Visual Impairment, Uncorrected Refractive Error, and Accelerometer-Defined Physical Activity in the United States FREE

Physical activity is an important predictor for many health outcomes.¹ Restrictions in physical activity have been associated with a decreased quality of life, higher morbidity, and higher mortality.^2- 10 Encouraging more physical activity may therefore provide important benefits.¹¹

Vision loss has been shown to affect several aspects of mobility, including balance, falls, and movement through mobility courses.^12- 20 The perceived risks of mobility may also further limit physical activity among patients with low vision.²¹ Previous studies of mobility in individuals with decreased vision have largely relied on self-reports or proxy reports, which are subject to reporting biases, or on physical activity performed in a laboratory setting. However, the relationships of these measures to real-world physical activity patterns are unclear. Thus, there is a need to objectively characterize the relationship between decreased vision and real-world physical activity using technology such as accelerometers.

To objectively evaluate real-world physical activity, the National Cancer Institute supported the use of accelerometers in the National Health and Nutrition Examination Surveys (NHANES) conducted during 2003-2004 and 2005-2006. Accelerometers have been validated as a measure of total energy expenditure and have been used as a preferred method for objective measurement of physical activity in several studies.^22- 30 Herein, we use NHANES visual acuity and accelerometer data to examine the relationship between decreased vision and objectively measured physical activity levels in adult Americans.

Some causes of decreased vision, uncorrected refractive error (URE) in particular, are easily correctable with minimal cost. Attention was therefore given to how physical activity is affected by URE as compared with visual impairment (VI), defined as decreased vision not resulting from URE.

The NHANES 2003-2004/2005-2006 protocols were reviewed and approved by the National Center for Health Statistics research ethics review board. Informed consent was obtained from all participants. The research adhered to the tenets of the Declaration of Helsinki.

Data were obtained from the 2003-2004/2005-2006 rounds of NHANES, a cross-sectional study chosen to reflect a representative sample of the US civilian, noninstitutionalized population through a complex, multistage probability design.³¹ Survey participants were interviewed in their homes and invited to undergo a comprehensive health examination in a mobile examination center, including visual acuity testing and initiation of a 1-week physical activity measurement trial using an accelerometer. Survey participants provided basic demographic data such as age, sex, and ethnicity.

Visual acuity was measured for each eye as previously described.³² Presenting visual acuity for each eye was assessed using the ARK-760 (Nidek Co Ltd), an autorefractor containing built-in visual acuity charts. Participants were asked to wear their usual distance vision correction, if any. The 20/50 line was presented first. If the participant was unable to read the 20/50 line, the 20/200 line was presented. Participants who could not read the 20/200 line had their visual acuity categorized as worse than 20/200. Participants able to correctly read at least 4 of the 5 characters for the 20/50 line were allowed to move to the next line of smaller characters. This continued until the participant missed 2 or more characters per line for 2 lines in a row. Presenting visual acuity was recorded as the last line for which 4 or more characters were read correctly. Visual acuity was not tested in participants who reported during the home interview that they had no light perception.

After presenting visual acuity was measured, corrective lenses were removed and the refraction of each eye was measured by the autorefractor. For eyes with presenting visual acuity worse than 20/25, corrected visual acuity was assessed using the measured refractive error correction. Visual acuity of the better-seeing eye was used to characterize visual impairment status. For participants with visual acuity data in only 1 eye, better-seeing eye visual acuity was taken as the acuity of the lone measured eye. When autorefraction results were missing from only 1 eye, we assumed that the visual acuity in that eye did not correct to 20/40 or better with refraction. Participants with missing presenting acuity in both eyes, or with visual acuity worse than 20/40 in both eyes with no autorefraction in either eye, were considered to have incomplete visual acuity data and were excluded from the analyses.

Subjects whose presenting visual acuity was 20/40 or better were classified as having normal sight. Individuals in whom presenting visual acuity was worse than 20/40, but postrefraction visual acuity was 20/40 or better, were characterized as having URE. Subjects whose visual acuity was worse than 20/40 even after autorefraction, or who reported no light perception (10 of 10 020 participants), were classified as having VI. The degree of decreased visual acuity was further classified as moderate (worse than 20/40 but better than 20/200) or severe (20/200 or worse).

Physical activity was measured by having participants wear an accelerometer (model 7164; Actigraph, LLC) over the right hip on an elastic belt for 7 days. Specific details of the accelerometer protocol have been previously described.³³ Participants were instructed to wear the device while awake, but not during times of swimming or bathing. Participants then returned the device, at which point data were downloaded and the accelerometer was evaluated to ensure it still met the manufacturer's calibration specifications. The accelerometer measured and recorded the intensity of vertical acceleration produced by locomotion or other activity over 1-minute intervals. Raw data were recorded as counts per minute and were used to define the activity level for each minute as sedentary/light or moderate/vigorous. One-minute intervals with counts of 2020 or more were defined as having a moderate or vigorous level of physical activity (MVPA).²⁵ Additionally, the device measured the steps taken over each 1-minute interval. Data on intensity of activity were made publically available for both the 2003-2004 and 2005-2006 rounds of NHANES, but step data were made available only for the 2005-2006 round.

Participants with invalid accelerometer data, defined using the SAS code available at http://riskfactor.cancer.gov/tools/nhanes_pam/, were excluded.^25,34 Excluded individuals included those with insufficient wear time and individuals whose accelerometer was found to be out of calibration when returned. Wear time for each day was defined after excluding all intervals of at least 60 consecutive minutes of zero activity intensity counts, with allowance for 1 to 2 minutes of counts between 0 and 100.²⁵ When an individual's wear time was less than 10 hours per day for at least 4 days of the week, that individual's accelerometer data were considered invalid.

Differences between groups defined by visual status were calculated using χ² analyses and univariate negative binomial regression models. Differences in physical activity were then compared across visual acuity status using multivariable negative binomial regression models. Negative binomial models were chosen because steps and minutes of MVPA were in the form of count data that failed tests of normality and displayed evidence of overdispersion.³⁵ Regression coefficients from negative binomial models represent rate ratios, which reflect the relative rate of events (ie, steps or minutes of MVPA) for each variable in the model compared with its control. Covariates included in multivariable models included age, sex, race, obesity (body mass index ≥30 [calculated as weight in kilograms divided by height in meters squared]), and education.^25,36 The following systemic diseases were also included as covariates: arthritis, congestive heart failure, chronic obstructive pulmonary disease/asthma, diabetes mellitus, and stroke. The presence of these systemic diseases was based on the participant's responses to the question “Has a doctor or other health professional ever told you that you had (specific disorder)?” If patients noted that they “didn't know,” they were categorized as not having that disorder. Analyses were restricted to individuals 20 years or older because self-reported comorbidity data were fully available only for this age range.

Summary measures of physical activity were generated in SAS version 9.2 (SAS Institute Inc) and subsequently analyzed using Stata version 11 (StataCorp).³⁷ All analyses used the examination weights provided with NHANES data sets to adjust for the complex NHANES design; 4-year and 2-year sample weights were used for analysis of MVPA and steps data, respectively.³¹

A total of 20 470 individuals participated in NHANES during the 2003-2004 and 2005-2006 periods. Among these participants, 10 020 (48.9%) were 20 years or older. Within this age group, 5722 participants (57.1%) had complete visual acuity and accelerometer data. In 2005 and 2006, when steps data were collected, there were 10 348 participants, of whom 4979 (48.1%) were 20 years or older. Within this age group, 2852 participants (57.3%) had complete visual acuity and accelerometer data.

Subjects who had complete data were older, differed in their racial/ethnic distribution, were less often obese, and more often had at least some college education when compared with individuals with incomplete visual acuity and/or accelerometer data (P ≤ .01 for all) (Table 1). Subjects with complete data were also significantly less likely to report a history of stroke but significantly more likely to report a history of arthritis (P = .01 for both). There were no significant differences in physical activity between those with complete and incomplete visual acuity data (Table 1).

Age, race/ethnicity, education, and the frequency of arthritis, congestive heart failure, diabetes, and stroke all varied significantly by vision status (P ≤ .01 for all) (Table 2). Compared with subjects with normal sight, subjects with VI were older, had less college education, and were more likely to report a history of arthritis, congestive heart failure, diabetes, and stroke (P < .05 for all). Subjects with URE, on the other hand, had less college education, were older, were more likely to be Mexican American or non-Hispanic black, and were more likely to report a history of diabetes or stroke as compared with subjects with normal sight (P < .05 for all). After adjusting for age, the comorbid conditions evaluated herein did not differ in frequency across vision status except for diabetes being more common in the URE group than the group with normal sight (P = .04; P >> .09 for all other pairwise comparisons).

In the 2003-2004/2005-2006 NHANES data, there was no meaningful difference in the average number of valid days of accelerometer data across those with normal sight (6.0 days), URE (5.9 days), and VI (5.9 days). The mean wear time of the accelerometer per valid day was 14.3 hours among participants with normal sight, as compared with 14.2 hours in subjects with URE (P = .49) and 14.3 hours in subjects with VI (P = .79).

No differences in physical activity measures were found between individuals with moderate or severe vision loss due to URE (P >> .10) (Figure 1). Individuals with moderately and severely decreased vision from URE were therefore analyzed together in all subsequent analyses. Individuals with different degrees of VI all had physical activity measures that were significantly lower than subjects with normal sight or URE (P < .01 for all), without demonstrating a clear dose response (Figure 1). Hence, in further analyses, individuals with moderate and severe VI were grouped together.

The average number of steps taken per day in individuals with normal sight was 9964 as compared with 9742 steps per day in individuals with URE (P = .57) and 5993 steps per day in individuals with VI (P < .01) (Figure 2). On average, individuals with normal sight accumulated approximately 24 daily minutes of MVPA, as compared with 23 daily minutes for individuals with URE (P = .77) and 9 daily minutes for individuals with VI (P < .01) (Figure 2). Individuals with VI took significantly fewer steps and engaged in fewer daily minutes of MVPA when compared with individuals with URE (P < .01 for all).

In multivariable models, participants with VI took 26% fewer steps per day (P < .01; 95% CI, 18%-34%) and engaged in 48% (P < .01; 95% CI, 37%-57%) less time in MVPA when compared with participants with normal sight (Table 3). Multivariable models also showed that physical activity outcomes were significantly worse in those with VI as compared with those with URE (19% fewer steps per day; P < .01; 95% CI, 5%-31%; 38% less time in MVPA; P < .01; 95% CI, 23%-50%). There were no significant differences in physical activity measures between those with URE and normal sight (Table 3).

Adult Americans with VI, but not those with URE, engage in less physical activity. The impact of VI on walking and physical activity was found to be equal to or greater than all systemic conditions examined, including obesity, chronic obstructive pulmonary disease, arthritis, stroke, and congestive heart failure. To our knowledge, these findings represent the first report of an association between VI and objectively measured physical activity and demonstrate that the cause of vision loss (URE vs nonrefractive causes) is highly relevant to physical activity measures. Decreased physical activity has been associated with numerous adverse health and quality of life outcomes,^2- 10 suggesting possible ties between VI and systemic well-being.³⁸ Previous work has shown that VI is associated with slower walking, errors when walking through mobility courses, and worse self-reported mobility.^15,19- 20 Our work extends these previous findings, demonstrating that individuals with VI also walk less and are less physically active. It also confirms previous studies showing that older age, female sex, and higher body mass index are associated with lower physical activity levels, while Mexican American ethnicity is associated with greater physical activity.^{25,28,36,39- 40}

Several plausible pathways link VI to lower physical activity.²¹ Individuals with VI may be less confident walking or may choose a less active lifestyle because of greater difficulty with social interactions.⁴¹ Visual impairment can also affect balance, leading to more frequent falls or engendering greater fear of falling.⁴² Physical activity is also generally greater away from home,⁴³ and VI may affect physical activity by making individuals more homebound, particularly when poor vision affects driving ability.^44- 45 Previous work did not consistently distinguish between nonrefractive and refractive vision loss, though the current work strongly suggests that this distinction is important when relating mobility outcomes to VI.

There are several possible reasons why URE, even when associated with presenting acuities worse than 20/40, was not observed to impact measures of physical activity. In uncorrected myopia, objects become clear as they come closer, which might be sufficient to allow mobility. Alternately, the URE group may be enriched for individuals who are able to function in the presence of their refractive error, minimizing the apparent impact of URE. A final possibility is that decreased visual acuity caused by organic disease fundamentally affects walking and physical activity levels more than decreased acuity from URE. The NHANES does not identify the cause of VI, making study of physical activity restriction with specific eye diseases an important area for future studies.

The health consequences of poor physical activity have been widely stated.^46- 47 However, the extent to which these health consequences are experienced by individuals with VI remains unclear. Visual impairment typically presents later in life⁴⁸ and may not persist long enough to cause systemic diseases that occur as a result of years of decreased physical activity.⁴⁹ Indeed, after age adjustment, we did not find that the presence of VI was associated with an increased prevalence of several systemic diseases, though the data available did not allow for subanalyses within individuals with long-term VI. Therefore, restriction of walking and physical activity resulting from VI may not profoundly affect the prevalence of systemic illness but instead have other impacts, such as worsening of existing disease, a greater reliance on others for tasks requiring mobility, greater social isolation, decreased strength and fitness,⁵⁰ and lower quality of life.^51- 52

The impact of VI on physical activity suggests that better systems are necessary to encourage walking and physical activity through low-vision rehabilitation. Currently, it is estimated that only 29% of low-vision rehabilitation entities/hospitals offer orientation and mobility training.⁵³ Only 20% of low-vision rehabilitation entities have mobility specialists⁵³ and they often do not stress physical activity as a priority. Our current results suggest that better systems are required to increase physical activity in individuals who are visually impaired, assuming greater physical activity does not translate into more falls or injuries.

Our study has several limitations. In the visual acuity examination portion of the study, there was a sizeable nonparticipation rate (12.8%), which may be due to insufficient time to participate, inability to cooperate with the visual acuity examination protocol, or equipment malfunction. Additionally, the refractive correction worn or not worn during visual acuity testing may not have matched the worn correction over the week of accelerometer testing. Considerable accelerometer data (39.2%) were also missing owing to a combination of nonparticipation and questions about data quality for some subjects. Some differences were noted between subjects who completed both the visual acuity and physical activity testing and subjects who failed to complete one or both of these measures, raising the possibility of bias introduced through selective nonparticipation. However, the rates of VI were similar in subjects with and without complete data, as were the rates of most of the diseases studied. This argues against selective nonparticipation among sicker subjects. Indeed, younger subjects were more likely to not complete accelerometer testing, suggesting that participation may have been less common among working individuals, which would unlikely bias findings in a positive direction.

While accelerometers are objective measures of physical activity, the values provided by them should not necessarily be taken as absolute. Specifically, the use of accelerometers to assess physical activity does not capture moments of upper extremity exercise nor swimming-related activities. In addition, because our analysis did not account for the decline in exercise capacity with age and used a single cut point value for evaluating moderate and vigorous activity, we may have underestimated the activity levels for older adults.²⁵ Moreover, previous work has noted that the accelerometers used in NHANES tend to overestimate the measurement of steps.⁵⁴

In summary, VI, but not URE associated with the decreased presenting visual acuity, is associated with lower levels of objectively measured physical activity, with an effect size comparable with or greater than serious medical conditions such as congestive heart failure, diabetes, or stroke. These findings highlight the potential adverse impact of VI on fitness, health, and quality of life. Individuals with VI are an important group to target with regard to increasing physical activity, and increasing physical activity levels without compromising safety (ie, falls) should be a focus of low-vision rehabilitation and mobility training.