Hypertensive Retinopathy in Afro-Caribbeans and Europeans
Prevalence and Risk Factor Relationships
Abstract The prevalence of hypertension is particularly high in people of black African descent throughout the world, and the consequences of hypertension, such as hypertensive heart and renal disease and stroke, are also more common. But there is little consensus on whether hypertensive retinopathy follows a similar pattern. We determined the prevalence of hypertensive retinopathy and its relationships with resting and ambulatory blood pressure in a population study of Afro-Caribbeans and Europeans aged 40 to 64 years in London, UK. Retinal photographs of 651 participants were graded for hypertensive retinopathy. Age- and sex-standardized prevalence of retinopathy was 11% (95% confidence interval, 8% to 14%) in Europeans and 21% (95% confidence interval, 16% to 26%) in Afro-Caribbeans (P<.001), respectively. This ethnic difference in prevalence was greatest in normotensive women (8% in Europeans versus 20% in Afro-Caribbeans, P<.001). Resting systolic pressure was 8 mm Hg higher in normotensive Afro-Caribbean compared with European women, but this could not fully account for the ethnic difference in the prevalence of retinopathy. Examination of the different relationships of age and resting and ambulatory blood pressures with hypertensive retinopathy showed that these relationships were strongest in European women and weakest in Afro-Caribbean women. We conclude that hypertensive retinopathy is more common in Afro-Caribbeans, particularly women, and that ethnic differences in resting blood pressure cannot fully account for this. The relatively weak relationship between resting and ambulatory blood pressures and retinopathy in Afro-Caribbeans suggests that factors other than blood pressure determine the high rates of hypertensive retinopathy in this group.
Populations of black African descent in the United Kingdom, United States, and Caribbean share a high prevalence of hypertension and high rates of mortality from conditions associated with hypertension.1 2 3 Rates of hypertensive end-organ damage should also be high. However, although it is generally accepted that black African populations have high rates of left ventricular hypertrophy4 and hypertensive renal disease,5 there is no clear consensus on whether hypertensive retinopathy follows a similar pattern.
Earlier studies have suggested that hypertensive retinopathy in black Africans is extremely rare, despite high blood pressure (BP) levels, and find no relationship between BP level and the likelihood of retinopathy.6 However, other studies suggest that retinopathy is common in people of black African descent. In hypertensive subjects, the prevalence of retinopathy has been estimated at well over 70%.7 One explanation for such discrepant findings may be that the latter study examined a highly selected clinic population, in which complication rates are expected to be high. But population-based studies also find a high prevalence of retinopathy, ranging from 60% to 85% in hypertensive subjects.8 9 These contradictory findings are therefore difficult to explain on the grounds of differences in population selection. Earlier studies have generally relied on funduscopy for the assessment of retinal changes, and none have assessed whether ethnic differences in the prevalence of retinopathy can be accounted for by differences in BP.
We examined these questions in a population-based study of Afro-Caribbeans in the United Kingdom. We have previously reported the high prevalence of hypertension in Afro-Caribbeans compared with Europeans10 ; our objective in the present section of the study was to determine whether the prevalence of hypertensive retinopathy was equally high in Afro-Caribbeans. We also assessed the relative contributions of age and BP in the two ethnic groups to hypertensive retinopathy and determined whether ambulatory BP was a better predictor of hypertensive retinopathy than resting BP.
The present study was part of a larger community-based survey examining hypertension in Afro-Caribbeans and Europeans; detailed methods have been described elsewhere.10 Briefly, 1164 Afro-Caribbean and European men and women aged between 40 and 64 years randomly selected from family doctor lists attended a health check, which included completion of a questionnaire detailing medical history, current drug therapy, and health-related behaviors. Resting BP was measured twice with subjects in a seated position after 5 minutes of rest with the use of a random-zero sphygmomanometer. Subjects were defined as hypertensive according to World Health Organization criteria: systolic BP greater than or equal to 160 mm Hg, or diastolic BP greater than or equal to 95 mm Hg, or on medication for hypertension.11 All other subjects were defined as normotensive. Every hypertensive subject and every alternate normotensive subject was asked to wear an ambulatory BP monitor (Takeda TM2420) for the 24-hour period after examination.10 Respondents attended one of two screening centers (Central Middlesex Hospital or Wembley Stadium Medical Centre) according to their area of residence. All participants (n=693) who attended the Central Middlesex Hospital screening center were considered eligible for retinal photography. Participants receiving treatment for glaucoma were not photographed because instillation of mydriatic drops was thought to be hazardous; this resulted in the exclusion of 10 people.
Retinal photographs were taken with a 45° field “nonmydriatic” camera (Canon) after the pupil had been dilated with 1% tropicamide. Two photographs were taken of each eye: a nasal and a macular field. These fields were chosen to ensure that adequate coverage of the disc and macular areas were obtained. Kodachrome daylight film was used (Kodak), and the film was printed as color transparencies. These transparencies were graded for hypertensive changes according to the classification of Keith et al.12 Grade I was taken as an arteriolar diameter less than or equal to 50% of the venous diameter, grade II as arteriolar narrowing plus arteriovenous crossing changes, and grade III as the above changes together with retinal hemorrhage or exudate formation. Films were graded twice, first by a general physician (P.S.S.) and second by the physician and an ophthalmologist (R.W.), both agreeing on a score for each subject. Data from the first grading were not available to the observers for the second grading. The two observers did not have access to clinical information on the subject, including the measured BP level. The κ coefficients of reliability for these duplicate assessments were 0.72 for Afro-Caribbeans and 0.75 for Europeans.
The mean of the two resting BP values was used in all analyses, including the values from those subjects on medication for hypertension. Age adjustment of continuous variables was performed by use of least-squares regression models; the mean of the variable was calculated as the value predicted in the model when age was held at its mean value. Prevalence rates were directly standardized to the age distribution of the study population that participated in the retinal photography section of the study. Complete ambulatory BP data were obtained from 242 subjects in this section of the study. Mean 24-hour ambulatory BP values were calculated for each individual and used in the analyses.
Mean age and resting BP values in the four sex/ethnic groups in this section of the study were compared with the entire study sample and were not found to be different.
Logistic modeling was used to examine the relative contributions of age and resting and ambulatory BPs on hypertensive retinopathy. Standardized relative odds were calculated by multiplying the beta regression coefficients by the standard deviation of each variable (age and resting and ambulatory BPs) from a model that included all these variables. The resulting odds ratio estimate allows the direct comparison of the independent effects of these variables to decide which has the greater effect on hypertensive retinopathy.
The relationship between BP and retinopathy was qualitatively similar with the use of either systolic, diastolic, or mean arterial BP. Therefore, results are presented for systolic BP only.
We were able to assign a hypertensive retinopathy grade to 95% (651 of 683) of all subjects. A greater proportion of Afro-Caribbeans (7%, 22 of 299) had ungradable photographs compared with Europeans (3%, 10 of 384). Most ungradable photographs were from normotensive subjects. People with diabetic retinopathy (11 in each ethnic group) and other retinal abnormalities were not excluded. Hypertensive retinopathy was confined to the milder categories, with only 4 subjects, 1 from each sex/ethnic group, displaying grade III retinopathy. These retinopathy grades were combined in subsequent analyses and were compared with subjects with no retinopathy.
The age- and sex-standardized prevalences of hypertensive retinopathy in all subjects were 11% in Europeans and 21% in Afro-Caribbeans (P<.001). This ethnic difference in prevalence generally persisted when the two sexes were examined separately by BP category (Table 1⇓). The ethnic difference in retinopathy prevalence was most marked in normotensive women (20% in Afro-Caribbeans compared with 8% in Europeans, P<.01).
The largest ethnic difference in resting systolic BP was also noted in normotensive women, with systolic BP being on average 8 mm Hg higher in Afro-Caribbean compared with European women (Table 1⇑). Logistic regression analyses were used to test whether this ethnic difference in BP could account for the observed difference in retinopathy prevalence. Age-adjusted odds of hypertensive retinopathy were 1.67 in Afro-Caribbean women compared with European women (P=.0009). This risk was considerably attenuated but not abolished when resting systolic BP was added to the model, with the odds of retinopathy becoming 1.42 in Afro-Caribbean women compared with European women (P=.03).
Examination of the standardized regression odds for all subjects who had retinal photographs shows that both age and resting BP had a stronger relationship with hypertensive retinopathy in Europeans than in Afro-Caribbeans (Table 2⇓). Thus, for an equivalent rise in age or BP, the risk of developing hypertensive retinopathy was greater for Europeans than for Afro-Caribbeans. The strongest relationships between BP and age with retinopathy were observed in European women; these relationships were weakest for Afro-Caribbean women.
For the subset of participants who wore an ambulatory BP monitor, the standardized regression odds show that the relationship between hypertensive retinopathy and predictor variables was again stronger in Europeans than in Afro-Caribbeans and that in both ethnic groups, ambulatory BP was generally a better predictor of hypertensive retinopathy than resting BP.
This is the first population-based study to report the relationships of resting and ambulatory BPs with hypertensive retinopathy in these two ethnic groups. We demonstrate that the prevalence of hypertensive retinopathy is higher in Afro-Caribbeans than in Europeans and that this difference is generally confined to the milder forms of retinopathy. Our prevalence estimates fall between the observations of researchers who have commented on the relative rarity of hypertensive retinopathy in people of black African descent6 and those who show strikingly high rates of hypertensive retinopathy in this population.8 9 We also confirm that the total prevalence of hypertensive retinopathy is greater in Afro-Caribbean women than men.9
There are several possible explanations for the high prevalence of hypertensive retinopathy in Afro-Caribbeans. The dark retinas of people of black African descent may result in an overestimation of the degree of arteriolar narrowing and hence a falsely high estimate of the prevalence of hypertensive retinopathy. But the ethnic difference in hypertensive retinopathy can be almost completely accounted for by the difference in the prevalence of retinopathy in normotensive women; Afro-Caribbean women had more than twice the prevalence of retinopathy compared with European women. It is therefore less likely that overestimation of vessel narrowing can completely account for the ethnic difference observed.
The reason for the high rate of retinopathy in Afro-Caribbean women may be associated with the marked ethnic difference in resting BP in normotensive women. However, we show that ethnic differences in resting BP cannot completely account for this ethnic difference in the prevalence of hypertensive retinopathy.
A further explanation is that resting BP does not adequately characterize the BP burden the body is exposed to and that measures of ambulatory BP might better approximate this. We show that equivalent changes in mean 24-hour ambulatory systolic BP had a greater effect on the likelihood of hypertensive retinopathy than resting BP, but our numbers were too small to test whether ambulatory BP differences could account fully for the ethnic differences in hypertensive reti- nopathy.
We also show that an equivalent rise in age or resting or ambulatory BPs increased the likelihood of retinopathy to a greater extent in Europeans than in Afro-Caribbeans. The relationship of age and BP with retinopathy was strongest for European women, who have the lowest prevalence of hypertensive retinopathy, and weakest for Afro-Caribbean women, who have the highest prevalence of hypertensive retinopathy. Therefore, retinal changes that are assumed to be the consequences of raised BP may have different causes in the different ethnic groups. The strong relationship between measures of BP and retinopathy in European women is an indication that these eye changes probably do reflect the effects of hypertension. In Afro-Caribbean women, the high prevalence of hypertensive retinopathy, in conjunction with the relatively poor relationship between BP and retinopathy, suggests that other factors play a more important role in determining these retinal changes. Selective mortality of Afro-Caribbeans with high BP and severe retinopathy might account for the less strong relationship between BP and retinopathy in this group but runs counter to the observation that the prevalence of retinopathy is greater in Afro-Caribbeans.
The assessment of hypertensive retinopathy is based on the classification of Keith et al.12 This classification has been criticized because factors other than BP, for example age, may have an equivalent effect on retinal changes. This classification is also assumed to be valid for all populations, but its appropriateness in different populations has not been formally tested. Thus, this classification may be less valid when two different ethnic groups are being compared in which risk factor relationships may differ. However, there is a relationship between age, resting and ambulatory BPs, and hypertensive retinopathy in Afro-Caribbeans, even though this is weaker than that found in Europeans. Although there is a fine distinction between the normal retina and early grades of hypertensive retinopathy, these changes do appear to predict a definite difference in vascular mortality.13 14 This would suggest that the high prevalence of hypertensive retinopathy observed in Afro-Caribbeans may be an early risk marker of mortality in that group.
We conclude that the milder forms of hypertensive retinopathy are more commonly found in Afro-Caribbeans than in Europeans and that this is most marked in normotensive women. Age is a powerful correlate of these changes, and ambulatory BP makes an important contribution, even when resting BP is taken into account. These relationships are less strong in Afro-Caribbean subjects, suggesting that factors other than age and BP produce the high rates of hypertensive retinopathy seen in this population and that the classification of Keith et al12 for hypertensive retinopathy may be less valid in population groups that are not predominantly European. The reasons for the high rates of hypertensive retinopathy changes in Afro-Caribbeans are at least partially due to the higher BP values of this group, but other factors that may result in the high rates of hypertensive retinopathy remain to be established.
This study was supported by a Medical Research Council programme grant (London, UK). We are grateful to the Jeffrey Kelson Diabetic Centre, Central Middlesex Hospital, for accommodation; general practices at Chalkhill, Craven Park, Brentfield, Law, St Raphaels Way, and Talbot Walk; the Family Health Services Authority (FHSA), Community Relations Council (CRC), Community Health Council (CHC), and other community organizations for advice and publicity; Angela Edwards for administrative support; and all members of the project team.
- Received August 19, 1994.
- Revision received October 4, 1994.
- Accepted January 20, 1995.
Office of Population Censuses and Survey. Mortality and Geography: A Review in the Mid-1980s. London, UK: HMSO; 1990. The Registrar-General’s decennial supplement for England and Wales, series DS No. 9.
National Center for Health Statistics. Health, United States 1988. Washington, DC: US Government Printing Office; 1989. US Dept of Health and Human Services publication PHS 89-1232.
Miall WE, Cochrane AL. The distribution of arterial pressure in Wales and Jamaica. Pathol Microbiol. 1961;24:690-697.
McDonough JR, Garrison GE, Hames CG. Blood pressure and hypertensive disease among Negroes and whites. Ann Intern Med. 1964;61(suppl 6):208-228.
Chaturvedi N, McKeigue PM, Marmot MG. Resting and ambulatory blood pressure differences in Afro-Caribbeans and Europeans. Hypertension. 1993;22:90-96.
World Health Organization Expert Committee on Arterial Hypertension. Arterial hypertension. WHO Tech Rep Ser. 1978;628:7-23.
Keith NM, Wagener HP, Barker NW. Some different types of essential hypertension: their cause and prognosis. Am J Med Sci. 1938;197:332-343.