Study procedure

All participants had their presenting visual acuity (PVA) tested using an Early Treatment Diabetic Retinopathy Study vision chart while wearing their habitual refractive correction. Best-corrected visual acuity was measured for those with PVA ≤20/40 in either eye. Presenting VI (PVI) was defined as PVA less than 20/40 in the better-seeing eye. The IOP was measured before mydriasis by a handheld tonometer (Tonopen; Mentor, Norwell, Massachusetts, USA) with three consecutive measurements of an achieved SE of <5%. Central cornea thickness (CCT) was evaluated using an ultrasound pachymetry (Echoscan US1800; Nidek, Corp). Height and weight were measured without shoes, using a standard calibrated scale. Body mass index (BMI) was calculated as the weight in kilograms divided by the square of the height in centimetres and was divided into three groups: underweight (BMI <18.5 kg/m²), normal to overweight (18.5 to 30 kg/m²) or obese (BMI ≥30.0 kg/m²). Diabetes mellitus and hypertension were based on self-reported history of a diagnosis and/or previous medication use.

Slit-lamp examination (TopconSL-8Z, Tokyo, Japan) with a 78-diopter lens was used to identify abnormalities of the anterior segment and posterior segment by an experienced ophthalmologist (MH). Detailed information of the gonioscopic examination in the Liwan Eye Study has been described previously.²² Briefly, all participants underwent slit lamp-based static and dynamic gonioscopy with a Goldmann-type, one-mirror lens (Haag Streit, Bern, Switzerland) at 25× magnification by the same experienced specialist-trained ophthalmologist (MH). Narrow angle and open angle were stratified by status of the iris insertion and recorded using five categories by the Shaffer system.²³According to the International Society of Geographical and Epidemiological Ophthalmology (ISGEO) classification, PAC suspect (PACS) was defined as simply an angle in which ≥270°of the pigmented trabecular meshwork cannot be seen without evidence of trabecular obstruction and glaucomatous damage. PAC was defined as eyes with PACS and features of peripheral anterior synechiae, elevated IOP, iris wholing or excessive pigment deposition on the trabecular surface, but no evidence of glaucomatous damage. PAC glaucoma (PACG) was defined as eyes with PAC and evidence of glaucomatous damage. Participants with PACS, PAC or PACG were grouped as PACD.

The optic disc was assessed using a 78-D lens at 16× magnification. The VCDR was used as key indicator of structural glaucomatous change. Visual field (VF) assessment was performed in those with a VCDR of >0.7 (97.5th percentile of the Liwan Eye Study) in either eye, VCDR asymmetry >0.2 or IOP of >21 mm Hg on a subsequent day. The definition of glaucoma was based on three levels of evidence using ISGEO criteria. The division of POAG and PACG was based on the gonioscopic results of narrow angle or open angle. If glaucoma status or VCDR was observed in both eyes, the eye with more severe status or larger VCDR value was used in the analysis.

Detailed data from the Chinese Centre for Disease Control and Prevention (CDC) were used to confirm the mortality of participants during the 10-year follow-up period. After providing the CDC with a list of names, age, year of birth, gender and latest address for the participants suspected of having passed away, based on which researchers at the CDC provided a corresponding list of ‘matched’ deaths with dates and causes. The causes of death recorded by the CDC were documented on the death certificates using the International Classification of Diseases, Ninth Revision.