The Handan Eye Study was a population-based longitudinal epidemiologic study of eye diseases conducted in Handan, Hebei province, northern China. In 2006–2007, 6830 people aged 30 years or older participated in the baseline study, which had been described in detail elsewhere [8, 9]. In 2012–2013, participants still surviving were invited to take part in a follow-up study, which had been described in detail too [10]. The study protocol was approved by the Beijing Tongren Hospital Ethics Committee (TREC2006-22), and written informed consent was obtained from all the participants, according to the Declaration of Helsinki.

In both the baseline and follow-up studies, the participants experienced a detailed eye examination that included visual acuity (VA), intraocular pressure (IOP), fundus photographs, and so on, and a physical examination including height and weight according to a similar protocol.

Presenting VA (i.e., wearing habitual correction, if any) was measured binocularly, then monocularly (right eye followed by left eye) according to the Early Treatment Diabetic Retinopathy testing protocol with a log MAR chart under standardized lighting conditions at 4 m. For subjects who could not see the chart at 4 m, VA was tested at 1 m, allowing acuities as low as 1/40 (0.025) to be tested. If no letters were identified on the chart, VA was assessed for the ability to count fingers, see hand movements, or perceive light. Subjective refraction was performed on all the participants whose VA was worse than 0.0 log MAR (Snellen 6/6 or 20/20) by trained and certified study optometrists. Autorefractor (KR8800; Topcon, www.global.topcon.com) readings were used as the starting point, and reconfiguration of sphere, cylinder, and axis was performed until the best corrected VA (BCVA) was obtained.

Intraocular pressure (IOP) was measured using the Kowa applanation tonometer HA-2 (Kowa Company Ltd., Tokyo, Japan) in cooperative people. Those who couldn’t cooperate were measured by a Schiotz tonometer or digital palpation [11]. AL, anterior chamber depth (ACD), and lens thickness (LT) were measured using a 10 MHz A/B-mode ultrasound device (Cine Scan, www.quantel-medical.com). The degree of lenticular opacity was graded by study ophthalmologists according to LOCS III with slit lamp microscopy (Topcon SL-2 F; www.global.topc.com) after pupil dilation.

At baseline, dilated 45° digital color fundus photographs of Early Treatment Diabetic Retinopathy standard field 1 [12] (centered on the optic disc, stereoscopic) and standard field 2 (centered on the macula, non-stereoscopic) were taken for each eye by trained and certified photographers. One-third of the fundus photographs were taken using a Topcon TRC-NW6S/7S (Topcon, www.global.topcon.com) camera (during the initial stages of the baseline study). And the other two-thirds were taken using a Canon CRDGi with a 20D SLR back (Canon, www.canon.com). In the follow-up study, a Canon CR 2 with a 20D SLR back (Canon, www.ca-non.com) was used for all the participants.

Height and weight were measured according to the standard operating procedure of a protocol by certified nurses at both baseline and follow-up study [13]. In addition, a detailed, interviewer-administered questionnaire including demographic information (educational level, marital status, last year’s personal income, and so on), medical conditions (history of diabetes, hypertension, drinking, and smoking), as well as family history of eye diseases, was used.

Myopia was defined as spherical equivalent (SE, sphere + 1⁄2 cylinder) less than − 0.5 diopters (D), and high myopia as SE less than − 5.0 D. Change SE was defined as SE at follow-up study minus SE at baseline study. Fundus tessellation related to myopia was defined as having well-defined choroidal vessels that can be observed clearly around the fovea as well as the arcade vessels [1]. Parapapillary atrophy was evaluated by the method mentioned by Jonas et al. [14]. Diabetes and hypertension were defined according to self-reports from participants (previously diagnosed). Educational level was divided into seven levels: illiteracy, half-illiteracy (< 1year), primary school (1–5 years), middle school (6–8 years), high school (9–11 years), mid-technical school, and college. BMI is equal to weight divided by height squared.

If a participant had at least one eye with fundus tessellation, then he was regarded as having fundus tessellation. A new-onset fundus tessellation was defined as the absence of fundus tessellation in both eyes of a participant at baseline while the presence of fundus tessellation in at least one eye of him/her at follow-up 6 years later (Fig. 1. A and B).

Comparative follow-up of fundus photographs. A: The fundus photographs of baseline without fundus tessellation. B: The subject developed fundus tessellation in the right eye after 6 years of follow-up

Fundus photographs at baseline and follow-up study were graded by a grader (C.X.L) who had been trained and certified for myopic maculopathy grading by a retinal specialist (K.O.M.) [15] For the photographs with fundus tessellation, the grader discussed with the other two graders (B.S.W and Y.X.F., trained and certified) and conformed the results. If there were still different opinions after discussion, a retinal specialist (K.O.M) reassessed the relevant photographs and made a final decision. Lastly, the specialist checked all the fundus photographs classified as fundus tessellation to confirm the final grading result. All graders were masked to information such as the participants’ refractive error and AL to minimize bias. Initially, photographs from the baseline and follow-up study were graded separately. Subsequently, for photographs diagnosed inconsistently in fundus tessellation between baseline and follow-up, side-by-side comparisons were made between these photographs; during the process, graders were masked to whether the photographs were taken at baseline or follow-up [16]. To determine intra-grader variability, the grader (C.X.L) randomly selected the images of 200 eyes of 100 participants and graded them twice in a masked manner with an interval of 2 weeks.

Statistical analyses were performed using SAS software (version 9.1.4, SAS Institute Inc.; Cary, North Carolina). Statistical analyses included the Wilcoxon test, χ2 test, and logistic regression. The prevalence and incidence of fundus tessellation were analyzed by person (person-specific), and the corresponding 95% confidence intervals (CIs) were reported. A binary logistic regression model was used to analyze the risk factors of new-onset fundus tessellation. First, univariate correlation analyses were carried out to identify ocular and systemic parameters associated with new-onset fundus tessellation. Then, multivariate logistic regression analysis was performed, with new-onset fundus tessellation as the dependent variable and the parameter with P < 0.1 in univariate analysis as the independent variable. If the participant had fundus tessellation in both eyes, the generalized estimation model was used to correct. The SMOTE model was used to correct the unbalanced dependent variable parameter [17,18,19]. The intra-grader weighted kappa value was 0.78 for the assessment of fundus tessellation.