143 742, approximately 40% of which was aged under 15 years and 4.9% was 65 years and over.Females constitute 53% of the population.The two main ethnic groups in the population are the Kassena (49%) and the Nankana (46%) 8 .
There is a district hospital located in Navrongo town and four health centres distributed regionally over the district.
The mean minimum distance to a health facility is 4.6 km.In addition to the static health facilities, there are communityhealth nurses who have been retrained and deployed into some communities in the district to offer door-to-door health services to the people.Estimates of the prevalence of high BP are lacking and there is also generally lack of mortality and morbidity data on cardiovascular diseases and strokes.
There are some hospital-based data but these are generally not reliable since death certification and causes of death are not consistently recorded.
Educational attainment in the district is low.About 66% of the population aged 15 years and above have never been to school and only 8% have attained senior secondary or higher levels of education 8 .

Study design and participants
Fieldwork was carried out in the KND from February to April 2007.A random sample of 600 participants was drawn from a population register, regularly updated by the Navrongo Demographic Surveillance System (NDSS), which is owned by the Navrongo Health Research Centre, a field station of the Health Research Unit of the Ghana Health Service.The NDSS is a continuous population registration system that assesses the demographic dynamics of the entire population of the KND.Every 90 days, all compounds in the district are visited to update vital events such as births, deaths, pregnancy, marriages and migration in and out.The sample size required (n = 531) was calculated according to an estimated prevalence of high BP of 5.9% 9 , with a precision of 2% and a confidence level of 95%.To this approximately 70 participants were added to compensate for refusals to participate, anticipated as approximately 10% to 15%.
The response rate was 95.7% with 207 males and 367 females completing the study protocol.The imbalance between the sexes in the random sample was due to females constituting the larger percentage of the population.The 26 non-respondents, all males, were either busy on their farms or not interested in participating.Participants with a previous diagnosis of high BP were not excluded from the study because they had either not been on any medication or had ceased taking their medication.Household salt consumption was also measured in 20 randomly selected households in the district.

Ethical approval was obtained from the local Institutional
Review Board of the Navrongo Health Research Centre.The traditional chiefs, community leaders, political and opinion leaders were approached prior to commencement of the fieldwork and community meetings were held to explain the purpose of the study.Individual affirmations of informed consent were obtained from residents willing to participate using informed consent forms.The consent forms were signed by those who were literate and marked with the left thumbprint of those who were not.
Five field workers and a supervisor with prior experience and training in epidemiological surveys of this nature were recruited and retrained in all the techniques of questionnaire administration, anthropometric and BP measurements, according to standardized protocols.The study team were all local to the study area and were fluent in the local languages.Before commencing the fieldwork, each of the techniques and instruments was tested on a small sample of the population and was shown to be practicable, sufficiently reliable and valid.Data were collected by interview using a simple structured questionnaire, and measurements were recorded on a standard form.For date of birth, an estimated date was assigned to those who were unable to give their exact dates of birth using the seasons, festivals and some historical indicators as guides.June 30th was allocated if the exact date and month could not be decided upon.All participants were visited in their homes by the field workers and subsequently invited on specific dates to attend the various health centres designated for data collection.
Weights were measured to the nearest kg in light clothing and without shoes using calibrated Salter scales (Salter Australia; Melbourne, VIC, Australia).Portable Leicester stadiometers (Invicta Plastics Ltd; Leicester, UK) were used to measure height to the nearest cm with the head in the Frankfurt position.Waist circumferences were taken with non-stretch tape measures and measured to the nearest 0.0 or 0.5 cm.Waist circumference was measured halfway between the lower border of the ribs and the anterior superior iliac spine.All the measurements were taken twice and the means used in the analyses.The ratio of the gap between the mean SBP of quartile 1 and quartile 4 on repeat measurement to that at baseline represents the RDR -the factor by which the standard deviation (SD) of the casual SBP measurements needs to be multiplied to determine the usual SBP distribution.
Salt consumption was also estimated in 20 single family households randomly selected from the survey area.Each family shared the same meals, ate the two or three main meals prepared and rarely bought ready-prepared meals.These were also mainly adult households; only three households contained children and these were aged over 10 years so they were counted as adults.A preliminary survey showed that most households bought just a small bag of salt, which was usually used within 5 to 7 days.Salt 10 samples of 0.5g/100 ml solutions of the rock salt sample (after grinding in a coffee grinder) for Na concentration and the mean was then calculated.The Na concentration in the rock salt was measured using the technique of flame photometry 13 .

Definitions
Casual high BP was defined as an SBP ≥140 mmHg and/or diastolic BP (DBP) ≥ 90 mmHg measured on a single occasion.The BMI was calculated as weight (kg) divided by height (m 2 ).Weight categories were defined based on the calculated BMI (kg/m 2 ) according to the World Health Organization 14 .Underweight is defined as BMI ≤18.

Statistical analysis
The data were coded and entered into a database using

Blood pressure levels
Overall prevalence of casual high BP was 19.3%.Table 2 shows age-specific mean (standard deviation) blood pressures for each sex.The SBP and DBP increased with age in both males and females.Tables 3 and 4 show age-specific mean SBP levels from reports of studies in various sub-Saharan African rural adult populations for comparison.

Univariate and multivariate associations with blood pressure
To assess factors associated with SBP, bivariate and multivariate regression models were developed.Linear regression analyses showed that SBP increased with increasing BMI, age and waist circumference.Systolic and diastolic BPs increased by 4.0 and 3.0 mmHg per decade, respectively.All regression coefficients were statistically significant (p <0.000).

Distribution of usual blood pressures and the regression dilution ratio
The overall mean SBP and DBP for the 89 participants changed from 125.66 (22.57)   6 shows the results of the analysis:

Salt consumption
The average amount of impure rock salt used per day per household was 46.53 g and the average household number was 3.92.The average amount of impure rock salt used per person per day therefore amounted to 13.07 g.The mean Na concentration for 0.5g/100 mL solution of rock salt sample was found to be 81.7 mmol/L.The average Na concentration being used per day per person came up to approximately 5.2 g which was equivalent to 12.5 g/day of (pure) salt.

Blood pressure levels
The findings of BP levels and a high prevalence of high BP in rural Northern Ghana are broadly consistent with reports from other sub-Saharan rural populations 2,3, The age-specific mean BP levels in this study were also consistent with the findings of other studies 19,20 .There was an increase in SBP with age.The increase in BP with age is consistent with reports from other studies conducted in other Ghanaian populations 1,19 .

Predictor variables for BP evaluated
Gone are the days when rural populations showed very little or no increase in BP with increasing age 21 .
The low mean BMIs for this rural population are consistent with the results of other studies conducted in sub-Saharan African rural populations 2,7,9,15-18 .Mean BMIs for females in other rural African populations ranged between 19.8 (3.0) and 23.9 (5.4) kg/m 2 , while that for males was between 19.8 (2.5) and 23.5 (3.1) kg/m 2 , which is also consistent with the finding in this study that mean BMI levels of males and females were not significantly different.

Regression dilution ratio
There were marked differences observed in the mean values at repeat measurements of participants sub-divided into quartiles on the basis of their baseline measurements.As expected, the baseline BP values had a wider distribution than the repeat BP measurements.
The RDR for SBP estimated was 0.57 which is comparable to the results of other studies 22 .The value of the RDR for SBP in this study indicates high within-person variability in SBP, which emphasizes the need to always take repeat measurements of BP after baseline in such population-based studies to correct appropriately for regression dilution.

Salt (Na) intake
The estimated amount of Na used per adult per day was 5.2 g, which is equivalent to 12.5 g of (pure) salt per day (on the assumption that all Na came from salt).This far exceeds current international recommendations of <5g salt per day 23,24 .Our estimates for salt consumption were subject to biases from the exclusion of stock cubes and salt-containing condiments for meal preparation, and failure to account for salt discarded in cooking water.
Salt intake has been associated with higher levels of BP and with more marked increases in BP with age [25][26][27] .In some developed countries, 80% of salt intake has been found to come from processed foods such as bread, pies, ready meals and takeaways 28 .As urbanization continues in sub-Saharan Africa, it is probable that salt consumption will increase due to increased patronage of processed foods.This rural populace still relies on a predominantly traditional 'healthy' diet but their use of salt in cooking, at the table and to preserve fish and meat is common.
There is reliable evidence that a reduction in salt intake reduces BP and that people of black African origin respond well to this strategy.Two short-term clinical trials conducted in Nigeria and Jamaica recently showed the effectiveness of a reduction in salt in lowering BP levels 23,29 .A community program conducted by Cappuccio et al 19 in the Ashanti region of Ghana showed that a reduction in average salt intake led to a small but worthwhile reduction in BP.Current recommendations on salt intake for the prevention of cardiovascular disease is <5 g per day (<2 g/day of Na) or, for persons with high BP, African Americans and middleaged or older adults, <3.6 g of salt per day 23,24 .

Study limitations
Our study was subject to multiple potential biases.The use of the automated BP device (Omron MX3 Plus) as opposed to the random-zero sphygmomanometer recommended by the WHO MONICA protocol 11  implying that BP lowering medication should be initiated earlier 34 .This is a topic for continuing investigation.

Implications for the study
Major implications for studies of this kind include difficulties in interpreting the proportion of the population with high BP, which hampers the comparability of studies, especially with regard to prevalence of high BPs, because of different definitions used and a weakening of the association of BP with risk.Despite the various implications, such studies are still needed in order to provide an indication of the distribution of BPs in populations that were previously thought to be at no risk of cardiovascular diseases and related mortalities.

Conclusion
Findings from the Northern Ghana Blood Pressure Survey show that prevalence of high BP and mean BPs are consistent with the findings of similar studies conducted recently in rural communities in sub-Saharan Africa.This cross-sectional survey provides a baseline against which future changes can be compared, considering the paucity of data on BP levels and factors associated with increased BP in this West African rural population.
This study identifies a need to prevent increasing BP contributing to a major increase in associated health burdens.
The problem of excessive salt consumption needs to be addressed.Because salt is added to food by the consumers themselves (they have little dependence on processed foods), health education programs are needed to specifically target reducing salt intake.
Blood pressure measurements followed administration of questionnaire and anthropometric measurements.Blood pressure was measured in the right arm with participants in the sitting position.Reliable automatic BP devices (Omron MX3 Pus; Omron Healthcare; Kyoto, Japan) which are validated against the reference method of sphygmomanometry (ie has passed the validation recommendations of the International Protocol of the European Society of Hypertension 10 ) were used by trained staff.Following the WHO MONICA methodology 11 , BP was measured twice and the mean of the two used in the analyses.The same trained staff made BP measurements in all locations.As a result of within-person variability in BP values, baseline BP measurements do not often reliably indicate the long-term average or usual BP and this can underestimate its association with an end outcome such as cardiovascular disease.This is known as the regression dilution effect or bias.To correct for this regression dilution effect, repeat measurements of BP in a representative sample of the study participants can be used to obtain an estimate of the attenuation factor known as the regression dilution ratio (RDR).The RDR was determined using a non-parametric method as described by Clarke et al 12 .Repeat measurement data was collected for 89 of the study participants 2 weeks after taking the baseline measurements.After an open invitation was extended to all study participants to attend for re-measurement after 2 weeks, 89 participants presented.These subjects were placed into four quartiles on the basis of their first measurement.The mean systolic BP (SBP) was then calculated for each of these quartiles at baseline and on repetition.The means of the repeat measurements provide unbiased estimates of the usual levels of BP in each quartile.

consumption over 4
days was measured by household inventories.The results, therefore, represented the usual salt consumption more adequately than a single 24 hour food record.Standardized weighing scales of 5 kg (accurate to 0.01 kg) were used to weigh household stocks, including their containers.The survey started with measuring opening stocks of salt before breakfast on the first day of surveillance.Closing stocks were measured on the fourth day before breakfast.Total consumption of each household was calculated as (inventory + new purchase) -remainder.The average salt consumption per day for each household was calculated by dividing the total consumption per each household by 3 days.The average salt consumption per individual per day was then calculated using the number of people in each household.A sample of the rock salt was sent to the Dunn Nutrition Unit (Diet & Cancer Group) of the Medical Research Council, Cambridge, UK to determine the Sodium (Na) concentration.Assays were conducted on

Table 1
shows the characteristics of the study population, anthropometrics and BP levels.In all, 75% of study participants were from the Kassim ethnic group; 72.0% were married and almost half (45.3%) had no formal education.Mean BMI was 21.78 (3.09) kg/m 2 ; 21.13 (2.27) and 22.15 (3.41) kg/m 2 for males and females, respectively.Mean BMIs did not vary significantly with age.Of all participants, 75.4% were in the normal weight category.

Table 4 : Age-specific mean systolic blood pressures (SD) in sub-Saharan rural African populations in published studies
-, No data available.

Table 5 : Multiple regression analysis of factors associated with systolic and diastolic blood pressure, according to sex
BMI, Body mass index; BP, blood pressure; SE, standard error.

Issues in the descriptive epidemiology of blood pressure
33432mmHg and/ or usual DBP ≥90 mmHg.Evans and Rose30proposed the seemingly pragmatic definition of high BP as the level of BP above which treatment does more good than harm.This definition, when applied to BP levels in isolation, also has its short-comings; the associated absolute risk is not taken into account.Though the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation and Treatment of High BP (JNC VII) still does not embrace an absolute risk approach in addition to the level of BP to guide treatment, compared with its previous version, the WHO and European guidelines now emphasise the importance of an absolute risk approach in cardiovascular disease interventions31,32.There have been suggestions about the need to set different thresholds for different populations because risk factor levels and disease risk varies in different populations.Very little data exists on what constitutes an appropriate threshold in African populations.Law and Wald33suggested that everyone at high risk of cardiovascular disease should be targeted instead of imposing risk-factor thresholds in cardiovascular disease intervention.They examined seven dose-response relations between physiological variables and risk of disease.It was realised that a given absolute change in a risk factor reduces the risk of disease by a constant proportion, irrespective of the starting level of the risk factor.The best predictors of risk of cardiovascular disease are the risk factors that cannot be changed, such as age and sex, and therefore interventions should be determined by a person's level of risk and not by the level of risk factors.Therefore BP lowering drugs should not only be limited to people with high BP because offering treatment only to those who have high values of the risk factors mean targeting a small percentage of the population destined to suffer a disease event.It has been suggested that there may be a lower threshold for target organ damage in African populations, Previous WHO guidelines gave the cut-off level for high BP as a usual SBP ≥160 mmHg and/ or a usual DBP ≥95 mmHg.This was changed in 1998 to a usual SBP