Tomany SC, Cruickshanks KJ, Klein R, Klein BEK, Knudtson MD. Sunlight and the 10-Year Incidence of Age-Related MaculopathyThe Beaver Dam Eye Study. Arch Ophthalmol. 2004;122(5):750-757. doi:10.1001/archopht.122.5.750
To examine the association of sunlight exposure and indicators of sunsensitivity with the 10-year incidence of age-related maculopathy (ARM).
Population-based cohort study.
We included persons aged 43 to 86 years at the baseline examinationfrom 1988 to 1990, living in Beaver Dam, Wis, of whom 3684 persons underwent5-year follow-up and 2764 underwent 10-year follow-up.
Data on sun exposure and indicators of sun sensitivity were obtainedfrom a standardized questionnaire administered at baseline and/or follow-up.We determined ARM status by grading stereoscopic color fundus photographsusing the Wisconsin Age-Related Maculopathy Grading System.
Main Outcome Measures
Incidence and progression of ARM.
While controlling for age and sex, we found that participants exposedto the summer sun for more than 5 hours a day during their teens, in their30s, and at the baseline examination were at a higher risk of developing increasedretinal pigment (risk ratio [RR], 3.17; 95% confidence interval [CI], 1.24-8.11; P = .01) and early ARM (RR, 2.14; 95% CI, 0.99-4.61; P = .05) by 10 years than those exposed less than 2 hoursper day during the same periods. In participants reporting the highest summersun exposure levels in their teens and 30s, the use of hats and sunglassesat least half the time during the same periods was associated with a decreasedrisk of developing soft indistinct drusen (RR, 0.55; 95% CI, 0.33-0.90; P = .02) and retinal pigment epithelial depigmentation(RR, 0.51; 95% CI, 0.29-0.91; P = .02). Participantswho experienced more than 10 severe sunburns during their youth were morelikely than those who experienced 1 or no burn to develop drusen with a 250-µmdiameter or larger (RR, 2.52; 95% CI, 1.29-1.71; P =.01) by the 10-year examination. No relationships were found between UV-Bexposure, winter leisure time spent outdoors, skin sun sensitivity, or numberof bad sunburns experienced by the time of the baseline examination and the10-year incidence and progression of ARM or its associated lesions.
Few significant relationships between environmental exposure to lightand the 10-year incidence and progression of ARM were found in the BeaverDam Eye Study. Consistent with results from the baseline and 5-year follow-upexaminations, significant associations were found between extended exposureto the summer sun and the 10-year incidence of early ARM and increased retinalpigment. A protective effect of hat and sunglasses use by participants whilein their teens and 30s against the 10-year incidence of soft indistinct drusenand retinal pigment epithelial depigmentation was also found, but only inthose who reported the highest amount of sun exposure during the same periods.
Age-related maculopathy (ARM) is the leading cause of vision loss inolder Americans.1- 6 Thepathogenesis of the disease is poorly understood,6,7 butit is thought to be multifactorial, involving both genetic8- 10 andenvironmental factors.11- 14 Withthe exception of antioxidant vitamins and zinc, few sight-saving therapiesare available to treat patients with this disease.15 Forthis reason, it is important to identify primary prevention strategies. Ithas been suggested that exposure to sunlight may play a role in the etiologyof ARM. Previous studies examining the relationship between sunlight exposureand ARM have produced inconsistent results.16- 22 Atthe time of the 5-year follow-up examination in the Beaver Dam Eye Study,19 persons who reported spending most of their summerleisure time outdoors during their teens and 30s were found to be at increasedrisk for incident early ARM (odds ratio [OR], 2.09; 95% confidence interval[CI], 1.19-3.65). This article examines the association of sunlight exposureand measures of sun sensitivity with the 10-year incidence of early ARM inthe Beaver Dam, Wis, population and further examines the relationships betweenthese sun exposure variables and additional ARM incidence and progressionoutcomes.
Methods to identify and describe the population have appeared previously.23- 31 Aprivate census of the population of Beaver Dam was performed from September15, 1987, to May 4, 1988, to identify all residents in the city or townshipaged 43 to 84 years. Of the 5924 eligible individuals, 4926 participated inthe baseline examination between March 1, 1988, and September 14, 1990.23 Of the population, 99% was white. Of those surviving,3684 (81.1%) participated in the 5-year follow-up examination between March1, 1993, and June 14, 1995. Comparisons between participants and nonparticipantsat the time of the baseline and 5-year follow-up examinations have appearedelsewhere.26 Additional analysis of sun exposure–relatedvariables showed that after adjusting for age and sex, participants were morelikely than nonparticipants to wear protective hats and sunglasses at baseline.
Prior to the 10-year follow-up examination on March 1, 1998, 350 participants(9.5%) from the 5-year examination died. Of the 3334 surviving participantsfrom the baseline and second examinations, 2764 (82.9%) participated in thesecond follow-up examination between March 1, 1998, and June 9, 2000. Of thenonparticipants, 1 could not be located, 42 (1.3%) permitted an interviewonly, 38 (1.1%) had moved out of the area and did not participate, 337 (10.1%)refused to participate, and 152 (4.6%) died after the start of the 10-yearfollow-up period prior to being examined. The mean time between the baselineand 10-year follow-up examinations was 10.1 ± 0.4 years, and the medianwas 10.0 years. Comparisons between participants and nonparticipants at the10-year follow-up examinations have been presented elsewhere.27 Additionalanalyses found participants less likely than nonparticipants to work outdoorsand spend time outside during the summer while in their 30s.
Similar procedures were used at both the baseline and follow-up examinationsand are described in detail elsewhere.24,25 Informedconsent was obtained from each participant. A standardized questionnaire wasadministered. Participants were asked about their residential history, timespent outdoors for leisure activities and work, and use of brimmed hats andsunglasses. At the 5-year follow-up examination, participants were asked aboutsun-related behaviors while in their teens (aged 13-19 years) and 30s (aged30-39 years).
An index of ambient UV-B exposure was constructed from the baselineresidential history, weighting the time spent outside Wisconsin by the ratioof the total ambient UV-B light present in that area to the level for 1 yearin Wisconsin (Wisconsin sun-years).18 The meanambient UV-B sun exposure was calculated by dividing the cumulative ambientsun exposure by age. Most participants had spent most of their lives in Wisconsin,resulting in a highly skewed distribution of mean annual exposure, so thisvariable was used to categorize participants into 2 levels of exposure: thosewith an index of 1.01 or lower and those with an index higher than 1.01. Betweenbaseline and follow-up, some people had migrated out of Beaver Dam to otherregions with higher UV-B levels. Therefore, an additional index was createdwith the same weighting procedure used for the historical data to captureadditional exposure during follow-up in areas with higher UV-B levels thanWisconsin.19
The amounts of time that participants reported using hats and sunglassesat the baseline examination were combined in a weighted fashion into levelsof increasing protection from UV-B light (none/low, moderate, and high).18 A high protection level was defined as wearing hats,sunglasses, or both at least half of the time spent outdoors. An index ofhat and sunglasses use while aged 13 to 19 years and 30 to 39 years was created.19 Participants who reported wearing hats or sunglassesat least half of the time when outside in the summer during both periods wereconsidered to have used them often and to have a high level of protection.People who reported rarely using sunglasses or hats during both periods wereconsidered to have a low level of protection. Participants who reported variouspatterns of use were considered to have intermediate levels of protection.
The amount of time participants spent outdoors in the summer at baselinewas reported as more than 75%, 25% to 75%, or less than 25% of the day. Theamount of time spent outdoors in the teens and 30s was obtained at the 5-yearfollow-up examination and reported as more than 5 hours, 2 to 5 hours, orless than 2 hours per day. These data were used to create a sun exposure index.Participants who reported being outdoors in the summer more than 5 hours aday in both their teens and 30s were considered to have high exposure; thosewho reported less than 2 hours of exposure during both periods, low exposure;and those with 2 to 5 hours per day or exposure that varied by age, intermediateexposure.19
A measure of cumulative sun exposure was similarly created by combiningbaseline summer sun exposure with that in the teens and 30s. Participantswho reported the highest level of exposure during all periods were categorizedas high exposure, those who reported the lowest level of ISD exposure duringall periods were categorized as low exposure, and all others were classifiedas moderate exposure. For those with moderate exposure, a variable indicatingthe period in which the participant spent the most time outdoors was created;in cases of ties, the earliest period with the highest value was selected.
The number of bad sunburns, or burns with pain lasting longer than 1day, experienced by a participant during youth (<21 years) and prior tothe baseline examination was ascertained through the questionnaire and classifiedas none, 1, 2 to 10, or more than 10 burns. Skin sun sensitivity was characterizedin response to a question asking participants how their skin reacted whenexposed to the sun for half an hour without sunscreen and was recorded as"skin burns but never tans," "skin burns, then tans," "skin never burns buttans," or "skin never burns or tans." A person was classified as a nonsmokerif he or she had smoked fewer than 100 cigarettes in his or her lifetime,as an ex-smoker if he or she had smoked more than this number of cigarettesin his or her lifetime but had stopped smoking before the baseline examination,and as a current smoker if he or she had not stopped smoking. A subject wasclassified as a current vitamin user if he or she had taken at least 1 vitaminper week in the month prior to the baseline examination, as a past vitaminuser if he or she had ever regularly taken vitamins at least once a week inthe past but not in the last month, and as never using vitamins if he or shehad never regularly taken vitamins at least once a week. Age was defined asthe age at the time of the baseline examination.
Stereoscopic 30° color fundus photographs centered on the disc (DiabeticRetinopathy Study28 standard field 1) and macula(standard field 2) and a nonstereoscopic color fundus photograph temporalto but including the fovea of each eye were taken. For purposes of this article,the 2663 people (3570 at the first 2 examinations) with at least 1 eye freeof confounding lesions (eg, retinal detachment or non–age-related chorioretinalscarring) at all 3 examinations (right eye: n = 2592 [3475 at the first 2examinations]; left eye: n = 2600 [3501 at the first 2 examinations]; botheyes: n = 2529 [3406 at the first 2 examinations]) are included in the analyses.
The Wisconsin Age-Related Maculopathy Grading System28 wasused to assess the presence and severity of lesions associated with ARM. Gradingprocedures, lesion descriptions, and detailed definitions for the presenceand severity as well as 10-year incidence of specific lesions have appearedelsewhere.26,31 Incidence impliesthe appearance of a lesion at follow-up when it was absent in any of the subfieldsthat could be graded at the baseline and follow-up examinations. Progressionimplies either the incidence of disease or the presence of a lesion at baselinewith a worsening at follow-up.26
Incidence was determined for maximum drusen size and type, increasedretinal pigment, retinal pigment epithelial (RPE) depigmentation, pigmentaryabnormalities (defined as RPE depigmentation or increased retinal pigment),signs of exudative macular degeneration, and pure geographic atrophy. Forexample, if none of the subfields had soft indistinct drusen at baseline andsoft indistinct drusen were present in 1 or more subfields at the 5- or 10-yearfollow-up examination, the eye would be considered to have incident soft indistinctdrusen.
Early ARM was defined by the presence of soft indistinct drusen or anytype of drusen associated with RPE depigmentation or increased retinal pigment.Late ARM was defined by exudative macular degeneration or pure geographicatrophy.
Progression for a participant was defined as an increase in maculopathyseverity on a 6-level scale. The cutpoints used were based on previous observationsin the Beaver Dam Eye Study of an increased risk of developing late ARM associatedwith larger areas of large soft drusen and pigmentary abnormalities.31 An increase in severity in either eye by 2 or moresteps from level 1 through 3, or 1 or more steps from level 4 to 5, was definedas progression. A detailed description appears elsewhere.14
We examined the relationships between risk factors at baseline and theincidence of each maculopathy lesion, the incidence of 2 end points of diseaseseverity (early and late ARM), and the progression of ARM. We used SAS statisticalsoftware (SAS Institute Inc, Cary, NC) to analyze the data. The analyticalapproaches used allowed persons who were right censored (not seen after the5-year examination as a result of death or nonparticipation) to contributeinformation to the estimates. Cumulative events were estimated using the Kaplan-Meier(product-limit) survival approach.32 Multivariaterisk ratios (RRs) and 95% confidence intervals (CIs) were calculated fromthe discrete linear logistic model.33 Age-and sex-adjusted models were constructed by outcome for each of the potentialrisk factors. Final models were then created by outcome for each risk factorby further adjusting the models for vitamin use and smoking history. Whenmultivariate models included eye-specific risk factors (such as iris coloror history of cataract), data were analyzed using the generalized estimatingequation approach.34 This method adjusts forcorrelations between the two eyes.
Table 1 shows the distributionof sunlight-related risk factors. Few participants reported spending mostof their days outdoors, either during the winter or summer (3.0% and 19.8%,respectively), at the baseline examination, and most (65.6%) reported spendingan intermediate amount of time outdoors while in their teens and 30s. Abouthalf of the participants reported wearing hats and sunglasses rarely or notat all, both at baseline and in their teens and 30s. Less than half had experiencedmore than 1 bad sunburn in their lifetime, and most persons had skin thateither burned and then tanned or tanned and never burned. Women reported spendingmore time indoors and being more likely to wear sunglasses and hats than men.They also reported fewer bad sunburns and had less UV-B exposure. Older participantsreported spending less time outdoors in both the summer and winter than youngerparticipants (data not shown). Older participants were more likely to reportnever or rarely wearing sunglasses in their teens and 30s and at baselineand to have experienced fewer bad sunburns, both in youth and at baseline,than younger participants (data not shown). Participants who experienced thegreatest number of bad burns in their youth and at baseline were more likelyto report wearing protective hats and sunglasses often and spending less than25% of their time outdoors during the summer than persons who experiencedthe fewest burns (data not shown).
Table 2 gives overall incidenceestimates, without adjustment, for early and late ARM and the progressionof ARM. No significant relationships were found between the amount of timespent outdoors for work or winter or summer leisure activities, either atbaseline or in the teens and 30s, and the incidence of early or late ARM orthe progression of ARM. Participants who reported wearing sunglasses and hatsduring their teens and 30s had a significantly lower incidence of early ARMthan those who never or rarely wore hats and sunglasses. This relationship,although not significant, held for baseline use of sunglasses and hats aswell. Participants who reported frequent use of hats and sunglasses at baselinewere also less likely to experience progression of ARM. Participants who experienced2 or more burns had a lower incidence of early and late ARM than those whoexperienced none or 1 bad burn. Participants with 2 or more bad sunburns duringyouth were also less likely to have progression than their counterparts. Nosignificant relationships were found between mean annual ambient UV-B exposureand incidence or progression of ARM. Participants who reported that theirskin tanned but never burned when exposed to the sun were more likely thanany of the other skin sensitivity groups to experience progression of ARM.
The RRs for sun-related factors and the 10-year incidence and progressionof ARM are presented in Table 3.All models are age and sex adjusted. Participants with the highest levelsof sun exposure during all 3 periods as well as those with moderate levelsof sun exposure during all 3 periods were significantly more likely to developearly ARM (RR, 2.20 and 1.94, respectively) and increased retinal pigment(RR, 2.99 and 2.42, respectively) by the 10-year follow-up examination thanthose reporting the lowest levels of sun exposure during all 3 periods. Participantsexposed to the summer sun for more than 5 hours a day both during their teensand 30s had a higher risk of developing increased retinal pigment (RR, 2.53;95% CI, 1.22-5.26; P = .01) and early ARM (RR, 2.02;95% CI, 1.08-3.79; P = .03) at 10 years than thosewho were exposed less than 2 hours per day during the same periods. Only amarginal relationship between level of sun exposure at baseline and the 10-yearincidence of increased retinal pigment was found (RR, 1.39; 95% CI, 0.94-2.04; P = .10). The relationship between cumulative sun exposureand increased retinal pigment and early ARM was further examined by addingpreviously identified risk factors (smoking and vitamin use) for ARM. Therelationships between sun exposure during the teens and 30s and incidenceof early ARM and increased retinal pigment remained (RR, 2.01 and 2.51, respectively),as did the relationships between high and moderate sun exposure during all3 periods and incident increased retinal pigment (RR, 2.40 and 2.95, respectively).The relationship between high sun exposure during all 3 periods and incidentearly ARM, however, was reduced to moderate significance. Other factors forwhich sun exposure could be a surrogate, including use of photosynthesizingdrugs, cataract, cataract surgery, aphakia, and iris color, were added toall significant models. None of these risk factors explained the relationshippreviously found (data not shown). For participants reporting variable sunexposure during the 3 periods, the relationships between the period in whichthe participant spent the most time outdoors and the 10-year incidence ofearly ARM and increased retinal pigment were examined. No significant relationshipswere found (data not shown).
As indicated in Table 3,participants who experienced more than 10 sunburns during their youth weremore likely than those who experienced 1 or no burn to have drusen at least250 µm in diameter (RR, 2.52; 95% CI, 1.29-4.94; P = .01) and soft indistinct drusen (RR, 1.71; 95% CI, 0.97-2.99; P = .06) by the 10-year examination. Those who experienced2 to 10 burns in their youth were significantly more likely to develop lateARM (RR, 2.32; 95% CI, 1.01-5.31; P = .05) but significantlyless likely to develop soft indistinct drusen (RR, 0.72; 95% CI, 0.51-1.00; P = .05) than participants reporting 1 or no burn. Noneof the participants who experienced more than 10 burns during youth (n = 177)developed late ARM by 10 years. All relationships besides that between softindistinct drusen and number of bad sunburns in youth held after further adjustingthe models for smoking status and vitamin use (risk of incident large drusenin those experiencing more than 10 burns: 2.40; 95% CI, 1.22-4.71; P = .01; risk of incident late ARM in those experiencing 1 to 9 burns:2.31; 95% CI, 1.00-5.32; P = .05).
After adjusting for age and sex, the use of hats and sunglasses by participantsat baseline or in their teens and 30s was not found to protect against the10-year incidence or progression of ARM. However, in those who reported havingthe highest summer sun exposure levels in their teens and 30s, the use ofhats and sunglasses at least half the time was associated with a decreasedrisk of developing soft indistinct drusen (RR, 0.55; 95% CI, 0.33-0.90; P = .02) and RPE depigmentation (RR, 0.51; 95% CI, 0.29-0.91; P = .02), but no significant associations were found betweenincreased retinal pigment (RR, 1.36; 95% CI, 0.76-2.42; P = .30), incident early ARM (RR, 0.83; 95% CI, 0.48-1.43; P = .50), incident late ARM (RR, 0.73; 95% CI, 0.22-2.37; P = .60), and progression of ARM (RR, 0.93; 95% CI, 0.55-1.59; P = .80).
In the Beaver Dam Eye Study, extended sun exposure was associated withthe 10-year incidence of early ARM and increased retinal pigment. Our resultsconfirm those from the baseline and 5-year follow-up examinations.18,19 At baseline, summer sun exposurewas related to increased retinal pigment in men, and at the 5-year follow-upexamination, summer leisure time spent outdoors while participants were intheir teens and 30s was significantly related to the incidence of early ARM.However, a study by Taylor et al17 found thatexposure to visible light was associated with late ARM, and other studieshave found no associations between sunlight exposure and ARM.20,22
If cumulative sun exposure is related to the incidence of increasedretinal pigment or early ARM, it is hypothetically the effect of exposureto visible rather than UV light. Previous studies have not found ARM to beassociated with cumulative UV-A or UV-B exposure16,17,22 butsupport associations between ARM and ocular exposure to visible blue light.17,35,36 The results of astudy by Winkler et al37 in which photoporphyricmice exposed to blue light developed sub-RPE fibrillogranular deposits, aswell as other animal studies,38,39 areconsistent with this hypothesis. Some studies have suggested that excessiveblue-light exposure results in photic damage to the RPE induced by photochemicalor photo-oxidative mechanisms,35,36 possiblycontributing to ARM.
A study by Delcourt et al22 found subjectswho used sunglasses regularly to have a decreased risk of soft drusen (OR,0.81; 95% CI, 0.66-1.00; P = .05). In the BeaverDam Eye Study, after adjusting for age and sex, no significant associationsbetween use of hats and sunglasses and 10-year incidence or progression ofARM were found. However, in those who reported the highest summer sun exposurelevels in their teens and 30s, the use of hats and sunglasses at least halfthe time was associated with a decreased risk of developing soft indistinctdrusen and RPE depigmentation but not with the 10-year incidence of increasedretinal pigment or early ARM, the 2 outcomes for which participants with excessivesun exposure were at significantly higher risk. Our lack of findings indicatethat the use of sunglasses with UV protection may not provide benefits againstthe development of some signs of ARM.
Results of analyses of the relationship between the number of bad sunburnsexperienced during youth and the incidence and progression of ARM were inconsistent.No participants with more than 10 bad burns during youth developed late ARM.However, those experiencing 2 to 10 burns during youth had a significantlyhigher incidence of late ARM than those with 1 or no burn. Participants experiencing1 or no burn in their youth were less likely than those experiencing morethan 10 burns, but more likely than those experiencing 2 to 10 burns, to developsoft indistinct drusen. One possible explanation for these inconsistent findingsis that participants who experienced the highest number of burns during youthwere more likely to take measures protective against sun exposure later inlife. A second possible explanation is that these statistically significantassociations are a type I error (ie, due to chance alone when no associationsexist). Our study found no relationships between number of severe burns experiencedat the baseline examination and the 10-year incidence or progression of ARM.This is consistent with the Blue Mountains Eye Study.21
No relationships were found between mean annual ambient UV-B exposureand incidence or progression of ARM at 10 years in the Beaver Dam Eye Study.This is consistent with our earlier findings,18 astudy by Delcourt et al,22 and the MarylandWatermen Study,17 which reported no significantassociations between UV-B exposure and late ARM. A study by Delcourt et al22 found that subjects exposed to high ambient solarradiation had a decreased risk of pigmentary abnormalities (OR, 0.61; 95%CI, 0.39-0.93; P = .02), but we did not find thisassociation.
Any conclusions or explanations regarding associations, or lack thereof,described in this article must be made with caution for several reasons. First,the concomitant incidence of some lesions (eg, exudative macular degeneration)may limit our ability to detect meaningful relationships. Second, the analysesconducted for this study include the examination of associations between manyrisk factors and outcomes. There is a distinct possibility that some statisticalassociations reported in our article are type I errors. Third, it is alsopossible that some relationships were not found or were attenuated becausepersons with these factors who developed ARM did not live to participate inthe follow-up examination. However, nonparticipation as a result of deathis unlikely to bias our findings because ARM has not been associated withmortality.40 Fourth, differences in the participationrates of survivors at the follow-up examinations by level of sun exposuremay have masked or accentuated some associations. Finally, it is difficultto capture an adequate history of sun exposure with a questionnaire. Recallof behaviors occurring long ago may be poor or inaccurate. Insensitivity maylimit our ability to find an association when one exists, and misclassificationcan result in spurious findings.
In summary, few significant relationships between environmental exposureto light and the 10-year incidence and progression of ARM were found in theBeaver Dam Eye Study. However, the power to detect relationships between sunlightexposure and late ARM is limited by the rarity of the disease. A significantassociation was found between extended exposure to the summer sun and the10-year incidence of early ARM and increased retinal pigment. This is consistentwith results from the baseline and 5-year follow-up examinations. A protectiveeffect of hat and sunglasses use by participants while in their teens and30s against the 10-year incidence of soft indistinct drusen and RPE depigmentationwas also found, but only in those who reported the highest amount of sun exposureduring the same periods. These results support previous findings indicatinga possible link between extended sunlight exposure and the incidence of earlyARM. Further follow-up is needed if conclusions about the relationship betweensunlight exposure and incidence of late ARM are to be reached.
Corresponding author: Ronald Klein, MD, MPH, Department of Ophthalmologyand Visual Sciences, University of Wisconsin–Madison, 610 N Walnut St,450 WARF, Madison, WI 53726 (e-mail: firstname.lastname@example.org).
Submitted for publication February 19, 2003; final revision receivedAugust 27, 2003; accepted October 1, 2003.
This study was supported by grant EY06594 (Drs R. Klein and B. Klein)from the National Institutes of Health, Bethesda, Md, and in part by the SeniorScientific Investigator Award (Dr R. Klein) and Lew R. Wasserman Award (DrCruickshanks) from Research to Prevent Blindness, New York, NY.
Reprints not available from the author.