DOI: https://doi.org/10.21203/rs.3.rs-1845619/v1
Objective: To investigate the association between glaucoma and mortality in the older population.
Design: Population-based, prospective cohort study.
Participants: Participants aged 45 years or older at baseline (47.9% male) were enrolled in 2011 for the China Health and Retirement Longitudinal Study (CHARLS). All-cause mortality of the participants was observed during seven years of follow-up.
Methods: The baseline data were collected in the 2011 CHARLS, and participants were followed up for seven years (until 2018). The risk of all-cause mortality was examined using Cox proportional hazards regression with age as the time scale, adjusting for significant risk factors and comorbid conditions.
Main outcome measures: Mortality, resulting from all causes.
Results: Among the 14,803 participants included, the risk of all-cause death was significantly higher among people with glaucoma than among those without glaucoma, after adjustment for other confounders (hazard ratio [HR]: 2.159, 95% confidence interval [CI]: 1.549-3.008). In a subgroup analysis based on the mean age of death, among those who were 75 years and older (n = 563), the risk of all-cause death was significantly higher in patients with glaucoma than in those without glaucoma (HR: 1.907, 95% CI: 1.249-2.911).
Conclusions: Participants with glaucoma were at an increased risk for all-cause mortality, especially those participants aged 75 years and above. Our findings revealed possible underlying mechanisms creating the association between glaucoma and all-cause mortality, and they highlighted the importance of glaucoma management to prevent premature death in middle-aged and older adults.
Glaucoma is a chronic, progressive disease characterized by structural changes to the optic nerve head with visual field loss [1]. Worldwide, the number of glaucoma patients has increased rapidly in recent decades, and it is expected to reach 112 million by 2040 [2]. As the most frequent cause of blindness [3, 4], glaucoma has a serious impact on the life expectancy of patients, especially for the elderly. A possible association between glaucoma and mortality has been emphasized since 1950 [5].
The relationship between glaucoma and mortality varies significantly in different studies. The Beijing Eye Study, using a multivariate model, observed an increased mortality rate in closed-angle glaucoma patients [6]. Furthermore, studies from Taiwan, analysing health insurance data, and from an Indian rural cohort both reported that primary open angle glaucoma (POAG) was associated with increased 10-year-mortality [7, 8]. However, a meta-analysis of nine studies presents no convincing evidence for an increased risk of all-cause mortality among those with POAG [9]. Similarly, there is no evidence of an association between POAG and survival in a Scandinavian cohort that was followed for up to 30 years [10]. Interestingly, in the population-based National Health and Nutrition Examination Survey (NHANES), a representative health survey conducted in the United States, glaucoma was related to increased mortality in an unadjusted Cox regression model, but the correlation no longer existed when the model was adjusted for age and sex [11].
In view of the anticipated increase in the incidence of glaucoma worldwide and the uncertainty of the evidence concerning the associated risk of death, there is a need for further research into the life expectancy of glaucoma patients [9, 12]. Therefore, our research evaluated the relationship between glaucoma and mortality through the population-based China Health and Retirement Longitudinal Study (CHARLS). We focused on an assessment of whether glaucoma is associated with increased mortality in subjects affected by comorbidities (such as diabetes or hypertension) and considered other potentially confounding factors.
The China Health and Retirement Longitudinal Study (CHARLS) is a nationally representative longitudinal survey of the Chinese middle-aged and elderly population (45 years old and above), approved by the Biomedical Ethics Review Committee of Peking University. The national public database covers a wide range of information, concerning family, health status and function, health care and insurance, work, retirement and pension and physical measurements. A multistage probability sampling method was applied through CHARLS, which covered 28 provinces, 150 counties/districts, and 450 villages or urban communities across China. A total of 17,705 individuals were recruited between June 2011 and March 2012 [13]. All participants underwent face-to-face computer-assisted personal interviews, which included a structured questionnaire, and they were followed up every two years. Physical measurements, including height, weight and blood pressure, were evaluated at every two-year follow-up, and blood samples were collected once in every two follow-up cycles. More details of the CHARLS are described elsewhere [14].
The presence of glaucoma was defined by a self-report answer to the question ‘Has a doctor/nurse/paramedical ever treated you for glaucoma?’ If the respondent answered ‘yes’ to this question, the conclusion was that they had glaucoma.
Participants enrolled in wave 1 (2011, baseline) were followed up in waves 2, 3 and 4 (2013, 2015 and 2018). Interview status (dead or alive) and date of death were recorded in wave 4. For those who had died from any cause, the survival time was calculated as the interval between the baseline and the date of death. If the exact date of death was not available, the survival time was defined as the median of the interval between the baseline time and the specific wave where the death information was received. For those who had not died during the follow-up period, their survival time was considered as the interval between two interview waves.
Information that included age, gender, education level (primary or below, middle school, high school, or college and above), marital status (married or partnered, or otherwise), smoking (yes or no), drinking (drinking more than once a month, drinking less than once a month, or none) was collected by trained interviewers using the 2011 CHARLS questionnaire. Respondents were categorized into different groups. Hypertension was defined as blood pressure ≥ 140/90 mmHg at baseline or a self-reported history of antihypertensive medication use. Dyslipidaemia was defined by having a self-reported dyslipidaemia history or as having more than one of the following: total cholesterol (TC) ≥ 6.2mmol/L, low-density lipoprotein cholesterol (LDLC) ≥ 4.1mmol/L, triglycerides (TG) ≥ 2.3mmol/L, and high-density lipoprotein cholesterol (HDLC) < 1.0mmol/L. Diabetes was defined as a self-reported diagnosis of diabetes or as a fasting plasma glucose ≥ 7.0 mmol/L or as glycated haemoglobin ≥ 6.5% [15].
A total of 8 ml of fasting blood samples was collected by trained nurses at township health centres, or at the local offices of the Chinese Centre for Disease Control and Prevention (CDC) after participants had fasted for at least eight hours overnight. The whole blood and centrifugal serum were transported to the CDC at -20℃ and stored at -80℃. The plasma glucose, lipids and glycosylated haemoglobin were then measured in a laboratory at Capital Medical University.
Continuous data were expressed as means ± standard deviation (SD) and analysed by student’s t-test for group comparisons, whereas categorical data were presented as numbers (n) with percentages (%) and analysed using the chi-square test for group comparisons. The overall significance of the univariate survival analysis was determined by the log-rank test using the Kaplan-Meier analysis. Cox proportional-hazards regression analyses were carried out to obtain hazard ratios (HR) and 95% confidence intervals (CI) according to glaucoma status and for risk factors significantly associated with mortality. The Cox proportional-hazards regression model was adjusted for potentially confounding factors. Subgroup analysis was conducted based on the average age at death, and the population was divided into two subgroups: those aged between 45 and 75 years and those over 75 years. All statistical analyses were conducted using the Statistical Analysis System (SAS) software version 9.4 (SAS Institute, Cary, NC, United States), and a two-sided P value < 0.05 was considered statistically significant.
Table 1 shows the baseline characteristics of the participants in our study. A total of 14,803 participants met the criteria for enrolment, of which 141 individuals (0.95%) were diagnosed with glaucoma. Compared to the participants with glaucoma, those without glaucoma were younger (66.6 ± 10.1, versus 59.0 ± 9.9, P < 0.001), were more likely to be male, educated, married or partnered, a smoker, more likely to consume alcohol, and less likely to suffer hypertension, dyslipidaemia or diabetes.
Variables |
Glaucoma |
P-value |
|
---|---|---|---|
Yes (n = 141) |
No (n = 14662) |
||
Mean age (yrs) |
66.6 (10.1) |
59.0 (9.9) |
< 0.001 |
Gender |
< 0.001 |
||
Male |
33.3% (47) |
48.0% (7039) |
|
Female |
66.7% (94) |
52.0% (7623) |
|
Education |
0.013 |
||
Primary or below |
80.9% (114) |
68.2% (9996) |
|
Middle school |
11.3% (16) |
20.6% (3014) |
|
High school |
5.7% (8) |
7.3% (1069) |
|
College or above |
2.1% (3) |
4.0% (583) |
|
Marital status |
< 0.001 |
||
Married or partnered |
78.0% (110) |
87.8% (12869) |
|
Otherwise |
22.0% (31) |
12.2% (1793) |
|
Smoking |
0.036 |
||
Yes |
31.2% (44) |
39.9% (5849) |
|
No |
68.8% (97) |
60.1% (8813) |
|
Drinking |
0.031 |
||
None |
77.3% (109) |
66.8% (9801) |
|
Drink but less than once a month |
5.7% (8) |
7.8% (1138) |
|
Drink more than once a month |
17.0% (24) |
25.4% (3723) |
|
Hypertension |
< 0.001 |
||
Yes |
36.9% (52) |
24.0% (3519) |
|
No |
63.1% (89) |
76.0% (11143) |
|
Dyslipidaemia |
0.003 |
||
Yes |
17.0% (24) |
9.0% (1315) |
|
No |
83.0% (117) |
91.0% (13347) |
|
Diabetes |
< 0.001 |
||
Yes |
14.2% (20) |
5.5% (803) |
|
No |
85.8% (121) |
94.5% (13859) |
|
Data are presented as % (N) or mean (standard deviation). Comparisons were for the glaucoma group with the No glaucoma group. |
Figure 1 presents the survival probability of the populations with and without glaucoma by Kaplan-Meier analysis. In general, across the whole age group, the survival probability of individuals diagnosed with glaucoma was significantly lower compared to the non-glaucoma population (Fig. 1A, P < 0.001). We further stratified patients by average age of death and found that survival probability was significantly associated with glaucoma status in people aged 75 years and older (Fig. 1C, P = 0.002), but not in people aged 45–74 years (Fig. 1B, P = 0.102).
During the seven-year follow-up, there were 1627 deaths among all participants. The participants who died with glaucoma were significantly older than those who died without glaucoma (75.94 years versus 68.99 years, P < 0.001), were less likely to be male (38.9% versus 58.9%, P = 0.016) and were less likely to smoke (30.6% versus 51.0%, P = 0.015) at baseline (Table 2). Considering the strong association between age and death over time, we further stratified by age according to average age at death to explore whether glaucoma status was associated with all-cause mortality. The characteristics of patients who died according to glaucoma status, and stratified by age group, are shown in Table 3. Among individuals who died aged between 45 to 74 years, there was a difference in age distribution between the glaucoma and non-glaucoma groups.
Variables |
Glaucoma |
P-value |
|
---|---|---|---|
Yes (n = 36) |
No (n = 1591) |
||
Mean age (yrs) |
75.94 (8.078) |
68.99 (10.887) |
< 0.001 |
Gender |
0.016 |
||
Male |
38.9% (14) |
58.9% (937) |
|
Female |
61.1% (22) |
41.1% (654) |
|
Education |
0.064 |
||
Primary or below |
94.4% (34) |
81.0% (1288) |
|
Middle school |
2.8% (1) |
13.6% (216) |
|
High school |
2.8% (1) |
2.7% (43) |
|
College or above |
0.0% (0) |
2.8% (44) |
|
Marital status |
0.137 |
||
Married or partnered |
61.1% (22) |
72.3% (1151) |
|
Otherwise |
38.9% (14) |
27.7% (440) |
|
Smoking |
0.015 |
||
Yes |
30.6% (11) |
51.0% (811) |
|
No |
69.4% (25) |
49.0% (780) |
|
Drinking |
0.102 |
||
None |
86.1% (31) |
69.6% (1108) |
|
Drink but less than once a month |
2.8% (1) |
5.2% (82) |
|
Drink more than once a month |
11.1% (4) |
25.2% (401) |
|
Hypertension |
0.585 |
||
Yes |
30.6% (11) |
34.9% (556) |
|
No |
69.4% (25) |
65.1% (1035) |
|
Dyslipidaemia |
0.396 |
||
Yes |
11.1% (4) |
8.8% (140) |
|
No |
88.9% (32) |
91.2% (1451) |
|
Diabetes |
0.297 |
||
Yes |
13.9% (5) |
8.9% (141) |
|
No |
86.1% (31) |
91.1% (1450) |
|
Data are presented as % (N) or mean (standard deviation). Comparisons were for the glaucoma group with the No glaucoma group. |
Variables |
45 to 74 Years of Age (n = 1064) |
75 + Years of Age (n = 563) |
|||||
---|---|---|---|---|---|---|---|
With glaucoma (n = 13) |
Without glaucoma (n = 1051) |
P-value |
With glaucoma (n = 23) |
Without glaucoma (n = 540) |
P-value |
||
Mean age (yrs) |
67.4(6.7) |
62.9(7.8) |
0.041 |
80.8(3.4) |
80.7(4.6) |
0.979 |
|
Gender |
0.443 |
0.086 |
|||||
Male |
53.8% (7) |
64.1% (674) |
30.4% (7) |
48.7% (263) |
|||
Female |
46.2% (6) |
35.9% (377) |
69.6% (16) |
51.3% (277) |
|||
Education |
0.080 |
0.774 |
|||||
Primary or below |
92.3% (12) |
73.8% (776) |
95.7% (22) |
94.8% (512) |
|||
Middle school |
0(0) |
19.4% (204) |
4.3% (1) |
2.2% (12) |
|||
High school |
7.7% (1) |
3.7% (39) |
0(0) |
0.7% (4) |
|||
College or above |
0(0) |
3.0% (32) |
0(0) |
2.2% (12) |
|||
Marital status |
0.361 |
0.411 |
|||||
Married or partnered |
92.3% (12) |
82.7% (869) |
43.5% (10) |
52.2% (282) |
|||
Otherwise |
7.7% (1) |
17.3% (182) |
56.5% (13) |
47.8% (258) |
|||
Smoking |
0.223 |
0.120 |
|||||
Yes |
38.5% (5) |
55.4% (582) |
26.1% (6) |
42.4% (229) |
|||
No |
61.5% (8) |
44.6% (469) |
73.9% (17) |
57.6% (311) |
|||
Drinking |
0.542 |
0.130 |
|||||
None |
76.9% (10) |
66.0% (694) |
91.3% (21) |
76.7% (414) |
|||
Drink but less than once a month |
7.7% (1) |
5.6% (59) |
0 (0) |
4.3% (23) |
|||
Drink more than once a month |
15.4% (2) |
28.4% (298) |
8.7% (2) |
19.1% (103) |
|||
Hypertension |
0.824 |
0.498 |
|||||
Yes |
30.8% (4) |
33.7% (354) |
30.4% (7) |
37.4% (202) |
|||
No |
69.2% (9) |
66.3% (697) |
69.6% (16) |
62.6% (338) |
|||
Dyslipidaemia |
0.484 |
0.790 |
|||||
Yes |
15.4% (2) |
9.6% (101) |
8.7% (2) |
7.2% (39) |
|||
No |
84.6% (11) |
90.4% (950) |
91.3% (21) |
92.8% (501) |
|||
Diabetes |
0.142 |
0.555 |
|||||
Yes |
23.1% (3) |
10.5% (110) |
8.7% (2) |
5.7% (31) |
|||
No |
76.9% (10) |
89.5% (941) |
91.3% (21) |
94.3% (509) |
|||
Data are presented as % (N) or mean (standard deviation). Comparisons were for the glaucoma group with the No glaucoma group. |
Risk factors associated with all-cause mortality are presented in Table 4 and Table 5. The multivariate Cox regression analysis indicates that individuals with glaucoma had an increased risk of all-cause mortality compared with those without glaucoma (HR: 2.159, 95% CI: 1.549–3.008). In addition, lifestyle factors, such as smoking and drinking, and chronic systemic diseases, including hypertension and diabetes, increased the risk of all-cause mortality. Subjects with a low education level also had an increased risk of all-cause mortality. Being married, or living with a partner, decreased the risk of all-cause mortality.
Variable |
HR (95%CI) |
P-value |
---|---|---|
Glaucoma: Yes vs. No |
2.159 (1.549,3.008) |
< 0.001 |
Gender: Male vs. Female |
1.878 (1.640,2.152) |
< 0.001 |
Education: Primary or below vs. College or above |
1.842 (1.360,2.494) |
< 0.001 |
Middle school vs. College or above |
1.027 (0.742,1.421) |
0.874 |
High school vs. College or above |
0.598 (0.394,0.909) |
0.056 |
Marital status: Married or partnered vs. Otherwise |
0.368 (0.329,0.411) |
< 0.001 |
Smoking: Yes vs. No |
1.314 (1.154,1.496) |
< 0.001 |
Drinking: Drink but less than once a month vs. Drink more than once a month |
0.720 (0.635,0.816) |
< 0.001 |
None vs. Drink more than once a month |
0.559 (0.446,0.700) |
< 0.001 |
Hypertension: Yes vs. No |
1.619 (1.455,1.801) |
< 0.001 |
Dyslipidaemia: Yes vs. No |
0.805 (0.672,0.965) |
0.059 |
Diabetes: Yes vs. No |
1.587 (1.327,1.897) |
< 0.001 |
Table 5. Cox proportional hazards regression models of all-cause mortality by glaucoma in different age group
Variable |
Participants 45 to 74 Years of Age (n = 13572) |
Participants 75 + Years of Age (n = 1231) |
||
---|---|---|---|---|
HR (95%CI) |
P-value |
HR (95%CI) |
P-value |
|
Glaucoma: Yes vs. No |
1.398 (0.807,2.419) |
0.232 |
1.907 (1.249,2.911) |
0.003 |
Gender: Male vs. Female |
2.052 (1.725,2.441) |
< 0.001 |
1.218 (0.981,1.513) |
0.074 |
Education: Primary or below vs. College or above |
1.767 (1.237,2.525) |
0.002 |
1.885 (1.054,3.371) |
0.033 |
Middle school vs. College or above |
1.296 (0.891,1.884) |
0.175 |
0.734 (0.334,1.616) |
0.443 |
High school vs. College or above |
0.711 (0.446,1.132) |
0.151 |
2.095 (0.673,6.519) |
0.202 |
Marital status: Married or partnered vs. Otherwise |
0.507 (0.432,0.596) |
< 0.001 |
0.823 (0.689,0.984) |
0.033 |
Smoking: Yes vs. No |
1.403 (1.192,1.652) |
< 0.001 |
1.274 (1.039,1.562) |
0.020 |
Drinking: Drink but less than once a month vs. Drink more than once a month |
0.745 (0.642,0.866) |
< 0.001 |
0.749 (0.595,0.943) |
0.014 |
None vs. Drink more than once a month |
0.561 (0.429,0.733) |
< 0.001 |
0.751 (0.492,1.146) |
0.185 |
Hypertension: Yes vs. No |
1.619 (1.416,1.850) |
< 0.001 |
1.224 (1.025,1.463) |
0.026 |
Dyslipidaemia: Yes vs. No |
0.818 (0.658,1.015) |
0.068 |
0.933 (0.670,1.299) |
0.680 |
Diabetes: Yes vs. No |
1.939 (1.576,2.385) |
< 0.001 |
1.003 (0.698,1.441) |
0.987 |
After being stratified by age, persons with glaucoma aged over 75 years showed an increased risk of all-cause mortality after considering age and concomitant conditions (HR: 1.907, 95% CI: 1.249–2.911, Table 5). The predictive factors for mortality among individuals aged over 75 years included less education, never having married or had a partner, smoking and hypertension. In the younger group, the factors predictive of mortality included being of male sex, less education, never having married or had a partner, smoking, drinking, hypertension and diabetes.
Overall, our current results based on a representative Chinese population strongly suggests that glaucoma is a high-risk factor for all-cause mortality in the middle-aged and elderly population, especially among people older than 75 years. In our study, among participants of all ages, the presence of glaucoma is accompanied by an increased risk of mortality resulting from all causes (HR: 2.159, 95% CI: 1.549–3.008), even after considering other conditions that might affect the risk of mortality, such as smoking, drinking, hypertension, dyslipidaemia and diabetes. When stratified by mean age of death, glaucoma significantly increases all-cause mortality in individuals in their mid-70s and above (HR: 1.907, 95% CI: 1.249–2.911). However, among younger individuals, of 45 to 74 years of age, those with glaucoma are not at higher risk compared with their same-age peers.
The results of our study are supported by many previous findings [6, 7, 16–21]. The National Health Interview Survey (NHIS) [21] reported that the probability of death from any cause occurring over a median of seven years of follow-up was higher (HR: 1.35, 95% CI: 1.19–1.53) among participants with glaucoma compared to those without glaucoma, even after adjustment for confounders. The NHIS survey also reported an increased risk of mortality from cardiovascular disease among participants with glaucoma. Similarly, a Taiwanese study [7] demonstrated higher mortality associated with POAG (adjusted HR: 2.11, 95% CI: 1.76–2.54), and it mentioned an association between glaucoma and a higher risk of acute renal failure. Moreover, the Beijing study, which examined 4356 subjects for glaucoma, suggesting that glaucoma, and angle-closure glaucoma in particular, might be associated with an increased rate of mortality among Chinese adults in Greater Beijing [6]. Our study supports and extends these findings reporting an association between glaucoma and mortality, and it enhances the understanding of glaucoma’s impact on the older population.
In contrast to our study, the NHANES study showed no statistical association between increased mortality and glaucoma after adjusting for age and sex [11]. Several studies confirmed an association between glaucoma and mortality in a univariate analysis but disproved the association after adjustment for confounders [8, 22]. In general, the relationship between glaucoma and mortality varies considerably across different studies, but this may have relevance to factors such as race, geography, sampling quantity and method. These factors may explain the differences in results between those studies and ours.
It is worth noting that our results elucidate the association between glaucoma and all-cause mortality in different age groups by showing that, in people older than 75 years, age is an important component of increased all-cause mortality from glaucoma. This may be because aging is one of the main factors that promotes the course of glaucoma and affects its treatment [23]. Aging will aggravate the severity of the disease and promote an increase in all-cause mortality.
Potential mechanisms that may explain the increased risk of mortality in adults with glaucoma include adverse treatment effects and exposure to risk factors known to increase the risk of glaucoma and major cause-specific deaths. First, adverse treatment effects from glaucoma surgery and medications are reported to be associated with high all-cause mortality rates: A retrospective cohort study from Korea indicated that all-cause mortality due to surgery for glaucoma was statistically significant (adjusted HR: 1.31, 95% CI: 1.05–1.62), both for open angle glaucoma and for angle closure glaucoma [24]. The risk of death from neurological disease was 2.7 times higher in older patients who underwent glaucoma surgery than in those who only received a diagnosis of glaucoma. Glaucoma medications can also cause severe side effects, including congestive heart failure (topical cholinergic agonists), increased blood pressure and tachyarrhythmias (topical adrenergic agonists) [25]. Topical administration of beta-blockers is one of the most common treatments for glaucoma, however, this is revealed to be associated with increased mortality. An Australia study in 2006 (relative risk [RR]: 2.14, 95% CI: 1.18–3.89) and research conducted in American in 2008 (RR: 1.91, P = 0.04) both showed higher mortality among glaucoma patients treated with topical timolol [26, 27]. This could be attributed to the drug’s effect on blood lipid levels, such as a 20% reduction in HDL and a 20–40% increase in triglycerides [28]. However, the conclusions were contradictory. An American retrospective cohort analysis indicated that use of any type of glaucoma medication had a statistically significant association with a 7% reduction of mortality (adjusted HR: 0.93, 95% CI: 0.90–0.95) [29]. Another American study, from the same year, showed similar results (HR: 0.26, 95% CI: 0.16–0.40) [25]. Use of topical beta-blockers did not seem to be associated with excess mortality [30]. Second, exposure to risk factors known to increase the risk of both glaucoma and major cause-specific deaths (e.g., neurological disease, cardiovascular disease) also possibly contribute to glaucoma-associated high levels of mortality. The association between neurologic diseases and glaucoma can be interpreted through the mechanisms that cause haemodynamic changes in the cerebral arteries and pathological substances, such as protein tau and amyloid-beta, that cause neurotoxicity and are also strongly associated with optic nerve damage in glaucoma [31]. It is suggested that stroke [32–34], Alzheimer’s disease [35–37] and Parkinson’s disease [38–40] are neurological diseases related to glaucoma. There was evidence that mortality in glaucoma patients was closely associated with cardiovascular events. A study in Australia indicated that cardiovascular mortality was 14.6% in patients with glaucoma as compared to 8.4% in non-glaucoma individuals, while further stratified analyses showed that cardiovascular mortality was higher among those previously diagnosed with glaucoma (RR: 1.85, 95% CI: 1.12–3.04) [26].
In addition, our study has suggested other independent risk factors associated with mortality, which include comorbidities, such as hypertension and diabetes, lifestyle factors, such as smoking and alcohol consumption, and social status, including marital and educational status. Regarding the relationship between hypertension and mortality, a meta-analysis in Japan of 13 cohort studies and a prospective cohort study in Shanghai (HR: 1.26, 95%CI: 1.02–1.55) suggested that high blood pressure increased the risk of all-cause mortality, and that this trend was more pronounced among younger individuals [41, 42], which was consistent with our study. This might be due to elevated blood pressure being associated with a range of cardiovascular outcomes including ischemic heart disease, myocardial infarction, ischemic stroke, sudden cardiac death, heart failure, atrial fibrillation and pulmonary embolism [43–45]. Our results indicate that diabetes significantly increase all-cause mortality, but only in the younger age group. This view is supported by several studies that pointed out that young-onset diabetes led to premature death, possibly due to an increase in complications such as heart disease, stroke and chronic kidney disease [46, 47]. In addition, low educational attainment, a non-married status, smoking and excessive alcohol consumption were also shown to be associated with higher mortality, which were consistent with our findings [21, 22, 48].
The strengths of our study include its nationwide representativity, the large elderly cohort with a high participation rate, the objective quality standards and the complete, adjudicated registry of deaths. As a nationally representative longitudinal survey, middle-aged and older population groups were included, with a high response rate. Moreover, matching using the global positioning system, data checking, recording and checking interviews, and calling participants back, were implemented at every stage of the study to ensure data quality and reliability.
Several study limitations should also be noted. First, we used self-reported measures of glaucoma, which, while subjective, had the advantage of feasibility at a low cost. To some extent, this avoided ambiguous diagnoses relating to elevated intra-ocular pressure or cup-to-disc-ratio, while ignoring other ophthalmic indicators, as occurred in some studies [11]. Second, we did not collect information on the use of antiglaucoma medication, which might have explained the increased risk of mortality. Therefore, there is a lack of data on the causes of death with the possibility that the findings might be explicable by confounders as yet unknown.
To summarize, the present study suggests that glaucoma may be associated with an increased rate of mortality in middle aged and elderly people in China. This provides an important reference for the design and evaluation of glaucoma treatment and for resolving patient management issues.
Acknowledgments
Ethics approval
CHARLS was approved by the Ethical Review Committee at Peking University.
Funding
This research was supported by the National Natural Science Foundation of China (82171076), Science and Technology Commission of Shanghai Municipality (20Z11900400), Shanghai Hospital Development Center (SHDC2020CR2040B, SHDC2020CR5014), and Shanghai Collaborative Innovation Center for Translational Medicine (TM202115PT).
Authorship contributions
XH, XZ, and XS conceived the cross-sectional study. MX and MZ were statistician and undertook the secondary analysis. WL and MZ checked the statistical methods and analysis results. XH, XZ and XS jointly drafted the manuscript, which was contributed to by MX and WL. All authors approved the final version of the manuscript.