A Novel Model for Cornea Power Shift in Eyes Wearing Orthokeratology Lenses With Different Back Optic Zone Diameters for Myopia Control

Abstract

Purpose: The purpose of this study was to quantify the corneal power changes after wearing orthokeratology lenses of different back optic zone diameters (BOZDs) and to propose a novel 4-parameter model capable of revealing the associations between each parameter and axial length growth (ALG).

Methods: A prospective self-controlled study was conducted between June 2022 and December 2023. One eye in each subject (N = 33) was randomly assigned to wear a lens with a BOZD of either 5 mm (5 oz) or 6 mm (6 oz). Axial lengths were measured at 6 and 12 months and ALG was calculated from those measurements. Corneal topography obtained at baseline, 6 months, and 12 months after lens-wearing was analyzed, and the central flatten region is considered as the treatment zone (TZ). The power change profile was fitted with a riverbank model with parameters describing the amplitude, location shift, slope, and base. A linear mixed model was used to evaluate the association between ALG and model parameters.

Results: ALG was significantly slower in eyes wearing 5 oz lenses (0.19 ± 0.15 vs. 0.26 ± 0.15 mm, P < 0.01). TZ sizes were smaller in the 5 oz group (6.76 ± 2.21 vs. 8.07 ± 2.01 mm², P < 0.01). For relative corneal refractive powers shift (RCRPS) profiles averaged across 360 degrees, greater amplitudes (4.70 ± 1.05 vs. 4.33 ± 0.99 diopter [D], P = 0.04) and smaller location shifts (1.97 ± 0.32 vs. 2.11 ± 0.27 mm, P < 0.01) were found in the 5 oz group. For RCRPS profiles calculated in eight different meridians, strong meridional modulations were found in the parameters in each group. Moreover, significant inter-group parameter differences (P < 0.01) were only found on the temporal side. Multiple regression showed that only location shift was significantly associated with ALG (P = 0.01).

Conclusions: Eyes wearing a 5 oz lens had significantly slower ALG, compared with the eyes wearing 6 oz lens. Such difference may be attributed to the smaller location shift on the temporal side.

Orthokeratology (OK) lenses are widely used in clinics for myopia control. The reverse geometry on its backside flattens the central cornea and steepens the mid-peripheral cornea.^1,2 Such changes cause a positive relative corneal refractive powers shift (RCRPS) on the mid-peripheral cornea and lead to increased myopic defocus on the peripheral retina, which hypothetically retards the axial length growth (ALG) in children with myopia.³ It can reduce ALG by 30% to 55% compared with children wearing single-vision spectacles.^1,2 

Previous reports have indicated that OK lenses with smaller back optic zone diameters (BOZDs) have higher efficacy in myopia control.^4–9 The exact mechanism underlying this difference remains unclear, and many potential contributing factors, including treatment zone size, decentration, RCRPS profile, etc., have been proposed. Even though multiple studies have investigated how RCRPS differs between the eyes wearing lenses with different BOZDs and how such differences relate to ALG, further improvement is still needed. To address these questions, the existing analytical methods must be improved in three regards. First, a model should made to be as close to mimic the realistic situation of the condition. Some existing models are overly simplified into one parameter.¹⁰ The summed power within the central cornea area ignores the spatial distribution of the power. The monomial fitting does not describe the RCRPS profile well, with many points deviating significantly from the fitted curve.^11,12 Other methods are too complicated, using up to 14 parameters.^5,13 It is impractical for clinicians to use these many parameters to evaluate corneal power changes.⁵ Second, the positive pressure from the central cornea and the negative pressure from the peripheral tear reservoir drive epithelial cells to migrate from the cornea center toward the mid-periphery.^14–16 An improved model should link the corneal power change to this transitional process. Parameters derived from the polynomial fitting may capture this shape well but provide no relatable physiological interpretation. Third, existing analyses use a single RCRPS profile to represent changes in corneal shape after OK lens wear. However, this approach is only valid when the profiles are homogeneous along the 360-degree meridians. This homogeneity assumption may not be true due to corneal asymmetry related to the treatment zone decentration.^13,17,18 A method should also consider the meridional variance and its relationship to axial growth. 

Therefore, this study proposes a new model that intends to overcome the drawbacks of existing methods. It has four parameters to capture an RCRPS profile's base, amplitude, location shift, and slope. By having these four parameters, we hope to offer a model that may potentially characterize the actual power change occurring on the cornea with OK lenses, providing better physiological interpretation over the single-parameter or complex multi-parameter model. Moreover, this model can be applied in multiple directions to reveal meridional variations of the RCRPS profiles. This new model will better describe the RCRPS difference between lenses with different BOZDs and thus potentially attribute the difference in ALG retardation effect to specific parameters. The findings from this study will provide a valuable tool for clinicians to evaluate lenses with various designs. 

Materials and Methods

Study Design

This 1-year, randomized, double-masked, and self-controlled trial was conducted at the Optometric Center of Tianjin Eye Hospital (Tianjin, China) from June 2022 to December 2023. Corneal refractive therapy (CRT) lenses (Paragon Vision Sciences, Gilbert, AZ, USA) of two BOZDs (5 mm, 5 oz or 6 mm, 6 oz) were fitted. For each enrolled subject, based on a random number list generated by SPSS, 1 eye was randomly assigned to wear a 5 oz lens and the other to wear a 6 oz lens. The detailed process of randomization and ensuring a double-blind study was described elsewhere.¹⁹ Randomization for lens assignment (5 oz vs. 6 oz) was conducted using SPSS Statistics 25.0 (IBM, Chicago, IL, USA) after eligibility for lens handling and care was confirmed. One unmasked coordinator managed the group allocation, whereas participants, examiners, and data analysts remained masked to the study throughout. Additionally, masking was applied to the assessment of treatment zone (TZ) size, decentration, and data analysis. The study adhered to the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of Tianjin Eye Hospital (KY202219). The study was also registered as a clinical trial with https://www.chictr.org.cn/ (ChiCTR2200061048). All children and their guardians were informed of the study and signed the informed consent forms. 

Participants

The inclusion criteria were: aged between 9 and 12 years; non-cycloplegic spherical equivalent refraction from −1.00 diopter (D) to −4.00 D; corneal astigmatism ≤1.50 D; anisometropia ≤1.00 D; and best-corrected monocular visual acuity (VA) of 0.0 LogMAR or less (20/20 or better). Exclusion criteria were abnormal binocular conditions, including strabismus; contraindications to contact lens wear, including a history of ocular inflammation or infection; corneal dystrophy; and systemic and ocular conditions that might affect myopic development. Based on the effect size reported in previous studies,⁴ a significance level of 0.05, and a power of 0.8, a sample size of 30 subjects was estimated to be sufficient. 

OK Lenses Design and Fitting

The OK lenses used in this study were CRT lenses manufactured using Paragon HDS100 material. It features a 4-zone design consisting of a spherical base curve, a sigmoid return zone with a width of 1.0 mm, a tangential landing zone, and a peripheral zone. All lenses were fitted by experienced optometrists, and the details of the lens fitting procedure were described elsewhere.¹⁹ Baseline data were measured before initiating CRT treatment. The OK lens parameter was modified only when significant lens decentration was observed or an unaided VA could not reach 20/25. Follow-up visits were scheduled at 1 day, 1 week, and 1, 3, 6, 9, and 12 months after lens wear. The participants and their guardians were trained on lens application, removal, and the daily care process before dispensing the lenses. 

Refraction

Refraction was performed at baseline. First, objective refraction under non-cycloplegic conditions was measured three times with an autorefractor (ARK-1, Nidek, Aichi, Japan), and the average values were recorded. Then, subjective refraction was conducted with a maximum plus for maximum VA. The spherical equivalent (SE) was calculated as the spherical power plus one-half cylindrical power obtained from subjective refractions and was used for analysis. Corneal keratometric values, restricted to a 3-mm diameter ring, were obtained using an auto refraction keratometer (ARK-1; NIDEK, Aichi, Japan) along the flattest (flat-K) and steepest meridians (steep-K). Follow-up visits were scheduled at 1 day, 1 week, and 1 and 12 months after lens wear. 

Axial Length Measurement

Axial length was measured using an ocular biometer (IOL-Master 500, Carl Zeiss, Germany) at baseline, 6 months, and 12 months after the initial lens dispatch. Five measurements were collected at each visit, and the average values were used for analysis. The ALG was the difference between each visit and the baseline. 

 

Corneal Topography Parameters

Corneal topography was obtained with a Medmont topographer (Medmont, International Pty. Ltd., Victoria, Australia) at baseline and at every follow-up visit. The topography maps from baseline, 6 months, and 12 months were included in the analysis. The Medmont instrument was regularly calibrated every day. Children were instructed to focus on the central fixation light to ensure precise alignment along the visual axis. Children were asked to blink between examinations to keep the tear film intact. Corneal topographic maps with a vertical diameter greater than 8 mm (coverage >85%) were selected for analysis. Each map contained 32 rings, with 300 points for each ring. 

Treatment Zone Size and Decentration

Tangential power maps were used to evaluate the TZ size and decentration. A difference map was calculated by subtracting the baseline map from the topography obtained at the 6-month and 12-month visits. The central region with a power reduction of more than 0.25 D was identified, and the boundary points were fitted into an ellipse using a custom program written in the R language (Supplementary Fig. S1A). The center of the fitted ellipse was taken as the center of the TZ. The distance between the TZ center and the corneal vertex was defined as the TZ decentration. Both major and minor axes and the orientation of the ellipse were also derived. The TZ size was calculated as p * major axis length*minor axis length, in units of mm². 

Relative Corneal Refractive Powers Shift

Axial power maps were used to analyze the RCRPS. First, a corneal refractive powers shift (CRPS) map was calculated as the difference between the baseline map and the 12 months after-treatment map. Then, the value of the corneal center vertex was subtracted from all values on the CRPS map to derive the RCRPS (Supplementary Fig. S1B).²⁰ 

RCRPS Profile Fitted into the Riverbank Model

The RCRPS was fitted into the riverbank model (Figs. 1A, 1B):  

\begin{eqnarray*} RCRPS = \textit{Amplitude} \, * \, \textit{erf} \, \left( {\frac{{{\rm{x}} - \textit{LocationShift}}}{{{\rm{S}}lope*\sqrt 2 }}} \right) + Base\end{eqnarray*}

in which x is the chord distance from the cornea center and the erf is the standard Gaussian error function.²¹ The base represents the flat portion of the RCRPS profile in the central cornea. As it moves from the cornea center toward the periphery, the RCRPS profile lifts up and forms a rising bank. The location shift represents the point that the RCRPS profile reaches the half-height, with a smaller location shift indicating a rising bank closer to the cornea center (purple line in Fig. 1D, left panel) and a larger location shift indicating a rising bank far away from the cornea center (orange line Fig. 1D, left panel). Amplitude represents the bank’s peak, with a larger value indicating a higher peak (see the purple line in Fig. 1D, middle panel) and a smaller value indicating a lower peak (see the orange line in Fig. 1D, middle panel). The slope represents the steepness of the bank near the location shift, with a larger value indicating a steeper rising bank (see the purple line in Fig. 1D, right panel) and a smaller value indicating a shallower rising bank (see the orange line in Fig. 1D, right panel). 

Figure 1.

 

Riverbank model analysis. (A) Riverbank model = a: amplitude; b: location shift; c: slope; and d: base level. (B) An example of an RCRPS profile fitted into the riverbank model. (C) A power map was divided into eight sections to show the meridional variation. (D) Examples illustrating how smaller or larger values of individual parameters affect the RCRPS profile. Left panel: Location shift. Middle panel: Amplitude. Right panel: Slope. The vertical grey line represents a pupil with a 2 mm radius. The shaded area in the middle panel indicates the summed RCRPS within the pupil (Sum4).

To reveal the meridional variation, the cornea power map was divided into eight sections, and the RCRPS profile in each section was fitted into the riverbank model. Moreover, to compare with previous studies, a commonly used parameter, the power accumulated within the central 4 mm cornea area (Sum4), was also included in the analysis. 

Statistical Analysis

For descriptive purposes, the mean and standard deviations (mean ± SDs) were calculated for all parameters after checking their normality with the Shapiro-Wilk test. Differences between the groups wearing the 5 oz or 6 oz lenses were tested with paired t-tests with post hoc Bonferroni correction. The meridian variation in summed RCRPS was tested with 1-way ANOVA. A linear mixed model analyzed the associations between the ALG and the riverbank model parameters. All analyses were performed using the R programming package (version 4.3.1). Any P < 0.05 was defined as statistically significant. 

Results

A total of 36 children were enrolled with 33 (92%) completing the 1-year study. There was no significant difference in baseline age, refractive error, corneal curvature, and axial length (all P > 0.05) between the eyes that wore OK lenses with different BOZDs (see the Table). 

Table.

 

Biometric Data of Subjects at Baseline

 

Refraction

In this study, both uncorrected visual acuity (UCVA) and best-corrected visual acuity (BCVA) were measured at 1 day, 1 week, 1 month, and 1 year after the initial lens wearing. VA and subjective refraction remained within normal limits, with no signs of overcorrection or undercorrection. At the 1-month follow-up, the UCVA was stable, and measured at −0.04 ± 0.08 LogMAR in the 5 oz group and −0.06 ± 0.05 LogMAR in the 6 oz group. Similar results were observed at the 1-year follow-up, with UCVA of −0.03 ± 0.10 LogMAR in the 5 oz group and −0.05 ± 0.06 LogMAR in the 6 oz group. The 1-year subjective SE was −0.05 ± 0.58 D in the 5 oz group and 0.01 ± 0.10 D in the 6 oz group. The detailed results are provided in the Supplementary Table S1. 

When the 2 eyes of the same subjects were compared, the eyes wearing the 5 oz lens had slower ALG. The line connecting the two eyes’ ALGs pointed upward (Fig. 2A). For 6 months of treatment, the ALG was 0.10 ± 0.11 mm (5 oz) vs. 0.16 ± 0.11 mm (6 oz), P < 0.01 (Supplementary Fig. S3B). Over 12 months of treatment, the ALG was 0.19 ± 0.15 mm (5 oz) vs. 0.26 ± 0.15 mm (6 oz), P < 0.01 (Fig. 2B). 

Figure 2.

 

The 12-months axial growth and treatment zone size. (A) A paired comparison of the two eyes was made within the same subjects. Red = eyes wearing the 5 oz lens; and blue = eyes wearing the 6 oz lens. A thin grey line connects the two eyes of a single subject. (B) Mean treatment zone size and ALG values for eyes wearing the 5 oz (red) and the 6 oz lenses (blue).

Treatment Zone Size and Decentration

Following 12 months of eyes wearing a 5 oz lens demonstrated a significantly smaller TZ size (6.76 ± 2.21 mm²) when compared with that of the eyes wearing a 6 oz lens (8.07 ± 2.01 mm², P < 0.01; Supplementary Fig. S2A). Both the major axis (1.55 ± 0.24 vs. 1.68 ± 0.23 mm, P < 0.01) and minor axis (1.36 ± 0.25 vs. 1.50 ± 0.22 mm, P < 0.01) of the TZs were shorter in the eyes wearing the 5 oz lenses. However, the ratios of the minor axis length over the major axis length were similar between the two groups (0.88 ± 0.06 for 5 oz vs. 0.89 ± 0.06 for 6 oz, P = 0.26), which indicated similar TZ shapes in both groups. There was also no difference in either decentration direction (187.03 ± 50.36 for 5 oz vs. 193.10 ± 41.56 degrees for 6 oz, P = 0.55) or magnitude (0.42 ± 0.12 for 5 oz vs. 0.43 ± 0.18 mm for 6 oz, P = 0.60; Supplementary Fig. S2B). In addition, there was a significant correlation between 1-year ALG and TZ size (R² = 0.16, P < 0.01), with greater ALG associated with larger TZ size. However, the large variability in the data should be noted, particularly the data points at the far ends. The 6-month results were consistent with the 1-year findings, as detailed in Supplementary Figure S3 and Supplementary Figure S4. A significant correlation was observed between the 6-month ALG and the TZ size (R² = 0.12, P = 0.002). 

RCRPS Fitted With the Riverbank Model

As the first step of analysis, the RCRPS profile was averaged over the entire 360 degrees and fitted into the riverbank model. Three significant differences were identified. First, the eyes wearing a 5 oz lens demonstrated a significantly greater amplitude (4.70 ± 1.05 D) when compared to those wearing a 6 oz lens (4.33 ± 0.99 D, P = 0.04; Fig. 3A). Second, the location shift was smaller in those wearing the 5 oz lenses (1.97 ± 0.32 vs. 2.11 ± 0.27 mm, P < 0.01; Fig. 3B). Third, the Sum4 value was greater in the 5 oz lenses (22.67 ± 8.18 vs. 18.99 ± 7.36 D, P = 0.02; Fig. 3E). However, there was no difference in either the slope (2.24 ± 0.42 vs. 2.20 ± 0.39 D/mm, P = 0.67; Fig. 3C) or base (0.02 ± 0.19 vs. 0.02 ± 0.30 D, P = 0.98; Fig. 3D) values. 

Figure 3.

 

Paired comparison of riverbank model parameters between eyes wearing the 5 oz and the 6 oz lenses in the same subjects. (A) The amplitude. (B) The location shift. (C) The slope. (D) The base. (E) Sum4. Red = eyes wearing the 5 oz lens and blue = eyes wearing the 6 oz lens.

A mixed-effect model was applied to explore the association between ALG and the riverbank parameters. Random intercepts were allowed for individual subjects to accommodate the fact that ALGs in the two eyes from the same subject are often correlated. The analysis also included age due to its known effect on ALG. In the initial step, ALG was regressed on age and one riverbank parameter. ALG was significantly associated with location shift (P < 0.01) and Sum4 (P = 0.04) but not amplitude (P = 0.18), slope (P = 0.63), or base level (P = 0.80). In further analysis, both the location shift and Sum4, along with age, were included in the model. Only the location shift was significantly associated with ALG (P = 0.01). 

Meridional Variation in RCRPS

In the second step analysis, the power map was divided into eight sections, and the RCRPS profile was calculated and fitted into the riverbank model for each section. The parameter values were not homogeneous in all meridians. The variations on the eight meridians were significant for amplitude (P < 0.01; Fig. 4A), location shift (P < 0.01; Fig. 4B), slope (P = 0.03; Fig. 4C), and Sum4 (P < 0.01; Fig. 4D). The variance of base level along the meridians was insignificant (P = 0.31). The difference between lens types was only significant at certain meridians (see the black stars in Fig. 4). Associations between parameters and ALG were mainly significant along meridians on the temporal side (see the orange stars in Fig. 4). Note that the location shift in the temporal meridians significantly differs between lenses and shows associations with ALG. 

Figure 4.

 

The meridian variation of riverbank model parameters for eyes wearing the 5 oz and the 6 oz lenses. (A) The amplitude among meridians. (B) The location shifts among meridians. (C) The slope among meridians. (D) Sum4 among meridians. The black stars indicate significant differences between lens types: **P < 0.01, *P < 0.05; and the orange stars indicate significant association with ALG: **P < 0.01, and *P < 0.05; red = 5 oz lens, and blue = 6 oz lens.

Discussion

In this study, a child’s 2 eyes were randomly assigned to wear either a 5 oz lens or a 6 oz lens. Such a design minimized the effects of age and environmental factors and helped to reveal the subtle relationship between RCRPS profile alterations and ALG. Compared with the eyes wearing a 6 oz lens, eyes wearing a 5 oz lens had smaller amplitudes and location shifts. More critically, they had significantly smaller ALG. Our study contributed to the literature in the following two aspects. 

Existing Models

Prior studies have provided several models to quantify the corneal power profile change. The monomial fitting model (y = xⁿ, see fig. 1, panel C in Zhang et al.’s study¹²) uses only one parameter: the power exponent (n). A lower power exponent represents a higher asphericity of the TZ.¹² However, the fitting curve does not capture the shape of the raw data (blue line; see fig. 1, panel C in Zhang et al.’s study¹²). We performed the monomial fitting on our data, and the mean R² value was 0.78, much lower than the R² value provided by the riverbank model (0.996). In the random mixed model analysis, n values were indeed significantly associated with ALG. However, the percentage of variance the monomial model could explain was less than half of the amount explained by the riverbank model. 

The polynomial fitting contains multiple items (y = A + Bx + CX² + Dx³ + ..Nxⁿ). It captured the shape of the profile much better (see fig. 1 in Zhang et al.’s study⁹). However, each coefficient (A to N) was not significantly associated with the ALG and presents no physiological interpretation. Instead, the x and y values representing the quarter, half, three quarters, and peak powers were measured from the fitting curves and used in the analysis. This defeats the purpose of modeling, which is supposed to provide insight into understanding potential mechanisms. The half Xmax value in the polynomial fitting was highly correlated with the location-shift value in the riverbank model and was significantly associated with ALG.^11,13 The slope of the rising bank was not covered in previous studies. 

A more descriptive method directly measures the local peaks and troughs along the horizontal and vertical meridians of the tangential map. This approach provides an extensive list of indexes (see fig. 1 in Guo et al.’s study⁴). Significant differences were found between OK lenses with 5 mm or 6 mm BOZDs in the following parameters: TZ area, horizontal TZ diameters, vertical TZ diameters, temporal peak power, nasal peak power, nasal defocus, and nasal slopes. However, only the horizontal or vertical TZ sizes were associated with ALG,⁵ which is consistent with our findings. 

An Improved Model

The migration of the epithelial cells from the center to the periphery formed a rising bank of the reverse zone. This process is similar to the gradual formation of a riverbank, which prompted the author to apply the formula describing the riverbank to corneal topographic changes.²¹ The model captured the RCRPS profile shape well. In 528 fitted curves, 66 eyes with 8 meridians for each eye, the mean R² value was 0.996 ± 0.010. Moreover, each parameter has a specific physiological meaning, an advantage over the polynomial fit used in previous studies. In such a model, the base and amplitude inform the maximal value of the magnitude of the RCRPS profile. The location of the slope of the rising bank is more informative in describing the power distribution. 

Our models’ parameters help isolate the effect of different aspects of corneal shape change on ALG. The single exponential component in the monomial model cannot differentiate the slope from the location shift. It mixed both aspects into asphericity. Although Sum4 was associated with ALG in univariate regression, including the location shift in the multivariable regression reduced the Sum4’s effect to insignificance. Therefore, the impact of Sum4 is secondary to the location shift. Our model also provides a way to delineate the RCRPS parameters related to a specific change in the lens design. As we know more about how each parameter is associated with axial growth, such an approach leads to effective lens design. 

Meridional Variance

Our study showed that RCRPS are not homogenously distributed along the 360-degree meridians. Instead, there are significant variations of RCRPS with the peak on the nasal side meridians and trough on the temporal side meridians in both groups. Such asymmetry is associated with TZ decentration.^13,22,23 Decentration may also explain why the model parameters only differed among the two groups in certain meridians. In both groups, a temporal-inferior decentration moved the nasal side reverse zone closer to the cornea center. Such a confounding factor mitigates the difference on the nasal side. However, the temporal-inferior decentration moves the temporal side reverse zone further away from the cornea center and amplifies the difference on the temporal side. Therefore, RCRPS on the temporal side of the cornea significantly differed between the two groups, whereas the nasal side did not. Consequently, only the temporal side summed RCRPS were significantly associated with ALG, not the nasal side. When monitoring the treatment progress with lenses of different BOZDs, clinicians would give more weight to the location shift on the temporal side. However, when cornea power shifts caused by lenses differing in other features are compared, the riverbank parameters and meridians demonstrating significant differences may vary. 

Connections to Previous Studies

This study’s findings about TZ and ALG agree very well with prior research. ALG was significantly smaller in the eyes wearing the 5 oz lenses.^4–9 The TZ size was significantly smaller in eyes wearing the 5 oz lenses.^4,7–9 ALG was significantly correlated with the TZ size.¹⁸ Moreover, the decentration amplitude and direction were similar to those reported in previous studies, with most decentrations located in the temporal-inferior quadrant.^13,18 Despite a relatively small sample size, these good agreements encourage us to believe that the same findings would emerge if the riverbank models were applied to the population in previous studies. 

Limitation

This study showed differential axial length growth in children treated with different optical zones, raising concerns about potential anisometropia with continued OK lens wear. However, no participant exhibited a binocular axial length difference exceeding 0.36 mm (possibly related to anisometropia) after 1 year²⁴ At the end of the 1-year study, all children were provided with OK lenses with a 5.0 mm back optical zone diameter to reduce interocular axial growth differences during binocular development. 

Conclusions

The riverbank model describes the RCRPS profile well with four easy-to-interpret parameters. Among them, a rising bank closer to the center of the cornea (smaller location shift) is significantly associated with smaller ALG in eyes wearing the OK lenses of different BOZDs. Moreover, the location shift on the temporal side, not other sides, was significantly smaller in eyes wearing smaller BOZDs. 

Acknowledgments

The authors thank Kevin Willeford for proofreading the manuscript and providing insightful comments on the interpretation of the results. 

Supported by the Tianjin Key Medical Discipline (Specialty) Construction Project (No. TJYXZDXK-016A), the National Program on Key Research Project of China (2022YFC2404502), and the open project of Institute of Optometry and Vision Science in Nankai University (NKSGY202304, NKSGP202304, and NKSGP202302). 

Disclosure: T. Wang, None; X. Chen, None; M. Li, None; H. Bi, None; X. Yang, None; M. Sun, None; Y. Liu, None; L. Li, None; B. Zhang, None