SEARCH

SEARCH BY CITATION

Keywords:

  • capsular glaucoma;
  • chronic simple glaucoma;
  • epidemiology;
  • exfoliation glaucoma;
  • intraocular pressure;
  • open-angle glaucoma;
  • optic disc haemorrhages;
  • pseudoexfoliation;
  • risk factor

Abstract.

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Purpose:  To study the effect of potential risk factors on the development of open-angle glaucoma (OAG) in a population in which pseudoexfoliation (PEX) is a common finding.

Methods:  In 1984–1986, a population-based survey of 760 people aged 65–74 years was conducted in the municipality of Tierp, Sweden. From 1988 to 2006, a follow-up study of the 530 people with normal visual fields has been in progress. To increase the cohort, 273 ophthalmic outpatients were enroled. Reliable visual fields were available for 679 people, representing 6 126 person-years at risk. A time-weighted mean intraocular pressure (IOP) for all visits was calculated.

Results:  Sixty-four subjects developed definite OAG, 29 of whom were exposed to PEX. Risk factors associated with OAG were higher age, a positive family history, increased IOP and PEX. The age-standardized rate ratio (SRR) was 14.8 times (95% confidence interval [CI] 7.92–27.8) greater in subjects with mean IOP ≥20 mmHg than in those with mean IOP <20 mmHg. When subjects with IOP <20 mmHg at baseline were affected by PEX, the SRR increased 5.01-fold (95% CI 1.97–12.8), compared with the unaffected group. However, when mean IOP at follow-up was taken into account, there was no relationship between OAG and PEX as a distinct risk factor. Among participants in the population survey, 69% of all cases were attributable to a mean IOP ≥20 mmHg.

Conclusion:  Increased IOP and PEX were serious risk factors for incident OAG. The effect of PEX was mediated by increased IOP.


Introduction

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Increased intraocular pressure (IOP) is a well-known risk factor for the development of open-angle glaucoma (OAG). Every follow-up study that has investigated the effect of IOP has found an association between increased IOP and OAG (Le et al. 2003; Müskens et al. 2007; Leske et al. 2008). Moreover, randomized trials have demonstrated positive results of pressure-reducing treatment of ocular hypertension (Kass et al. 2002) and early manifest glaucoma (Heijl et al. 2002). Other established risk factors for OAG are higher age (Le et al. 2003), a family history of OAG (Wolfs et al. 1998), myopia (Grødum et al. 2001) and pseudoexfoliation (Ekström 1993).

Pseudoexfoliation (PEX) is an age-related disorder, characterized by the production and accumulation of a fibrillar material in the anterior segment of the eye (Ritch & Schlötzer-Schrehardt 2001). Common sequence variants in the LOXL1 gene, involved in elastin formation, have been found to confer susceptibility to OAG (Thorleifsson et al. 2007). Although PEX is attracting increasing attention, longitudinal data regarding PEX and OAG are scarce. In a previous report from Tierp (Ekström 1993), PEX was identified, for the first time, as an independent risk factor for the development of OAG. In the Visual Impairment Project, Le et al. (2003) found an increased risk for subjects with PEX. Finally, in a long-term follow-up of a cohort in northern Sweden, Åström et al. (2007) demonstrated a fourfold increased risk for participants exposed to PEX.

In the present investigation, potential risk factors for OAG were studied. The significance of optic disc haemorrhages was also evaluated. The investigation took the form of a population-based 20-year follow-up study.

Methods

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The Tierp Glaucoma survey

In 1984–1986, a population survey was conducted in the municipality of Tierp, south central Sweden. Its target population comprised 2 429 residents 65–74 years of age. The size of the sample was limited to about one-third of the target population. Of the eligible number of 838 residents, 760 underwent a screening examination, as described in detail elsewhere (Ekström 1996). Briefly, an interview was held, covering medical history and family history of glaucoma. Information was also obtained from medical records. Intraocular pressure was measured with single Goldmann applanation readings. If the difference between the two eyes exceeded 2 mmHg, a control measurement was made. In this event, the second reading was defined as the IOP for that person. Visual fields (VFs) were tested with the Competer 350 automated perimeter (Bara Elektronik AB, Lund, Sweden), using supraliminal threshold-related test logic (Heijl 1985). After perimetry, the pupils were dilated, and binocular assessment of the optic disc and gonioscopy were undertaken.

The cohort

A total of 549 participants in the population survey had normal and reliable VFs. Of these, 15 had received treatment for OAG and 4 had undergone cataract surgery. The remaining 530 residents were invited to attend a follow-up study. The study was approved by the Human Subjects Committee at the Faculty of Medicine, University of Uppsala. The tenets of the Declaration of Helsinki were observed. To increase the cohort, 14 patients diagnosed with ocular hypertension at the Eye Department in Tierp in 1984–1986 were included (Fig. 1). A further number of 259 people, participating in a case–control study in 1988–1995 (unpublished data), were also recruited. Those enroled were in the age range of 65–74 years. They underwent a baseline examination identical to that of the population survey.

image

Figure 1.  Flow chart showing how the cohort of 679 individuals was derived. OH = ocular hypertension. VFs = visual fields.

Download figure to PowerPoint

Altogether, 803 people fulfilled the VF criteria. Of these, 15 were defined as ineligible for the study and 47 were lost (Fig. 1). Twenty-five people (3.4% of those eligible) were unable to participate. Interpretable VFs were missing for 17 subjects. For the purpose of this study, at least 3 years of follow-up was required before new cases of OAG were accepted (immortal person-time). Consequently, data for 20 participants who reached an end-point within 3 years from baseline were removed. The remaining 679 subjects constituted the study cohort, whose characteristics are presented in Table 1.

Table 1.   Characteristics of the cohort, by age and gender.
Age groupNo. of people (n = 679)Person-years (n = 6 126)
Female (%)Male (%)Female (%)Male (%)
  1. Mean follow-up time: 9.0 years (standard deviation 4.3).

65–69 years253 (62)167 (61)2 481 (66)1 548 (65)
70–74 years152 (38)107 (39)1 252 (34)844 (35)
65–74 years405 (100)274 (100)3 733 (100)2 392 (100)

Follow-up examinations

Follow-up started after the baseline examination and ended in December 2006. The study protocol included determinations of best-corrected visual acuity, IOP measurement and automated perimetry. Visual fields were tested on the Competer in the same way as in the population survey. Study end-points were noninterpretable or abnormal VFs in either eye, initiation of pressure-reducing treatment, intraocular surgery and nonparticipation. Details are given in a previous report (Ekström 1993). A VF defect (VFD) had to be confirmed in two consecutive tests. If a VF was abnormal, a detailed examination was undertaken to reveal the cause. For those who did not reach an end-point, perimetry was repeated every second or fourth year (up to 2006). In addition to a protocol, data were entered in special glaucoma case records. Simultaneous stereo colour photographs (Carl Zeiss GmbH, Jena, Germany) of the optic discs were taken of subjects with suspected OAG.

Concepts and definitions

Consistent with the concept of Foster et al. (2002), chronic simple glaucoma and capsular glaucoma were classified as OAG. Angle-closure glaucoma presupposed a history of chamber angle closure; secondary glaucoma was of traumatic, neovascular or inflammatory origin. Capsular glaucoma was distinguished from chronic simple glaucoma by the presence of PEX. Normal-tension glaucoma (NTG) was defined as a variant of OAG in cases where not more than one IOP reading above 21 mmHg had ever been recorded and no readings above 24 mmHg. For a diagnosis of OAG, a reproducible VFD was a prerequisite, consistent with glaucoma and not explicable on other grounds, with the supra-threshold test logic. Only optic disc haemorrhages observed at ophthalmoscopy are noted.

Classification of open-angle glaucoma

Patients diagnosed with OAG according to the supra-threshold test logic underwent manual Goldmann perimetry and repeated visual field testing using Competer threshold test logic. Threshold fields were sent to University Hospital, Trondheim, Norway, for grading by an ophthalmologist otherwise unconnected with the study. According to impairment of abnormal test points or development of VF loss over time, each patient was classified as having either progressive or nonprogressive disease. Patients deemed to have progressive disease were regarded as definite OAG cases.

The glaucoma case records of patients with nonprogressive disease, or missing threshold fields, were reviewed by an ophthalmologist at Uppsala University Hospital. Data pertaining to the optic discs were evaluated together with automated or manual VFs. In addition, photographs were examined. Before the review, all information with the exception of visual acuity and optic disc characteristics was masked. The patients were classified as having definite OAG (or not). Consequently, a diagnosis of OAG was established in this study in two stages. First, OAG cases were identified on the basis of VF testing. Secondly, among subjects diagnosed with OAG, the progressive cases described above and the cases classified as OAG by the reviewer constituted the final number of cases.

Statistical methods

The number of person-years at risk of developing OAG was summed in each subgroup of exposed and nonexposed individuals. Exposure-specific incidence rates provided for the calculation of rate ratios. Age-standardized rate ratios (SRRs) were estimated, using the exposed part of the cohort as a standard. Follow-up time was calculated from baseline to the last visit with interpretable VFs in any eye. To avoid bias, immortal person-time was excluded from follow-up time in all analyses (Rothman & Greenland 1998).

Subjects with PEX in either eye at baseline were defined as exposed to PEX. The highest pressure of the two eyes was used in the statistics. A time-weighted mean IOP for all visits was calculated, using weights proportional to the interval between two measurements. This method was applied in previous reports (Ekström 2000, 2008). A chi-square test of significance for incidence rate differences across strata was performed according to a method described by Rothman (1986). The fraction of OAG cases attributable to a mean IOP ≥20 mmHg during follow-up was estimated (Rothman 2002). In the calculations, age-adjusted risk ratios based on cumulative incidences were used.

The relationship between increased IOP, PEX and OAG was further studied with multivariate hazard ratios derived from Cox proportional hazards models. Mean IOP was included in the models as time-dependant covariates to account for the effect of IOP over time. The proportionality assumption was confirmed according to a method suggested by Hill & Lewicki (2006). Statistica 8.0 (StatSoft, Inc., Tulsa, OK, USA) was used for multivariate analyses.

Results

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

By the end of the study, 64 cases of OAG had been detected, 34 with capsular and 30 with chronic simple glaucoma, all of whom were ultimately classified as definite cases. Five of the capsular glaucomas were not exposed to PEX at baseline. Three cases of capsular and 10 of chronic simple glaucoma were diagnosed as NTG. Table 2 provides SRRs for OAG at baseline. Increased IOP, PEX and a positive family history were associated with OAG. People aged ≥70 years experienced a 1.97-fold (95% confidence interval [CI] 1.21–3.22) increased risk, compared with those <70 years.

Table 2.   Associations of potential risk factors and incident open-angle glaucoma.
Baseline characteristicsSRR (95% CI)
  1. * Spherical equivalents.

  2. SRR = age-standardized rate ratio, CI = confidence interval, OAG = open-angle glaucoma, IOP = intraocular pressure.

  3. International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) characters are given in parenthesis.

Female gender0.77 (0.46–1.30)
Positive family history of OAG2.04 (1.04–4.00)
Medical history factors
 Smoking, ≥35 years1.16 (0.63–2.14)
 Diabetes mellitus (E10–E14)0.63 (0.23–1.74)
 Blood-pressure-lowering treatment0.96 (0.57–1.62)
 Ischaemic heart disease (I20–I25)0.85 (0.42–1.73)
 Arteriosclerosis (I70–I72)1.59 (0.50–5.09)
 Obstructive lung disease (J42–J45)0.49 (0.12–2.00)
Ocular factors, any eye
 IOP ≥20 mmHg6.83 (3.99–11.7)
 Pseudoexfoliation5.68 (3.47–9.31)
 Myopia (≤−1 dioptre)*0.87 (0.27–2.76)

The results presented in Table 3 suggested interaction between increased IOP and PEX at baseline, i.e. the joint effect of increased IOP and PEX was greater than the sum of their individual effects. A chi-square test supported a significant difference in incidence rates across IOP categories (χ2df=2 = 6.21; p = 0.04). In subjects with IOP <20 mmHg, PEX was associated with a fivefold excess risk of OAG, compared with the reference (no exposure) group. In Table 4, baseline IOP is replaced by mean IOP at follow-up visits. In this case, PEX alone was not associated with OAG. When excluding subjects in the intermediate pressure category, interaction between mean IOP ≥25 mmHg and PEX was evident (χ2df=1 = 5.71; p = 0.02). In the whole cohort, a mean IOP ≥20 mmHg increased the SRR for OAG 14.8-fold (95% CI 7.92–27.8). In the part of the cohort that participated in the population survey, 69% of all cases were attributable to a mean IOP ≥20 mmHg.

Table 3.   Age-standardized rate ratios for open-angle glaucoma (OAG) associated with increased intraocular pressure (IOP) and pseudoexfoliation (PEX) at baseline.
Baseline IOPPEX
YesNo
  1. * Reference group.

≥25 mmHg41.1 (16.6–102)10.6 (3.96–28.2)
20–24 mmHg19.4 (8.97–41.9)5.49 (2.60–11.6)
<20 mmHg5.01 (1.97–12.8)1.00*
Table 4.   Age-standardized rate ratios for open-angle glaucoma (OAG) associated with increased mean intraocular pressure (IOP) at follow-up and pseudoexfoliation (PEX) at baseline.
Mean IOPPEX
YesNo
  1. * Reference group.

≥25 mmHg55.9 (25.9–121)25.8 (11.3–59.0)
20–24.99 mmHg9.90 (3.38–29.0)6.76 (2.91–15.7)
<20 mmHg2.27 (0.497–10.4)1.00*

Cox proportional hazards models included age, mean IOP, an interaction term, PEX and gender. The final model is shown in Table 5. Consistent with results of the standardized analysis, interaction between mean IOP ≥25 mmHg and PEX more than doubled the relative hazard. Inclusion of PEX in a model containing mean IOP and the interaction term did not add any further information.

Table 5.   Cox regression model relating fixed and time-dependent covariates to development of open-angle glaucoma.
CovariateHR (95% CI)
  1. * Age at baseline, continuous variable.

  2. Interaction term.

  3. Reference category.

  4. HR = hazard ratio, CI = confidence interval, IOP = intraocular pressure, PEX = pseudoexfoliation.

Fixed
 Age (per year)*1.15 (1.05–1.26)
 Mean IOP ≥25 mmHg, PEX2.38 (1.87–3.03)
Time-dependent (per 10 years)
 Mean IOP <20 mmHg1.00
 Mean IOP 20–24.99 mmHg3.92 (2.13–7.22)
 Mean IOP ≥25 mmHg15.4 (4.52–52.1)

Optic disc haemorrhages were found in 22 subjects, 12 of whom developed OAG. Haemorrhages were associated with a 15.0-fold (95% CI 6.05–37.1) increased risk in men and a 2.55-fold (95% CI 1.01–6.49) increased risk in women; men with haemorrhages were 3.41 (95% CI 1.08–10.7) times more likely to be diagnosed with OAG, than were women. A marked covariation between increased IOP, haemorrhages and OAG was demonstrated; the SRR in subjects exposed to haemorrhages and mean IOP ≥20 mmHg increased 25.3-fold (95% CI 10.3–62.4), compared with the no exposure group.

Discussion

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Spanning over 20 years, this study is the longest follow-up study on risk factors for OAG based on automated perimetry to be reported so far. It is the second study of its kind conducted in a population in which PEX is a common finding. Higher age, a positive family history, increased IOP and PEX were associated with incident OAG. At baseline, IOP ≥20 mmHg increased the risk sevenfold, while PEX increased the risk sixfold.

The significance of increased IOP has been documented in other studies. In the Barbados Eye Studies, every increase of baseline IOP by 1 mmHg was associated with a 12% increased risk (Leske et al. 2008). Follow-up of the Visual Impairment Project in Australia revealed an excess risk of ‘at least probable’ OAG of 10% for each 1 mmHg increase at baseline (Le et al. 2003). Whereas PEX is extremely uncommon in Barbados (Leske et al. 2002), PEX increased the risk 11-fold in the Australian study.

In general, cohort studies on risk factors for OAG report an effect of increased IOP at baseline. At rescreening, a substantial proportion of subjects classified as exposed may actually have an IOP within the normal range, while subjects with normal IOP at baseline have increased IOP (Ekström 1993). To avoid this misclassification of exposure, mean IOP at every visit was used in the analyses. In addition, mean values were adjusted to account for the effect of aggregated pressure readings within a short period of time.

Total attributable fraction is the proportion of cases in the population attributable to a specific exposure: the larger the proportion of exposed cases, the greater the attributable fraction. In the absence of bias, attributable fraction is the proportion of disease burden caused by the exposure. When excluding those recruited to the original cohort, the proportion of OAG cases attributed to a mean IOP ≥20 mmHg was estimated to 69%.

Pseudoexfoliation at baseline was identified as an independent risk factor for OAG. When subjects with IOP <20 mmHg were affected by PEX, the risk increased fivefold, compared with the unaffected group. However, when mean IOP at follow-up was taken into account, there was no association between PEX as a separate risk factor and OAG. This finding indicates that the effect of PEX was mediated by increased IOP, as suggested in a previous report (Ekström 2000). The result concurs with the understanding of PEX as a disorder obstructing the outflow of aqueous humour.

Interaction between PEX and increased IOP was established. Exposure to PEX and a mean IOP ≥25 mmHg increased the risk of OAG 56-fold, compared with the reference group. The finding in Tierp is of great public health importance. An intensified effect of increased IOP in the presence of PEX is a likely explanation for the high prevalence of OAG, reported from Nordic countries (Jonasson et al. 2003; Ekström & Alm 2008).

Multivariate analyses using Cox hazards models identified high age, increased IOP and interaction between increased IOP and PEX as risk factors for OAG. In agreement with the results of the stratified analysis, interaction explained more of the variation in the data than PEX as a distinct risk factor. Unfortunately, competing risks affected the analyses. In particular, data on subjects who received pressure-reducing treatment were strongly correlated to OAG data. Therefore, conclusions based on Cox hazards models should be interpreted with some caution.

To increase the statistical power of the study, the original cohort of people participating in the population survey was enlarged by adding patients seeking medical attention at the Eye Department in Tierp. These measures will never bias the result as long as the identification of OAG was independent of the exposure under study, which was the case in this study. In fact, OAG incidence in people exposed to increased IOP and PEX was somewhat lower among those recruited to the cohort (data not shown).

Apart from the already established risk factors, this study failed to identify any association between baseline characteristics and OAG. None of the medical diagnoses increased the incidence. However, exposed cases were generally few in numbers, limiting the ability to reveal unknown risk factors. Similarly, there was insufficient power to study the effect of myopia. Smoking did not increase the risk.

Optic disc haemorrhages found at ophthalmoscopy were closely associated with OAG. In view of all the unanswered questions regarding haemorrhages, they should be conceived of as a sign of the disease rather than a risk factor for the disease. Notably, men with a haemorrhage were 3 times more likely to develop OAG than women with haemorrhage. To the best of our knowledge, such a finding has not been reported elsewhere. The result seems to be in conflict with that reported by Bengtsson et al. (2008), which suggested lower odds ratios for the presence of disc haemorrhages for men with manifest glaucoma. Bearing in mind the relatively small numbers of subjects with haemorrhages, the finding in Tierp should be regarded with a certain reservation.

The strengths of this study include the community-based approach, the long follow-up time, the high participation and the use of computerized perimetry to identify cases. The classification of OAG was based on the decision of ‘masked’ observers. The limitations are the tolerably small size of the cohort and the lack of optic disc photography at study visits. Only people in the age range 65–74 years were included.

To sum up, the cohort study in Tierp provides estimates of risk factors for OAG over a 20-year period. Increased IOP and PEX were the only risk factors of importance. A mean IOP ≥20 mmHg accounted for approximately 70% of OAG morbidity in the population. The effect of PEX was mediated by increased IOP.

Acknowledgements

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The Swedish Medical Research Council, Crown Princess Margaretha’s Foundation for the Visually Impaired, the Glaucoma Research Fund at the Department of Ophthalmology, Uppsala University Hospital and Uppsala County Council provided financial support for this study. Professor Albert Alm performed the review of OAG cases and Dordi Austeng, MD, undertook grading of threshold fields.

References

  1. Top of page
  2. Abstract.
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References