Five-year incidence of any late age-related maculopathy lesions by age and baseline smoking status.
Mitchell P, Wang JJ, Smith W, Leeder SR. Smoking and the 5-Year Incidence of Age-Related MaculopathyThe Blue Mountains Eye Study. Arch Ophthalmol. 2002;120(10):1357–1363. doi:10.1001/archopht.120.10.1357
To assess the relationship between baseline smoking and the 5-year incidence of late and early age-related maculopathy (ARM) in an older population cohort.
The Blue Mountains Eye Study examined 3654 participants aged 49 years or older during 1992 to 1994 and then 2335 survivors (75.1%) after 5 years. Retinal photographs were graded using the Wisconsin Age-Related Maculopathy Grading System. Those with any ARM lesions at either examination were regraded in detail using a side-by-side method similar to that developed for the Beaver Dam Eye Study. We also used similar definitions for incident ARM lesions. Smoking status was recorded at interview.
Age-standardized incidence rates for any late ARM lesions were 3.1%, 1.2%, and 1.4%, respectively, among baseline current, past, or never smokers. Corresponding age-standardized incidence rates for early ARM were 10.6%, 8.2%, and 9.3%, respectively. The mean age for cases with incident late ARM was 67 years for baseline current smokers, 73 years for past smokers, and 77 years for those who had never smoked (P = .02). After adjusting for age, current smokers, compared with never smokers, had an increased risk of incident geographic atrophy (age-adjusted relative risk [RR], 3.6; 95% confidence interval [CI], 1.1-11.3) and any late ARM lesions (RR, 2.5; 95% CI, 1.0-6.2). Current smokers had an increased risk of incident retinal pigmentary abnormalities (RR, 1.7; 95% CI, 1.1-2.7), with the risk higher in men (RR, 2.8; 95% CI, 1.4-5.6).
Five-year incidence of early ARM, late ARM, and ARM lesions.
In this cohort, persons who were current smokers had an increased risk of 5-year incident late ARM lesions and retinal pigmentary abnormalities. Current smokers developed late ARM at a significantly earlier age than never or past smokers.
DURING THE past decade, findings from multiple population-based studies1- 5 have demonstrated strong cross-sectional associations between smoking and age-related maculopathy (ARM). Consistently, the strongest associations have been found with current smoking. These studies have found either a weak association or no association with past smoking. Recently, a study6 that pooled data across populations from the United States, the Netherlands, and Australia provided strong and consistent cross-sectional evidence of a relationship between smoking and late ARM, also termed age-related macular degeneration (AMD) by the International AMD Epidemiological Study Group.7 This report indicated a 3-fold increased prevalence of AMD among current smokers.
To date, limited longitudinal data have been reported in support of these prevalence findings. Data from both the Physicians Health Study8 and the Nurses9 Health Study demonstrated a significantly higher incidence of AMD and visual loss in current smokers, with a similar magnitude of increased risk found for the 2 studies. The Beaver Dam Eye Study10 is the only population-based study yet to report the relationship between smoking and incident ARM. This study had insufficient power to assess the relationship with AMD but found a statistically significant higher risk of incident early ARM lesions, including large drusen. The Beaver Dam Eye Study report also presented some evidence of a dose-response relationship (pack-years of smoking) for early ARM. The purpose of the present report is to assess the relationship between smoking status at the baseline examination and the 5-year incidence of ARM lesions and constituents in the Blue Mountains Eye Study (BMES) cohort.
The BMES is a population-based survey of vision and common eye diseases in an urban population aged 49 years or older who were residents of 2 postal codes of the Blue Mountains region, west of Sydney, Australia. The survey methods and procedures have been previously described.11,12 The study was approved by the Western Sydney Area Health Service Human Ethics Committee, and written, informed consent was obtained from all participants. A detailed questionnaire was administered, and participants underwent a comprehensive eye examination after pupil dilation.
Baseline smoking status was assessed during a face-to-face interview. Participants were asked whether they had ever smoked and, if so, at what age they started smoking. If they had quit, participants were asked the age they had stopped smoking and the duration of this cessation. Participants were also asked what form their tobacco intake took (cigarettes, hand rolled, cigars, or pipe) and the quantity per day usually smoked. Pack-years were calculated from the total time a person had smoked, multiplied by the usual daily cigarette-equivalent intake, divided by 20.4 Current smokers were defined as participants who reported that they currently smoked or had stopped smoking less than 1 year before the examination.
At both examinations, 30° stereoscopic retinal photographs of the macula and other retinal fields of both eyes were taken, as described previously,11 using a Zeiss FF3 fundus camera (Carl Zeiss, Oberkochen, Germany). Photographs of both eyes were obtained in 98% and of at least one eye in 99% of those who participated at both the baseline and follow-up examinations. Details of the photographic grading of ARM lesions performed in the BMES were previously reported.11 Grading closely followed the Wisconsin Age-Related Maculopathy Grading System methods13 and was followed by a side-by-side grading of the baseline and 5-year photographs, when lesions were identified at either examination. Assessments of both intergrader and intragrader reliability showed good agreement for identifying ARM lesions.11
All cases were adjudicated by the senior author (P.M.). Late ARM was defined to include the 2 late lesions, geographic atrophy (GA) involving the fovea11 and neovascular ARM. The international ARM classification7 does not require GA lesions to involve the foveal center. For compatibility, we further defined any late ARM lesion as the presence of any GA or neovascular ARM lesions. Early ARM was defined as absent late-stage ARM lesions and presence of either (1) large(>125-µm diameter), indistinct soft or reticular drusen or (2) both large, distinct soft drusen and retinal pigmentary abnormalities (hyperpigmentation or hypopigmentation)7 within the area of the superimposed grading grid.13 Definitions for ARM lesion incidence closely followed those developed by Klein et al14 for the Beaver Dam Eye Study, modified to use fewer categories for drusen size.
Person-specific 5-year incidence rates were calculated. Incident late ARM was defined as the appearance at follow-up of neovascular ARM or GA involving the foveal center in either eye of persons in whom no late ARM lesion was present at baseline. An incident late ARM lesion was defined as the appearance at follow-up of neovascular ARM or GA (whether or not the foveal center was involved) in either eye of persons in whom no late ARM lesion was present at baseline. Incident early ARM was defined by the appearance at follow-up of either indistinct soft or reticular drusen or the copresence of distinct soft drusen and retinal pigmentary abnormalities in either eye of persons in whom no late or early ARM was present at baseline. Incident early lesions, such as retinal pigmentary abnormalities and indistinct or distinct soft drusen, were defined by the appearance at follow-up of these lesions in either eye of persons in whom no late ARM lesion, early ARM, or corresponding lesions were present at baseline. Cases with only distinct soft drusen or retinal pigmentary abnormalities at baseline, not initially classified as early ARM, who then developed complementary lesions that together comprised the criteria for early ARM are included as incident early ARM cases in this report.
We used SAS statistical software (SAS Institute Inc, Cary, NC) for all analyses, including t test and χ2 statistics. The ARM incidence rates by baseline smoking status were directly age standardized to the 5-year follow-up study population (BMES II) using baseline age. Age-adjusted relative risk (RR) calculations used the Mantel-Haenszel method across 10-year age group strata. Age- and sex-adjusted odds ratios (ORs) were estimated using logistic regression models. Interactions between age and smoking status and between sex and smoking status have been checked in all logistic regression models, and no significant interaction was found. The RRs, ORs, and 95% confidence intervals (CIs) are shown.
Baseline eye examinations of 3654 residents (BMES I) were conducted during 1992 to 1994, with an overall participation rate of 82.4%. All members of the cohort were invited to attend the 5-year examinations during 1997 to 1999. There were 543 participants who died before BMES II commenced. Of the 3111 survivors, 2335 (75.1%) were reexamined, whereas 383 (12.3%) had moved from the study area and 393 (12.6%) refused to participate. The mean follow-up was 5.1 years (median, 4.9 years; range, 3.0-7.8 years).
The BMES I participants who had moved from the area or who refused to participate in BMES II were slightly older at baseline (9.3% vs 5.1% for the age group ≥80 years), more likely to have reported difficulty in walking(7.9% vs 3.2%), more likely to rate their health as poor (4.3% vs 2.0%), and more likely to have reported being current smokers (18.2% vs 13.0%). They were less likely to own their home (81.7% vs 89.6%) or to have achieved qualifications after leaving school (44% vs 50%) (Table 1).
Among participants reexamined after 5 years, 303 (13.0%) were current smokers and 802 (34.4%) were past smokers. During the 5 years, 26 participants(1.1%) developed late ARM, including 5 who developed neovascular ARM in one eye and GA in the fellow eye. Age-specific incidence rates for late ARM in persons younger than 60 years, 60 to 69 years old, 70 to 79 years old, and 80 years or older were 0.14% (95% CI, 0.004%-0.80%), 0.64% (95% CI, 0.24%-1.40%), 2.4% (95% CI, 1.3%-4.1%), and 5.4% (95% CI, 2.0%-11.4%), respectively. For the slightly broader definition, age-specific incidence rates for any late ARM lesions in persons younger than 60 years, 60 to 69 years old, 70 to 79 years old, and 80 years or older were 0.14% (95% CI, 0.004%-0.80%), 0.74%(95% CI, 0.3%-1.5%), 3.5% (95% CI, 2.1%-5.4%), and 6.3% (95% CI, 2.695%-12.600%), respectively. Corresponding age-specific incidence rates for early ARM in these age groups were 3.1% (95% CI, 1.9%-4.6%), 7.6% (95% CI, 6.0%-9.5%), 18.3% (95% CI, 15.0%-22.0%), and 14.8% (95% CI, 8.1%-23.9%), respectively.
The mean age for cases with incident late ARM (or any late lesions) was 67 (70) years for baseline current smokers, 73 (73) years for past smokers, and 77 (76) years for those who had never smoked (late ARM: P = .02; any late lesions: P = .06). Table 2 and Figure 1 show the incidence rates for any late ARM lesions by baseline age and smoking status and suggest that current smokers were more likely to develop late ARM lesions in their 60s to 70s, approximately 10 years earlier than the typical age of onset for this disease. Age-standardized incidence rates for any late ARM lesions were 3.1% for those who were current smokers at baseline, 1.2% for past smokers, and 1.4% for persons who had never smoked.
Twenty-three participants (1.0%) developed incident neovascular ARM in at least one eye and 17 (0.7%) developed incident GA, including 8 cases with foveal involvement and 9 with new GA lesions outside the foveal center, not yet classified as incident late ARM. Age-standardized incidence rates for late ARM or any late ARM lesions and age- and sex-adjusted RRs for ARM lesions among smokers at baseline compared with those who had never smoked are shown in Table 3. After adjusting for age, the risk of developing GA for current smokers was approximately 3.5 times the risk for persons who reported having never smoked. The risk of developing any late ARM lesions for current smokers was 2.5 times the risk for persons who reported having never smoked.
The mean age for incident early ARM was 64 years for current smokers at baseline and 70 years for past smokers and for those who had never smoked(P = .002). Table 2 shows early ARM incidence rates by baseline age and smoking status. These data similarly suggest that current smokers were more likely to develop early ARM in their 50s and 60s than at older ages. Table 3 shows age-standardized incidence rates for early ARM or individual early ARM lesions together with age- and sex-adjusted RRs for current smokers compared with never smokers. Current smokers also had more than a 50% increased risk of developing retinal pigmentary changes compared with never smokers.
Associations between baseline smoking status and incident ARM were examined among men and women separately (Table 4). All associations, particularly those for late ARM lesions and retinal pigmentary abnormalities, appeared to be evident in men but not in women. To further explore whether the sex difference found was due to the difference in smoking exposure between men and women, we examined the difference in pack-years of smoking between men and women among past and current smokers. Among ex-smokers, there was a difference in pack-years of smoking between men (mean, 35 pack-years) and women (mean, 21 pack-years) (P = .001). Among current smokers, a smaller difference was found between men (mean, 38 pack-years) and women (mean, 31 pack-years) (P = .04).
No significant association was found between pack-years of smoking and the incidence of either early or late ARM (data not shown), except for incident hypopigmentation in both men and women (age- and sex-adjusted OR, 1.008; 95% CI, 1.001-1.014) or in men (age-adjusted OR, 1.01; 95% CI, 1.002-1.017).
In this study, current smokers were more likely to develop various ARM lesions than past smokers and never smokers. Not only were current smokers at greater risk of ARM, but they were also more likely to develop ARM at a substantially earlier age than never or past smokers. These findings are consistent with previous reports1,4,6,10,15 examining associations between smoking and ARM and strongly support the hypothesis that current smoking is the major environmental risk factor for ARM. Pooled data across 3 large, population-based, cross-sectional studies demonstrated a strong and consistent association between smoking status and prevalent ARM.6
Using the strong cross-sectional association found between late ARM and current smoking from the BMES and population-attributable risk calculations, we previously estimated that smoking may cause or contribute to up to 20% of blindness among Australians aged 50 years or older.15
This is only the second population-based prospective study to assess the association between smoking status and incident ARM. The Beaver Dam Eye Study10 reported similar findings from the United States. Although both the Beaver Dam Eye Study and our study have only borderline statistical power, particularly for incident late ARM, the findings from both studies are statistically significant and are of the same order of magnitude. Both studies are comparable in their outcome measures and definitions and similar in their ascertainment of exposure. The outcome measures used in both studies have been extensively tested and found to have excellent reliability and validity11 and were subsequently cross-graded to ensure negligible misclassification for diagnosis of AMD.6
An important limitation in our study is the loss to follow-up of almost 25% of survivors. Furthermore, those lost to follow-up were more likely to be older and to have been current smokers at the baseline examination compared with those who returned for the 5-year follow-up examinations. It is possible that older smokers who did not participate at follow-up may have had fewer vision problems and considered the study not relevant. This possibility, however, does not seem likely, because most of those who were detected to have early ARM at follow-up had no visual symptoms attributable to ARM, so it seems unlikely that the participants could have differentiated themselves by early ARM status. In addition, our study revealed that smokers were much more likely to develop ARM at an earlier age. This suggests that we would be less likely to detect an equally strong association at older ages if a real association existed that would be biased toward the null by nonresponse.
The overall effect of the losses to follow-up is much more likely to be a bias toward a lower estimate of association between smoking and incident ARM, because we have lost both those more likely to develop ARM and smokers, decreasing our statistical power to detect an association. Our findings are therefore more likely to represent an underestimation of the relationship.
Our findings suggest that the ARM risk from smoking may be stronger in men than in women. The numbers of current male and female smokers at risk of late or early ARM were similar (Table 4). The differences in risk magnitude between male and female current smokers were not completely explained by differences in smoking exposures, suggesting that there could be a real difference between men and women in terms of the effect of smoking on ARM. If the development of ARM secondary to smoking exposure is mediated through vascular factors, then it could be considered in a similar manner to cardiovascular disease. Because men are at greater risk of cardiovascular disease than women, then by analogy we might expect men to have a larger increase in their ARM risk secondary to smoking than women, if ARM was a manifestation of cardiovascular disease. Although some evidence implicates a vascular pathogenesis for ARM,16 available data are inconsistent.17,18
Although current smokers had a significantly greater risk of ARM, no statistically significant association was found between past smoking and incidence of ARM or between pack-years of smoking and ARM. This finding matches most previous population-based reports of the relationship between smoking and prevalent ARM, in which the major association from all studies was with current smoking. The association between past smoking history and incident ARM is likely to be biased toward the null, because our population represents an elderly survivor cohort that excluded subjects with prevalent ARM at baseline. Despite this potential bias, a dose-response relationship between smoking and ARM seems unlikely. This has important implications for both the pathogenesis of ARM and in regard to clinical advice provided to older patients. It is also in keeping with a finding reported by the Macular Photocoagulation Study that people who continued to smoke had a substantially higher risk of recurrent choroidal neovascularization in the first year after successful photocoagulation treatment compared with nonsmokers (RR, 1.8; P =.02).19
Although consistent with a possible cardiovascular pathogenesis, our findings suggest that a more short-term smoking effect on macular tissue may be operating in ARM than is the case in coronary heart disease. Our results(and the results of the Macular Photocoagulation Study) support the possibility that the risk of incident late ARM may decrease rapidly after cessation of smoking.
This information has considerable public health importance not just for ophthalmology but also for the likely success of health promotion campaigns that attempt to prevent the uptake or increase the cessation of smoking. It is highly applicable to the development of strategies to encourage older people to stop smoking. This could be particularly important if further data support our contention that the risk of incident ARM decreases quickly after smoking cessation. Older smokers and their medical advisers often struggle to find compelling reasons why they should give up smoking in their advanced years. The possibility of sudden death from myocardial infarction or stroke may be less persuasive to elderly smokers than the substantially increased likelihood of blindness and loss of independence. Antismoking campaigns often suffer from self-exemption among smokers20: although people may understand the risks and may even believe the warnings, they consider that the risk burden is for others and that they personally are exempt. Such behavior may be accentuated among those who have survived to old age unscathed by their smoking.
Finally, the magnitude of the potential benefit to future vision from smoking cessation is likely to be large compared with the 25% reduction in risk of visual impairment achieved by at-risk persons taking supplements that contain high-dose antioxidants and zinc, recently reported by the Age-Related Eye Disease Study.21
Our findings provide further support for the notion that smoking may be the strongest risk factor for ARM, other than age6 and genetic factors.22 In conclusion, our study has demonstrated a significantly higher risk of incident ARM among current smokers than in past or never smokers. An important finding was that smokers were more likely to develop ARM at a substantially younger age. The increased risk of incident ARM in this study was somewhat higher among smoking men than among smoking women.
Submitted for publication November 26, 2001; final revision received June 10, 2002; accepted June 25, 2002.
This study was presented in part at the 2000 Annual Meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Fla, May 1, 2000.
This study was supported by grant 974159 from the Australian National Health and Medical Research Council, Canberra, Australia.
Corresponding author and reprints: Paul Mitchell, MD, PhD, Centre for Vision Research, Department of Ophthalmology, University of Sydney, Hawkesbury Road, Westmead, New South Wales, Australia 2145 (e-mail: firstname.lastname@example.org).