Sleep Patterns in Children With Blindness: A Comparison With Normally Sighted Peers

Abstract

Purpose: Studies showing problematic sleep patterns in blind and visually impaired children are often based on (parent) self-report. The purpose was to compare sleep patterns of blind children to normally sighted peers using objective measures.

Methods: In this cross-sectional study, 100 blind (best-corrected visual acuity <3/60) and 100 age- and gender-matched normally sighted children aged 7 to 17 years wore a digital activity monitoring device for 1 week. Sleep quantity (i.e., total sleep time and total time in bed) and sleep quality (number of awakenings, latency, efficiency, wake after sleep onset [WASO], and sleep fragmentation index) were measured. Adjusted linear regression analyses were used to model group differences in sleep parameters.

Results: Data of 163 children were included. Blind children spent significantly less total time in bed in minutes (β, −31; 95% confidence interval, −56 to −6) and had a lower total sleep time (−41; −66 to −17), smaller number of awakenings (−2.8; −4.5 to −1.0), a lower WASO (−10; −16 to −5), and a more efficient sleep pattern (1.5; 0.1 to 2.8) compared to normally sighted children.

Conclusions: Although sleep quantity and recommended hours of sleep per night were lower among blind children than normally sighted children, their sleep quality was better. This contradicts findings of self-report studies and warrants further studies to measure sleep objectively. Further, the discrepancy between previous findings and our findings regarding sleep quality may be explained by the house rules of the boarding schools attended by blind children, which may facilitate improved sleep hygiene.

Sleep is a key aspect of human activity and of equal importance to well-being as wakefulness. Sleep is a fundamental part of childhood development as it is essential for cognitive, behavioral, and neurologic development.^1–3 This, in turn, has an impact on memory development and academic achievements.^4–6 Sleep is directly related to effective learning and better physical and mental health.^7,8 The number of hours of sleep humans need depends primarily on a person's age and the type of physical activity they engage in. The American Academy of Sleep Medicine (AASM) has recommended that children aged 6 to 12 years should regularly sleep 9 to 12 hours per 24 hours and teenagers aged 13–18 years 8 to 10 hours.⁹ Sleep quality is equally important in the sleep cycle. It has been reported that an estimated 25% of the pediatric population has (behavioral) sleep problems during childhood and adolescence such as bedtime refusal, frequent nighttime waking, sleep walking, or parasomnias.^10,11 Sleep problems are commonly considered a global public health issue.^12,13 

Blindness and visual impairment (VI) are among the leading causes of disabilities in children and are considered a priority condition by the World Health Organization (WHO).^14,15 It is important to understand sleep problems among children with VI as they have attentional, behavioral, and physical health problems, and issues with their circadian rhythms are well known.¹⁶ Studies have shown that around 80% of children with blindness experience abnormal sleep patterns and sleep disturbances, mainly because their sleep–wake cycle is altered due to a lack of light perception (LP).^17,18 Ocular light exposure is an important factor in the regulation of the circadian rhythm, and individuals with no LP (NLP) may therefore experience a desynchronized circadian rhythm, resulting in poor sleep and daytime dysfunctioning.^19,20 

Among the children with VI, sleep disturbances are more common in children with vision loss at the extent of those having only limited LP or NLP. Studies using self-reported sleep questionnaires have found increased daytime napping, night awakenings, and longer time to fall asleep in children with anophthalmia, microphthalmia, and optic nerve dysfunctions.^21,22 

Quantity and quality of sleep can be measured in various ways, including subjective methods such as sleep diaries and self-reported sleep questionnaires. There are also objective measures that can be obtained using actigraphy and polysomnography.^23–25 An actigraph is a small noninvasive digital activity monitoring device that is particularly useful in pediatric populations in whom reliance on the parent-reported questionnaires may have limited accuracy of a child's actual sleep behavior. Former studies have reviewed that sleep measurements using actigraphy correlate highly with other methods of assessing sleep such as polysomnography, sleep diaries, and questionnaires.^26,27 The use of actigraphy in sleep studies in children is gaining popularity to the extent that methods such as polysomnography are now slowly replaced. The AASM states that “actigraphy is indicated for delineating sleep patterns, and to document treatment responses in normal infants and children (in whom monitoring by polysomnography can be difficult to perform), and in special pediatric populations.”²⁸ Moreover, studies have recommended combining sleep diary data with actigraphy measurements as the best method of choice for sleep analysis.^29,30 

Studies on sleep patterns in blind and visually impaired children are often based on self- or parent-reported questionnaires.^31–33 These studies have shown that children with blindness and VI have more sleep disturbances than normally sighted children. To date, only a few studies investigated sleep patterns in children with VI or blindness using actigraphy.^34,35 Two studies using actigraphy concluded that there were no differences between sleep parameters in blind, VI, and normally sighted children. However, these studies carry several limitations. The sample size of one study was rather small, comparing 20 blind youngsters (age 10–15 years) to deaf and normally sighted, normally hearing children.³⁵ A second larger study compared 58 younger children between 4 and 11 years with VI to normally sighted children. In 62% of children in this study, however, the severity of vision loss was not established. In our study, we aimed to investigate sleep behaviors in blind children and adolescents compared to matched normally sighted peers from the same school and grade level, using a larger sample size. In addition, we aimed to investigate the association between vision-related factors and sleep patterns such as the severity of blindness in terms of LP and the causes of blindness. The findings of this study help contribute to the scant available information on sleep behavior in children with VI. The study will help increase awareness on sleep disorders, importance of sleep screening, and potential treatment modalities of these disorders, including counseling and medication. 

Methods

This cross-sectional study is a part of the Nepal Pediatric Visual Impairment Study, which aims, apart from sleep behaviors, to investigate causes of childhood VI and blindness, the children's level of participation, physical activity, and quality of life. 

Ethical Consideration

The study was approved by the ethical review board of the Tilganga Institute of Ophthalmology and then forwarded to Nepal Health Research Council, where it received final approval. The study adhered to the tenets of the Declaration of Helsinki. 

Informed consent was provided by the parents of the children, and assent was provided by the children themselves. 

Participants

Children aged 7 to 17 years, studying in integrated schools for the blind in and around Kathmandu valley, were recruited. One hundred participating blind children were age and gender matched to 100 normally sighted children studying in the same schools and who were in the same grade level. Exclusion criteria for this study were children having a developmental delay, cognitive impairment, or other disabilities aside from VI. 

Criteria for Visual Impairment and Blindness

Vision loss was classified by the WHO categories of VI,³⁶ where blindness was defined as best-corrected visual acuity (BCVA) <3/60 in the better eye with two groups: BCVA <3/60 to LP and NLP. The visual field was not assessed. Normally sighted participants were defined as having a BCVA of 6/12 or better in the better-seeing eye. 

Procedure and Measurements

All participants underwent an ocular examination including a vision assessment, refraction, and anterior and posterior segment examination. Sociodemographic and clinical data (i.e., age, gender, living situation [hostel/boarding school versus living with family], history of use of melatonin or any other medication that would alter sleep behavior, diagnosis, and visual acuity categories) were collected. 

Participants’ sleep outcomes were measured using an ActiGraph (wGT3X-BT; ActiGraph, LLC, Pensacola, FL, USA). It was worn on the wrist of the nondominant hand for 1 week to record physical movements associated with daily activities and sleep. Children were advised not to remove the device except during bathing. After 1 week, the device was collected and the stored data were downloaded within the Actilife software environment (Version 6.13.3). 

We used the “Tudor-lock” algorithm to calculate the “automated sleep period” and “wear time” for each child. It is an auto sleep period detection algorithm that classifies each epoch in the ActiGraph as asleep or awake. This algorithm has been validated in children.³⁷ 

Sleep diaries were used to complement the data gathered by the ActiGraphs. All children were asked to keep a sleep diary that included daily recording of “in bed time,” “out of bed time,” and “screen time prior to bedtime.” The participants were asked to fill in the diary by themselves or with help from parents, caregivers, or friends if needed. 

The sleep parameters were calculated using the “Sadeh” algorithm, which is considered appropriate for use in children and adolescents.³⁸ In the children whose sleep diary was complete, the “in bed time” and “out of bed time” were manually included in the software consistent with the sleep diary. The data were exported in a csv file for each participant. 

Seven sleep parameters were obtained: 

• 1. Total time in bed: the time between sleep onset and sleep offset
• 2. Total sleep time (TST): the total number of minutes scored as “asleep,” which represents the time between sleep onset and offset while removing the wake after sleep onset (WASO)
• 3. Number of awakenings: the number of different awakening episodes
• 4. WASO: the total number of minutes the participant was marked awake after sleep onset occurred
• 5. Latency: the time taken to fall asleep after “lights-out.” Latency was calculated in minutes using the in bed and out bed time provided by the child in the sleep diary
• 6. Efficiency: the percentage of time the child sleeps in relation to the amount of time child spends in bed was calculated by dividing TST by total time in bed
• 7. Sleep fragmentation index: the measure of sleep quality based on restlessness, where zero represents “good sleep,” and the higher the number, the poorer the quality of sleep

The first two parameters represent the sleep quantity and the last five the quality of sleep. 

To further explore the circadian rhythm of children with NLP, 1-week data of sleep time and latency were analyzed in these children and compared with a random selection of age- and gender-matched children with category 3 blindness (3/60 to LP). 

Statistical Analysis

Every child included in the study had at least four and a maximum of seven observations (days) for every sleep parameter. Children with data representing less than 4 days of at least 10 hours per day were excluded from further analysis. Means across these 4 to 7 days were calculated in Excel for every child and for every parameter. Cohen's d effect sizes were calculated to get an initial idea of the magnitude of mean differences of the sleep parameters between children with blindness and normal sight. 

After testing assumptions (normality and multicollinearity) and removing outliers where necessary, differences in sociodemographic and clinical data between children with blindness and normal sight were calculated using independent samples t-tests and chi-square tests. Also, linear regression analyses were performed for every sleep parameter as the dependent variable. The determinant was “vision status” (blind versus normal sight). Every model was adjusted for potential confounding by age and gender. “Living situation” was highly correlated (Spearman r: 0.85) with vision status and was not included in the regression models. In addition, linear regression models were repeated also for the “diagnosis categories” as the determinant (retina versus whole globe versus other categories, the latter representing diagnoses related to lens, cornea, uvea, and optic nerve). These models were adjusted for potential confounding by age and gender. A P value of <0.05 (two-sided) was considered statistically significant. Finally, to further investigate the role of the perception of light on sleep, we compared children with NLP to children with LP and used their mean 4- to 7-day data. Out of the children with NLP (n = 11), 7 had entered sleep diary data from which we could calculate their mean in bed time and latencies, and we compared each child with NLP with a “similar” child with LP. Similarity was based on age and gender, and the selection of the children with LP was made at random. As this was an additional analysis in a very small group of children, we computed graphs showing differences between these children, performed a nonparametric test (Mann–Whitney U), and reported z scores. All data were analyzed in IBM SPSS Statistics Version 27 (SPSS, Inc., Chicago, IL, USA). 

Results

Available Data, Demographics, and Clinical Characteristics

Out of 200 participants, data of 163 (83.5%) participants were included in the study. The 37 excluded participants were 33 children for whom the sleep data were not detected by the Actilife software and another 4 children who did not wear the ActiGraphs for at least 4 days, 10 hours per day. The included data consisted of 78 children who were normally sighted and 85 children who were blind. There were no statistically significant differences in terms of age or gender between “excluded” and “included” participants (respectively, P = 0.54 and 0.46). 

Table 1 shows the demographical and clinical characteristics of the blind and normally sighted participants. The most common etiology of blindness was due to retinal diseases. In children with NLP (n = 11), blindness was due to whole-globe and optic nerve disorders in, respectively, nine and two children. A significant difference was found in the “living condition” of children between two groups wherein blind children were more likely to reside in a hostel/boarding school compared to their normally sighted peers (P < 0.001). 

Table 1.

 

Demographics and Clinical Characteristics of Blind and Normally Sighted Children

 

As an additional descriptive analysis, we also explored whether the total group of blind and normally sighted children followed the recommended sleep durations according to their age groups (total minutes in bed and TST). Younger children (7–12 years) and adolescents (13–17 years) spent a mean (SD) of a 7.6 (1.02) hours and 7 (1.07) hours in bed, respectively. Mean (SD) TSTs in these two age groups were 6.7 (1.04) hours and 6.2 (1.09) hours, respectively. This was lower than the recommended 9 to 12 hours of sleep in children aged 7 to 12 years and 8 to 10 hours in those aged 13 to 17 years. 

Out of 163 responders, sleep diary data from 106 children were available (55 blind and 51 normally sighted children). The sleep diaries were used to calculate latency and to explore screen time habits (i.e., watching television or playing with their devices such as mobile phone or tablet before bedtime). We found that 36% of children with VI and 47% of normally sighted children watched television or used their devices before bedtime. 

Latency data were not normally distributed and were therefore log-transformed. Descriptive information for all seven sleep outcome measures was calculated for the blind and normally sighted children (Table 2). Small effect sizes were found with lower total sleep times, higher latencies, lower sleep efficiencies, lower sleep fragmentation indices, and lower numbers of awakening for blind children compared to normally sighted children. Medium effect sizes were found with a lower total time spent in bed and a lower WASO for blind children versus normally sighted children, meaning they were fewer minutes awake after sleep onset. 

Table 2.

 

Comparison of Different Sleep Parameters Between Blind and Normal-Sighted Children

 

The crude regression analysis presented in Table 3 revealed that blind children spent less time in bed, had a lower total sleep time (almost three-fourths of an hour), a lower WASO (fewer minutes awake before sleep onset) score, and a smaller number of awakenings compared to normally sighted children. No significant differences were found between blind and normally sighted children among the other three sleep parameters. 

Table 3.

 

Regression Analysis of Sleep Parameters Comparing Blind to Normally Sighted Children

 

The linear regression models that were adjusted for potential confounders showed a similar pattern. In addition, after adjustment for age and gender, blind children showed significantly more efficient sleep pattern compared to normally sighted children. Further, after adjustment, differences in latency and sleep fragmentation were on the verge of significance as well. 

No statistically significant differences in sleep parameters were found between causes of blindness within the group of blind children (Supplementary Table). 

Finally, we compared the 7 (out of 11) children with NLP to the age- and gender-matched children with LP on the parameters of total sleep time and latency. Four children with NLP had no sleep diary data. Child-pairs were between the ages of 7 and 17 years, and there were 4 boys and 3 girls. Figure A shows that the total sleep time between children with NLP versus LP was comparable. Regarding latency, it appeared that one child (No. 4 in Fig. B) with NLP had an extremely high latency compared to the child with LP. However, one child with LP (No. 5) also appeared to have a slightly higher latency than its counterpart. Differences between children with NLP versus LP on these two sleep parameters did not seem to be structural, and z scores were not significant (P = 0.65, z = −0.44, for total sleep time and P = 0.52, z = 0.64 for latency). 

Figure.

 

(A) Total sleep times for 7 blind children with NLP and 7 blind children with LP matched by age and gender. (B) The latencies between 7 blind children with NLP and 7 blind children with LP matched by age and gender.

Discussion

In our study, we explored sleep patterns of blind children using actigraphy data and compared these to gender- and age-matched data of normally sighted peers. All children wore ActiGraphs for 1 week. We found statistically significant differences in most sleep parameters between the two groups. Importantly, the blind children in our study had a lower sleep quantity (total sleep time and time in bed) compared to the normally sighted children. Concerningly, the total sleep time in both blind and normally sighted children in both the younger and older age groups was between 1 and 4 hours shorter than the recommended amount of sleep by the AASM.⁹ Yet blind children had a higher sleep quality compared to normally sighted children in our study (i.e., less awakenings, a lower WASO, and a more efficient sleep pattern). These differences in sleep quality and sleep quantity between blind and normally sighted children are inconsistent with the two similar studies previously conducted, both of which did not find statistically significant differences in sleep parameters between groups.^34,35 Reema et al.³⁵ concluded that their lack of significant findings may be explained by differences between methods (i.e., subjective versus objective) and the age groups studied. 

Indeed, studies among blind and visually impaired children that used subjective methods such as sleep questionnaires found more sleep disorders and a lower sleep quality in blind children compared to normally sighted children.^31–33 Hayton et al.³⁴ concluded, for example, that the discordance between subjective and objective measures, as shown by very low correlations (0.05 to 0.15) in the visually impaired group, could reflect the heightened awareness by caregivers of sleep problems in children with blindness. Those reasons may partly reflect why we found a better sleep quality in blind children. 

Further, the discrepancy between our findings and the aforementioned studies might be due to the relatively small sample sizes in the previously conducted studies. Consequently, the studies might have been underpowered to detect any statistically significant differences. However, the mean differences between the visually impaired and normally sighted children seemed genuinely smaller. Therefore, the difference between our studies on the measured objective sleep parameters could also be due to other factors, such as using another type or the duration of use of the ActiGraphs, the different age, or the severity of the VI of the participants. In our study, all children had a worse VI with a BCVA of the better eye <3/60 to NLP, whereas in the study by Hayton et al.,³⁴ of the six children (10%) who were assessed as blind, four had NLP. 

Yet, our analysis did not show a structural pattern of differences in latency and total sleep time between children with NLP and LP, but our study sample of children with NLP was admittedly very small (n = 7). Hence, studies including a larger group of children with NLP seem warranted. Previous reports have shown sleep problems and disturbed circadian rhythms in blind patients.^39,40 The effect of light on the human circadian rhythm, sleep, and mood is well known.⁴¹ On the other hand, few recent studies have explored the possible role of intrinsically photosensitive retinal ganglion cells (ipRGCs) in rodents and human beings that are stimulated by light even in the absence of rods and cones. These ipRGCs participate in light responses, which include circadian entrainment, pupillary light reflex, modulation of sleep/alertness, and mood.^42,43 Similarly, Flyn et al.⁴⁴ have studied patients with LP and NLP with different eye diseases and found that the etiology of blindness in addition to the LP status is related to an individual's ability to process the circadian light signal. 

We found out of 11 children with NLP, 9 had a whole-globe anomaly and two had optic nerve disease. The role of ipRGCs in the sleep pattern of children needs further exploration.⁴⁴ Sleep hygiene in the institutionalized children with blindness might have contributed to less disturbed sleep patterns in children with NLP, but results should be interpreted with caution. 

The first environmental factor that may have impacted the high sleep quality of the blind children in our study is a potentially greater level of sleep hygiene. Most blind children in our study lived in boarding schools that have strict bedtime routines and limited screen time prior to bedtime. Increased sleep hygiene is known to have a positive impact on sleep quality.^45,46 In view of the potential negative health effects of a lack of sleep, both quantitatively and qualitatively, educating parents and caregivers about the importance sleep hygiene, as also suggested by Hayton et al.,³⁴ is a relevant next step in improving the sleep of blind children, if needed. 

The second environmental factor that may have impacted the sleep quality and quantity of the children included in the study is that the current study was conducted in a low-income country (Nepal), while the Hayton et al.³⁴ and Reema et al.³⁵ studies were conducted in higher-income countries (i.e., United Kingdom and Saudi Arabia). There are known differences in culture, nutrition, physical activity levels, and pressure with regards to academic achievements among children in different geographical locations,⁴⁷ some of which are known to impact sleep.⁴⁸ Therefore, comparisons with previous studies should be made with these differences in environmental factors in mind. 

A limitation of our study was the lack of the available of sleep data for 16.5% of our participants, even though only 4% of children did not wear the ActiGraphs for the duration as set by our study design. The missing data might have been due to ActiGraphs being damaged by improper handling, a lack of instructions, or understanding. 

Despite our instruction to administer sleep diaries for 7 days, which is used by most other studies,⁴⁹ compliance was low. Only 65% of children included in the analysis had complete sleep diary information, resulting in missing latency data and other qualitative insights in 35% of children. 

In conclusion, our study shows that in a large group of Nepalese gender- and age-matched children, blind children have a lower sleep quantity and greater sleep quality compared to their normally sighted peers. However, the extent to which the level of VI, sleep hygiene, or environmental factors contributes to these differences remains unclear. Another important finding of our study is that the sleep duration of all groups of children was far shorter than the recommended sleep duration for their respective age groups. Future studies are therefore recommended to continue to use objective sleep measures and collect additional data, which may account for environmental factors that can impact sleep and why discrepancies between objective and subjective measures seem persistent. This may include determining cortisol and melatonin levels, which might point toward the mechanisms behind (dys)functional sleep patterns in blind children. Educational programs for parents, caregivers, and children regarding recommended amounts of sleep and practices of sleep hygiene are suggested to address the short sleep duration in children. 

Finally, this is the largest comparative study on sleep behavior in children using objective measurement devices. Since sleep behavior in children is less often explored, our study will help increase the awareness about the importance of sleep hygiene importance of sleep screening and potential treatment modalities of sleep disorders, including psychosocial counseling and medication. 

Acknowledgments

The authors thank Suman S. Thapa, Pradeep Banjara, and Rina Budathoki in the research department of the Tilganga Institute of Ophthalmology and all the principals, teachers, and students of integrated schools for the blind. 

Disclosure: S. Adhikari, None; R.M.A. van Nispen, None; M. Poudel, None; F. van Rens, None; E.B.M. Elsman, None; Y.D. van der Werf, None; G.H.M.B. van Rens, None