Sleep disorders affect over 1 billion people worldwide (1). Poor sleep quality is linked to cardiovascular diseases, metabolic disorders, cognitive impairments, and increased mortality risk, thus constituting an integral component of holistic health (2, 3). Individuals with visual impairment experience sleep disturbances at rates 2–5 times higher than the general population (4, 5). This problem is particularly pronounced in patients with diabetes mellitus (DM). These patients face a dual challenge: high prevalence of visual impairment (6), affecting 21.7% of patients due to diabetic retinopathy and other complications, and high prevalence of sleep disorders (7), affecting 55-86% of patients, including obstructive sleep apnea and insomnia. However, despite the clinical significance of this relationship, a critical research gap remains. No studies have systematically examined the associations between visual acuity (VA), sleep quality, and sleep duration in diabetic populations while considering diabetic retinopathy staging as a potential modifying factor (8).

Visual impairment may affect sleep quality through reduced light input disrupting circadian rhythms (9) and dysfunction of photoreceptor cells including intrinsically photosensitive retinal ganglion cells (ipRGCs) (10). Few studies have examined the relationship between visual acuity (VA) and sleep quality (11–14). Most existing research has focused on patients with visual impairment (VI), overlooking individuals with VA decline that does not meet clinical VI thresholds. Furthermore, the effects of varying degrees of VA decline on sleep quality remain poorly understood.

The Fushun Diabetic Retinopathy Cohort Study (FS-DIRECT) (15) offers a unique opportunity to bridge existing research gaps. We specifically focused on patients with type 2 diabetes as this population has a higher prevalence of both visual impairment and sleep disturbances. We aimed to examine whether visual acuity affects sleep quality even during mild visual decline before reaching clinical visual impairment thresholds, and whether the association between VA and sleep parameters in diabetic populations is independent of diabetic retinopathy status.

We hypothesized that (1): decreased visual acuity would be associated with poorer sleep quality and altered sleep duration, with this relationship being non-linear and potentially exhibiting threshold effects; (2) the association would be modified by diabetic retinopathy severity; and (3) even mild visual decline not meeting clinical visual impairment criteria would significantly impact sleep parameters.

In our investigation, we observed that diminished VA correlates significantly with poorer sleep quality, particularly affecting aspects such as sleep latency, sleep disturbances, and daytime dysfunction. Notably, no significant association was discerned between VA and sleep duration, whether short or long.

These findings align with several previous studies. Earlier research has shown a connection between VI and poor sleep quality (12, 14), specifically affecting sleep latency (25), premature morning arousals, sleep disturbances (25, 26), and daytime dysfunction (25). However, it is pertinent to note that these studies predominantly relied on self-reported VI or utilized non-standardized sleep questionnaires. Additionally, many failed to adjust for potential confounding variables, which could influence the observed associations.

A contributing factor to sleep disturbances associated with declined VA could be reduced light input, often resulting from ocular diseases like cataracts. Light is instrumental in the synchronization of homeostatic and circadian rhythms, crucial for the optimization of sleep patterns (9). The reduced light perception could compromise the adaptability of the suprachiasmatic nucleus (SCN) to the diurnal rhythm (27, 28). The SCN regulates the phase shift and secretion levels of melatonin, influencing the sleep-wake cycle. This maladaptation precipitates a desynchronization of sleep-wake cycles, culminating in compromised sleep quality.

Additionally, suboptimal sleep quality in individuals with diminished VA might also stem from dysfunctional photoreceptor cells and associated mechanisms. The rod-cone photoreceptor system, in conjunction with the melanopsin system, plays a pivotal role in mediating the effects of light on sleep (10, 29). This mediation primarily occurs through the intermediation of intrinsically photosensitive retinal ganglion cells (ipRGCs) (10, 29, 30). Ocular pathologies, including DR (31), glaucoma (32), age-related macular degeneration (AMD) (33), and optic disc disorders, can markedly impair sleep quality by detrimentally affecting these cells. Furthermore, the increased likelihood of poor sleep quality continued to be evident in patients with diminished VA, even after adjusting for these ocular conditions. This finding firstly substantiates the robustness of our results. Secondly, it suggests that factors beyond photoreceptor cell damage contribute to the poor sleep quality associated with diminished VA.

VI may affect sleep quality through several pathways. These include reduced outdoor activity (34), increased vigilance due to safety concerns (35, 36), and potentially greater exposure to sleep-disrupting blue light from adaptive devices (37).

In our research, we found a J-shaped relationship between VA and sleep quality. The lowest point of this curve occurred at approximately 0.5 LogMAR. The probability of poor sleep quality increased as VA worsened, up to this point. Beyond 0.5 LogMAR, the probability plateaued. Our study is the first to report this J-shaped association. Currently, no existing studies are available for direct comparison. Even a minor decline in VA can negatively influence sleep quality, even before reaching the threshold of clinical VI.

Specifically, patients with VA ranging from 0.1-0.2 LogMAR demonstrated a 1.7-fold increased likelihood of poor sleep quality compared to those with normal vision. Importantly, this identifies a pre-clinical threshold where mild visual decline compromises sleep quality before meeting formal visual impairment criteria. Our findings have important implications for clinical practice in ophthalmology and optometry. They suggest the need for a more integrated approach to patient care. Eye care professionals should consider screening for sleep issues in high-risk patients with visual impairments. This could facilitate timely referrals when necessary. A proactive strategy would include early interventions such as sleep hygiene education, phototherapy, and referrals to sleep specialists. This approach aligns with the growing focus on preventive care in healthcare systems. Patient education about how visual impairment affects sleep quality offers additional benefits. It may improve adherence to vision care treatments and encourage sleep-enhancing behaviors. Our findings also emphasize the importance of collaboration between eye care professionals and sleep specialists. This is especially valuable during the early stages of vision decline. Previous research supports this integrated approach. A study by Smith et al. (38) showed improved sleep quality in cataract patients after intraocular lens replacement. This highlights the potential benefits of timely intervention for treatable eye conditions.

Our findings also suggest a critical threshold at approximately 0.5 LogMAR, marking the clinical demarcation between non-VI and VI. The 0.5 LogMAR threshold represents a critical functional boundary where individuals typically begin experiencing significant difficulties with daily activities. Beyond this threshold, sleep quality in individuals tends to stabilize across the VA spectrum.

The loss of association beyond the 0.5 LogMAR threshold suggests a fundamental shift in sleep regulation mechanisms. This threshold pattern suggests different underlying mechanisms across the visual acuity spectrum. We propose that individuals without VI may primarily regulate their circadian rhythms through light perception via the eyes. In contrast, those with VI may rely more on non-visual sensory input.

Specifically, for individuals with better visual acuity, circadian rhythm regulation appears to depend primarily on the traditional rod-cone photoreceptor system, where progressive visual decline leads to proportional sleep quality deterioration. However, beyond the 0.5 LogMAR threshold, a fundamental shift occurs toward alternative regulatory pathways. Patients with VI may transition to greater reliance on the intrinsically photosensitive retinal ganglion cell-melanopsin pathway for circadian photoentrainment, which maintains basic rhythm regulation even in profound visual loss (39, 40). Additionally, non-visual sensory compensation through auditory (41) and thermal (42) environmental cues may partially replace diminished visual input. However, we don’t fully understand how the rod-cone and melanopsin systems work together at different levels of visual acuity. This gap in knowledge highlights the necessity for more comprehensive experimental research to corroborate these hypotheses. Given the dramatically varying effects of light therapy on sleep across individuals (43), gaining a deeper insight into this interplay is also crucial for refining the application of phototherapeutic interventions. Sleep interventions for people with VI might therefore need to differ from those with mild VI.

While our findings suggest that visual acuity may influence sleep quality, we must also consider the possibility of reverse or bidirectional causality. Poor sleep quality could potentially affect visual function through several biological mechanisms. Sleep deprivation and poor sleep quality increase oxidative stress and inflammatory markers such as C-reactive protein and interleukins, which may contribute to retinal damage over time (44). Chronic sleep disturbances have been associated with impaired vascular regulation and increased risk of microvascular complications (8), which could adversely affect the retinal microvasculature. The cross-sectional nature of our study prevents definitive causal inference, and longitudinal studies are necessary to establish the temporal relationship between visual decline and sleep disturbances.

Findings regarding the correlation between VA and sleep duration present a contentious landscape. The Korea National Health and Nutrition Examination Survey (KNHNES) (45) posited that VI correlates with both short and long sleep durations. However, Peltzer et al. (14) reported no significant association between these variables. Our findings support the latter conclusion. We found no relationship between VA and sleep duration when analyzing VA as a segmented variable. Such findings imply that deep sleep duration, rather than total sleep duration, may more significantly influence sleep quality. Relying solely on sleep duration as a metric is insufficient for assessing overall sleep health.

Several methodological strengths enhance the reliability of our findings. Our large sample size enabled comprehensive analysis across the visual acuity spectrum in real-world settings, supporting sophisticated statistical approaches including restricted cubic spline modeling and multiple sensitivity analyses. Importantly, we incorporated depression adjustment based on emerging evidence of its mediating role in sleep-health relationships, ensuring more accurate estimation of the direct association between visual acuity and sleep quality (46).

Our study has several limitations. First, its observational nature prevents establishing causal relationships. Second, we did not screen for conditions such as obstructive sleep apnea, REM sleep disorders, or measure melatonin levels. This limited our comprehensive assessment of sleep health. Moreover, sleep duration was assessed using PSQI Component 3, which relies on a single-item self-report measure susceptible to recall bias and may not capture night-to-night variability in sleep patterns. Third, we excluded individuals without VA records, which could introduce selection bias. We chose not to use multiple imputation for these missing data because individuals without VA tests often have poorer vision. This violates the “missing at random” assumption required for imputation. We cannot rule out unmeasured confounding factors such as socioeconomic conditions and genetic variation. However, our homogeneous study population and extensive data on known risk factors likely reduced potential confounding. Finally, our focus on patients with diabetes mellitus limits the generalizability of our results to broader populations.

In conclusion, our study elucidates a J-shaped association between VA and sleep quality within a substantial community-based population. We observed that the risk of poor sleep quality commences even with a slight decline in VA. Notably, the positive relationship between VA and sleep quality diminished in individuals with VI. This insight into the independent role of VA on sleep is pivotal, offering significant implications for demystifying the intricate interplay among ambient light exposure, photoreceptor cell functioning, and the regulation of sleep.