Association Between Alcohol Consumption Patterns and Glaucoma in Japan

Abstract

Précis:

In this case-control study of the Japanese population, including 3207 glaucoma cases, alcohol consumption patterns such as frequency and quantity showed a positive association with glaucoma prevalence.

Purpose:

To examine the association between alcohol consumption patterns and glaucoma.

Subjects and Methods:

This case-control study evaluated 3207 cases with glaucoma and 3207 matched controls. Patients over 40 years of age were included from 1,693,611 patients admitted to 34 hospitals in Japan. Detailed alcohol consumption patterns (drinking frequency, average daily drinks, and total lifetime drinks) were obtained, as well as various confounding factors, including smoking history and lifestyle-related comorbidities. Conditional logistic regression models were used to calculate odds ratios (ORs) and 95% CIs for glaucoma prevalence.

Results:

Drinking frequency showed an association with glaucoma for “a few days/week” (OR, 1.19; 95% CI, 1.03–1.38) and “almost every day/week” (OR, 1.40; 95% CI, 1.18–1.66). Average daily drinks showed an association for “>0–2 drinks/day” (OR, 1.16; 95% CI, 1.03–1.32). Total lifetime drinks showed an association for “>60–90 drink-year” (OR, 1.23; 95% CI, 1.01–1.49) and “>90 drink-year” (OR, 1.23; 95% CI, 1.05–1.44). As alcohol consumption levels differed considerably between men and women, additional analyses were conducted separately for men and women. Among men, drinking frequency of “a few days/week” and “almost every day/week,” average daily drinks of “>0–2 drinks/day” and “>2–4 drinks/day,” and total lifetime drinks of “>60–90 drink-year” and “>90 drink-year” had an association with glaucoma. Conversely, among women, neither drinking frequency, average daily drinks, nor total lifetime drinks were associated.

Conclusions:

Both the frequency and quantity of alcohol consumption were associated with glaucoma. Further research on gender differences is warranted.

Glaucoma is a leading cause of blindness worldwide.¹ Globally, it is estimated that the total number of glaucoma patients will reach 111.8 million by 2040.² Despite the seriousness of the disease, lowering intraocular pressure (IOP) is currently the only treatment option.³ Several epidemiologic studies attempted to investigate clinically modifiable risk factors other than IOP, such as drinking, smoking, exercise, and other lifestyle habits.^4–6 Among them, we investigated the association between drinking habits and glaucoma. Given the worldwide increase in alcohol consumption,⁷ the comorbidities associated with alcohol consumption might increase. Therefore, the effect of alcohol on glaucoma is intriguing.

Several studies have reported an association between alcohol consumption and glaucoma development.^8–17 Based on these multiple reports, the latest meta-analyses suggested a possible association between alcohol consumption and glaucoma⁴ but added that further validation studies are desirable due to the methodological heterogeneity among each report. As there have been many reports showing no association between alcohol consumption and glaucoma,^11–17 the effect of alcohol on glaucoma remains unclear. One reason for the variance in the reported results may be the differences in the effects of the degree of alcohol consumption. For instance, alcohol consumption has harmful effects, mainly due to heavy exposure to cardiovascular disease (CVD) and diabetes.^18–20 It is also possible that sensitivities to alcohol vary by race, which may lead to differences in the results reported previously. In Asian populations, this relationship is also controversial.^21,22 For Japanese populations, one report in a Japanese population found no association between alcohol consumption and glaucoma.²³ Further investigation in Asian populations is warranted.

In this large case-control study, we examined the association between alcohol exposure and glaucoma prevalence using data from the Inpatient Clinico-Occupational Database of the Rosai Hospital Group (ICOD-R), a nationwide, multicenter hospital-based inpatient registry database throughout Japan. Detailed drinking patterns, such as drinking frequency, average daily drinks, and total lifetime drinks, were obtained, revealing how alcohol affects glaucoma in a dose-dependent manner. A better understanding of the effects of alcohol consumption on glaucoma will provide evidence for clinical guidelines in the management of glaucoma.

SUBJECTS AND METHODS

We conducted this study using the ICOD-R, a large-scale survey by the Japan Organization of Occupational Health and Safety (JOHAS), which includes approximately 250,000 cases per year from 34 regional core hospitals, as described elsewhere.^24–28 The ICOD-R is a detailed investigation of diseases, lifestyles, and working conditions, including medical chart information confirmed by physicians. Clinical diagnoses were coded according to the International Statistical Classification of Diseases and Related Health Problems, 10th revision (ICD-10). Unlike medical claims data, the ICOD-R has high diagnostic accuracy and studies based on ICD-10 codes from the ICOD-R have been reported.^29,30 Lifestyle-related factors such as alcohol consumption, smoking habits, and past medical history were obtained through interviews based on a formatted questionnaire at each hospital. All information in the ICOD-R was registered with the health information manager at each hospital. Informed consent was obtained from all participants in written form. This study adhered to the tenets of the Declaration of Helsinki. The study conformed to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist, Supplemental Digital Content 5, http://links.lww.com/IJG/A841, which is used to improve the quality of reporting of observational studies.³¹ Access to the dataset was granted by a research agreement between the JOHAS and the researchers. This study was approved by JOHAS (approval no. R1-006) and the Research Ethics Committee of Tokai University School of Medicine (approval no. 18R-309).

The inclusion criteria for this study were patients admitted to a JOHAS Group hospital between April 1, 2005, and March 31, 2020, leaving 3,469,498 hospitalization records of 1,693,611 individuals. As shown in Figure 1, patients over 40 years of age were included, and we excluded patients with missing or inappropriate responses to the drinking status questionnaire, which is the main exposure of this study.

FIGURE 1.

Flow diagram. Among 3,469,498 hospitalization records of 1,693,611 individuals, patients over 40 years of age were included, and those with missing or inappropriate responses to the drinking status questionnaire were excluded. Cases of glaucoma were extracted based on the ICD-10, and one matched control subject for each case was selected. Finally, we enrolled 3207 glaucoma cases and 3207 controls. ICD-10 indicates International Statistical Classification of Diseases and Related Health Problems, 10th Revision. ID indicates identification numbers specific to each patient.

Cases were selected from glaucoma patients (ICD-10, H40.1, and H40.9) to include most cases of primary open angle glaucoma. Controls were selected based on the methodology described in previous studies.^24,26 The inclusion criteria of our controls were patients admitted for treatment of infectious and parasitic diseases (A00–B99), diseases of the ear and mastoid process (H60–H95), diseases of the skin and subcutaneous tissue (L00–L99), diseases of the genitourinary system (N00–N99), pregnancy, childbirth, and the puerperium (O00–O99), and certain conditions originating in the perinatal period (P00–P96). Among these patients, those who had previous ophthalmic histories (H00–H59) were excluded. We randomly selected one control subject for each case, matched by sex (male or female), age (same strata in 5-year intervals), admission date (same year), and admitting hospital (34 hospitals). None of the controls were matched more than once.

To assess the extent of exposure to alcohol consumption, we used the same methodology as described in a previous paper.³² After obtaining 3 variables (drinking frequency, average daily drinks, and duration of drinking) through the questionnaire, 3 alcohol consumption patterns were evaluated: drinking frequency, average daily drinks, and total lifetime drinks. Drinking frequency was categorized into 4 groups (never, former, a few days/week, and almost every day/week). As the questionnaire was modified between 2005 and 2020, the categories of drinking frequency were combined so as not to differ from their actual values. Average daily drinks were collected as continuous values (“standard drink”=10 g of pure ethanol; a measurement to assess the amount of alcohol consumed, accounting for the variances in alcohol content)³³ and then categorized into 4 groups (never, >0–2, >2–4, and >4 drinks/day), which signifies the average consumption specifically on the days when alcohol was consumed. The total lifetime drinks were calculated by multiplying the average daily drinks (drinks/day) by the duration of drinking (years). We then categorized the patients into 5 categories according to their drink-year levels (never, >0–40, >40–60, >60–90, and >90 drink-years). Furthermore, we generated a weekly index approximating total drinks per week by multiplying the drinking frequency values (never=0, former=0, a few days/week=1, almost every day/week=2) and the average daily drinks, as total drinks per week is a common indicator.³⁴ Treating the weekly index as a continuous value, we categorized the patients into 4 categories (never, low, middle, and high). In this classification, we separated the current drinkers into 3 quartiles of the weekly index (low, middle, and high) for both men and women.

Confounding variables included smoking history (never, former, or current) and lifestyle-related comorbidities (yes, no) such as hypertension, diabetes, hyperlipidemia, CVD, and obesity, in addition to the variables adjusted for matching (age, sex, admission year, and hospital admission).

Odds ratios (ORs) and 95% CIs for glaucoma prevalence were estimated for the 4 variables of alcohol consumption (drinking frequency, average daily drinks, total lifetime drinks, and weekly index) using conditional logistic regression models with multiple imputations. Since our analytic sample had 9.9% of missing variables for smoking history, we conducted multiple imputations and generated 5 imputed datasets for the missing data using multiple imputations by the chained equations method.³⁵ For Model 1, sex, age, admission date, and hospital admission were matched. For Model 2, smoking history and lifestyle-related comorbidities (hypertension, hyperlipidemia, diabetes, CVD, and obesity) were adjusted. The linear trend for the association between alcohol consumption patterns (drinking frequency, average daily drinks, and lifetime daily drinks) was evaluated by considering these variables as continuous variables. A test for interaction was conducted to examine the interaction by sex. An interaction term was generated by multiplying the dichotomized variable of sex (male=0, female=1) with categories of alcohol consumption patterns (as continuous variables, assigned ordinal numbers for each level) and added to Model 2. In the sensitivity analysis, these analyses were performed for patients without diabetes or CVD, patients aged 40–70 years, and patients over 70 years old. α was set at 0.05, and all P values were 2-sided. All analyses were performed using the Statistical Analysis System (SAS) Software version 9.4 (SAS Institute).

RESULTS

As shown in Figure 1, we included 3,469,498 records of 1,693,611 individuals in the 2015–2019 fiscal year and finally enrolled 3207 glaucoma cases and 3207 controls among these records. The baseline characteristics of patients and controls are shown in Table 1. The prevalence of diabetes and CVD was significantly higher in glaucoma cases (χ² tests, P<0.001 for both), and there were no significant differences in other factors between the two groups.

TABLE 1.

Characteristics of Glaucoma Cases and Controls

Characteristics	Controls, N (%)	Glaucoma, N (%)	P
Men	1616 (50.4)	1616 (50.4)	1.00
Age, y	73.7±10.3	73.7±10.2	0.93
>70 y old	2263 (70.6)	2263 (70.6)	1.00
Hypertension, yes	1140 (35.5)	1141 (35.6)	0.98
Diabetes, yes	475 (14.8)	613 (19.1)	<0.001
Hyperlipidemia, yes	310 (9.7)	334 (10.4)	0.32
Obesity, yes	298 (9.3)	282 (8.8)	0.49
Cardiovascular diseases, yes	27 (0.8)	141 (4.4)	<0.001
Smoking history, yes	1204 (37.5)	1212 (37.8)	0.38

Continuous variables are shown as mean (SD).

Categorical variables are shown as number (percentages).

t test was performed for continuous variables and χ² test for categorical variables.

The ORs for glaucoma prevalence according to the alcohol consumption category in the total are shown in Figure 2 and Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/IJG/A837, and ORs of confounding factors are shown in Supplemental Table 2, Supplemental Digital Content 2, http://links.lww.com/IJG/A838. Drinking frequency was associated with glaucoma for “a few days per week” and “almost every day per week.” Average daily drinks showed a relation for “>0–2 drinks per day,” but not a significant association for higher alcohol exposure. Total lifetime drinks showed a relation for “>60–90 drink-year,” and “>90 drink-year.” Overall, those who drank more frequently or in larger quantities had a higher relation. The trend of higher ORs for each alcohol consumption pattern was robust, as P values for the trends were <0.0001, 0.09, and 0.003 for drinking frequency, average daily drinks, and total lifetime drinks, respectively.

FIGURE 2.

Odds ratios of glaucoma prevalence in both sexes. This figure shows the multivariable odds ratios for glaucoma prevalence according to alcohol consumption categories in the total population. Drinking frequency was associated with a significant risk of glaucoma for “a few days per week” and “almost every day per week.” Average daily drinks showed a significant risk for “>0–2 drinks per day,” but not significant for higher alcohol exposure. Total lifetime drinks showed a significant risk for “>60–90 drink-year,” and “>90 drink-year.” Overall, those who drank more frequently or in larger quantities had a higher risk.

Tests for interaction by sex were conducted, considering that alcohol consumption levels differed considerably between men and women in our dataset. As the P values for interaction were 0.10, 0.87, and 0.78, respectively, for drinking frequency, average daily drinks, and total lifetime drinks, we could not dismiss the possibility of interaction by sex, especially for drinking frequency. Thus, we conducted additional separate analyses for men and women.

The ORs for glaucoma prevalence according to alcohol consumption categories for men and women are shown separately in Table 2. Among men, drinking frequency of “a few days per week” and “almost every day per week,” average daily drinks of “>0–2 drinks per day” and “>2–4 drinks per day,” and total lifetime drinks of “>60–90 drink-year” and “>90 drink-year” had a relation with glaucoma. Conversely, among women, neither drinking frequency, average daily drinks, nor total lifetime drinks were associated with glaucoma.

TABLE 2.

Odds Ratios for Glaucoma Prevalence by Alcohol Consumption Patterns in Men and Women

			aOR (95% CI)
	Controls, N (%)	Glaucoma, N (%)	Model 1	Model 2
Men
Drinking frequency
Never	424 (26.2)	355 (22.0)	1	1
Former	349 (21.6)	299 (18.5)	1.06 (0.85–1.32)	1.04 (0.83–1.30)
A few days/week	439 (27.2)	455 (28.2)	1.28 (1.05–1.57)	1.28 (1.04–1.58)
Almost every day/week	404 (25.0)	507 (31.4)	1.59 (1.29–1.95)	1.59 (1.29–1.97)
Trend			P<0.0001	P<0.0001
Average daily drinks
Never	424 (26.2)	355 (22.0)	1	1
>0–2 drinks	802 (49.6)	862 (53.3)	1.32 (1.10–1.58)	1.33 (1.11–1.60)
>2–4 drinks	288 (17.8)	300 (18.6)	1.30 (1.03–1.63)	1.29 (1.02–1.63)
>4 drinks	102 (6.3)	99 (6.1)	1.21 (0.88–1.65)	1.19 (0.86–1.66)
Trend			P=0.09	P=0.09
Total lifetime drinks
Never	424 (26.2)	355 (22.0)	1	1
>0–40 drink-year	207 (12.8)	202 (12.5)	1.19 (0.93–1.53)	1.17 (0.91–1.50)
>40–60 drink-year	166 (10.3)	173 (10.7)	1.28 (0.99–1.65)	1.29 (0.99–1.69)
>60–90 drink-year	230 (14.2)	251 (15.5)	1.37 (1.07–1.75)	1.38 (1.08–1.76)
>90 drink-year	589 (36.4)	635 (39.3)	1.35 (1.11–1.64)	1.34 (1.10–1.64)
Trend			P=0.003	P=0.003
Women
Drinking frequency
Never	1207 (75.9)	1192 (74.9)	1	1
Former	164 (10.3)	147 (9.2)	0.92 (0.73–1.17)	0.91 (0.72–1.16)
A few days/week	180 (11.3)	212 (13.3)	1.20 (0.96–1.50)	1.23 (0.98–1.53)
Almost every day/week	40 (2.5)	40 (2.5)	1.02 (0.64–1.63)	1.06 (0.66–1.70)
Trend			P=0.27	P=0.20
Average daily drinks
Never	1207 (75.9)	1192 (74.9)	1	1
>0–2 drinks	354 (22.3)	372 (23.4)	1.07 (0.90–1.28)	1.09 (0.91–1.30)
>2–4 drinks	27 (1.7)	18 (1.1)	0.69 (0.38–1.26)	0.68 (0.36–1.27)
>4 drinks	3 (0.2)	9 (0.6)	2.99 (0.81–11.08)	3.33 (0.88–12.61)
Trend			P=0.50	P=0.40
Total lifetime drinks
Never	1207 (75.9)	1192 (74.9)	1	1
>0–40 drink-year	201 (12.6)	196 (12.3)	0.99 (0.79–1.23)	1.01 (0.81–1.26)
>40–60 drink-year	61 (3.8)	55 (3.5)	0.91 (0.62–1.34)	0.92 (0.63–1.36)
>60–90 drink-year	61 (3.8)	66 (4.1)	1.09 (0.76–1.58)	1.14 (0.78–1.65)
>90 drink-year	61 (3.8)	82 (5.2)	1.38 (0.97–1.96)	1.42 (0.99–2.04)
trend			P=0.15	P=0.11

Model 1: Conditional logistic regression matched for sex, age, admission date, and hospital.

Model 2: Additionally adjusted for smoking history, hypertension, hyperlipidemia, diabetes, and obesity.

P for trend was calculated by converting each drinking status into continuous variables.

aOR indicates adjusted odds ratio.

Figure 3 exhibited the multivariable ORs for glaucoma prevalence according to the weekly index class in men (Fig. 3A) and women (Fig. 3B). Among men, the weekly index consistently showed a relation in the low, middle, and high classes. In contrast, among women, the weekly index showed a relation only for the middle class.

FIGURE 3.

Odds ratios of the weekly index in men and women. This figure shows the multivariable odds ratios for glaucoma prevalence according to the weekly index class in men (A) and women (B). In both sexes, the weekly index consistently showed a significant risk for glaucoma in the low, middle, and high classes. Among men, the weekly index also showed a significant risk for the low, middle, and high classes. Among the women, the weekly index had a significant risk only for the middle class.

The results of the sensitivity analysis are shown in Supplemental Figures A–F, Supplemental Digital Content 3, http://links.lww.com/IJG/A839. First, we performed additional analyses in cases with advanced glaucoma indicated for surgery (n=455) (Supplemental Fig. A, Supplemental Digital Content 3, http://links.lww.com/IJG/A839). As a result, the association between alcohol and glaucoma is greater in these patients. The results of the other analyses presented below were consistent with those of the main analysis. Due to the high prevalence of diabetes and CVD in glaucoma cases, a strong influence of diabetes and CVD on glaucoma has been suggested. Thus, we performed additional analyses in patients without diabetes and without CVD (Supplemental Figs. B, C, Supplemental Digital Content 3, http://links.lww.com/IJG/A839). As patients with missing values (n=675) might have had a bias in the distribution of covariates, we performed additional analyses in complete cases (Supplemental Fig. D, Supplemental Digital Content 3, http://links.lww.com/IJG/A839). Finally, analyses stratified by age under and over 70 years were performed (Supplemental Figs. E, F, Supplemental Digital Content 3, http://links.lww.com/IJG/A839). Furthermore, robust results were obtained for the case definition where only cases coded as H40.1 were included (Supplemental Table 3, Supplemental Digital Content 4, http://links.lww.com/IJG/A840).

DISCUSSION

In this large-scale case-control study, we enrolled 3207 glaucoma cases and 3207 matched controls and evaluated the association between alcohol consumption and glaucoma prevalence. As a result, the ORs for glaucoma prevalence were more significant for heavy alcohol consumption, and a positive dose-response association was observed. These results are particularly robust among men.

The association between alcohol and glaucoma was remarkable at higher drinking frequency. In previous reports, drinking frequency was well associated with mortality,³⁶ and reducing drinking frequency improved prognosis, even for heavy drinkers.³⁴ These reports suggest the importance of liver holidays and support our finding that drinking frequency is particularly detrimental. Regarding quantitative effects, the results for total lifetime drinks indicated that quantitative accumulation over time is crucial. However, the effect of the average daily drink intake was not statistically significant. We should be careful that the substantial effects would considerably differ even in the same daily drinks, depending on whether the alcohol was consumed once a week or every day. To deal with this problem, the total number of drinks per week is a common indicator.³⁴ However, we only obtained information such as “almost every day,” “a few days a week,” or “never drink.” Instead, we generated a weekly index that approximates the total drinks per week. Even when compared with the drinking frequency or average daily drinks, the weekly index yielded more robust results. In brief, alcohol consumption had a dose-response effect on glaucoma in both frequency and quantity.

Several studies have reported that alcohol consumption is a risk factor for glaucoma.^4,8–10 However, these results could not be simply compared as definitions of exposure and outcomes are different among studies. Although many case-control studies^{13–17,22,23} and cohort studies^10–12,21 have been conducted, the conclusions are controversial and thus, we compared our results with other case-control studies (Table 3). In a recent meta-analysis,⁴ Asians showed a strong association between alcohol consumption and glaucoma, which was consistent with our results in a Japanese population. In comparison to these reports, there are several other features in our study that might have led to differences in the results. First, we included a large sample size of over 3000 patients with a wide variety of alcohol consumption levels, which increased the statistical power. Second, we obtained detailed alcohol consumption patterns, such as frequency and average daily drinks, which might explain why alcohol was harmful, at least in high exposure.

TABLE 3.

Comparison of Previous Case-Control Studies in Different Countries and Races

Reference	Race	Case vs. control (N)	Exposure	OR (95% CI)
Renard et al13	American (white)	339 vs. 339	>3 drinks per day	0.81 (0.29–2.31)
			Heavy drinking	1.41 (0.82–2.45)
Leske et al15	American (white)	122 vs. 190	Drinking history	1.22 (0.66–2.24)
Leske et al16	American-African	67 vs. 219	Drinking history	0.80 (0.34–1.88)
Kaimbo et al17	African	40 vs. 104	Drinking history	0.96 (0.39–2.40)
Charliat et al14	European	175 vs. 175	Current drinking	1.00 (0.57–1.73)
Chiam et al22	Asian	180 vs. 1427	>2 d per week	1.27 (0.53–3.03)
The current study	Japanese	3207 vs. 3207	Almost everyday	1.40 (1.18–1.66)
			>4 drinks per day	1.13 (0.84–1.52)

OR indicates odds ratio.

In this study, the results for men alone were more robust than those for both sexes, with the ORs for glaucoma prevalence being more significant. In contrast, the results in women were not statistically significant. For this reason, especially among women, the alcohol consumption reported in surveys tends to be less than actual consumption,³⁷ introducing a potential information bias that could skew the results. In addition, sample size bias by alcohol consumption levels in women may have led to the differences in the results between sexes. Furthermore, biological or social differences between sexes might have influenced the results. Regarding biological differences, there might be a sex difference in IOP elevation due to alcohol consumption, which increases IOP in the long term.^38,39 As high IOP is a direct factor of glaucoma,³ it is plausible that there are sex differences in glaucoma prevalence. In addition, CVD caused by alcohol consumption can affect glaucoma.^18,19,40 As the concomitance of CVD and glaucoma was particularly high in men,⁴⁰ regarding CVD as an intermediate factor might explain sex differences in glaucoma induced by alcohol. In addition, social differences between sexes in drinking habits have been reported, such as the pace of drinking and the type or amount of alcohol consumed.⁴¹ For example, blood alcohol levels are less likely to increase in women with a slower drinking pace,⁴¹ which might lead to differences in results between men and women.

The reason why our study showed a relatively strong association between alcohol consumption and glaucoma might be that Asians, including the Japanese, are susceptible to alcohol. We hypothesized that there would be differences among races in their ability to metabolize alcohol. Acetaldehyde is a metabolite of alcohol that induces apoptosis and cytotoxicity in various organs through reactive oxygen species and impairment of mitochondrial function.⁴² Aldehyde dehydrogenase-2 (ALDH2) mitigates cellular and organ damage by metabolizing acetaldehyde.⁴³ In some individuals, the lower ALDH2 activity elicited by genetic variance may aggravate the cytotoxicity of acetaldehyde.⁴⁴ Meanwhile, Asians have genetic mutations in ALDH2 more frequently than other races.^45,46 Low ALDH2 activity in Asians could lead to vulnerability to acetaldehyde. In flushers, which have a low capacity to metabolize acetaldehyde, alcohol is associated with IOP elevation.⁴⁷ As elevated IOP is a definite risk factor for glaucoma,³ acetaldehyde may be an indirect threat to glaucoma. If acetaldehyde affects glaucoma, it is reasonable that alcohol consumption is strongly associated with glaucoma in Asian populations that metabolize acetaldehyde less effectively.

As for the mechanism of how alcohol consumption affects glaucoma, several mechanisms would be assumed. First, alcohol has been reported to elicit vascular diseases, such as hypertension, diabetes, and CVD.^18–20,48 These vascular diseases and atherosclerosis were associated with elevated IOP,^49–51 and high IOP is a definite risk for glaucoma.³ Furthermore, these vascular diseases have been also reported to be directly related to glaucoma,^40,52,53 and among them, our results regarding diabetes and CVD are consistent (Table 1, Supplemental Table 2, Supplemental Digital Content 2, http://links.lww.com/IJG/A838). Second, alcohol provokes neuropathy through reactive oxygen species and thiamine deficiency.⁵⁴ In particular, alcohol has been reported to be related to optic neurotoxicity and retinal nerve fiber layer thinning, possibly associated with glaucomatous neuropathy.^55–57 As we do not have detailed data regarding toxic neuropathy, further study on this mechanism is warranted. Finally, the drinking frequency may be particularly relevant. Generally, IOP fluctuations affect the incidence and progression of glaucoma.^58–61 IOP has been reported to decrease immediately after alcohol consumption.⁶² Hence, IOP fluctuations due to frequent alcohol consumption may lead to a high prevalence of glaucoma.

A sensitivity analysis of advanced glaucoma cases indicated for surgery showed a strong association between alcohol consumption and glaucoma. The results showing that advanced glaucoma patients had a stronger association with alcohol suggest a relation between alcohol and glaucoma progression. Further study into the effects on glaucoma progression was warranted. In addition, the result in patients with primary open angle glaucoma (H40.1) was also consistent with the main analysis.

Our study has several strengths: (1) we included a large sample size of over 3000 patients with a wide variety of alcohol exposures, leading to increased statistical power. (2) As this was a nationwide multicenter study, it automatically limited the bias of characteristics by region. In addition, matching by the hospital could adjust for differences in medical standards by region or hospital. (3) Our study performed multivariable analyses accounting for a variety of confounding factors, such as lifestyle-related diseases and smoking history. (4) Clinical information on the ICOD-R is reliable and filled out by physicians at each hospital.

Our study also has several limitations: (1) nondifferential misclassification of diseases for both cases and controls may have introduced selection bias in either direction (toward or away from the null). Due to the lack of detailed labels, we could not exclusively screen for primary open angle glaucoma in this study. Particularly, H40.9 may include other types of glaucoma. To deal with this problem, we conducted sensitivity analyses, including only H40.1 cases or only cases indicated for glaucoma surgery. These results were robust and in the same direction. (2) Our study is a hospital-based study including only inpatient data, which is prone to selection bias due to differences in characteristics such as frequent systemic disease complications. Therefore, to mitigate the bias, we selected controls from patients with no alcohol-related hospitalizations. In addition, we conducted sensitivity analyses to consider the potential heterogeneity in the patients’ background; the results of these analyses were robust. (3) The validity of the results in women was undermined by the small number of women in classes with high alcohol exposure. Further studies on the interaction by sex are warranted. (4) Data on diet or exercise were not obtained, and these confounding factors could not be considered. (5) Data on IOP or visual field test were not obtained, and the exact information of glaucoma was unknown. (6) Data on types of alcohol such as wine or beer were not obtained, and we could not evaluate the impact of these differences.

In conclusion, our study examined detailed alcohol consumption patterns and revealed a positive association between alcohol consumption and glaucoma prevalence. Frequency and quantity of alcohol consumption were influential because robust results were shown in the analysis including both. The results of this study will encourage further research on how drinking habits affect glaucoma incidence and progression. We believe the current study is a crucial step toward elucidating the epidemiology and pathogenesis of glaucoma.

Supplementary Material

ijg-32-968-s001.pdf ^{(76.2KB, pdf)}

ijg-32-968-s002.pdf ^{(69.5KB, pdf)}

ijg-32-968-s003.pdf ^{(1.1MB, pdf)}

ijg-32-968-s004.pdf ^{(75.4KB, pdf)}

ijg-32-968-s005.pdf ^{(116.6KB, pdf)}