Intraocular Pressure in General and Diabetic Populations From Southern China: the Dongguan Eye Study

Abstract

Purpose: To investigate the distribution and risk factors for intraocular pressure (IOP) among general and diabetic populations in Southern China.

Methods: The study participants aged 40 years or older were enrolled from the Dongguan Eye Study, a population-based cross-sectional study from September 2011 to February 2012. Systemic and ophthalmic examinations were performed, and diabetes status was screened based on the American Diabetes Association diagnostic criteria (2010). IOP was measured by a noncontact tonometer per standardized protocol. Regression analyses were used to assess the association between potential risk factors and IOP.

Results: A total of 2112 subjects were included with a median age of 55 years. IOP for general population showed a near normal distribution with an average of 15.58 ± 3.27 mm Hg. Multiple regression analyses revealed that higher IOP was significantly correlated with younger age, higher body mass index (BMI), shorter height, higher blood pressure (BP), higher fasting blood glucose (FBG), higher low-density lipoprotein cholesterol (LDL-C), lower high-density lipoprotein cholesterol (HDL-C), and thicker central corneal thickness (CCT). There was no association between diabetes status and IOP after adjusting for possible confounders. IOP for diabetic participants showed a right-skewed distribution. Risk factors for IOP elevation in diabetes included female, younger age, higher BP, higher LDL-C, lower HDL-C, and thicker CCT.

Conclusions: The present study identifies risk factors for elevated IOP in general and diabetic populations. Younger age and lower HDL-C, as well as higher BP, LDL-C, and CCT were significant factors contributing to higher IOP, especially in the female diabetic population.

Intraocular pressure (IOP) is the fluid pressure inside the eye and an important ophthalmic physiological parameter. High IOP is widely acknowledged as the most important risk factor for glaucoma, and IOP reduction therapy is the only proven effective treatment.^1,2 Thus it is of pragmatic significance to understand the distribution and risk factors of IOP for glaucoma prevention and prognosis. Many factors, such as age,^3–6 body mass index (BMI),⁷ blood pressure (BP),^5,8 blood glucose,^9,10 central corneal thickness (CCT),⁴ have been reported to associate with IOP, but their results were not entirely consistent in all studies, and the potential risk factors in their analysis were failed to account due to lack of data. Therefore, population-based studies with larger sample size and detailed information are needed to better understand these issues. 

To be noted, diabetes has become a global epidemic problem. It has been estimated that there were 451 million (age 18–99 years) people with diabetes in 2017, and these figures were expected to increase to 693 million by 2045.¹¹ It remains equivocal whether diabetic populations have different distribution or risk factors for IOP, and the association of diabetes with glaucoma has still been controversial, despite the fact that people with diabetes are twice likely to develop glaucoma compared with nondiabetes.¹² Therefore, data on IOP distribution and risk factors in diabetic populations are needed to clarify the relationship between glaucoma and diabetes and plan effective prevention strategies. 

In view of the above issues, we conducted a large scale cross-sectional study based on the Dongguan Eye Study (DES) to investigate: (1) the distribution of and risk factors for IOP in general Southern Chinese population; (2) the difference in IOP distribution between diabetic and nondiabetic populations and risk factors for IOP in diabetes. 

Methods

Design, Procedures, and Participants of DES

The DES, a large population-based cross-sectional study in Southern China, was established to investigate the frequency and risk factors for visual impairment and major vision-threatening eye diseases in an adult rural population. This study complied with the Declaration of Helsinki, and was approved by the Medical Ethics Committee of Dongguan People's Hospital and Research Ethics Committee of Guangdong Provincial People's Hospital. Each step in the study was verbally explained to the participants, and informed consent was obtained from each participant. 

The methodology in detail has been reported elsewhere.^13,14 In brief, residents aged 40 years or older from Hengli Town of Dongguan were recruited from September 2011 to February 2012. A total of 11,357 participants were eligible for inclusion and 8952 (78.8% of eligible) participants were successfully enrolled for the systemic and ophthalmic examinations. Demographic, socioeconomic status, and health- and vision-related quality of life data were collected during interviews. Height, weight, waist and hip circumference, heart rate, and BP were measured per standardized protocol. Waist-hip ratio was defined as the quotient of waist circumference divided by hip circumference. BMI was calculated as weight in kilograms divided by height in meters squared. Body surface area (BSA) was calculated using the following formula: BSA (m²) = 0.0061 × height (cm) + 0.0124 × weight (kg) − 0.0099. Venous blood was collected and laboratory tests, including fasting blood glucose (FBG), hemoglobin A1c, serum total cholesterol (TC), triglyceride (TG), high-density lipoprotein-cholesterol (HDL-C), and low-density lipoprotein-cholesterol (LDL-C), were conducted. A comprehensive ophthalmic examination included visual acuity, autorefraction, slit-lamp examination, IOP measurement, ocular biometry, gonioscopy, fundus photography, etc. 

Every third person examined as part of the DES was systematically sampled to undergo an ocular biometric measurement. Subjects with glaucoma or other ocular diseases affecting measurement (e.g., pterygium, corneal opacity, severe trichiasis, strabismus, nystagmus, severe cataract, and lens subluxation) or history of ocular surgery (including cataract surgery) and trauma or unable to cooperate with the examination were excluded. A total of 2602 individuals were invited to take part in this examination, and 2112 (81.2% of invited) participants with available data on ocular biometry were included in this study. Moreover, according to the American Diabetes Association diagnostic criteria (2010),¹⁵ 370 of the 2112 study participants were found to have type 2 diabetes with a corresponding prevalence of 17.52%. 

Measurement of Ocular Biometry

The Lenstar LS900 optical biometer (Haag-Streit, Koeniz, Switzerland) was used to measure ocular biometry. Data on axial length (AL), CCT, anterior chamber depth (ACD), lens thickness (LT), corneal curvature, horizontal corneal diameter, and pupil diameter were obtained for analysis. 

Measurement of IOP

IOP was measured by a noncontact tonometer (NCT, AT80, Topcon, Japan) for each eye. Three measurements were taken for each eye, and a retest was performed after 5-minute rest if the standard error of the three measurements exceeded 2 mm Hg. The final data of each eye was recorded as the average of three valid independent readings. When IOP was not measurable (e.g., due to corneal scar), it was estimated by digital tonometry and excluded in this study. IOP measurements were normally undertaken between 9 AM to 11 AM, and instrument calibration for the NCT was performed after finishing examination for each village. 

Quality Control and Statistical Analysis

All data analysis was performed using SPSS software (version 20.0; SPSS, Inc., Chicago, IL, USA)_. Measurements from the right eye were selected for analysis because of the high correlation between the two eyes (Spearman rank correlation analysis). Distributions of IOP and other examination parameters were assessed for normality with Kolmogorov-Smirnov test and were considered significantly different from normal when the P value was ≤ 0.10. Normally distributed continuous variables were presented as mean ± standard deviation, and nonnormally distributed continuous data were presented as the median (interquartile range [IQR]). Differences between groups were compared using unpaired t-test/1-way ANOVA (normally distributed variables) or Wilcoxon/Kruskal-Wallis test (nonnormally distributed variables), as appropriate. Age was divided into four age groups 10 years apart: 40 to 49, 50 to 59, 60 to 69, and ≥70 years. Demographic and laboratory variables were also divided into subgroups for association analysis. Trend analysis was used to assess the association between IOP and age. Univariable and multiple linear regressions were used to assess the associations between potential risk factors and IOP. A 2-tailed P value of less than 0.05 was considered statistically significant for all analysis. 

Results

Baseline Characteristics of Participants

A total of 2112 participants were included in the current analysis with a median age of 55 years (IQR, 47–62 years), 90.3% of the participants were aged 40 to 69 years old and 58.3% were female. Table 1 presents the baseline demographic, laboratory characteristic, ocular biometric parameters, and IOP for participants and nonparticipants of the current study. Most of systemic parameters were significant differences between the two groups, but without marked clinical significance, possibly due to the large sample size and the exclusion criteria. In view of strong correlations between the eyes for all biometric parameters and IOP (all with P < 0.001; Supplementary Table S1), the characteristics of ocular biometric parameters and the distribution of IOP from the right eye were analyzed. The mean IOP for the general population was 15.58 ± 3.27 mm Hg. IOP showed a near normal distribution with the majority of participants having an IOP value within the normal clinical range of 10 to 21 mm Hg (skewness: 1.139 [0.053], kurtosis: 5.849 [0.106]) (Fig. 1A). 

Table 1

 

Baseline Demographic, Laboratory Characteristic, Ocular Biometric Parameters, and IOP for Participants and Nonparticipants of the Current Study

 

Figure 1

 

Distribution of IOP for the general participants (right eye). (A) All participants, (B) female participants, and (C) male participants. All participants: skewness is 1.139 (0.053), kurtosis is 5.849 (0.106); Females: skewness is 0.589 (0.070) and kurtosis is 1.212 (0.139); Males: Skewness is 1.645 (0.082) and kurtosis is 9.404 (0.165).

IOP Distribution and Risk Factor in the General Population

The mean IOP was 15.60 ± 3.08 mm Hg in female participants and 15.55 ± 3.53 mm Hg in male participants, respectively. The distribution of IOP approximated normal distribution in both sexs while showing a slightly bigger variance in female participants (Figs. 1B, 1C). IOP for female and male participants were similar within the same age group, while IOP in younger age groups were higher compared with older age groups. Trend analysis showed a significant decreasing trend of IOP with increasing age (Supplementary Table S2, P for trend < 0.001). 

Univariable analysis of the association between IOP and potential risk factors showed that subjects with presence of diabetes or hypertension, or with younger age, higher BMI, bigger waist-hip ratio, higher TG, higher LDL-C, lower HDL-C, thicker CCT, lower LT, higher relative lens position (RLP), and a pupil diameter level of 3.5 to 4.0 mm had higher IOP (Supplementary Table S3). 

The parameters with P value less than 0.10 in univariable analysis and those considered risk factors for glaucoma in literature^16–18 were selected into multivariable analysis (Table 2). The results showed that sex was not significantly associated with IOP. IOP tended to decrease with the increase of age, height, HDL-C, but tended to increase with the increase of BMI, SBP, DBP, FBG, LDL-C, and CCT in adjusted analyses. The age- and sex-adjusted mean IOP of the diabetic participants was significantly higher than that of nondiabetes (mean difference 0.849 [0.486–1.213] mm Hg; P < 0.001), but when adjusting for systemic factors (BMI, height, SBP, DBP, FBG, HDL-C, and LDL-C) at the same time, there was no association between diabetes status and IOP (Table 2, model 1, model 3). Moreover, although lower R1 tended to have higher IOP when adjusted for sex, age, and ocular parameters, there was no correlation when adjusted for systemic and ocular factors. The AL, ACD, LT, lens/axial length factor (LAF), RLP, pupil diameter, and spherical equivalent, however, showed to be not related to IOP after adjusting for possible confounders. The model 3 in Table 2 displayed better fitness when adjusting for both systemic and ocular factors (adjusted R² = 0.204) as compared to adjusting systemic factors (model 1: adjusted R² = 0.116) or ocular parameters (model 2: adjusted R² = 0.115). 

Table 2

 

Multiple Linear Regression Models of IOP in the Right Eye of All Persons Among Participants of the Current Study

 

IOP in the Diabetic Population

Table 3 shows the baseline demographic, laboratory characteristic, ocular biometric parameters, and IOP for participants with or without diabetes. The mean IOP for diabetic population was 16.09 ± 3.29 mm Hg. Compared with nondiabetic participants, those with diabetes were older and had a higher BMI, BSA, waist-hip ratio, BP, heart rate, FBG, and blood lipids levels. Furthermore, participants with diabetes had smaller AL, ACD, horizontal corneal diameter, and pupil diameter, but larger LAF. Of note, the CCT and spherical equivalent had no significant difference between the participants with and without diabetes, while the LT was higher in the diabetic participants with a borderline significance (P = 0.051). IOP showed a right-skewed distribution among diabetic participants and a nearly normal distribution among nondiabetic participants (Fig. 2). 

Table 3

 

Baseline Demographic, Laboratory Characteristic, Ocular Biometric Parameters, and IOP for Nondiabetic and Diabetic Participants

 

Figure 2

 

Distribution of IOP in diabetic (A) and nondiabetic (B) participants (right eye). Diabetic participants: skewness is 0.745 (0.127) and kurtosis is 2.225 (0.253); Nondiabetic participants: skewness is 1.234 (0.059) and kurtosis is 6.796 (0.117).

Univariable analysis of the association between IOP and potential risk factors among diabetic participants showed that subjects with presence of hypertension, or with younger age, higher BMI, bigger waist-hip ratio, higher TG, higher LDL-C, and thicker CCT had higher IOP (Supplementary Table S4). 

As shown in Table 4, after adjusting for the potential confounders, multivariable analysis showed that significantly higher IOP was found in female, younger age, higher BP, higher LDL-C, lower HDL-C, and thicker CCT. The age- and sex-adjusted mean IOP was not associated with FBG in diabetic participants (mean difference 0.037 [−0.076, 0.151] mm Hg; P = 0.517). The model 3 displayed the best fitness (adjusted R² = 0.257) among the three models. 

Table 4

 

Multiple Linear Regression Models of IOP in the Right Eye of Diabetic Persons Among Participants of the Current Study

 

Discussion

Based on larger samples and more comprehensive data, our study found that IOP showed a near normal distribution among general population and right-skewed distribution among diabetes. There was no association between diabetes status and IOP after adjusting for possible confounders. IOP would decrease with older age, greater height, and higher HDL-C, and increase with higher BMI, SBP, DBP, FBG, LDL-C, and thicker CCT in the general population. Risk factors for IOP elevation in diabetic population might be related to female sex, younger age, higher SBP and DBP, higher LDL-C, lower HDL-C, and thicker CCT. The other adjusted ocular biometric parameters, such as AL, ACD, LT, LAF, RLP, pupil diameter, and spherical equivalent, showed to be not related to IOP in both populations. 

In our study, the mean IOP is slightly lower than that of the Beijing Eye Study conducted in Northern China (15.58 ± 3.27 mm Hg versus 16.11 ± 3.39 mm Hg), possibly due to lower CCT (0.529 ± 0.030 mm versus 0.556 ± 0.033 mm).¹⁹ In similarly aged groups, the reported IOP in Chinese participants seems to be higher than that of Japanese and Korean participants in which a mean IOP is reported from 11 to 14 mm Hg.^20–23 However, most of the reports in white and black people vary and show a higher^24,25 or similar IOP^26,27 compared with ours. An association study of IOP from 12 population-based studies across Europe showed that mean IOP ranged from 13.6 mm Hg in the Rotterdam Study II to 16.0 mm Hg in the EPIC-Norfolk Eye Study.²⁸ Difference in ethnics, age, body statures, obesity, CCT, and measurements of IOP may in part explain the difference across studies. 

The distribution of IOP in the Southern Chinese people aged 40 years or older is similar to that found in other Asian people, with a negative age-IOP relationship in cross-sectional analysis.^{5,20–22,29,30} There is no consensus on the direction of association between IOP and age in the literature, with studies reporting increasing IOP,^31,32 decreasing IOP, no association of IOP²⁶ with older age, or an inverted U-shaped relationship.²⁸ Possible reasons for this inconsistency are differential associations by population, or a nonmonotonic relationship between age and IOP such that different studies of different aged participants yield different results.²⁸ Although hormonal difference may play a role in sex difference of IOP, and several studies found that female participants had higher IOP than male participants,^2,5,29 our study only observed the similar result in the diabetic population, while there was no sex difference in IOP distribution in the general population. 

Most risk factors for IOP in our study are consistent with the literature. Higher BMI has been consistently reported as a risk factor for higher IOP, through its effect on blood viscosity and episcleral venous pressure.^{20,21,33–35} A significant decrease in IOP with greater height after adjustment for possible confounders was found in our general population, which was consistent with a meta-analysis of pan-European study.²⁸ The mechanism underlying lower IOP in taller people is not clear, but may be related to the distance between the eye and the heart.²⁸ As shown in our study, BP is the most commonly reported risk factor for high IOP as higher BP could lead to an increase in ocular ultrafiltration and decrease in aqueous liquid outflow.^{8,20–22,29,30,34–40} Contrary to the results that higher IOP is significantly correlated with history of diabetes mellitus in other reports,^{20,29,34,35,41–43} our study found that IOP is not related to diabetes status in general population or to FBG in diabetic population, which suggested that diabetes mellitus or higher glucose might be not a risk factor for high IOP or glaucoma. We also identified that IOP was positively correlated with LDL-C and negatively correlated with HDL-C, while Yokomichi et al.⁸ found that IOP was positively correlated with HDL-C and TG levels in a Japanese population. Although increased osmotic gradient and fibronectin accumulation related with diabetes and obesity have been suggested to play a role in IOP variation, the real association and underlying mechanism need further study. It is worth noting that, as our findings show, dyslipidemia is often closely related to abnormal BMI, BP, and blood sugar. Thus, these factors should be considered together when the relationship between blood lipids and IOP is analyzed, which can provide clues for screening of elevated IOP and related disease. 

CCT has been demonstrated to be an important ocular contributor to IOP elevation, and the correlation between CCT and IOP in this study is similar to that for other populations.^{4,20,33,35,40} It has been reported that the mean IOP increased by approximately 2.3 mm Hg for every 100-μm increment in CCT by ultrasound pachymetry,⁴ while IOP increased 0.7 mm Hg per 20-μm thicker CCT (2.1 mm Hg/100 μm) in our general population. With respect to ophthalmic parameter, corneal curvature was another cornea-related factor influencing the IOP readings besides CCT. Our result shows that the larger the corneal curvature radius (R1), the lower are the IOP readings, suggesting that a relatively flat cornea as compared to a relatively steep cornea needs less force to be flattened to an area of a defined size.⁴⁴ In addition, other biometric factors possibly associated with glaucoma, such as AL, ACD, LT, LAF, RLP, pupil diameter, were not associated with high IOP in our both general population and diabetic subgroup. 

Given the ever-increasing prevalence of diabetes and the higher risk of glaucoma in diabetic population, risk factor analysis for IOP among this subgroup is of important public health significance. Unfortunately, rare data are available for the Chinese population to date, and whether risk factors for IOP among diabetes are different from the nondiabetic population is largely unknown. Different from a nearly normal distribution in the nondiabetic population, IOP shows a right-skewed distribution in our diabetic population. SBP, DBP, and LDL-C, rather than FBG, were more positively correlated with higher IOP in the diabetic population, which mechanism might be related to metabolic syndrome in general.³⁸ Although diabetes was generally considered to be associated with higher IOP and thicker CCT, CCT contributed a small proportion of mediating effect to the total effect of diabetes on IOP.¹⁰ It is considered that diabetes-related autonomic dysfunction and corneal stiffening caused by glycation-induced corneal collagen cross-links in diabetes may contribute to increase IOP.⁴⁵ 

Strengths of the current study include a population-based design, a large sample size, and the availability of multiple potential risk factors (systemic and ocular parameters) for IOP. However, some limitations need to be noted. First, the cross-sectional design is subject to cohort effect, and our conclusion still needs further validation from longitudinal cohorts. Second, inclusion method and exclusion criteria in this study might lead to selection bias. Third, IOP was measured by NCT instead of Goldmann tonometer (Gold standard), but NCT had been reported to have a good reliability within the normal IOP range.⁴⁶ Forth, our results can only represent IOP during the measuring period, and may not be able to reflect the circadian IOP fluctuations.⁴⁷ 

In conclusion, the distribution of IOP in our study is similar to that in the majority of Asian population-based reports. In the general population aged 40 years or older from Southern China, higher IOP was significantly correlated with younger age, lower height, and HDL-C, and higher BMI, BP, FBG, LDL-C, and CCT. Female sex, thicker CCT, and higher BMI, BP, HDL-C, and LDL-C may be more likely to have higher IOP in diabetic population. By identifying specific characteristics associated with IOP, those results have implications for detecting persons most likely to have higher IOP. Because high IOP is a major risk factor for glaucoma, it would be reasonable to speculate that similar associations would be found for both conditions. Of cause, the real association between these IOP-related risk factors and glaucoma still need further investigation. 

Acknowledgments

The authors appreciate the great support offered by the government of Hengli Town for this study. We thank the staff of Hengli Hospital for their work relating to the survey. 

Supported by the National Natural Science Foundation, Beijing, China (81371031), Guangdong Natural Science Foundation (2017A030313609), Project of Guangdong Medical Science and Technology Research Foundation, Guangzhou, China (A2018401), Guangzhou Science and Technology Project, Guangzhou, China (Z032012245, 201704020048). The funding organizations had no role in the design or conduct of this research. 

Disclosure: Y. Cui, None; X. Yang, None; G. Zhang, None; H. Guo, None; M. Zhang, None; L. Zhang, None; J. Zeng, None; Q. Liu, None; L. Zhang, None; Q. Meng, None