Prevalence of Primary Open Angle Glaucoma in a Rural Adult Chinese Population: The Handan Eye Study

Abstract

Purpose.: To estimate the prevalence and associations of primary open angle glaucoma (POAG) in a rural population of northern China.

Methods.: In a rural county in Handan, China, 6716 adults residing in 13 villages were randomly selected and participated in the study. All participants completed a comprehensive eye examination, including intraocular pressure (IOP), slit-lamp examination, and fundus evaluation. Visual fields were obtained with on glaucoma suspects.

Results.: Sixty-five persons (1.2%) had POAG, with an adjusted prevalence of 1.0% (95% confidence interval [CI], 0.7% −1.3%) in those aged 40 years and older. Sex was not significantly associated with POAG (P > 0.05). Age (odds ratio [OR], 1.9; 95% CI, 1.4–2.5, for each 10-year increase), IOP (OR, 1.5; 95% CI 1.2–2.0 for each 5-mm Hg increase), axial length (OR, 1.3; 95% CI, 1.1–1.6), and moderate myopia (3.1–6.0 D; OR, 4.7; 95% CI, 1.6–13.5) increased the risk for POAG in multivariate analysis. The mean IOP of persons with POAG was 16.3 ± 3.5 mm Hg, and 90% of them presented with an IOP ≤21 mm Hg. Of those with POAG, 4.5% were blind from glaucoma in at least one eye.

Conclusions.: Approximately 1% of adults aged 40 years and older living in rural China have POAG. As seen in other populations, increasing age, higher IOP, greater axial length, and having myopia were associated with POAG. Given the rapid aging and myopic shift (acquired myopia) in China's population, POAG is likely to increase in prevalence in the coming decades.

Although it was once believed that primary open-angle glaucoma (POAG) was rare among Chinese persons, ¹ two recent urban surveys reported its prevalence to be similar to that seen in persons of European descent. He et al. ² examined 1504 subjects aged 50 years and older in Liwan District, Guangzhou, and found a 2.1% prevalence rate of POAG. ² Xu et al. ³ examined 4451 subjects from Beijing urban and suburban areas aged 40 years and older and reported a prevalence of POAG of 1.7%. Similar rates were reported among Chinese Singaporeans. ⁴  

More than half the Chinese population lives in the countryside. To improve our understanding of the burden of eye diseases on rural residents of China, we conducted the Handan Eye Study (HES). ⁵ This article specifically addresses the age- and sex-specific prevalence and associations of POAG in this rural population. 

Methods

Study Design

The study population consisted of self-identified Han Chinese, 30 years or older, living in a rural county of Handan, China. Details of the study design, sampling plan, and baseline data are reported elsewhere. ^5,6 In brief, the study adhered to the Declaration of Helsinki, ethics approval was obtained from the Beijing Tongren Hospital Ethical Committee, and written informed consent was obtained from all participants. Residents of Yongnian County, Handan, Hebei province, aged 30 years or older were randomly selected using a clustered sampling technique with probabilities proportionate to the size of population in each cluster. In Yongnian County, 90% of the population is farmers, and 98% is Han. Per capita annual net income in this rural area is 3468 Yuan ($510), which is similar to the average income (3255 Yuan; $478) of those living in rural areas throughout mainland China. ⁷  

Clinical Data

All eligible persons were invited and scheduled for a detailed eye examination and questionnaire interview, which were performed in a standardized manner at the HES centralized clinic (located in the Yongnian County Hospital). The examination included visual acuity, autorefraction, and subjective refraction; questionnaires, anterior segment examination with a slit lamp; blood pressure, height, weight, waist-hip ratio, intraocular pressure (IOP), gonioscopy, and dilated examination and photography of the fundus including sequential stereoscopic optic nerve photographs. Other tests carried out in the clinic have been described in detail previously. ⁵ Subjects who did not present for a complete examination at the central clinic had all the tests described above performed at a clinic that was set up in the local community for this purpose. 

Optic Nerve Evaluation

The optic nerve was evaluated using a 78-diopter (D) or 90-D lens at 16× magnification after pupil dilation. The vertical cup-to-disc ratio (VCDR) was used as the key index of structural glaucomatous change. Measurement of VCDR excluded peripapillary atrophy and the scleral ring of Elschnig. The margins of the cup were defined by stereoscopic view as the point of maximum inflection of the vessels crossing the neuroretinal rim. Standard photographs for VCDR from 0.1 to 1.0 in 0.1 increments were used in the grading process. ² Although the final diagnosis was based on fundus photographs of the optic nerve, the clinical examination of the optic nerve head was one factor used to determine whether a subject was glaucoma suspect and required additional testing (including gonioscopy and visual field testing). 

Those with any of the following conditions were identified as glaucoma suspect: IOP >21 mm Hg, VCDR ≥0.6 (95th percentile of the HES population), VCDR asymmetry ≥0.2, optic disc hemorrhage, visible retinal nerve fiber layer defect, deposits at the pupil margin consistent with pseudoexfoliation syndrome, or pigment deposition on the cornea consistent with pigment dispersion syndrome. Everyone who was glaucoma suspect was asked to come to the central clinic from August 2007 to October 2007 for a definitive examination, including visual field testing and gonioscopy. 

Stereoscopic optic disc photographs were evaluated by three glaucoma specialists (YBL; YJ and TY [see Acknowledgments]) using a stereoscopic viewer (Screen-Vu Stereoscope; PS Manufacturing, Portland, OR). The status of the optic nerve was categorized as “definite glaucoma,” “probable glaucoma,” “possible glaucoma,” and “not glaucoma.” VCDR, notching of the neural rim, localized or diffuse loss of the neural rim, presence of neural rim tissue ≤0.1, and presence of a nerve fiber layer defect were all documented. Those felt to have definite glaucoma, probable glaucoma, and possible glaucoma by either of the specialists were presented to a panel of glaucoma specialists to make the determination. 

Visual Field Evaluation

Every 10th person examined as part of the HES was systematically sampled to undergo a visual field test using the 24–2 Swedish Interactive Testing Algorithm (SITA) fast program with a Humphrey visual field analyzer (750i; Carl Zeiss, Jena, Germany). In addition, all with suspect glaucoma or angle closure glaucoma underwent SITA Standard visual field tests. 

The visual field was retested 20 minutes later if the glaucoma hemifield test result was outside the normal limits or was borderline or if the test was unreliable (i.e., fixation loses >20%, false positives >33%, or false negatives >33%). 

Final Glaucoma Diagnosis

Three senior glaucoma specialists from China (NLW; ZR and ML [see Acknowledgments]) reviewed disc photographs/clinical records for vertical cup/disc ratios and visual fields and categorized subjects as having definite, probable, possible, or no glaucoma based on consensus. 

A second, independent review of the findings was carried out by a glaucoma specialist (DSF), who also classified the patients according to the same definitions. If the two classifications differed, a final independent adjudication was conducted by another glaucoma specialist (HDJ [see Acknowledgments]), and this final assignment was used to determine who had glaucoma. Those graded with definite or probable glaucoma using the review described were ultimately classified as having glaucoma. ^{8 –11} Glaucoma was also diagnosed as present in persons in whom the optic nerve was not visible because of media opacity, visual acuity was <20/400, and IOP was >99.5th percentile; or in whom visual acuity was <20/400, and the eye had evidence of previous glaucoma filtering surgery; or if medical records were available confirming glaucomatous damage. 

POAG subjects had to have no secondary cause of glaucoma and open angles on gonioscopy. Open angles were defined as present when the posterior trabecular meshwork was visible in at least 180° of the angle with static gonioscopy and no peripheral anterior synechiae were found with indentation gonioscopy using either a Goldmann gonio-lens or a Zeiss 4-mirror lens. For subjects with unilateral pseudophakia or aphakia, the angle of the contralateral eye was used to define the angle status. For those with bilateral pseudophakia, aphakia, or both, we could not classify the angle status of the eye; it was therefore defined as underdetermined. ¹²  

Assessment and Definitions of Risk Factors

A detailed interviewer-administered questionnaire was used that included information on occupation, marital status, education level, smoking (current/past /never), alcohol use (current/past/never), whether the participant had previously been diagnosed with diabetes mellitus, hypertension, stroke or heart disease, and the numbers of family members previously diagnosed with glaucoma. For this analysis, educational level was categorized as illiterate and literate (less than 1 year of formal education), primary school level (1–6 years of education), middle school level (7–9 years of education), and high school and above (10 years or more). Marital status was classified as two categories: single or widowed, and married. 

Hypertension was defined as systolic blood pressure ≥140 mm Hg or diastolic blood pressure ≥90 mm Hg at the time of the examination or a previous physician diagnosis. Blood pressure was taken with the participant seated and after 5 minutes of rest by using a digital automatic blood pressure monitor (OMRON model Hem-907 automated oscillometric sphygmomanometer; OMRON, Tokyo, Japan) according to a protocol similar to that used in the Multi-Ethnic Study of Atherosclerosis. ¹³ Blood pressure and heart rate were measured twice, 5 minutes apart. A third measurement was made if the systolic blood pressure differed by >10 mm Hg or the diastolic by >5 mm Hg. The mean between the two closest readings was then taken as the blood pressure of that person. The same calculation was used for heart rate. 

Diabetes was defined in participants with a known history of diabetes (known diabetes) or fasting plasma glucose ≥7.0 mM at examination for the HES (newly diagnosed diabetes). ¹⁴ Anthropometric measurements including weight, height, and waist measurements were obtained using standardized techniques. The body mass index (BMI) was calculated as weight divided by height squared (kg/m²), and waist/hip ratio (WHR) was calculated as waist circumference divided by hip circumference. 

Central corneal thickness (CCT) was measured using an ultrasound pachymeter (UP1000; Nidek, Inc., Tokyo, Japan). Axial length was measured using a 10-MHz A/B-mode ultrasound device (Cine Scan; Quantel Medical, Paris, France), using a hard-tipped, corneal contact probe mounted on a slit lamp. All measurements had standard deviations of anterior chamber depth (ACD) <0.13 mm. 

Refraction was measured using an autorefractor (KR8800; Topcon, Tokyo, Japan). ¹⁵ Myopia was defined as spherical equivalence more myopic than −0.5 diopter. ¹⁵ Myopia was categorized as −0.51 to 3.0 D, −3.1 to −6.0 D, −6.1 to −8.0 D, and −8.1 D and above. For analysis, we used the mean value of the two eyes. ¹⁶  

For comparisons with other studies using the definition developed by the International Society for Geographical and Epidemiological Ophthalmology (ISGEO), we also present the prevalence of POAG using the ISGEO definition. ¹⁷ Briefly, the highest level of evidence requires optic disc and visual field evidence (VCDR or asymmetry ≥97.5th percentile and reproducible glaucomatous field defect). In the second category, if the visual field test could not be performed satisfactorily, a severely damaged disc (VCDR or asymmetry ≥99.5th percentile) was considered compatible with glaucoma. The third level of evidence specifies that if the optic discs cannot be examined because of severe media opacity, subjects who are blind (corrected visual acuity <0.05) and have undergone previous glaucoma surgery or have very high intraocular pressure are classified as having glaucoma. ^2,12  

Statistical Analysis

The prevalence of POAG was calculated as the ratio of the number of persons with POAG to the total number of persons who were examined in the central clinical or temporal village clinic, and the age-specific rates were used to standardize to the Chinese national population census of 2000. If a person had glaucoma in at least one eye, that person was defined as having glaucoma. Blindness was defined as best-corrected visual acuity <0.05. Low vision was defined as best-corrected visual acuity ≥0.05 but <0.3. Visual field criteria for low vision and blindness were not used in this study. Univariate analysis was performed, adjusted by age and sex. Those factors found to be associated at P < 0.05 in the univariate analysis were then added to the multivariate analysis. Statistical significance was present for P < 0.05. Statistical analysis was carried out using predictive analysis software (SPSS version 14; SPSS, Inc., Chicago, IL). 

Results

We excluded 114 subjects who were examined at home because of the limitations of domiciliary examinations in determining visual impairment. Included in this study were 6716 subjects who were examined in a centralized or village clinic and for whom data on glaucoma diagnosis were available. Of these, 6534 persons (97.3%) had VCDR records in both eyes, and 112 (1.5%) had VCDR data in only one eye. In addition, 6424 (95.7%) had disc photograph in both eyes, and 113 (1.5%) had disc photographs in one eye. Eleven percent of the disc images were of poor quality (11.2% in right eyes, 10.8% in left eyes) largely because of cataract; 6411 subjects (95.5%) had IOP measurements by Goldmann applanation tonometry. 

The correlation of VCDR between the two eyes was 0.81 (P < 0.001). Therefore, we present the right eye data on the distribution of VCDR. The mean ± SD was 0.41 ± 0.11, the median was 0.40, the 97.5th percentile was 0.70, and the 99.5th percentile was 0.80. The 97.5th percentile of asymmetry of cup/disc ratio was 0.2, and the 99.5th percentile was 0.3. 

The correlation of IOP between the eyes was 0.83 (P < 0.001). Using right eye data only, the mean IOP of the right eye was 15.0 mm Hg, the 97.5th percentile was 20.3 mm Hg, and the 99.5th percentile was 23.7 mm Hg. 

Of the 1465 subjects who were classified as glaucoma suspect based on the initial screening, 1166 (79.6%) attended a second visit to undergo a definitive examination for glaucoma, including visual field testing and gonioscopy. Those who returned for the second visit were slightly younger (P = 0.055) and had a higher rate of history of hypertension (P < 0.05). They also had a larger VCDR at the initial visit (P = 0.004) and higher IOP (P < 0.05). No difference was found regarding sex, education, marital status, corneal thickness, axial length, history of diabetes, and history of glaucoma (P > 0.05; Table 1). 

Table 1.

 

Characteristics of Participants and Nonparticipants at the Second Visit

Characteristics	Participants (n = 1166)	Nonparticipants (n = 299)	P *
Age, y	54.4 ± 10.8	56.1 ± 13.3	0.055
Male	543	146	0.263
Female	623	153
Education
<1 y	185	46	0.825
1–6 y	590	160
7–9 y	352	83
≥10 y	39	10
Marital status
Single/widow	111	36	0.196
Married	1055	263
Hypertension
Absent	509	151	0.034
Present	657	148
Diabetes
Absent	988	206	0.861
Present	87	19
History of glaucoma
Absent	988	288	0.933
Present	87	10
IOP†	15.6 ± 3.6	15.0 ± 3.4	0.029
CCT, μm‡	534.9 ± 32.4	535.1 ± 32.9	0.947
Axial length, mm‡	22.8 ± 0.9	22.7 ± 0.8	0.392
VCDR§	0.6 ± 0.1	0.5 ± 0.1	0.004

 BP, blood pressure.

*   t-test or chi-square test.

†  Used the data in the eye with higher IOP.

‡  Used the mean values of the two eyes.

§  Used data in the eye with higher VCDR.

Table 2 summarizes the completeness of the data for glaucoma diagnosis (only data from right eye are presented). Approximately 30% of the subjects who were glaucoma suspect had unreliable visual fields or no visual fields. However, approximately 75% of these subjects who had unreliable or no visual field had a clear optic nerve image for review, and another 20% of these subjects underwent clinical optic disc evaluation. Less than 5% of study subjects had no optic nerve information or useful visual field information. 

Table 2.

 

Completeness of Data for Classification in the Handan Eye Study

Disc Image Quality	Clinical Disc Evaluation	Reliable Visual Fields				Total n (%)
		1 Visual Field	≥2 Visual Field	Unreliable	No
Very good	Available	320	344	79	121	868 (59.2)
	NA	1	1	2	0
Good	Available	115	124	65	53	357 (24.4)
	NA	0	0	0	0
Poor	Available	43	57	45	40	189 (12.9)
	NA	1	0	0	3
No image	Available	15	15	6	4	51 (3.5)
	NA	0	0	8	3
Total, n (%)		495 (33.8)	541 (36.9)	205 (14.0)	224 (15.3)	1465 (100)

  n = 1465, right eye. NA, not available.

One hundred four diagnoses were made of definite glaucoma, 52 of probable glaucoma, and 78 of possible glaucoma. Among persons with definite glaucoma, 67 (1.0% of those 30 years of age and older) had POAG, 41 (61.2%) were women, and 26 (38.8%) had bilateral optic nerve damage. The disc was able to be assessed by photographs in 65 persons (97.0%). Those with POAG had a mean VCDR of 0.7, with a range of 0.4 to 1.0, and 91% (60/65) had a mean VCDR ≥0.6. At least one reliable VF was present in 39 subjects (58.2%) with a mean and median MD of −9.9 ± 7.2 and −8.2 dB. Three subjects (4.5%) had received previous diagnoses of POAG. Three eyes were blind from glaucoma (3.2% by eye, 4.5% by person), and four more had low vision in at least one eye (4.3% by eye, 6.5% by person). Two persons had bilateral low vision, and none were bilaterally blind from glaucoma. 

The mean IOP of POAG patients was 16.1 ± 3.5 mm Hg (range, 10.0–27.5 mm Hg), and 90% (56/62) of those newly diagnosed with POAG had an IOP that was 21 mm Hg or lower (Fig. 1). The mean CCT was 531.9 ± 31.0 μm (range, 441–610 μm), and CCT was similar among subjects with POAG (530.3 ± 30.8 μm in the right eye, 532.6 ± 28.7 μm in the left eye) and without POAG (535.2 ± 33.2 μm in the right eye and 536.6 ± 28.7 μm in the left eye) (P > 0.05). 

Figure 1.

 

The cumulative percentage of subjects whose intraocular pressure was ≤21 mm Hg among those with POAG in the Handan Eye Study (n = 62). Three had a history of medical or surgical treatment; two missed IOP data were excluded. ▴, 56 of 62 (90.3%) patients had higher IOP ≤21.

The prevalence of POAG was 0.16% in men and 0.4% in women for those aged 40 to 49 years and increased to 3.0% in men and 2.9% in women for those in their 70s (Table 3). The prevalence of POAG was also increased with higher IOP. Among persons with IOP ≤15 mm Hg, 16 of 3409 (0.5%) had POAG. The rate increased to 1.2% (24/2038) among persons with IOP >15.1 mm Hg and ≤17.5 mm Hg, 2.2% (20/922) among persons with IOP >17.5 mm Hg and ≤20.0 mm Hg, and 2.6% (5/194) among persons with IOP >20 mm Hg. The risk was approximately five times higher among persons with IOP >20 mm Hg than among persons with IOP ≤15 mm Hg. 

Table 3.

 

Prevalence of POAG in the Handan Eye Study by Age and Sex

Sex	Age (y)	n	POAG by Expert Consensus		POAG by ISGEO Definition
			n	% (95% CI)	n	% (95% CI)
Male
	30–39	560	0	0	5	0.9 (0.1–1.7)
	40–49	604	1	0.2 (0–0.5)	14	2.3 (1.1–3.5)
	50–59	1142	12	1.1 (0.5–1.6)	26	2.3 (1.4–3.1)
	60–69	549	6	1.1 (0.2–2.0)	14	2.6 (1.2–3.9)
	70–79	234	7	3.0 (0.8–5.2)	10	4.3 (1.7–6.9)
	80+	28	0	0.0 (0.0–0.0)	1	3.6 (0.0–10.4)
		3117	26	0.8 (0.5–1.2)	70	2.2 (1.7–2.8)
Female
	30–39	676	2	0.3 (0–0.7)	4	0.6 (0.0–1.2)
	40–49	721	3	0.4 (0–0.9)	8	1.1 (0.3–1.9)
	50–59	1315	13	1.0 (0.5–1.5)	20	1.5 (0.9–2.2)
	60–69	550	13	2.4 (1.1–3.6)	15	2.7 (1.4–4.1)
	70–79	307	9	2.9 (1.0–4.8)	8	2.6 (0.8–4.4)
	80+	30	1	3.3 (0–9.8)	0	0.0 (0.0–0.0)
		3599	41	1.1 (0.8–1.5)	55	1.5 (1.1–1.9)
Total*
	30+	6716	67	0.7 (0.5–0.9)	125	1.7 (1.4–2.0)
	40+	5480	65	1.0 (0.7–1.3)	98	2.3 (1.9–2.7)
	50+	4155	61	1.5 (1.1–1.9)	78	2.5 (2.0–3.0)

*  Standardized to China National Population Census 2000.

In a logistic regression model adjusted for sex, older age was strongly associated with POAG. Using the 40- to 49-year-old group as a reference population, the odds ratio (OR) increased from 3.4 (95% confidence interval [CI], 1.2–9.8) for the 50- to 59-year-old group to 9.6 (95% CI, 3.2–28.6) for the 70 year of age and older group. However, only one case was identified among those 80 years of age and older. The OR was 1.5 (95% CI, 1.2–1.8) for every 5 mm Hg of higher IOP. Myopia was also strongly associated with POAG. For mild myopia (−0.51∼−3.0 D), the OR was 1.9 (95% CI, 1.0–3.4); for moderate myopia (−3.1∼−6.0), the OR rose to 4.5 (95% CI, 1.7–12.1); for high myopia (−8.1 D and greater), the OR was 5.2 (95% CI, 1.2–22.9). Other parameters such as sex, educational level, marital status, smoking, alcohol use, height, weight, WHR, BMI, blood pressure, pulse rate, hypertension, diabetes, number of family members with glaucoma, and CCT were not associated with POAG after adjustment for sex and age. 

In multivariate logistic regression analysis, every 10 years increasing of age (OR, 1.9; 95% CI, 1.5–2.5), every 5 mm Hg higher IOP (OR, 1.16; 95% CI 1.2–2.0), every 1 mm longer axial length (OR, 1.3; 95% CI, 1.0–1.6), and moderate myopia (−3.1∼−6.0 D; OR, 4.7; 95% CI, 1.6–13.5) remained independent risk factors for POAG (Table 4). 

Table 4.

 

Multivariate Logistic Regression for Risk Factors for POAG in the Handan Eye Study

Characteristics	OR (95% CI)
Age, 10 y	1.9 (1.5–2.5)‡
Axial length, mm*	1.3 (1.0–1.6)‡
Intraocular pressure, 5 mm Hg†	1.5 (1.2–2.0)‡
Myopia, D*
>0.50	Reference
−0.51 to ∼ −3.0	1.5 (0.8–3.0)
−3.1 to ∼ −6.0	4.7 (1.6–13.5)‡
−6.1 to 8.0	0
−8.1	1.8 (0.17–19.3)

*  Used the mean values of the two eyes.

†  Used the data in the eye with higher IOP.

‡   P < 0.05.

Using the ISGEO definition, 125 diagnoses (1.9%) of POAG would have been made (101 in category 1, 23 in category 2, and one in category 3). The age-sex standardized prevalence of POAG using the ISGEO scheme is summarized in Table 3. The discrepancy between glaucoma diagnosis using the HES consensus definition and the ISGEO scheme is summarized in Table 5. 

Table 5.

 

Discrepancy in Glaucoma Diagnosis by Study Consensus and ISGEO Scheme

Glaucoma by Study Consensus	Glaucoma by ISGEO Scheme				Total
	Category 1	Category 2	Category 3	Not Glaucoma
Definite	43	12	12	37	104
Probable	12	3	0	37	52
Possible	5	3	0	70	78
Not glaucoma	71	14	0	6397	6482
Total	131	32	12	6541	6716

Discussion

The prevalence of POAG in this rural farming population was 0.7% among adults aged 30 years and older and 1.0% for those aged 40 years and older. The prevalence of POAG was lower than that reported in the Liwan Study ² from urban Guangzhou and was lower than that reported in both the urban and peri-urban populations studied in the Beijing Eye Study. ¹⁸ This difference in prevalence may, in part, be due to our limited ability to adequately assess for glaucoma in the oldest persons in Handan. Prevalence rates among the oldest old are substantially higher in most reported studies, ^{2,19 –23} and the fact that we only identified a low prevalence in this population and had a poor response rate may indicate a bias in our estimate. Another explanation for the lower prevalence seen in the present study could be how glaucoma was defined. In our study, we used consensus of glaucoma specialists that might have been more conservative. The Liwan Study ² used the ISGEO approach, the Beijing Eye Study used the opinions of single experts for glaucoma diagnosis ³ and the ISGEO approach ¹⁸ (Table 6). Finally, the lower prevalence may reflect a true difference in prevalence rates in rural settings and may point to the possibility that urban living in some way predisposes to higher rates of glaucoma. 

Table 6.

 

Age-Specific Prevalence of Open-Angle Glaucoma as Reported in Selected Population-Based Studies

Studies	Ethnic Group	Age Groups (y)
		40–49	50–59	60–69	70–79	80+	40+	50+
Handan Eye Study	Chinese	0.3	1.02	1.73	2.68	1.4	1.0†	1.5†
Handan Eye Study*	Chinese	1.7	1.9	2.6	3.3	1.7	2.3†	2.5†
Liwan study* 2	Chinese	—	0.8	1.7	2.2	8.3	—	2.1†
Beijing Eye Study 3	Chinese	0.62	0.84	1.78	5.63		1.78†
Beijing Eye Study* 18	Chinese	1.8	1.6	3.3	5.7		2.7†
Tanjong Pagar Study* 4	Chinese			NA			2.4
Baltimore Eye Study 24	White	0.18	0.32	1.53	3.33	1.94	1.44
Singapore Malay Eye Study* 25	Malay	1.7	2.5	3.6	5.2		2.5
Bangkok, Thailand* 26	Thai			NA			2.3
Chennai Glaucoma Study 20	Indian	2.25	3.57	4.08	5.89	10.2	3.51
Aravind Eye Survey 27	Indian	0.3	1.6	1.8	2.9		1.2
Tajimi Study 28	Japanese	2.0	2.7	4.7	8.2	6.0	3.9
Dahaka Division Study* 29	Bangladesh	NA					2.5
Blue Mountains Eye Study 19	White		0.4	1.3	4.7	11.4	3.0
Visual Impairment Project 21	White	0.1	0.6	1.9	5.2	6.2	1.7
Baltimore Eye Study 24	White	0.18	0.32	1.53	3.33	1.94	1.44
Baltimore Eye Study 24	Black	1.27	4.15	6.19	8.88	12.87	4.97
Barbados Eye Study 22	Black	1.4	4.1	6.7	14.8	23.2	6.8
Los Angeles Latino Eye Study 23	Latino	1.32	2.92	7.36	14.72	21.76	4.74

 NA, not available.

*  Adjusted by China National Population Census 2000.

†  Using the definition recommended by ISGEO.

We might have underestimated the prevalence of POAG because of how we screened for the disease. We used clinical evaluation of the optic nerve as one of the referral criteria, a cup/disc ratio ≥0.6, and asymmetry of a cup/disc ratio ≥0.2 as cutoffs for glaucoma suspect persons to undergo additional examination. Among those identified as glaucoma suspect, 495 had a VCDR <0.6 and an asymmetry of cup/disc ratio <0.2 but were selected for other reasons, such as narrow angles or IOP >21 mm Hg. Five of these persons (1.0%) were ultimately determined to have POAG. If the rate were half this in the rest of the population who had VCDR <0.6 and asymmetry of the cup/disc ratio <0.2, the overall crude prevalence would have increased to 1.36%. A second potential cause of underestimation of glaucoma prevalence is the methodology used to define the disease. Requiring consensus initially among one set of ophthalmologists and then requiring a second pair of ophthalmologists to agree with the initial grading led to a reduction in the number determined to have definite or probable glaucoma. A third factor of the underestimation of glaucoma prevalence was that only approximately 80% of the suspects were returned to the second visit for definitive glaucoma. The suspects who did not return for visual field testing had a lower detection rate of glaucoma (4.3% vs. 6.9%) because visual field data were less complete. We therefore agree that this 20% rate might have led to a slight underestimation of the prevalence rate. 

We have presented an alternative approach to assessing glaucoma prevalence using the ISGEO recommendations. This approach likely led to an inflated estimate of the prevalence of glaucoma. Using the ISGEO approach in our study, the prevalence increased to 2.3% for those aged 40 years and older, twice the rate found using expert consensus. Nearly three-fourths of the cases of POAG determined using the ISGEO approach were not recognized as glaucoma by the experts reviewing all available data. The Beijing Eye Study used both methods, and the prevalence was increased by 1.0% (1.7%–2.7%) using the ISGEO approach. 

Although the ISGEO system allows for glaucomatous optic neuropathy when there is a thin rim, it is largely driven by the cup/disc ratio and bases the criteria on percentile cutoffs. However, the current algorithm does not accommodate variations in disc size that heavily influence the cutoffs. ³⁰ Because of statistical considerations, using a cutoff for cup/disc ratio pushes glaucoma prevalence toward approximately 2.5%. In the present study, the 97.5th percentile of cup/disc ratio was 0.7, but the mean cup/disc ratio of persons with POAG was 0.7 as well, and 32.3% of subjects with POAG by expert consensus had a cup/disc ratio <0.7 based on evaluation of the digital images of the optic nerve. A similar finding was reported in the Los Angeles Latino American Eye Study, which also relied on consensus to determine the presence of glaucoma. In that study the cup/disc ratio was <0.7 in 50% of patients with POAG, whereas the 97.5th percentile of cup/disc ratio was also 0.7. ²³  

Age and IOP were strongly associated with POAG, as has been reported numerous times in the past. ^{19,31 –41} Myopia was also strongly associated with a higher risk for POAG, as has been reported in previous research. ^{27,42 –44} Mitchell ¹⁶ reported an OR of 3.3 for moderate to high myopia (95% CI, 1.7–6.4) in the Blue Mountains Eye Study. Xu et al. ⁴² also reported myopia exceeding −6 D was associated with glaucomatous optic neuropathy. In the present study, we found similar results with the risk for POAG 1.9-fold for mild myopia and 5.2-fold for high myopia (−8.1 D and greater) in univariate analysis. However, after adjustment for age, axial length, and IOP, only moderate myopia was associated with POAG. 

An alternative explanation for why myopes are more frequently categorized as having glaucoma is the possibility that retinal changes may cause visual field defects and the optic nerves can become difficult to assess. ^{45 –48} However, in a randomly recruited sample of 137 young males with myopia from Singapore National Service enlistees, only one subject had a significant field defect. ⁴⁹ It is unlikely that the association with myopia and glaucoma is simply attributable to field loss caused by retinal changes. 

More than 90% of those with diagnoses of POAG had statistically normal IOP. This has important implications for glaucoma screening. The proportions of those with POAG with an IOP <22 mm Hg varies greatly in population-based prevalence studies, from 30% to 92%. ^{2,3,19,21,25,26,28,32,50 –54} The findings in the present study are similar to those of another population-based study in Chinese, the Liwan Eye Study (85%). ² Given that high IOP is frequently a reason for referral to an ophthalmologist, clinic-based studies tend to underestimate the proportion with low-pressure glaucoma and to create a false impression that glaucoma at lower IOP is uncommon when in fact it is the norm. ⁵⁵ Based on our results and those of many others, the use of 21 mm Hg as a meaningful number in the diagnosis and management of glaucoma should be abandoned. The high proportion of POAG with statistically normal IOP cannot be interpreted as indicating that IOP is not a risk for optic nerve damage. IOP was strongly associated with POAG in the present study and those with the highest IOP had the greatest risk, as seen in Table 2. Additionally, intervention studies ^{38 –41} have documented that IOP lowering decreases the rate of glaucoma development and progression. 

Fewer than 1 in 20 persons with POAG had received previous diagnoses. Low rates of diagnosis were also found in an urban Chinese population. These low rates of diagnosis are certainly related to the lack of regular screening for eye diseases in China. Furthermore, it is probable that many doctors rely on IOP to identify glaucoma, but <10% of those with glaucoma had IOP >21 mm Hg. However, the vision loss caused by POAG in this population was less than previously reported in other studies. The rate of monocular blindness was 4.5%, which was lower than that reported among Chinese in the Liwan Study (17%) and the Beijing Eye Study (12%) and among Singapore Chinese (27%). The reason for this lower rate of vision loss among this population is not clear. Four subjects with glaucoma were blind in one eye; another had low vision, but it was uncertain whether POAG was the diagnosis because assessment of the angle was not available. Had these cases been confirmed as POAG, the rate would have been similar to that reported previously. Another possible explanation is that persons with glaucoma and lower IOP may not develop blindness at the same rate as those with higher IOP because most of the incidences in this study occurred at lower IOP. ⁵⁶ In the Tajimi study, in which 92% of subjects with POAG had IOP <21 mm Hg, the rate of unilateral blindness from glaucoma was also low (1.7%). ²⁸ The lower rate of severe visual loss may be one of the underlying factors for the low ratio of POAG in hospital-based reports in China. ^57,58  

In conclusion, approximately 1% of adults aged 40 years and over living in rural China had POAG, and 90% of them had IOP ≤21 mm Hg. One in 10 of the subjects was visually impaired in at least one eye. Older age, higher IOP, myopia, and greater axial length are associated with POAG. Given the rapid aging and the increasing prevalence of myopia in China's population, POAG is likely to become a large public health problem. 

Footnotes

 Supported by National Basic Research Program of China (973 Program) Grant 2007CB512201 from the Ministry of Science and Technology of the People's Republic of China; the program of Health Policy for blindness prevention from the Ministry of Health of the People's Republic of China; and in part by the Key Technologies R&D Program and Grant 2006-10903 from the Bureau of Science and Technology of Handan City, Hebei Province, China. Additional support was provided by Beijing Tongren Hospital and the key discipline fund of the Bureau of Health, Handan City, Hebei Province, China.

Footnotes

 Disclosure: Y.B. Liang, None; D.S. Friedman, None; Q. Zhou, None; X. Yang, None; L.P. Sun, None; L.X. Guo, None; Q.S. Tao, None; D.S. Chang, None; N.L. Wang, None

The authors thank Meiyu Li (Peking University First Hospital), Youqin Jiang (2nd Xiangya Hospital, Central South University), Tiancai Ye (Zhongshan Ophthalmic Center, Zhongshan University), Zeqin Ren (Peking University People's Hospital), and Henry D. Jampel (Wilmer Eye Institute, Johns Hopkins University) for evaluating the disc images/visual fields for glaucoma diagnosis.