Astigmatism is a highly prevalent refractive error globally, often occurring alongside myopia or hyperopia. Population-based studies indicate that astigmatism prevalence varies by ethnicity, with East Asian school-aged children exhibiting higher rates compared to other groups.1,2 Eyelid pressure is likely to play a role in the development of corneal astigmatism.3 In Asian populations, increased eyelid tension has been proposed as a potential factor contributing to the higher prevalence of astigmatism.4 As eyelid pressure changes with child development, the long-term progression patterns of astigmatism remains unclear.

Longitudinal studies on astigmatism progression in children have reported inconsistent findings. For instance, a cross-sectional comparison by Huynh et al. observed relative stability in Australian children,5 whereas a longitudinal study by Harvey et al. documented significant changes in Tohono O'odham Native American children, with the direction of progression influenced by age and baseline astigmatism magnitude.6 While this discrepancy could be attributed to either methodological or ethnic factors, a prospective longitudinal study by Tong et al. provides evidence for the role of ethnicity.7 Their work found that the progression of refractive astigmatism varied among Asian ethnicities, with Chinese children exhibiting the most rapid increase.7 However, the long-term progression pattern in Chinese children, a population known for its high prevalence of astigmatism,8 is not well established. Thus, extended longitudinal data are needed to define the trajectory of refractive astigmatism in this cohort.

Beyond astigmatism progression, its relationship with myopia development is yet to be elucidated. Existing longitudinal studies have produced conflicting findings. Fan et al. reported that the presence of astigmatism was accompanied by a faster myopic shift in preschool children over a 55.7-month period,9 suggesting a link in early childhood. This aligns with early evidence suggesting that infantile astigmatism may predict higher refractive error severity later in life.10 In contrast, Chan et al. found no such association in school-aged children after a 1-year follow-up,11 a duration that may be limited for capturing long-term relationships in this age group. Therefore, extended longitudinal studies in a large school-aged cohort are needed to clarify this relationship.

The incidences of astigmatism and myopia in school-age children are alarmingly high in southern China,1,12 yet the longitudinal trajectory of astigmatism and its interaction with myopia development remain poorly characterized in real-world optometric practice. To address this gap, we analyzed longitudinal optometric data from a large cohort of children in a major hospital setting, aiming to quantify long-term astigmatism changes using both clinical measures and vector notation (J₀ and J₄₅). This combined approach was chosen because clinical measures offer valuable insights into the actual clinical impact of astigmatism, while vector notation enables a precise analysis of how astigmatism changes along different orientations to provide a more complete understanding of its progression pattern. We also evaluated the impact of baseline astigmatism on SE progression, with the goal of enhancing the understanding of the interplay in long-term refractive development.

A total of 2946 patients were included in the analysis (57.3 % male; 42.7 % female). The mean age of the patients at the initial visit was 8.72 ± 1.21 years. There were 156 six-year-olds, 382 seven-year-olds, 595 eight-year-olds, 802 nine-year-olds, and 1011 ten-year-olds. All patients were 16 years old at the final visit. The mean duration of follow-up was 7.40 ± 1.23 years. Twenty nine percent of the patients had refractive evaluation at least once a year, 91.3 % had refractive evolution at least once every two years, and 97.3 % had refractive evolution at least once every 3 years. Analysis was performed on right eye data due to the strong interocular correlation of astigmatism (r = 0.927) and SE (r = 0.940, both p < 0.001).

Changes in astigmatism and its influencing factors

Table 1 displays astigmatism parameters, axis distribution, and SE at baseline and age 16. The proportion of WTR astigmatism decreased significantly from 96.1 % to 90.8 %, while oblique astigmatism increased notably from 2.8 % to 7.6 % (p < 0.001). In contrast, the change in ATR astigmatism (from 1.1 % to 1.6 %) was not statistically significant (p = 0.084) (Fig. 1). Children with low astigmatism showed a significant cylindrical power increase from 1.10 D to 1.32 D (+0.018 D/year, p < 0.001), whereas those with high astigmatism demonstrated reduction from 3.77 D to 3.35 D (−0.048 D/year, p < 0.001). Moderate astigmatism cases remained stable (−0.009 D/year, p = 0.067). Concurrently, both vector components inhibited gradual changes: J0 (with-the-rule) decreased from 0.98 to 0.92 (−0.008 D/year), while J45 (oblique) increased from 0.02 to 0.05 (+0.005 D/year) (both p < 0.001).

Table 1.

Summary of Sample Characteristics at baseline and the assessment at 16 years of age.

Variable	Initial visit	16 years of age	P value
Mean Cyl (SEM)
Low astigmatism	−1.10 (0.01)	−1.32 (0.01)	<0.001
Moderate astigmatism	−2.15 (0.01)	−2.10 (0.027)	0.067
High astigmatism	−3.77 (0.02)	−3.35 (0.03)	<0.001
Mean J0 (SEM)	0.98 (0.01)	0.92 (0.01)	<0.001
Mean J45 (SEM)	0.02 (0.01)	0.05 (0.01)	<0.001
WTR (No. %)	2831 (96.1)	2675 (90.8)	<0.001
ATR (No. %)	32 (1.1)	47 (1.6)	0.084
OBL (No. %)	83 (2.8)	224 (7.6)	<0.001
SE (SEM)	+0.04 (0.06)	−3.29 (0.07)	<0.001

SD: Standard deviation, SEM: standard error of measurement, Cyl: cylinder, WTR: with-the-rule axis, ATR: against-the-rule axis, OBL: oblique axis, SE: spherical equivalent.

Fig. 1.

Changes in astigmatism axis classification patterns between initial examination and 16-year follow-up. The proportion of with-the-rule (WTR) astigmatism decreased from 96.1 % to 90.8 %, while against-the-rule (ATR) astigmatism increased from 1.1 % to 1.6 % and oblique astigmatism increased from 2.8 % to 7.6 %.

The ANCOVA revealed significant effects of baseline astigmatism severity on Cyl and J0 slopes (Table 2, both p < 0.001). From low to high baseline astigmatism groups, the Cyl slope progressively decreased (0.018 D/y, −0.009 D/y, and −0.048 D/y, respectively), while the J0 slope showed a similar declining trend (0.007 D/y, −0.010 D/y, and −0.035 D/y, respectively). In contrast, the J45 slope remained comparable across all severity groups (p = 0.586), with values of 0.004 D/y, 0.005 D/y, and 0.006 D/y for low, moderate, and high astigmatism, respectively. Additionally, astigmatism axis (WTR, ATR, or oblique) showed no significant effect on Cyl slope (p = 0.537), with corresponding values of −0.008 D/y, 0.010 D/y, and −0.009 D/y.

Table 2.

Mean slopes of astigmatism measurements and SE (D/Y) stratified by the amount of astigmatism and SE at baseline (Mean (SEM)).

Amount of astigmatism at baseline	n	Cyl slope	J0 slope	J45 slope	SE slope
0.75 D to ≤1.50 D	1327	0.018 (0.002)	0.007 (0.002)	0.004(0.001)	−0.567 (0.008)
1.75 to <3.00 D	910	−0.009 (0.003)	−0.010 (0.002)	0.005 (0.001)	−0.529 (0.010)
≥3.00 D	709	−0.048 (0.003)	−0.035 (0.002)	0.006 (0.002)	−0.442 (0.011)
p		<0.001	<0.001	0.586	<0.001
Amount of SE at baseline
≥+3.00 D	494	−0.014 (0.004)	−0.018 (0.003)	−0.005 (0.002)	−0.489 (0.014)
+0.50 to <+3.00 D	430	−0.021 (0.004)	−0.017 (0.003)	0.006 (0.002)	−0.499 (0.014)
> −0.50 to < +0.50 D	628	−0.013 (0.003)	−0.011 (0.003)	0.004 (0.002)	−0.544 (0.012)
−3.00 to −0.50 D	1022	−0.003 (0.003)	−0.005 (0.002)	0.005 (0.001)	−0.531 (0.010)
<−3.00 D	372	0.021 (0.005)	0.006 (0.003)	0.015 (0.002)	−0.559 (0.016)
p		<0.001	<0.001	<0.001	0.002

SEM: Standard error of the mean; Cyl: cylinder; SE: spherical equivalent. A positive slope for astigmatism measurements indicates an increase in value, and a negative SE slope indicates a myopic shift.

The ANCOVA also revealed significant effects of SE on the slope of astigmatism measures (Table 2). Myopic groups had more positive Cyl slopes (−0.003 D/y to 0.021 D/y) compared to emmetropic (−0.013 D/y) and hyperopic groups (−0.014 D/y to −0.021 D/y, p ≤ 0.001). Moderate-high myopia (0.021 D/y) was more positive than low myopia (−0.003 D/y, p < 0.001). For J0, myopic groups (−0.005 D/y to 0.006 D/y) had more positive slopes than hyperopic groups (−0.017 D/y to −0.018 D/y, p ≤ 0.002). For J45, moderate-high hyperopia (−0.005 D/y) was more negative than other groups (0.004 D/y to 0.015 D/y, p ≤ 0.001).

Changes in SE and its influencing factors

SE decreased significantly from the first visit to the last visit (from +0.04 to – 3.29D, p < 0.001, Table 1). The ANCOVA revealed significant effects of baseline astigmatism magnitude on the mean SE slope (Table 2). The SE slopes for low, moderate, and high astigmatism groups were −0.567D/y, −0.529 D/y, and −0.442 D/y, respectively, showing a decreasing trend with increasing astigmatism magnitude (p < 0.001).

Additionally, SE at baseline significantly affected the SE slope, with myopic and emmetropic groups (−0.531 to −0.559 D/y) having more positive slopes than hyperopic groups (−0.489 to −0.499 D/y, p = 0.002). However, the astigmatism axis type (WTR, ATR, or oblique) did not significantly affect the SE slope, with values of −0.544 D/y, −0.546 D/y, and −0.535 D/y for WTR, ATR, and oblique, respectively (p = 0.966).

To investigate the related factors associated with slope of Cyl and SE, multivariate linear regression analysis was applied to examine age, SE and Cyl at the baseline, and gender (see Table 3). The correlation between the initial Cyl or SE and the slope of them were consistent with the above ANCOVA results. Interestingly, gender was related to the rate of change in Cyl (p < 0.001), with boys showing an increase (0.004 D/y) and girls showing a decrease (−0.017 D/y). Figs. 2 and 3 show the association between initial cylinder refraction or SE with the slope of astigmatism measurements and SE, respectively.

Table 3.

Multivariate linear regression analysis of the associations of astigmatism and SE slopes with sex, age, and Cyl/SE values at initial visit.

	Cyl slope		SE slope
Characteristics	β (95 % CI)	p	β (95 % CI)	p
Initial Cyl	−0.026 (−0.029, −0.023)	<0.001	0.047 (−0.057, −0.038)	0.001
Initial SE	−0.004 (−0.005, −0.003)	<0.001	0.008 (−0.011, −0.005)	0.023
Age	0.000 (−0.003, 0.002)	0.879	−0.044 (0.035, 0.053)	<0.001
Sex	−0.022(−0.028, −0.016)	<0.001	0.003(−0.025, 0.053)	0.776

Cyl: cylinder, SE: spherical equivalent, CI: Confidence Interval. A positive slope for astigmatism measurements indicates an increase in value, and a negative SE slope indicates a myopic shift.

Fig. 2.

Scatter plots show the relationship between cyl at the initial visit and the slopes of astigmatism measurements and SE. A positive slope for astigmatism measurements indicates an increase in value, and a negative SE slope indicates a myopic shift. Cyl: cylinder, SE: spherical equivalent. A. The relationship between the initial cylinder refraction and cylinder refraction slope (r = −0.316, p < 0.001). B. The relationship between cylinder refraction at the initial visit and J0 slope (r = 0.280, p < 0.001). C. The relationship between cylinder refraction at the initial visit and J45 slope (r = −0.012, p = 0.525). D. The relationship between cylinder refraction at the initial visit and SE slope (r = 0.173, p < 0.001).

Fig. 3.

Scatter plots show the relationship between SE at the initial visit and the slopes of astigmatism measurements and SE. A positive slope for astigmatism measurements indicates an increase in value, and a negative SE slope indicates a myopic shift. Cyl: cylinder, SE: spherical equivalent. A. The relationship between the SE at the initial visit and cylinder refraction slope (r = −0.088, p < 0.001). B. The relationship between SE at the initial visit and J0 slope (r = −0.078, p < 0.001). C. The relationship between SE at the initial visit and J45 slope (r = −0.129, p < 0.001). D. The relationship between SE at the initial visit and SE slope (r = 0.079, p < 0.001).

To our knowledge, this study represents the largest longitudinal cohort with over 7 years of follow-up investigating astigmatism dynamics in Chinese school-age children. We identified a baseline-dependent divergence: low astigmatism progressed while high astigmatism regressed over time. Notably, more negative baseline SE values predicted faster astigmatism progression, whereas higher baseline astigmatism attenuated SE changes toward myopia.

Similar to earlier studies, WTR astigmatism was predominant in the present study.8,18 which can be partly attributed to factors such as lid pressure and natural corneal anatomy.19 Leung et al. demonstrated significant shifts in astigmatism axis patterns from childhood (3–10 years) to older age (60 years) in a Hong Kong clinical population, with WTR decreasing from 92.6 % to 2.7 % and ATR increasing from 2.9 % to 79.7 %, while oblique astigmatism showed more modest changes.8 Our study tracked astigmatism axis changes during school-age years (6–10 to 16 years), revealing that WTR and oblique astigmatism changed in opposing directions with similar magnitude, whereas ATR showed minimal increase. When considered alongside Leung et al.'s findings,8 these results suggest a developmental sequence where the WTR axis first transitions toward oblique during adolescence before ultimately shifting to ATR in later adulthood.

Results on astigmatism development in children are inconsistent, with limited long-term longitudinal data. Our stratified analysis by baseline astigmatism revealed distinct progression patterns. In children with low baseline severity, we observed progression, aligning with Tong et al.'s longitudinal findings in Asian children (7–9 years)7 but contrasting with the cross-sectional stability described by Huynh et al. in Australian children (6–12 years),5 both of which primarily examined low astigmatism, indicating the role of ethnic differences. The progression magnitude in our cohort was somewhat smaller than that of Tong et al. (0.018 D/year vs. 0.03 D/year), a difference potentially explained by the older age of our participants. Children with high baseline astigmatism in our cohort exhibited reduction, diverging from the progression observed by Sherrill et al. in younger American children (3–11 years),6 a difference that may reflect both ethnic and age-related factors. Children with moderate astigmatism, however, remained stable throughout the observation period.

Beyond baseline astigmatism magnitude, gender may also influence astigmatism progression. Previous studies have reported inconsistent findings regarding gender differences: some found higher astigmatism prevalence in girls,18,20 others in boys,21,22 while several reported no significant differences.23,24 In our cohort, the gender distribution was equal, and we observed that astigmatism decreased in girls but increased in boys. These findings align with Tong et al.'s report of higher 3-year cumulative incidence in boys,7 as well as Otabor et al.'s finding that males exhibit significantly higher mean cylindrical power,25 but contrast with Zhao et al.'s multivariate analysis showing no gender association.26 Further research is needed to clarify these gender-related patterns.

Additionally, myopic children were found to had increased astigmatism, while nonmyopic children had reduced astigmatism in the current study. Nevertheless, J0 and J45 also tended to progress more in children with higher myopic SE. These findings align with Tong et al.'s report of five-fold higher astigmatism incidence in myopic versus non-myopic children,7 and with recent evidence of accelerated astigmatism and myopia progression during COVID-19 home confinement.27,28 Collectively, these results support a consistent association between myopia progression and astigmatism development."

An important objective of this study was to evaluate the effects of basic refraction, especially astigmatism, on the development of spherical equivalent. A greater change in SE was observed in children with myopia compared to those with hyperopia, which implied that the degree of myopia progression in children was greater than the degree of hyperopia regression. This is consistent with previous studies reporting that myopic children have a greater myopic shift than those without myopia.9,29 However, a higher magnitude of astigmatism at baseline was associated with less myopia progression in the current study, which seems to contradict previous studies. Fulton et al.30 found that younger children (< 3 years) with astigmatism tended to have increased myopia. Fan et al.9 also reported that the presence of astigmatism, particularly with increasing astigmatism, appeared to predispose preschool children to progressive myopia. Therefore, astigmatism is thought to blur visual images and thus may promote the development of myopia. It is important to note that all of these studies involved preschoolers.

Our findings demonstrate that higher baseline astigmatism is associated with slower myopic progression in school-aged children, consistent with Chan et al. who reported greater myopic shifts in children without astigmatism.11 While the precise mechanisms remain unclear, animal studies offer potential biological explanations. Vyas and Kee found that optically imposed astigmatism in chickens resulted in less myopia development compared to spherical lenses alone.31 They proposed that astigmatic blur may activate differential growth regulation mechanisms, potentially redirecting ocular growth toward the least myopic image plane. Although our human clinical data cannot directly confirm these biological mechanisms, the observed association between higher baseline astigmatism and slower myopia progression warrants further investigation. Future studies combining precise optical measurements with controlled interventions should explore whether similar biological mechanisms operate in human children, and how these mechanisms might interact with known optical drivers of myopia like retinal defocus.

It is important to note that, although these trends were statistically significant, their clinical relevance remains limited. The annual progression of astigmatic components was very small in magnitude. The total average astigmatism change did not exceed 0.50 D, and the protective effect of high astigmatism on myopia progression was <0.25 D per year. Both values fall below conventional thresholds for clinical significance. Therefore, while these findings do not provide direct clinical guidance, they offer valuable insights the longitudinal progression of astigmatism and its relationship with spherical equivalent development.

This study has several limitations. First, although the follow-up period was long, the retrospective design and irregular follow-up intervals limited our ability to assess age-specific differences in astigmatic progression. Second, the sample was drawn from optometry clinics, where astigmatism levels tend to be higher than in the general population, potentially affecting generalizability. Third, the absence of objective biometric data—such as corneal curvature and axial length—precluded deeper analysis of the structural factors underlying astigmatic changes.

In summary, this longitudinal analysis offers exploratory insights into the progression of astigmatism and its interaction with SE change in Chinese school-age children. We found that lower baseline levels tended to worsen, while higher levels generally improved. Coexisting myopia was associated with accelerated astigmatism progression, and conversely, higher astigmatism showed a modest protective effect against myopia development. It should be emphasized that all observed effects were small in magnitude, indicating limited immediate clinical applicability.

This work was supported by Wenzhou Municipal Science and Technology Bureau (Grant number Y20210974), the National Key Research and Development Program of China (2023YFC3604001) and the Zhejiang Provincial Health Science and Technology Program (2025HY0598).