Hypertension treatment and control in sub-Saharan Africa: the epidemiological basis for policy
BMJ 1998; 316 doi: https://doi.org/10.1136/bmj.316.7131.614 (Published 14 February 1998) Cite this as: BMJ 1998;316:614
This article has a correction. Please see:
- Richard S Cooper, chairman (rcooper{at}orion.it.luc.edu)a,
- Charles N Rotimi, assistant professora,
- Jay S Kaufman, research associatea,
- Walinjom FT Muna, chief of cardiologyb,
- George A Mensah, associate professor of medicinec
- a Department of Preventive Medicine and Epidemiology, Loyola University Stritch School of Medicine, Maywood, IL 60153, USA
- b University of Health Sciences, Yaounde, Cameroon
- c Medical College of Georgia, Augusta, GA, USA
- Correspondence to: Dr Cooper
- Accepted 22 October 1997
Although enormous challenges persist in the control of infection in sub-Saharan Africa, non-communicable diseases are also important threats to the health of adult Africans. 1 2 Controversy exists, however, over the priority these conditions deserve in the competition for scarce resources. It has recently been argued that hypertension treatment, for example, should not be attempted in sub-Saharan Africa given the high costs.3 Unfortunately, these discussions take place in an information vacuum, since it is impossible to define the burden of chronic conditions in societies where health statistics are unavailable.4 Cohort studies may serve as a proxy for vital statistics and give approximate answers to questions on the usefulness of treatment for chronic disease.5 Hypertension is particularly suited to this model because it is easily diagnosed, highly prevalent, and information on outcomes is plentiful.
Although the relative risk of a cardiovascular event in people with high and normal blood pressure is similar in Africa and the United States, the absolute risk is up to 13 times greater in Africans
Summary points
In sub-Saharan Africa it is difficult to formulate and justify policy on treating chronic conditions such as hypertension as there are no health statistics from which to judge likely costs and benefits
Cohort studies on hypertension in Nigeria and Zimbabwe and epidemiological information show that between 10 and 20 million people in sub-Saharan Africa may have hypertension and that treatment could prevent around 250 000 deaths each year
Taking account of both relative risk and absolute risk of a cardiovascular event or death, a systolic pressure of 160 mm Hg is recommended as a threshold for treatment in Africa
The reduction in population attributable risk associated with treatment could be 2% in Africa compared with 0.15% in the Unites States—some 13 times higher
“Number needed to treat” analysis shows that the costs of drugs to prevent one death would be $1800 (£1104) in Africa and $14 000 to $1m (£8589 to £613 496) in the United States
This evidence challenges the assertion that treatment for hypertension should not be a health priority in sub-Saharan Africa
Hypertension is the most common cardiovascular condition in the world and the problem of defining a strategy for control confronts all societies. Hypertension is fully treatable, but social conditions in Africa make the implementation of blood pressure control programmes difficult. Lack of a clear strategy based on evidence has undermined further these efforts. We outline here the epidemiological data on hypertension that are available to guide health policy in sub-Saharan Africa.
Burden of hypertension in sub-Saharan Africa
Sub-Saharan Africa is a diverse region comprising 47 countries. It is home to approximately 480 million people.6 Among the elite in African society, the model for hypertension control currently in force in Europe and the United States would be entirely appropriate. However, most Africans (fully 75%) live in rural areas and are marginally integrated into the cash economy, while around 20% live in extreme poverty in cities. The challenge lies in developing effective strategies for these sections of society.
Distribution of hypertension
The prevalence of hypertension is low in rural Africa,7 and a graded increase is seen in the urban poor and working class.8–4 Comparing prevalences in studies is difficult, however, as sampling and measurement vary. We recently completed surveys in three communities using a common sampling and measurement protocol (table).13 In southwest Nigeria, blood pressure in villagers rose modestly with age compared with values in residents of urban areas. Seven per cent of the rural sample had hypertension (defined as blood pressure greater than 160/95 mm Hg or antihypertensive treatment). High blood pressure was more common among the urban poor from Ibadan in Nigeria (17%), and substantially more prevalent in salaried workers in Harare, Zimbabwe (26%). Figure 1 shows the gradual upward shift in the distributions of blood pressures across these groups.
Mean blood pressure (mm Hg) and age in three cohort studies in Africa
Probability density distribution of age adjusted systolic blood pressure in three African communities.
Estimates of preventable deaths
These data represent the principal social strata of African society. They include the range of previous estimates and provide a reference point for considering the burden of hypertension. The estimated distribution of the African population between the three sectors was 75% rural, 20% urban poor, and 5% urban salaried workers and elite. Based on a sub-Saharan population of 500 million, half of whom are older than 25, a hypertension prevalence of 5-10% yields 10-20 million cases. If annual mortality is 2%, and 5% of deaths result from hypertension, then approximately 250 000 deaths each year are preventable.
Risk factors
In Africa, as elsewhere, obesity and sodium intake are risk factors for hypertension. 12 14–4 In industrialised societies such as the United States, obesity accounts for 25% of cases of hypertension. However, the relative leanness of Africans means that the contribution of obesity to high blood pressure is only around 10%.15 Psychosocial factors in hypertension have been studied little. Instruments for measuring these factors in African societies have not been developed. No trials of preventive measures that have reduced risk factors for hypertension have been reported from Africa. Drug treatment is therefore the only proved option at present.
Impact of hypertension on mortality
Although the sequelae of hypertension are predictable,18 the net impact of high blood pressure on all cause mortality is not. Given the high all cause mortality in Africa, and the small proportion of people who reach an age where sequelae are common, the contribution of hypertension is uncertain. We know of only one prospective study that has been published. In three years of follow up of 1200 adults from the rural district of Igboora-pa, Oyo State, Nigeria, the relative risk of death in people with hypertension was 1.6,19 a value observed in many other studies.20 Mortality in this community was high (2.8% per year) and hypertension was associated with a large attributable risk of about 2% per year. It seems unlikely, however, that all of the attributable risk in people with hypertension resulted solely from cardiovascular diseases. Chronic diseases are a predisposing factor for fatal infection, and this could lead to short survival in patients with stroke, heart failure, or renal insufficiency. Under these circumstances prevalence surveys would underestimate disease burden, and prospective risk estimates would exaggerate the cause specific role of hypertension. If this analysis is correct, this interaction between chronic and acute conditions changes considerably the framework within which the value of treating chronic disease in Africa should be viewed.
Mortality in African adults is unknown and probably varies considerably between regions from 1% to 2.5% per year.21 Community and hospital studies suggest that 5% to 15% of people die from cardiovascular diseases—mainly stroke and congestive heart failure resulting from hypertension. 2 5 22 23 By combining the data on prevalence and relative risk summarised above, we can also estimate the deaths attributable directly to hypertension from the Igboora-pa study. The annual mortality in people over age 25 was 2.8%, and hypertension was associated with a relative risk of 1.6. Calculation of the population attributable risk confirms that about 5% of deaths can be attributed to hypertension. Given that half of all deaths occur in adults, the overall contribution of hypertension would therefore be around 2.5%. By comparison, a study of global disease burden ascribed 5.8% of deaths at all ages to hypertension.24
Potential for hypertension control with drugs
Who is a candidate for antihypertensive treatment? An answer to this critical question requires information on projected benefit, feasibility, and cost effectiveness of treatment.24–4 While practical considerations will be paramount in the end, it remains essential to describe the medical consequences of the decision that is taken. The calculations presented here are preliminary ones, intended to place the value of hypertension treatment in context. In particular, they provide a counter argument to the view of some that treating hypertension is not cost effective in Africa, and that support should be removed.3
Analysis by “number needed to treat”
Data on the benefit of drug treatment for hypertension in industrialised countries probably understate its impact. Observational studies in westernised societies since the 1960s do not reflect the natural history of the disease, given its widespread treatment.29 In trials, some patients in the placebo arm cross over to treatment. 30 31 Indeed, early placebo trials contradict the impression from later trials on “mild” hypertension, and should be considered as the background for policy decisions. In the first Veterans Administration cooperative trial, one of 73 patients taking treatment and 27 of the 70 control subjects had a cardiovascular event.32–4 The excess in absolute event rates was 28% per year. Under these conditions, four patients would need to be treated each year to prevent one cardiovascular event. In the second Veterans Administration trial, in which patients with diastolic pressures of 90-114 mm Hg were enrolled, the annual cardiovascular event rates were 5.5% in the treated group and 16.7% in the control group, and death rates were 1% and 3% respectively. The corresponding number needed to treat values would be nine each year to prevent a cardiovascular event and 50 to prevent a death. In patients with milder hypertension, however, reductions in mortality were substantially smaller, and the number needed to treat rose to over 1000. 25 35
Individual attributable risk, number needed to treat, and population attributable risk in relation to thresholds for hypertension treatment. Smoothed plots were derived from analysis of adult mortality in Igboora-Pa, Nigeria19
Proposed guidelines for treatment
Choosing an appropriate threshold for treating blood pressure is problematic in Africa. Because resources are scarce, an argument exists for raising the cut off point to reduce the cost. An objective decision, however, requires information that enables benefits to be quantified. The relative risk is the usual basis for determining the treatment threshold. In most studies, the relative risk in people whose blood pressure is greater than 140/90 mm Hg compared with those who have normal blood pressure is around 1.6, and increases 50% with each succeeding 10 mm Hg increase in systolic pressure.21 If the probability of an event is low, the absolute benefit of treatment is small, no matter what the relative risk. Annual death rates in cohorts in the United States are 0.15% and 0.30% for people with normal and high blood pressures respectively.21 These rates produce the same relative risk as cardiovascular event rates in people with normal and high blood pressure in Africa (2% and 4% respectively). In the United States, the reduction in attributable risk associated with treatment could be 0.15% at most, while in Africa it could be 2%—some 13 fold higher. This order of magnitude difference in attributable risk is the central problem in assessing the potential benefit of treatment in Africa.
Absolute versus relative risk
Absolute levels of risk that warrant treatment in Africa are probably in the range of 1%-3%. In the preliminary findings of the prospective study in rural Nigeria, mortality was 2.8% per year in people with normal blood pressure, but 5.1% per year in those with hypertension.19 If treatment for hypertension eliminated all the excess, then the number needed to treat per year to prevent a death would be 43. Obviously, this estimate is based on the optimistic assumption that 100% control could be achieved, and is an upper bound of what is possible. However, as noted, the Veterans Administration trials gave similar estimates.32–4
In other regions of the world, recommendations for treatment are based on relative risks. With the exception of New Zealand, where absolute levels have been introduced,36 blood pressure thresholds are either 160/95 mm Hg or 140/90 mm Hg.36 It is assumed that adopting these thresholds in Africa would achieve equivalent benefit. While this is a reasonable assumption in relation to relative risk, the situation in terms of attributable risk could be very different. Taking into account both relative and absolute risk, we believe that a systolic pressure of 160 mm Hg is still currently justified as the threshold. We calculate that the annual mortality in people with blood pressures above this would be around 4%, compared with 1%-2% in those with normal pressures. The number needed to treat to prevent one death would therefore be around 50 each year, and the number needed to treat to prevent a serious cardiovascular complication would be lower.
Making a compromise
Although benefit at the level of the individual rises as the treatment threshold increases, the benefit at the population level falls (fig 2). A compromise is required which makes treatment worthwhile for individuals, yet still has an impact on public health. The healthcare system must allocate funds to urgent priorities. All of these decisions ultimately require epidemiological data incorporated into the decision making model. An important attribute of the algorithm shown in fig 2 is its potential to determine a useful level of absolute risk (or the number needed to treat) and to calculate from this the blood pressure that should be used as the cut off point. With this approach it is not necessary to rely on relative risk thresholds adopted by external expert panels.
Costs in relation to numbers needed to treat
Given the data on reductions in mortality associated with diuretic drugs and β blockers, a strong argument exists for using these as standard treatment 30 31 37 Costs in Nigeria are 10-15 cents (US$) per tablet, yielding annual expenditure of $36 (£22) for treatment with one drug only. Assuming a number needed to treat of 50, the cost of drugs alone to prevent one death would be $1800 (£1104) in Africa. In the United States, however, the number needed to treat per year to prevent a death is 1354 for people of similar age, and costs for drugs alone to prevent one death range from $14 000 to $1 million (£8589 to £613 496), depending on the drug used.25
Conclusions
Complex problems confront any attempt to design a public health strategy to control chronic disease in developing countries, particularly Africa. The obligation for the health professions is to assess potential benefit to patients in the local context, without imposing external standards. For example, the rationale for dismissing the value of hypertension treatment was based on the costs estimated in the United States, which are unlikely to apply in Africa.3 Only after the health benefits have been defined can useful discussions take place on the social and political possibility that such treatment will be made available. While acknowledging the obstacles to implementing chronic care for asymptomatic conditions in Africa today, the spectre of low cost effectiveness should not foreclose the scientific debate. Otherwise concern over chronic disease in developing countries is little more than hand wringing. Empirical evidence challenges the assertion that hypertension treatment should not be a health priority, suggesting instead that investment in an organised care system would reap large gains in adult health.
Acknowledgments
Funding: This work was supported by grants from the NHLBI (HL 45508 and HL 47910).
Conflict of interest: None.
References
Log in
Log in using your username and password
Log in through your institution
Subscribe from £173 *
Subscribe and get access to all BMJ articles, and much more.
* For online subscription
Access this article for 1 day for:
£38 / $45 / €42 (excludes VAT)
You can download a PDF version for your personal record.