Article Text

Download PDFPDF

Glaucoma-associated long-term mortality in a rural cohort from India: the Andhra Pradesh Eye Disease Study
  1. Rohit C Khanna1,2,
  2. Gudlavalleti V S Murthy3,4,
  3. Pyda Giridhar1,2,
  4. Srinivas Marmamula1,2,5,
  5. Hira B Pant4,
  6. Ghanshyam Palamaner Subash Shantha6,
  7. Subhabrata Chakrabarti2,
  8. Clare E Gilbert3,
  9. Gullapalli Nageswara Rao1,2
  1. 1 Allen Foster Community Eye Health Research Centre, Gullapalli Pratibha Rao International Centre for Advancement of Rural Eye Care, L V Prasad Eye Institute, Hyderabad, India
  2. 2 Brien Holden Eye Research Centre, L V Prasad Eye Institute, Hyderabad, India
  3. 3 Department of Clinical Research, International Centre for Eye Health, London School of Hygiene and Tropical Medicine, London, UK
  4. 4 Indian Institute of Public Health, Hyderabad, India
  5. 5 Wellcome Trust, Department of Biotechnology (DBT) India Alliance, L V Prasad Eye Institute, Hyderabad, India
  6. 6 Division of Cardiovascular Medicine, Roy and Lucille J. Carver College of Medicine, University of Iowa Hospitals and Clinics, Lowa City, Lowa, USA
  1. Correspondence to Dr Rohit C Khanna, L V Prasad Eye Institute, Hyderabad 500034, India; rohit{at}


Aim To evaluate glaucoma-associated mortality in a rural cohort in India.

Methods The study cohort comprised individuals aged 40 years and above who took part in the Andhra Pradesh Eye Disease Study (APEDS1) during 1996–2000. All participants underwent detailed comprehensive eye examination. Glaucoma was defined using International Society of Geographic and Epidemiologic Ophthalmology criteria. This cohort was followed up after a decade (June 2009 to January 2010; APEDS2). Mortality HR analysis for ocular risk factors was performed using Cox proportional hazards regression after adjusting for sociodemographic, lifestyle and clinical variables.

Results In APEDS1, 2790 individuals aged more than or equal to 40 years were examined. 47.4% were male. Forty-five participants had primary open angle glaucoma (POAG) and 66 had primary angle closure disease (PACD). Ten years later, 1879 (67.3%) were available, 739 (26.5%) had died and 172 (6.2%) had migrated; whereas 22 of the 45 (48.8%) with POAG and 22 of the 66 (33.3%) with PACD had died. In univariate analysis, a higher mortality was associated with POAG (HR 1.9; 95% CI 1.23 to 2.94), pseudoexfoliation (HR 2.79; 95% CI 2.0 to 3.89), myopia (HR 1.78; 95% CI 1.54 to 2.06) and unit increase in cup:disc ratio (HR 4.49; 95% CI 2.64 to 7.64). In multivariable analysis, only cup:disc ratio remained independently associated with mortality (HR 2.5; 95% CI 1.3 to 5.1). The association remained significant when other ocular parameters were included in the model (HR 2.1; 95% CI 1.03 to 4.2).

Conclusions This is the first longitudinal study to assess the association of glaucoma and mortality in a rural longitudinal cohort in India. Increased cup:disc ratio could be a potential marker for ageing and would need further validation.

  • Glaucoma
  • Mortality

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Glaucoma is the leading cause of irreversible blindness and visual impairment (VI) globally.1 The number of glaucoma cases worldwide was estimated to be 64.3 million in 2013, which is likely to increase to 76 million by the year 2020 and 111.8 million by 2040.1 In India, approximately 11.2 million adults above 40 years are estimated to have glaucoma: 6.5 million with primary open angle glaucoma (POAG) and 2.5 million with primary angle closure glaucoma (PACG).2 Several studies have shown association of cataract with mortality.3–7 However, the association between glaucoma and mortality is inconsistent with some studies reporting an association8 9 while others have not.10–16 Most of these studies were undertaken in Caucasian populations and most focused on POAG—only one study investigated a black population.16 In the only study from Asia, there was a strong association between mortality and PACG but not with POAG.17 Reports on the association of mortality with raised intraocular pressure (IOP) and pseudoexfoliation (PXF) are also inconsistent with some showing an association10 16 whereas others have not.12 18–20 In this study, the association between mortality and all forms of glaucoma (POAG and primary angle closure disease (PACD)), as well as IOP, PXF and cup:disc ratio (CDR) was investigated in a very well-characterised population-based sample of adults recruited to the Andhra Pradesh Eye Disease Study (APEDS).

Materials and methods

Details of the methodology for APEDS, which was conducted between 1996 and 2000, are described elsewhere.3 Data on mortality were obtained during an exercise undertaken between June 2009 and January 2010, the purpose of which was to identify as many participants as possible from the original APEDS, and recruit them for subsequent re-examination. For brevity, the original APEDS survey will be called APEDS1 and the present study will be referred to as APEDS2.

Detailed protocols for APEDS1 have been published.21 22 In brief, standard examination included distance and near visual acuity at presentation (PVA) and best corrected visual acuity following refraction measured for each eye separately using logMAR (logarithm of minimum angle of resolution) charts. Detailed ocular examination included slit lamp biomicroscopy, IOP measurement by Goldmann applanation tonometry and gonioscopy with an NMR-K-2-mirror. Participants with suspicion of angle closure underwent laser iridotomy before dilated examination. Lens opacities were graded using the Lens Opacities Classification System III and the Wilmer classification23 24 after pupil dilation. Optic discs were evaluated using a 78 dioptre (D) lens and the peripheral fundus was examined with a 20 D lens. Visual fields were assessed with automated Humphrey Visual Field (HVF) Analyzer for those with any of the following features suggestive of glaucomatous disc damage: vertical CDR of 0.65 or more in either eye; asymmetry of CDR between eyes (≥0.2); notch; haemorrhage; rim <0.2 in any quadrant; nerve fibre layer defect; peripapillary chorioretinal atrophy (alpha or beta); and glaucomatous optic atrophy and non-glaucomatous optic atrophy. HVF was also performed if the IOP was ≥22 mm Hg in either eye or if there was a difference in IOP of ≥6 mm Hg between eyes. Visual field analysis was repeated if it was unreliable. Anderson’s criteria were used to determine glaucomatous visual field defects: a field defect that correlated with optic disc damage and met two of the three Anderson’s criteria was considered to be significant.

Study definitions

These too have been described in our previous publication.3 21 22 Blindness was defined as presenting visual acuity (PVA) less than 6/60 or central visual field less than 20° in the better eye.25 VI was defined as PVA less than 6/18-6/60 or equivalent visual field loss.21 The lens was examined after fully dilating the pupils. Because different types of cataract frequently coexist, for analysis we considered pure nuclear, pure cortical, pure posterior subcapsular cataract (PSC) and mixed cataract. Those with total cataract were categorised as mixed cataract and those having undergone cataract surgery (unilateral/bilateral) formed a separate group.26 Age-related macular degeneration (ARMD) was defined using the International Classification and Grading System27 and diabetic retinopathy was defined using a modification of the standard classification system.28 Glaucoma was defined using International Society of Geographic and Epidemiologic Ophthalmology criteria.29–31

Hypertension was defined as a history of high blood pressure diagnosed by a physician and/or current treatment with antihypertensive medications and/or a blood pressure reading of ≥140/90 mm Hg. Diabetes was defined as a history of diabetes and/or taking diabetic medication and/or diabetic retinopathy was detected on clinical examination. The duration of diabetes since diagnosis was also documented. Body mass index (BMI) was calculated from the measured height and weight according to the formula weight (in kilograms) divided height (in metres) squared. WHO categories of BMI were used, that is, underweight (BMI<18.5), normal (18.5≤BMI<25), overweight (25≤BMI<30) and obese (BMI≥30).32 For smoking, participants were categorised as never smoked, former smoker and current smoker. Current and former smokers were those who had smoked for a minimum of 1 year. Participants who had never smoked, or had smoked for less than 1 year were considered to be ‘never smokers’.26

Methods used in APEDS2

Extensive changes have taken place in the urban and semiurban Hyderabad over the last decade, and the original urban area could not be delineated. Hence, the tracing exercise, undertaken from 2009 to 2010, was limited to the three rural clusters. The purpose of tracing the original participants was as follows: to assess the mortality rate among those at APEDS1, and to determine the factors at APEDS1 that predicted subsequent mortality, for example, lens status, VI, ARMD and glaucoma (POAG and angle closure disease), IOP, PXF and CDR, after adjusting for confounders.

The associations of mortality and VI, lens status and ARMD have already been published.3 The purpose of this study is to explore whether glaucoma (POAG and angle closure disease), IOP, PXF and CDR are associated with a different mortality risk.

During APEDS1, 7771/8832 (88%) of those enumerated in 70 clusters were examined between 1996 and 2000; 2790 (35.9%) were aged 40 years and above. Details of the tracing exercise and assessment of the cause of mortality have already been described.3 In brief, names and addresses of APEDS1 participants were extracted from the APEDS1 database. Field investigators visited each cluster to trace APEDS1 participants. A pilot study was undertaken to standardise data collection instruments. Following training, surviving participants who could be traced were interviewed. In households where APEDS1 participant(s) had died or migrated, a structured questionnaire was administered to the present household head to collect information on the cause and/or the reason of death and the year of death or migration, as applicable. Among those who died, time to death since the examination in APEDS1 was determined using date of examination in the database and reported year of death. In situations where the entire household had migrated, these questions were administered to neighbours. As formal death certificates were not available, the cause of death was based on verbal autopsy.

The study was approved by the Institutional Review Board of Hyderabad Eye Research Foundation, LV Prasad Eye Institute, and adhered to the tenets of the Declaration of Helsinki. As most participants were not literate, verbal consent was obtained after explaining the purpose of the study in presence of the head of the village.

Data analysis

Data were analysed using STATA V.11 (StataCorp, 2011). For the continuous outcome variables (IOP and CDR), Student’s t-test was used. For categorical data (eg, age group, gender, and so on), Fisher’s exact test was used. Cox proportional HRs were used to assess associations between mortality and glaucoma (POAG and angle closure disease), family history of glaucoma, IOP, CDR, PXF, myopia and hyperopia.33 Two models were used. In model 1, data were adjusted for age, gender, level of education, diabetes, hypertension, BMI and smoking status. In model 2, ocular variables likely to affect mortality risk (VI, pure nuclear cataract, pure cortical cataract, pure PSC, mixed cataract, history of cataract surgery and ARMD) were added to the model 1.3 Tests of significance for survival curves were assessed using the log-rank test. Multicollinearity between variables was assessed using variance inflation factors, and proportionality of the model was tested based on Schoenfeld residuals. Interactions for age and gender, age and CDR, and gender and CDR were assessed.


The interval between APEDS1 and APEDS2 ranged from 10 to 12 years (mean 11 years; SD 0.81 year). Information was obtained on the status of all 2790 individuals aged 40 years and above examined during APEDS1: 739 (26.5%) had died by APEDS2; 172 (6.2%) had migrated and 1879 (67.4%) were still living in the area (ie, they were ‘available’). 1322/2790 (47.4%) were male.

Migration was higher in women (52.3%), and was higher in one of the rural areas (Mahabubnagar, 9.1%). There was no significant difference in the mean age of those available in APEDS2 (mean 51.9 years; SD 8.9 years) compared with those who had migrated (mean 52.7 years; SD 10.9 years; p=0.34), but those who had died were significantly older at APEDS1 compared with those available or who had migrated (mean 62.3 years; SD 10 years; p<0.001) (table 1). Similarly, there were more deaths in men, illiterates, those with hypertension and diabetes, current smokers and those with lower BMI or overweight and obese (table 1).

Table 1

Sociodemographic, lifestyle and systemic risk factors for mortality in participants available in APEDS2

Table 2 shows the distribution of all forms of glaucoma, family history of glaucoma, PXF, myopia and hyperopia, and other clinical traits among those available and those who had died. There was a higher prevalence of POAG, PXF and myopia among participants who had died. Similarly, those who had died had a higher mean CDR (0.37, 95% CI 0.36 to 0.38) than those alive (0.34, 95% CI 0.33 to 0.35) and the difference was statistically significant (p<0.001). However, there was no difference (p=0.43) in the mean IOP between those alive (15.2 mmHg, 95% CI 15.1 to 15.3) and those who had died (15.3 mmHg, 95% CI 15 to 15.6).

Table 2

Distribution of glaucoma (POAG and angle closure disease), family history of glaucoma, PXF,myopia and hyperopia, by mortalitystatus

Table 3 shows the univariable as well as multivariable analysis of ocular risk factors with mortality. In univariate analysis, there was a higher hazard of mortality in those with POAG (1.9, 95% CI 1.23 to 2.94), PXF (2.79, 95% CI 2.0 to 3.89), myopia (1.78, 95% CI 1.54 to 2.06) and for every unit increase in CDR (4.49, 95% CI 2.64 to 7.64) (table 3). A higher risk of mortality was not associated with family history of glaucoma, unit increase in IOP, hyperopia and PACD. In multivariable analysis, mortality was higher with each unit increase in CDR, that is, the mortality hazard increased by 2.5 (95% CI 1.3 to 5.1) in model 1, and by 2.1 (95% CI 1.03 to 4.2) in model 2 (table 3). There was no association of mortality with POAG, PACD, family history of glaucoma, IOP, PXF, myopia or hyperopia. The association did not change when age was used as a continuous variable or with stepwise regression (data not shown). No interactions were observed between age and gender, age and CDR, or gender and CDR (data not shown). Proportionality of the model was tested based on Schoenfeld residuals which showed a p value of more than 0.05 (in models 1 and 2), suggesting that we accept proportional if hazards.

Table 3

Univariable and multivariable association of ocular factors with mortality in two different models

After adjustment for age and gender, life table graphs of the probability of death by follow-up time showed no association for POAG (HR 1.23; 95% CI 0.8 to 1.9), primary angle closure (PAC) and PACG (HR 1.06; 95% CI 0.5 to 2.25), primary angle closure suspect (HR 1.32; 95% CI 0.8 to 2.2) and PXF (HR 1.2; 95% CI 0.84 to 1.7). However, life table graphs of the adjusted probability of death with different baseline cup-disc ratios showed significant associations for three different cut-offs of cup-disc ratio, that is, 0.35:1 (HR 1.25; 95% CI 1.1 to 1.45), 0.5:1 (HR 1.27; 95% CI 1.01 to 1.6) and 0.7:1 (HR 1.6; 95% CI 1.2 to 2.2). Figures 1–3 show life table graphs of the unadjusted probability of death by follow-up time with three different baseline cup-disc ratios (0.35:1, 0.5:1 and 0.7:1).

Figure 1

Survival curves for cumulative proportion of mortality at cut-off of cup-disc ratio of 0.35:1.

Figure 2

Survival curves for cumulative proportion of mortality at cut-off of cup-disc ratio of 0.5:1.

Figure 3

Survival curves for cumulative proportion of mortality at cut-off of cup-disc ratio of 0.7:1.


There is conflicting evidence whether glaucoma is associated with mortality.5 8 10 11 13 15–17 Most of the previous studies were on Caucasian populations and only included participants with POAG. Some of these studies demonstrated an association in univariable analysis, but this did not remain significant after adjusting for confounders.5 10 11 13 15 The present study had similar findings for POAG. The US National Health Interview Survey reported an association between glaucoma and mortality8 but their definition of glaucoma was self-reported and likely subject to enrolment bias. The variability across studies could be due to variation in study definitions, design, sample size, adjustment for confounders, as well as the population or ethnic groups studied.

Some studies have reported an association between POAG and cardiovascular mortality, that is, the Barbados study and the Blue Mountain Eye Study.14 16 A meta-analysis supported an association with cardiovascular mortality but not for all causes of mortality.10 In the present study, it was not possible to collect reliable information on the cause of death and we were unable to explore this association.

Unlike Beijing Eye Study, we did not find an association between PAC and PACG with mortality.17 This could be attributed to factors specific to the population or study design or the confounders adjusted in analysis as described above. As far as PXF was concerned, we found an association in univariable analysis but the association was not significant in multivariable analysis, as in other studies.19 20

The eye as a model of ageing has been previously described.34 We have also observed a consistent and significant association between vertical CDR and mortality which has not been described before. This finding raises the possibility that nerve fibre loss may be a marker of ageing, particularly as it has been associated with neurodegenerative diseases such as Alzheimer’s and Parkinson’s disease,35–37 and neuroimaging abnormalities of the central nervous system (CNS) have been reported in patients with glaucoma.38 Although glaucoma is undoubtedly a neurodegenerative condition of retinal ganglion cells, there is controversy concerning whether glaucoma is a primary neurodegenerative condition of the CNS or a primary optic neuropathy with secondary effects in the CNS.35 Unfortunately, we did not measure nerve fibre layer thickness in the present study.

The major strengths of our study are that it was a population-based sample with long-term follow-up and high participation rates, and a standard definition of glaucoma was used. Limitations are lack of data on the causes of death and the possibility that the findings may be explained by unknown confounders. We also did not collect information on the use of antiglaucoma medications. Apart from that, we did not measure disc size as cup-disc ratio is related to disc size as well as neurodegeneration. Hence, a structural association is also possible.

In conclusion, our data do not support an association between glaucoma and an increased risk of all-cause mortality. However, there was an association with increasing CDR. Based on our findings, after adjusting for disc size, the association between nerve fibre layer thickness and mortality is to be explored.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.


  • Contributors RCK: contributions to the conception and design of the work, acquisition, analysis and interpretation of data; drafting the work and revising it critically and final approval of the version published. GVSM, CEG, GNR: contributions to the conception and design of the work and interpretation of data; revising it critically and final approval of the version published. PG: contributions to acquisition of data; revising it critically and final approval of the version published. SM, HBP, GPSS: contributions analysis and interpretation of data; revising it critically and final approval of the version published. SC: contributions to interpretation of data; revising it critically and final approval of the version published.

  • Funding Sightsavers International.

  • Disclaimer The sponsor or funding agency has no role in design or conduct of this research.

  • Competing interests None declared.

  • Patient consent Obtained.

  • Ethics approval The study was approved by the Institutional Review Board of Hyderabad Eye Research Foundation, L V Prasad Eye Institute, and adhered to the tenets of the Declaration of Helsinki. As most participants were not literate, verbal consent was obtained after explaining the purpose of the study in presence of the head of the village.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement Part of the data involved in writing the manuscript can be shared on request.