We found a prevalence of late AMD of 1.2% (95% CI, 0.9–1.6) in people aged 60 and older in a population-based study in North and South India. Age-specific prevalence was similar in both centers. The large size of our study ensured narrow 95% CIs around the prevalence rates. The upper range of our CI (1.6) was lower than the estimate of 3% ± 0.4% assumed for our sample size calculations.
A limitation of our results was the proportion of ungradable images, especially in the oldest age groups where AMD prevalence is likely to be the highest. If it is to have a sizeable impact on the overall prevalence estimates, the prevalence of late AMD in people with ungradable images would have to be considerably higher than that found in people with gradable images. For example, we calculated that the age-specific prevalence of late AMD in those with ungradable images would have to be six times higher than in those with gradable ones to achieve an overall prevalence of 2.9%. In the 260 people originally ungradable who had postoperative gradable fundus images, there was only one case (0.4%) of late AMD; 66% had no AMD; and 29% had grade 1, 5% grade 2, and 0.4% (n = 1) grade 3. The results suggest a slightly lower prevalence of early and late AMD than that in the group with gradable images. However, the number of people with postoperative images is too small for any conclusions to be drawn. The modeled estimates including those with ungradable images increased the overall prevalence only slightly (from 1.2 to 1.3), but had a larger effect on the prevalence in the oldest age group (80+), from 2.5 to 3.5.
Response bias may also have led to a lower estimate of AMD in our sample. The oldest age group (80+) was less likely to attend the clinical examination (390/636 enumerated, 61% response) than the youngest age group (60 to 64 years, 2144/2618 enumerated, 82% response) and less likely to have gradable fundus images. Of the enumerated population aged 80 and older, gradable fundus images were available for only 31%. Response bias is suggested by the decline in the age-specific prevalence from 3.2% at 75 to 79 years (95% CI, 1.2–5.3) to 2.5% (95% CI, 0.4–4.7) at 80+ years, whereas, based on studies in Western populations, a substantial increase in the rate in the 80 and older group might be expected. In a study in Western populations in which results were pooled, the prevalence rate in the women rose from 3.4% at 75 to 79 years to 16.4% at 80+ years. Comparable figures for the men were 3.97 and 11.9, respectively.7
Very similar statistics were reported in the EUREYE study, in which the prevalence of AMD increased from 3.6% at ages 75 to 79 to 12.2% at age 80+.5
Unfortunately, we have no information to ascertain the number of missed cases of AMD in the 80+ group. At other age groups younger than 80 years, our results are comparable to those for Western studies.5,7
In the modeled estimates including all nonattendees and those with ungradable images, AMD prevalence showed a linear increase in age, with the highest prevalence of 4.0% estimated for the 80+ age group. The results from modeling missing data on AMD prevalence suggest that our observed prevalence in the 80+ group was underestimated but it is unlikely that in our study, there is a substantial increase in prevalence in the 80+ group as observed in Western populations.
The prevalence of late AMD was lower in the present study than we found in our feasibility study ().22
The feasibility study was conducted only in North India and included people aged 50 years and older with a smaller number of individuals in the older age groups (n
= 817 aged 60 and older). The differences may be due to sampling error, chance, or unknown biases. Two other population-based studies in India have reported results for late AMD. Both studies recruited participants in the age range 40 years and older. In the Aravind Comprehensive Eye Study (ACES),20
the prevalence of late AMD (ascertained by clinical grading at the slit lamp) was 1.4% in those 60 to 69 years of age and 2.0% in those 70+ years. In the Andhra Pradesh Eye Study (APES),21
the prevalence of late AMD (fundus photography) was 3.4% at 60 to 69 years and 3.7% at 70+ years. No information was provided on AMD prevalence in the oldest age groups (i.e., 75–79 and 80+) in ACES and APES, but the number of people 70+ (around 350 in each study) was small. The number in the oldest age groups (80+) was not reported, but was likely to be very small. Information on response rates by age was also not available from the two studies. ACES reported that those without retinal data (mainly due to media opacities) were older than those with data, but no information on ungradable images was reported for the APES, although 39% were reported to have cataract.
Prevalence of Late AMD in Studies in India
In common with studies in Western populations, we found that CNV was the more frequent type of late AMD (83%). This finding was in contrast to the result from the APES and ACES studies, which reported that a higher proportion of AMD was GA (95% in APES and 79% in ACES). Both APES and ACES took place in South India. In our study, proportionately more CNV than GA was reported for both the South India center (24 CNV of 30 late AMD) and the North India center (20 CNV of 23 late AMD). We categorized 10 cases in which GA was present on grading as CNV since, where GA and CNV are coexistent, we consider that the GA is secondary to CNV. In our feasibility study we also found that most late AMD was CNV (14/15 cases). The ACES reported cases of GA but not CNV at young ages (40–49 and 50–59) which is very unusual compared with Western countries, where late AMD is usually not seen below the age of 60. Grading was performed at the slit lamp, a technique that is more prone to error than grading from fundus images, which allow for careful review. Fundus images were used in the APES, but the prevalence by age of GA and CNV was not reported separately. In APES, the prevalence of late AMD was also unusually high at younger ages and was most likely to be due to GA, as most cases in the study (67/71) were GA. It is possible that the higher GA observed in other studies in APES and ACES were due to misclassification of early AMD as GA. It is also possible that we missed some cases of GA, because the darker pigmented retinas of our Indian participants may make it more difficult to see GA.
Comparison of early AMD between studies is more problematic because of variations in methods and reporting. The INDEYE study and the INDEYE feasibility study22
used identical methods of grading as the EUREYE study facilitating comparison between the studies. Early AMD grade 1 (small soft distinct drusen only or pigmentary irregularities only) was observed in 36.4% (95% CI, 32.7–40.3) of the EUREYE participants 65 years of age and older and in 40.2% (95% CI, 37.5–42.7) in the comparable age range of INDEYE participants. Early AMD grade 2 (large soft indistinct (≥125 μm) or reticular drusen only or soft distinct drusen with pigmentary irregularities) was observed in 10.1% (95% CI, 8.9–11.4) in EUREYE, with similar findings at the South India center (8.9%; 95% CI, 7.6–10.1), and slightly lower results in North India (6.3%; 95% CI, 4.3–8.3). For early AMD grade 3 (soft indistinct or reticular drusen with pigmentary irregularities) the results were different between EUREYE and INDEYE, although in both studies, the prevalence was low, 2.4% in EUREYE (95% CI, 1.8–3.1) and 0.2 (95% CI, 0–0.5) in South India, with only one person with this grading in North India. In our feasibility study in North India, we also found a low prevalence (1.1% in the 60 to 69 age group and 0.8% in the 70+ age group). Moreover, whereas in EUREYE there was a positive increase in AMD3 with age, the trend with age was inverse in INDEYE and in the feasibility study. In the ACES the prevalence of large drusen with pigmentary irregularities was higher, ranging from 1.3% at 40 to 49 years to 4.9% at 70+ years.
Our results show that the prevalence of late AMD is comparable to that in Western populations, at least in the 60 to 79 age group. The high proportion of ungradable images, along with the lower response rate in the oldest age group (80+), make it likely that the prevalence in this group was underestimated.