Risk factors of primary open-angle glaucoma: the rst case-control hospital-based study in Vietnam

Purpose: To identify risk factors for primary open-angle glaucoma (POAG) in a Vietnamese population. Methods: A matched case-control study was conducted at Vietnam National Institute of Ophthalmology. Cases were patients clinically diagnosed with POAG. Controls were caregivers of patients with other eye diseases, free from any ocular and systemic abnormalities, and were selected and then matched with the cases by gender and age. Cases and controls were examined by a glaucoma specialist to conrm presence or absence of POAG. Data on demographic, ophthalmic and medical conditions were collected via interview at time of examination. Associations between POAG and potential factors were examined by univariate conditional logistic regression analysis (OR – Odd Ratios) and multivariate conditional logistic regression (AOR – Adjusted Odd Ratios) with level of signicance is 0.05. Results: In total, 128 cases with POAG and 128 controls were recruited. Multivariate conditional (xed-effects) logistic regression analysis resulted in several signicant risk factors, including hypertension AOR =4.7 (95% CI: 1.8–12.0; p=0.002); family history of glaucoma AOR = 6.4 (95% CI: 1.3 – 32.2; p = 0.026); over–the–counter eye medication AOR = 3.1 (95% CI: 1.5 – 6.5; p = 0.006), and a protective role for marriage AOR = 0.2 (95% CI: 0.1 – 0.7; p = 0.006) and weekly exercise AOR = 0.3 (95% CI: 0.1 – 0.8; p = 0.021). Conclusions: Hypertension, family history of glaucoma and over–the–counter eye medication were risk factors while marriage and weekly exercise were protective factors for POAG in Vietnamese population.


Background
Glaucoma is globally one of the leading causes of irreversible blindness, and therefore, is a worldwide public health problem. A systemic review and meta-analysis by Tham et al indicated that the prevalence of glaucoma in the population between 40 and 80 of age was 3.54% (95% CI: 2.09 -5.82). In Africa, primary open angle glaucoma (POAG) was predominant type of glaucoma with a prevalence of 4.20% (95% CI: 2.08-7.35) while the most popular type of glaucoma in Asia was primary angle closure glaucoma (PACG) with 1.09% (95% CI: 0.43-2.32). In 2013, it was projected that the glaucoma population ages from 40 to 80 years would rise to 76 million in 2020, and 118.8 million in 2040. [1] POAG is considered a silent thief of vision because of its chronic natural course and asymptomatic manifestation. Oftentimes glaucoma patients do not realize their ocular problem until the vision is severely impaired. Therefore, screening and early diagnosis is becoming more vital in glaucoma prevention. In addition to several unchangeable risk factors that have been mentioned in the medical literature such as age and family history of glaucoma [2], modi able risk factors relating to lifestyle habits are receiving more attention from researchers. To contribute to the knowledge of glaucoma epidemiology in Asia and around the world, we conducted this study to assess select risk factors for POAG in a Vietnamese population.

Methods
This was a matched case-control study conducted in Vietnam National Institute of Ophthalmology (VNIO) from February to June 2016. VNIO is located in Hanoi, the capital of Vietnam, and the catchment area covers the whole North and Middle Vietnam.

Cases
All consecutive patients over 18 years old with newly -diagnosed POAG presenting during the study period were included as cases in the study. The diagnostic criteria of POAG were: the open angle in gonioscopy, without (normal -tension glaucoma) or with intraocular pressure ≥ 22 mmHg, the glaucomatous optic disc change, and the corresponding glaucomatous visual eld defect.Intraocular pressure was measured three times in each eye, with the median value documented as the pressure in that eye. Intraocular pressure of a patient was de ned as the highest pressure between the two eyes. Glaucomatous optic disc changes included focal or diffuse neuroretinal rim thinning, localized notching, disc hemorrhage and asymmetry between both eyes and retinal nerve ber layer defects consistent with visual eld defects. Glaucomatous visual eld defects were con rmed by two of the following three criteria on two consecutive appointments: the presence of a cluster of three points on a pattern deviation probability plot with a probability < 5% in a typical location, one of which had a probability < 1%; a PSD with a probability < 5%; or a glaucoma hemi eld test result outside the normal limits.
A glaucomatous diagnostic procedure was carried out by a glaucoma specialist. The indirect ophthalmoscope and funduscopy camera Nidek AFC -230 with dilation were used for capturing the optic nerve head. Glaucomatous visual eld assessment was performed by a Humphrey R Field Analyzer II -750i series (24 -2, SITA; Carl Zeiss Meditec, Dublin, California). The reliability requirement for visual eld assessment was xation loss of 20% or less, false -positive and false -negative responses of 33% or less. Intraocular pressures were measured by Goldmann applanation tomometry. Patients with pseudoexfoliation syndrome, pigmentary dispersion syndrome and other types of secondary glaucoma, cornea diseases as well as any previous ocular surgery or medical history of other eye diseases or systemic diseases that may damage the visual eld were excluded from the study.

Controls
Controls adults over 18 years old who were were caregivers of surgically-indicated cataract patients in other departments without glaucoma or ocular hypertension. The selection criteria for control group were: peak intraocular pressure < 21 mmHg, vertical cup to disc ratio < 0.5, cup to disc ratio asymmetry < 0.2, no visual eld defect, no medical history of other eye diseases or systemic diseases that may damage the visual eld. Controls were assessed by the same glaucoma specialist of the case group with the same protocol and were recruited if they met all inclusion criteria and matched cases by sex and age (± 5 years).

Risk factors
Data on demographic, medical and ophthalmic conditions were obtained through interview and clinical examination. Demographic characteristics included age, gender, education occupation, habitation and marriage. Medical information included diabetes mellitus, hypertension, former and current smoking (> 1 cigarette/day for more than 25 years), alcohol intake (≥ 10 g alcohol/month for ≥ 5 years or more) and weekly exercise (≥ 75 minutes of vigorous -intensity physical activity or ≥ 150 minutes of moderateintensity physical activity according to 2016 recommendation of WHO). [3] Ophthalmic variables consisted of family history of glaucoma, presence of any refraction errors (myopia, hyperopia and astigmatism), over-the-counter (OTC) ophthalmic medication ≥ 1 time/year and routine eye check < 1 time/2 year.

Statistical analysis
Univariate conditional logistic regression was applied to analyze the association between POAG and potential risk factors. After adjustment for age, factors signi cantly associated with POAG (level of signi cance α = 0.05) were included in multivariate conditional logistic regression. Data were analyzed with SPSS online version (provided by The University of Sydney) and SAS statistical software version 9.4 (provided by University of Colorado).

Ethical approval
This study strictly follows the tenets of Declaration of Helsinki. All participants signed the consent forms. The Ethical Committee of Vietnam National Institute of Ophthalmology and Hanoi University of Public Health approved our method.

Results
We recruited 128 cases of POAG and 128 controls into our study. Table 1 shows some demographic characteristics of both cases and controls. There were more males than females, and over 50% of the sample lived in cities. The majority of participants had not received a bachelor degree (81.2% for cases and 74.2% for controls) and had married (79.7% for cases and 90.6% for controls) (  ( Table 2 would be here) Table 3 demonstrates the univariate analysis of risk factors for POAG. All variables (except age) were then adjusted for age in the analysis procedure. As a result, marriage status, hypertension, alcohol intake, weekly exercise, family history of glaucoma, OTC eye medication and routine eye examination had statistically signi cant relation to POAG (p<0.05). Routine eye examination was no longer signi cant after adjusting for age (p=0.170).
( Table 3 would be here) Table 4 and gure 1 show results of the multivariate logistic model of risk factors for POAG. Except for alcohol intake (p = 0.148), the risk factors remained signi cantly associated with POAG (p<0.05). Indeed, the model demonstrated that people having family history of glaucoma would have 6.4 times of risk for POAG higher than people without this glaucoma history (95% CI: 1.3 -32.2). Similarly, hypertension people had higher risks for POAG than non-hypertension people (OR=4.7, 95% CI: 1.8 -12.0) and people with OTC eye medication also had 3.1 times (95% CI: 1.5 -6.5) of higher risk for POAG than those without OTC eye medication. Having married and weekly exercise are protective factors for POAG as these characteristics had OR less than 1. Figure 1 summaries these results in the forest plot of OR and their 95%CI.

Discussion
Primary open angle glaucoma has proven associations with a wide range of factors including intraocular pressure, African-American race and older age especially after the age of 40 years. [2,4] To the best our knowledge, the present study provided the rst comprehensive assessment of glaucoma epidemiology in Vietnam. Vietnam is a typical country of emerging and dynamic Asian economics and it is likely that these results would apply to other similar Asian countries as well. We detected signi cant associations between POAG and hypertension, family history of glaucoma, OTC eye medication, marriage and weekly exercise.
We found that POAG cases were almost ve times more likely than controls to have hypertension compared to controls, which has been mentioned in previous studies conducted in Europe, Australia, China and South East Asia [5][6][7][8][9]. Optic nerve head nutrition is remarkably affected by ocular perfusion pressure (OPP) which relates with systemic blood pressure (mean brachial arterial pressure -MAP) and intraocular pressure (IOP) through the equation OPP = 2/3.MAP brachial -IOP. [10] When hypertension is recalcitrant, atherosclerosis will occur and cause the drop of local perfusion pressure. On the other hand, if hypertension treatment is aggressive with day and night medications, patients can suffer from nocturnal systemic hypotension. In both scenarios, POAG will keep progressing although IOP is wellcontrolled due to the decreased OPP. This encourages a tight connection between ophthalmologists and cardiologists for a combined follow up.
Not surprisingly, cases were more than six times more likely to have a family history of glaucoma. This classical risk factor has been persistently con rmed through a variety of studies and populations. [11][12][13] Li et al demonstrated that familial concordant risks were standardized incidence ratio (SIR) = 3.25 (95% CI: 3.16 -3.35) for POAG. [13] According to O'Brien et al, positive family history was correlated with increased risk of POAG, age-adjusted OR = 3.4 (95% CI: 2.8 -4.1). [12]. With the evidence of Myocilin mutations in advanced POAG and copy number variations of TBK1 in normal -tension glaucoma, highrisk people should be recommended for family screening for early glaucoma detection. [14,15] Our result that OTC eye medication was three times higher among cases than controls is a unique nding from our study. We did not collect details on the speci c types of ophthalmic medications that were being used by study participants. We inquired about any OTC eye medications with the exception of saline 0.9% and lubricant. From our experience in Vietnam, the most commonly used OTC eye medication is steroid drops to self-treat red or itchy eye problems. While ophthalmic steroid drops are typically regulated in developed countries like the United States, these medications are often readily available and not regulated in developing countries like Vietnam. In this environment, medications are commonly dispensed by pharmacies without prescription. Although not speci cally ophthalmic medication, one study in Hanoi found that although 60% of pharmacy staff reported that they would not dispense corticosteroids, when assessed in practice, 98% of the pharmacies dispensed corticosteroids without a prescription. [16] Almost one-third of patients in our study population had used over-the-counter eye medications at least once in the prior year, which demonstrates how prevalent this issue is in Vietnam. Steroid drops have been shown to increase IOP and induce glaucoma, ocular hypertension, and cataract. Although glaucoma medications are also readily available in Vietnam, awareness and utilization of these medications is much less common and likely does not contribute to this high prevalence of OTC medication in our study. However, due to the lack of speci c information on the type of eye medication used, this nding warrants further study.
Our analysis demonstrated that POAG cases were less likely to engage in weekly exercise. This nding was consistent with the bene t of immediate IOP reduction effect after exercises such as walking, [17] running a 42 km marathon [18] and nishing a 110 km march carrying a 20 kg back pack. [19] Some theoretical mechanism involving lowered blood pH, increased blood plasma osmolarity and blood lactate [20], autonomic changes [21] and decreased aqueous production due to dehydration [22] have been introduced. Nonetheless, other evidences have proven other IOP elevation effects of exercise relating to valsava manoeuver, expiratory effort, muscular effort and body position. [23][24][25][26][27][28] In a recent populationbased cross-sectional study, Lin et at found that subjects committed to vigorous exercise 7 days a week had higher risk of having glaucoma than those with three-day exercise per week, and glaucoma was more prevalent in the high intensity of exercise compared to the group of moderate intensity.
[29]This disputation could be interpreted that the study of Lin et al had more speci c strati cations of physical activities regarding the number of day per week and the level of intensity whereas our study only examined this factor in general. Taken all together, these studies may suggest that POAG patients should maintain moderate-intensity activities and avoid aggressive exercises, especially those which can induce valsalva effect, however, our study did not delineate between levels of physical activity.
In this study, marriage was signi cantly more common among controls compared with cases of POAG. This association has been identi ed in other research as well, with the risk of having open angle glaucoma increased 1.39 times (95% CI: 1.03 -1.870) in unmarried people in the United States [30] and glaucoma risk perception was signi cantly associated with marital status in a Ghana study (χ 2 = 41.293, p = 0.0001).
[31] Although the reason for this association is not clear, one hypothesis is that this relationship could be because married people have a greater awareness of their disease and/or are given better care from their partners, which may lead to lower prevalence rates of glaucoma. Finally, our study did not show an association between POAG and alcohol intake in multivariate analysis. This result has been con rmed in other studies as well. [32][33][34] Our study has several strengths and limitations. One limitation to this research was recall error that could exist for both cases and controls. There is no reason to believe that cases and controls had different recall, therefore this limitation is unlikely to in uentially bias our study results. Second, our cases recruited from a hospital which likely sees more severe cases of POAG, therefore, patients may not be representative of all POAG cases in Vietnam. There are some strengths to our study. First, case -control is a suitable design to study risk factors for POAG, an incurable disease with chronic progression that is not too common in Vietnam. Second, Vietnam National Institute of Ophthalmology, as an a liated hospital of Hanoi Medical University, is one of the best eye care facilities across the country with standardized care, state-of-art equipment and outstanding specialists where the majority of patients in Northern Vietnam visit annually. We also randomly re-input 10% of research forms to ensure the accuracy of our data. In addition, by matching cases and controls, we eliminated confounding by gender, limited confounding by age, and gained e ciency to identify associated risk and preventative factors for POAG. However, matching by age and sex does preclude us from assessing these factors as contributors to POAG risk.

Conclusion
This is the rst study that we are aware of that speci cally looks at risk factors for POAG in Vietnam. Our study ndings indicate that many known risk factors for POAG remain associated with POAG in our Vietnamese population. In addition, we found that use of over-the-counter eye medication is a novel risk factor that warrants more research. In conclusion, our ndings enrich the epidemiology knowledge of glaucoma in Vietnam speci cally, and in similar countries in Asia. Public health programs in developing regions and countries can bene t from our data to su ciently tackle the mission of glaucoma screening, primary care and prevention. The protocol of this study was approved by the Scienti c and Ethical Committee in Biomedical Research, Hanoi University of Public Health and Vietnam National Institute of Ophthalmology. All human subjects in the study were asked for their consent before collecting data, and all had complete rights to withdraw from the study at any time without any threats or disadvantages.