Outdoor/sports activity was significantly different between the subjects who became myopic
and the emmetropes
, with the difference present as early as 4 years before myopia onset. In contrast, each of the other individual activities became significantly different either at onset or after onset. Diopter-hours were significantly greater in became-myopic
children 1 year before onset, but the difference was small compared with the differences seen at onset and after onset. If increased near work is important as a cause of myopia, 1 year yields only a small window of time for the purpose of predicting myopia. For example, differences in refractive error and axial length between those who became myopic
were seen earlier before onset.10 Became-myopic
subjects began to diverge from emmetropes
3 years before myopia onset (as defined by −0.75 D or more in each principal meridian) for axial length and 4 years before reaching this criterion for refractive error. Hispanic children may be the one exception in which near work activity has some predictive utility. It should be noted that recruitment efforts targeted a given ethnicity at a single site. Therefore, our results related to ethnicity may not be representative across the United States, and some of what has been attributed to ethnicity may in fact be related to region or site instead.
The magnitude of the difference between the became-myopes
and the emmetropes
for outdoor/sports hours was small. Given that the protective effect of outdoor/sports activity has been found in several studies, it may be that small differences are a potent factor in inhibiting the onset of myopia. For example, in children with two myopic parents, performing sports or outdoor activity in the highest quartile (over 14 hours per week) compared with the lowest quartile (0–5 hours per week) reduced the probability of myopia onset from 0.60 to 0.20, respectively.6
This difference is similar to the statistically significant associations for physical activity and myopia progression in a cohort study of university students.8
Dirani et al.18
found a significantly lower odds ratio for both outdoor leisure activities and outdoor sports activities in an analysis of the SCORM, unlike an earlier SCORM analysis that found no association with outdoor activity.5
They credit the use of a more detailed survey that contained multiple questions, although our single question about each activity yielded a similar result. Reports from Saw et al. (IOVS
2008;49:ARVO E-Abstract 1551) and Rose et al.19
indicate that the protective effect may not arise from any specific physical activity so much as simply being outdoors. Because our question did not distinguish indoor from outdoor sports, it is possible that indoor sports were included by parents in this estimate. If the actual association is outdoor exposure and the indoor sports were unrelated as reported in other papers, it would be anticipated to attenuate the result toward the null.
Reported seasonal effects on myopia progression lend credence to the visual implications of outdoor experiences, although perhaps not in the way originally proposed.20,21
For example, Fulk et al.20
found that most children had slower myopic progression in the summer. Goss and Rainey21
also found that myopia progression in children during the 6 months that include summer was slower by half compared with the 6 months that include school attendance.
The distance viewing that accompanies being outdoors is a potential explanation for the preventive association of outdoor/sports activities. Brief exposures to distance stimuli or minimal defocus powerfully counteract the elongating effects of hyperopic defocus or deprivation in several animal species including the tree shrew,22,23
These studies provide evidence that perhaps distance clarity is the “stop” signal for growth, requiring only minutes or a few hours of clarity to offset the risk of a much longer exposure to a “grow” signal.
Retinal levels of dopamine increase during the day and decrease at night in a circadian rhythm, indicating a potential correlation with light exposure.27
Dopamine has been postulated to be related to eye growth, suggesting that more outdoor exposure could affect dopamine levels and therefore eye elongation, as shown in animal studies of form deprivation.28–31
Light-induced change in retinal dopamine levels remains a potential explanation for the inhibitory effects on refractive development associated with outdoor exposure.
Some individual visual activities did not differ between the became-myopes
until the onset of myopia (for reading, studying and computer/video games hours) or 1 year after onset (for TV hours). Of these four activities, three of these (reading, computer/video games, and TV) had the most robust associations, continuing to differ between the two groups as long as 3 to 5 years after onset. Differences after onset but not before onset lead to the question of whether there is a change in behavior in response to myopia development. Is a child more likely to opt for indoor activities after beginning to wear glasses? This hypothesis is consistent with two existing cross-sectional studies that have described increased amounts of near work activities among those already myopic compared with those who are not, but found no increased risk before onset.5,6
Unfortunately, the literature does not address behavior in myopes or behavior changes in relation to wearing glasses.
As stated in our previous paper6
and in a SCORM report,32
the hypothesis that increased reading represents a corresponding decrease in outdoor/sports activities is not supported by CLEERE data; that is, there does not appear to be a substitution of reading for outdoor/sports activity, either before or after myopia onset, as evidenced by the lack of negative correlation between outdoor/sports activity and any of the near work activities. This result is supported by a report from Harrison and Narayan,33
who reported that students who spent 1 to 2 hours per week in sports and 1 to 2 hours in other activities (such as club activities) had increased rather than decreased odds of studying 3 or more hours per week.
There are several potential areas for bias in these data. The first is that the responses are elicited from the parents rather than the children. Given the ages of these children, the parents are the most reliable source of information, as other studies in myopia and other research areas indicate. Saw et al.5,34
also used parent-reported data from their baseline visit, while the Sydney Myopia Study used subject-reported data from the baseline visit.35,36
There is, of course, the concern that parents may not know what their children are doing outside of school hours. One would suspect that parents' inability to track their children's activities would be truer of older children than of younger children. We posited in an earlier paper6
that there may be some differential recall bias, because parents often transport their children to sporting activities and therefore might be more likely to remember hours engaged in sports.
It is possible that using computer/playing video games may differ across the study, by virtue of the length of this longitudinal study. This is a valid observation that affects longitudinal studies of all kinds. A look at some of the literature may indicate that a large difference over the past 10 years in the type of game used (increase in hand-held games) versus the time spent (less than an hour among those playing) and in the increased number of children engaging in this activity.37
The reported data over time in the CLEERE Study indicate an upward trend over the course of the study, with the increase being no more than 1.5 hours on average per group; however, this increase occurs in both those who became myopic and the emmetropes, leaving the difference between the two groups relatively unbiased over time.
Another limitation is related to there being only one administration of the survey per year, as opposed to multiple administrations per year. Most studies lack longitudinal visual activity data, making it difficult to determine what activities are currently performed. Perhaps the more pertinent problem is activities during summer vacation versus during the school year. Deng et al.38
presented data that parents reported on the time children spent in various visual activities during both the school year and summer. Their results showed that decreased outdoor activity was seen in myopes during the school year compared with nonmyopes, but myopes did not differ from nonmyopes in the time spent reading, studying, or on the computer. During the summer there was no significant difference between myopes and nonmyopes for any visual activity.
Another potential issue deals with the recall bias that may be involved with the use of questionnaires. Other options that have been used in myopia research include diaries and the experience sampling method (ESM), where subjects carried pagers and responded by calling in to report their visual activity when paged.39,40
The biggest drawbacks to either of these methods are the burden on the respondent and the complex logistics.
Rah et al.39
used the ESM to quantify children's daily visual activities. In comparison to a questionnaire completed by the children, the ESM responses differed only in the amount of time spent in conversation. When comparing parents' responses to ESM, differences were found in conversation time and chores time. The sample was small, however, and so some comparisons had low statistical power.
Saw et al.40
compared 4 days of diaries completed by parents within a few weeks of a baseline clinic visit with questionnaire responses. They reported data from four different times of year. A small sample was pilot tested for reliability, and they found that total near work (a sum of the near work variables) had an ICC of 0.87, although the ICC for total weekend near work hours was lower (0.33). Otherwise, TV hours had the lowest ICC (0.47), with reading and writing together at 0.97, and video games at 0.80. Comparing the interview using the questionnaire to the diary showed an ICC for total near work of 0.50 and for reading and writing of 0.55. The remaining ICCs were in the range of 0.67 (TV hours) to 0.90 (computer hours). The questions asked were similar to those in the CLEERE survey, with the exception of separating computer and video games, breaking out weekday from weekend, and attempting to include examination and vacation time. The relation between the time the diaries were completed and the questionnaire is unclear, as the 4 days of diaries were completed within a few weeks of the questionnaires. How this correlated to four different points in the year was not specified.
An additional potential source of bias may arise from the use of a brief questionnaire as opposed to a more comprehensive one. In addition to the variation of the CLEERE questionnaire that Saw et al. used, the Sydney Myopia Study35,36
had a baseline questionnaire that asked detailed physical activity questions on the amount of time per day spent in various activities. For the near work activities, information per day was solicited by choosing one of four check boxes: none; less than 1 hour; 1 to 2 hours; and 3 or more hours. While the benefits of parsing out sports from being outdoors are definitely an advantage of a detailed questionnaire, one would assume that the likelihood of finding any association would increase using a more general question. A significant finding might result from the pooling of multiple items. It would seem less likely that there would be an absence of an association that in fact exists.
Follow-up of our emmetropes with necessary data was shorter than that of myopes and therefore may have affected the growth curves. Data were available across the entire age range for the emmetropes, so the modeling applied should lead to robust results. Some of the follow-up problems were intrinsic to the study design, where there was a staggered entry of subjects during the initiation of the study and a cutoff at eighth grade,13
meaning that some of the identified emmetropes had the opportunity for only three visits (those enrolled as sixth graders). In addition, the older emmetropic subjects (at entry) were probably more likely to remain emmetropic. It could be that emmetropes had a lack of incentive to continue to participate. There were approximately 260 13-year-old emmetropes and approximately 100 14-year-old emmetropes who were included in the analysis. These subjects represent the eighth graders. The strict definition of emmetropia made it more likely to exclude subjects when small measurement variability occurred. Although this criterion limited the sample size, we do not believe that it introduced a systematic bias.