Original Article

Home , Ophthalmoscopy, Visual field test

Clinical and Experimental Ophthalmology 2007; 35: 237–243 doi: 10.1111/j.1442-9071.2007.01457.x Original Article

Glaucoma screening: analysis of conventional and telemedicine-friendly devices

Sajeesh Kumar PhD,1 Antonio Giubilato FRANZCO,2 William Morgan FRANZCO,2 Ludmila Jitskaia FRANZCO,2 Chris Barry MMedSci,2 Max Bulsara PhD,3 Ian J Constable FRANZCO2 and Kanagasingam Yogesan PhD1 1Centre of Excellence in e-Medicine and 3School of Population Health, University of Western Australia, Nedlands, and 2Lions Eye Institute, Perth, Western Australia, Australia

ABSTRACT INTRODUCTION Purpose: Portable, telemedicine-friendly devices offer novel Glaucoma is an optic nerve neuropathy that leads to progres- opportunity for screening and monitoring glaucoma in the sive visual field loss and affects millions of people worldwide remote and rural regions of the world.This study examines with prevalences between 1% and 3%.1 Family history, older the effective combination of telemedicine-friendly screening age and being of African descent place an individual at devices for detection of glaucoma in relation with conven- increased risk for glaucoma. The most frequent diagnosis is primary open angle glaucoma (POAG).1 Widespread screen- tional, hospital-based devices. ing is critical for early diagnosis, treatment and limiting the Methods: A total of 399 eyes were screened with incidence of glaucoma-associated blindness. Meanwhile, telemedicine-friendly devices and conventional, hospital- individuals living in rural or remote areas have limited access based devices such as ophthalmoscope, tonometer and to optometrists or ophthalmologists hence to glaucoma tests.2 However, recent advances in compact, portable, easy- perimeter. to-train, automated ophthalmic diagnostic devices and Results: Combination of age and family history of glaucoma increasing spread of the Internet-based telecommunications alone has a sensitivity of 35.6% (specificity 94.2%, area under offer novel opportunities for telemedicine-based glaucoma screening and monitoring.3 Semiskilled, rural, ancillary the curve 0.81, correctly classified 81.1%) and an addition of health-care workers could easily be trained to connect these telemedicine-friendly or conventional visual field tests opti- devices to teleophthalmology system, perform screenings mized the sensitivity to 91.1% (specificity 93.6%, area under and send the results electronically to specialists in urban the curve 0.95, correctly classified 93%). Analysis indicates centres for second opinion and management (Fig. 1).4 good agreement between vertical cup-to-disc ratio by oph- Meanwhile, conventional, hospital-based glaucoma thalmoscopy and digital image reading. An addition of testing itself is a controversial issue as no single test has intraocular pressure test does not change sensitivity (35.6%) shown an adequate balance of sensitivity, specificity and and specificity (94.2%). positive predictive value (PPV).5 The three most common screening tests for glaucoma are tonometry (for intraocular Conclusion: This study indicates that evaluations of cup-to- pressure [IOP]), ophthalmoscopy (for vertical cup-to-disc disc ratio and visual field, using telemedicine-friendly devices, ratio [VCDR]) and perimetry (for visual field). Currently are most useful tools in screening for glaucoma.When used very little is known about the effective combination of these together these devices may be an alternative for conven- devices especially for telemedicine-based glaucoma tional glaucoma screenings. screening. Hence, the present study aims to specify effective combination of telemedicine-friendly devices for glaucoma Key words: glaucoma, screening, telemedicine. screening. These are portable, screening-friendly devices

᭿ Correspondence: Dr Sajeesh Kumar, Lions Eye Institute, Centre of Excellence in e-Medicine, University of Western Australia, 2 Verdun Street, Nedlands, WA 6009, Australia. Email: [email protected] Received 24 May 2006; accepted 24 November 2006.

a bc

Figure 1. Usage of telemedicine-friendly portable devices – (a) frequency doubling technology perimeter, (b) Pulsair tonometry and (c) non-mydriatic fundus camera – for glaucoma screening in remote rural regions in Western Australia. Data collected by rural, ancillary health care workers are sent electronically to specialists in urban centres for second opinion and management. that take less time, could be performed in a rural telemedi- meet the probability value associated with the Humphrey cine centre with less specialized personnel and could be global indices, corrected pattern standard deviation, P < 0.01 easily integrated to telemedicine screening systems. and the mean deviation was equal to or worse than -13 dB. For the FDT screening test, 17 locations, including a central point, were evaluated. The result at each location is com- METHODS pared with a normative database that grades a point as abnor- The study included informed, consenting patients attending mal starting at the P < 0.05 level. the general eye examination clinic at Lions Eye Institute, All devices were calibrated for optimal performance as University of Western Australia. Although all patients were recommended by the manufacturers. To reduce the risk recruited consecutively and offered the chance to partici- of observer bias, the order of both conventional and pate, patients were excluded from the study if they recently telemedicine-friendly tests were conducted at random. The had an eye surgery or had been already diagnosed with sensitivity and specificity rates for each testing mode were glaucoma. Patients had both eyes tested. VCDR was calculated for conventional and telemedicine-friendly observed through images taken with a digital, portable, non- devices. Age above 45 years was taken as risk group for mydriatic retinal camera (NM-200D, Nidek, Tokyo, Japan). glaucoma. Family history was ascertained by interviewing Image acquisition process was repeated when the original the participants whether any first-degree relatives (parents, image quality was unsatisfactory. The better digital image siblings and children) had been diagnosed with glaucoma. was stored in the built-in memory of the camera and later Specialists determined the presence or absence of glaucoma down loaded to a personal computer. Conventional evalua- in accordance with specific glaucoma screening protocol of tion of VCDR was performed by glaucoma specialists (AG, Lions Eye Institute that acknowledged glaucoma presence WM and LJ) using direct ophthalmoscope (Keeler Instru- as VCDR > 0.5, IOP > 21 mmHg, abnormal visual field ments, Broomall, PA, USA) after pupilary dilation. Asymme- related to glaucoma and or disk asymmetry >0.2. Identifi- tries in VCDR between both eyes were noted if it was >0.2 cation of glaucoma by the specialists, using conventional VCDR. hospital-based method, was taken as gold standard. To Intraocular pressure was measured in all the eyes with the prevent bias or confounding the optic disc photos as well telemedicine-friendly non-contact tonometer (Pulsair-Easy as visual field tests were graded separately, in masked and Eye, Keeler, PA, USA) and with the conventional, clinic- randomised fashion. based, Goldmann applanation tonometer (Haag Streit AT To investigate the agreement between the tests a plot of 900, Mason, OH, USA). In both methods, three measure- the differences between the methods against their mean was ments of IOP in each eye of all the patients were taken and carried out as described by Bland and Altman.6,7 To examine the arithmetic average was made. All the measurements were the relationship formally we used the Spearman rank corre- taken in sitting position. lation coefficient (r) between the absolute difference and the Telemedicine-friendly perimetry was performed with average. IOP of 21 mmHg was used as the threshold for Humphrey Frequency Doubling Technology (FDT) Visual glaucoma.8 Multivariate logistic regression (after controlling Field Instrument (Carl Zeiss Meditech, Dublin, CA, USA) for age, sex and family history) was used to obtain area under using the C-20-5 screening protocol. Conventional hospital- the curve (AUC), sensitivities and specificities. PPV and based perimetry was performed with Humphrey Automated negative predictive value (NPV) were calculated using 3.13% Field Analyser II (Carl Zeiss Meditech) using 24-2 threshold. prevalence of glaucoma in Australia.9 All statistical analysis Both Humphrey visual field (HVF) and FDT tests were was performed using Stata version 8 (Stata Statistical Soft- graded normal or abnormal according to visual field damage. ware: Release 9.0 [2005], Stata Corporation, College HVF test results were graded as abnormal, if the eye failed to Station, TX, USA). This study conformed to the standards of

Table 1. Patient demography on age, gender and family history of glaucoma

Glaucoma Gender Family history of glaucoma Frequency Mean age Ϯ SD No Female No 63 59.3 Ϯ 12.9 Yes 21 60.9 Ϯ 12.2 Male No 64 56.1 Ϯ 11.6 Yes 8 58.0 Ϯ 14.5 Yes Female No 10 75.2 Ϯ 11.9 Yes 17 71.8 Ϯ 11.1 Male No 14 67.3 Ϯ 12.2 Yes 4 72.3 Ϯ 11.7

Table 2. Age, VCDR and IOP indices for patient with and without glaucoma 0.30

Variables With glaucoma With glaucoma Mean Mean 0.20 (SE) (SE) Mean + 2SD Age, years 71.3 (1.81) 58.0 (0.99) 0.10 VCDR 0.62 (0.03) 0.26 (0.01) IOP 16.5 (0.55) 14.4 (0.27) 0.00 Mean IOP, intraocular pressure; VCDR, vertical cup-to-disc ratio.

–0.10 Mean – 2SD the Human Research Ethics Committee of the University of Western Australia. –0.20

RESULTS Difference of VCDR ophthalmoscopy and digital image –0.30 Altogether, 402 eyes of 201 consenting patients were screened (111 women, 90 men, mean age 61.04 years). Two 0.00 0.20 0.40 0.60 0.80 1.00 Average of VCDR ophthalmoscopy and digital image eyes had cataract and one eye had miosis and these three eyes were excluded from the study. Thus, total eyes studied Figure 2. Plot of the differences between the conventional verti- were 399. Although 50 participants (25%) reported family cal cup-to-disc ratio (VCDR) by ophthalmoscopy and VCDR by history of glaucoma, none had previously been diagnosed digital photograph reading. The mean bias is -0.008 Ϯ 0.0577. The with glaucoma. Ophthalmologists identified POAG in 45 Spearman rank correlation coefficient between the absolute differ- people (Tables 1,2). ence (r) = 0.21. Analysis indicates good agreement between VCDR by ophthalmoscopy and digital image reading (Fig. 2). There is favourable in place of ophthalmoscopy – VCDR (sensitiv- was no significant difference between the proportion of ity 67.4%, specificity 93.6%, AUC 0.94, CC 87.9%) to patients who had FDT compared with proportion who had determine glaucoma. HVF (P = 0.832, using McNemars test). If the practitioner already has all the aforementioned A plot of the differences between the conventional clinic- devices, the test for disk asymmetry will be ideal (sensitivity based IOP readings by Goldmann applanation tonometer and 93.2%, specificity 93.6%, AUC 0.95, CC 93.5%) compared telemedicine-friendly Pulsair tonometer is shown in Figure 3. with the test for VCDR (sensitivity 83.7%, specificity 96.8%, Table 3 indicates age or family history of glaucoma alone AUC 0.98, CC 93.9%). has a sensitivity of 35.6% (specificity 94.2%, AUC 0.81, correctly classified [CC] 81.1%) and an addition of HVF test DISCUSSION optimizes the screening sensitivity to 91.1% (specificity 92.2%, AUC 0.94, CC 92%). With the ageing population in the world, glaucoma will be Correspondingly, analysis in Table 4 indicates age, family more prevalent. Because of glaucomas generally long pre- history and telemedicine-friendly FDT tests will optimize clinical phase, reproducible, patient-friendly diagnostic the sensitivity to 91.1% (specificity 93.6%, AUC 0.95, CC techniques may permit earlier detection of glaucomatous 93%). An addition of IOP test does not change sensitivity damage, which can then offer patients earlier treatment pos- (35.6%) and specificity (94.2%). sibilities and better outcomes. Telemedicine reportedly offers Analysis further indicates telemedicine-friendly FDT an economic method for glaucoma screening with portable (sensitivity 83.7%, specificity 96.8%, AUC 0.98, CC 93.9%) equipment operated by less expensive and ubiquitous health

Table 3. Comparison of different combination of tests with con- 10.00 ventional, hospital-based devices for glaucoma detection

Variables AUC Sen% Spec% CC% Mean + 2SD Age + sex + family history Age + family history 0.81 35.6 94.2 81.1 5.00 +VCDR 0.94 69.8 94.2 88.9 +IOP 0.83 42.2 93.6 82.1 +Disk asymmetry 0.87 51.1 96.2 86.1 +HVF 0.94 91.1 92.2 92.0 Mean Age + sex + family history + VCDR +IOP 0.95 67.4 94.2 88.4 0.00 +Disk asymmetry 0.96 76.7 94.8 90.9 +HVF 0.97 86.1 94.1 92.4 Age + sex + family history + IOP Disk asymmetry 0.88 53.3 93.6 84.6

Difference of IOP PulsairDifferenceIOP and GAT of + Mean – 2SD +HVF 0.95 88.9 92.9 92.0 Age + sex + family history + disk asymmetry –5.00 +HVF 0.95 88.9 92.9 92.0 Age + sex + family history + VCDR + IOP +Disk asymmetry 0.95 79.1 94.2 90.9 5.00 10.00 15.00 20.00 25.00 30.00 +HVF 0.98 88.4 96.1 94.4 Average of IOP Pulsair and GAT Age + sex + family history + VCDR + disk asymmetry +HVF 0.98 83.7 95.4 92.9 Figure 3. Plot of the differences between the conventional Age + sex + family history + IOP + disk asymmetry intraocular pressure (IOP) readings by applanation tonometry and +HVF 0.96 88.9 94.2 93.0 IOP by telemedicine-friendly digital tonometer. The mean bias is Age + sex + family history + VCDR + IOP + disk asymmetry Ϯ 1.508 2.7164. The Spearman rank correlation coefficient between +HVF 0.98 83.7 96.1 93.4 the absolute difference (r) = 0.17. GAT, Goldmann Applanation Tonometer. AUC, area under the curve; CC, correctly classified; HVF, Hum- phrey visual field test; IOP, intraocular pressure; Sen, sensitivity; Spec, specificity; VCDR, vertical cup-to-disc ratio. workers in the rural regions.10–14 It should be noted that although individual tests may have less-than-ideal sensitivities and specificities, these can be increased by combining and scanning laser polarimetry may in some cases permit diagnostic devices in parallel, thereby providing a more reli- earlier glaucoma diagnosis.17 Therefore their role in screen- able screening programme.15,16 In the absence of any single ing, especially when coupled with telemedicine, appears test for screening of glaucoma, the current study evaluated promising and remains to be studied. Furthermore, diverse combinations of devices and tests towards an effec- telemedicine-based use of gonioscopy also needs to be tive screening strategy. studied. A dilated eye examination with direct ophthalmoscopy by Glaucoma (POAG) also leads to characteristic patterns of an ophthalmologist has a reported sensitivity of 59% and a visual field defects, which is valuable for glaucoma detection, specificity of 73% for detecting and classifying optic disc but diagnostic testing with conventional hospital-based changes associated with glaucoma.5 Qualitative evaluation of perimetry (HVF) is time-consuming, hence not feasible for digital images of the optic disk is more sensitive, which also mass screening.18 Both HVF and FDT perimeters have been allows for precise measures of disk parameters (e.g. vertical reportedly successful in detecting glaucoma eyes.19 Portable and horizontal cup-to-disk ratios, neuroretinal rim width). In FDT perimetry in the current setting proved sensitive (91. this study, VCDR observation using telemedicine-friendly 1%) for glaucoma detection. Both HVF and FDT perimetry non-mydriatic camera was fairly sensitive to glaucoma detec- exhibited identical PPV of 99.7% (Table 5). Several investi- tion (Table 3). Eliminating dilation and electronically gators have found that FDT has both sensitivity and speci- archiving and transmitting retinal images are striking features ficity in the 90th percentile range, making it an excellent of these telemedicine-friendly devices. Although conven- screening tool.20–23 Whereas Cioffi et al. suggests repeat tional, subjective, dilated ophthalmoscopy by specialists testing to be beneficial, FDT offers definite advantages for often encounters difficulty in detecting glaucomatous optic early glaucoma detection.24–31 disc progression,5 the archiving of retinal images by tele- Pulsair tonometry is a user-friendly device and its IOP medicine devices assists in identifying optic disc progression. readings are digitally displayed that could be easily inte- Recent technological advances, however, have produced grated to a telemedicine system to transfer the data to an new instruments that can objectively quantify the optic urban diagnosing centre. However, this study indicates that nerve head and the retinal nerve fibre layer. Confocal scan- the effectiveness of IOP measurement as a screening tool for ning laser ophthalmoscopy, optical coherence tomography POAG appears to be limited. There is no single cut-off value

Table 4. Comparison of different combination of tests with develop glaucoma.18,34,35 Earlier studies reported correspond- telemedicine-friendly devices for glaucoma detection ing readings by Pulsair and Goldmann tonometer.36 The Pulsair tonometer is relatively easy to use by technicians and Variables AUC Sen% Spec% CC% has numerous special applications in ophthalmic practice Age + sex + family history (e.g. measuring IOPs for paediatric patients and those with Age + family history 0.81 35.6 94.2 81.1 compromised corneas). +VCDR 0.94 67.4 93.6 87.9 In the current study, the telemedicine-friendly Pulsair +IOP 0.81 35.6 94.2 81.1 +Disk asymmetry 0.83 45.5 96.8 85.5 tonometer was relatively less sensitive for glaucoma detec- +FDT 0.95 91.1 93.6 93.0 tion (sensitivity 35.6%, specificity 94.2%, NPV 16.6%, PPV Age + sex + family history + VCDR 97.8%). And it is suggested that even if measured optimally, +IOP 0.94 67.4 93.6 87.9 increased IOP alone is a poor predictor of glaucoma35 +Disk asymmetry 0.95 64.3 93.6 87.3 although it remains the most important risk factor.32 +FDT 0.98 83.7 96.8 93.9 The high incidence of glaucoma detected in the study Age + sex + family history + IOP population could be due to the reference standard +Disk asymmetry 0.83 45.5 96.8 85.5 (IOP > 21 mmHg) and the study participants from an eye +FDT 0.95 91.1 93.6 93.0 clinic. The analysis is reported per number of eyes, rather Age + sex + family history + disk asymmetry +FDT 0.95 93.2 93.6 93.5 than per patient. Although a positive family history of glau- Age + sex + family history + VCDR + IOP coma may help to identify those at risk, it is subject to recall, +Disk asymmetry 0.95 64.3 93.6 87.3 selection and survival bias as well as community underdiag- +FDT 0.98 83.7 96.8 93.9 nosis of glaucoma.37 Age + sex + family history + VCDR + disk asymmetry Meanwhile, deciding about effective glaucoma screening +FDT 0.98 85.7 96.8 94.4 can be complex.38–40 In March 2005, the US Preventive Ser- Age + sex + family history + IOP + disk asymmetry vices TaskForce found insufficient evidence to recommend for +FDT 0.95 93.2 93.6 93.5 or against screening adults for glaucoma.5 Recently, Maier Age + sex + family history + VCDR + IOP + disk asymmetry 39 +FDT 0.98 85.7 96.8 94.4 et al. presented a meta-analysis of the glaucoma trials. POAG reportedly had an ‘effective’ population attributable risk per- AUC, area under the curve; CC, correctly classified; FDT, fre- centage (PAR), a measure that reflects the public health impor- quency doubling technology visual field test; IOP, intraocular pres- tance of a disease of 16%, perhaps high enough to be sure by Pulsair tonometer; Sen, sensitivity; Spec, specificity; VCDR, considered a public health problem and justify inclusion as a vertical cup-to-disc ratio by digital camera. target disease in the Vision 2020 programme. However, the logistics and opportunity costs of diagnosis and treatment would probably prevent inclusion of POAG in public health Table 5. Sensitivity, specificity, NPV, PPV of both conventional budgets of most developing countries.41 Recently, the use of and telemedicine-friendly devices (calculated with 3.13% prevalence of glaucoma) FDT and the non-mydriatic digital camera was reported as cost- and time-effective way to monitor changes over Devices Sen% Spec% NPV% PPV% time.42,43 In this scenario, telemedicine promises both cost- 3 Conventional effective and clinically effective solution for global eye care. IOP 42.2 93.6 15.6 98.0 While acknowledging that the current study sample was from VCDR 69.8 94.2 28.0 99.0 an hospital-based population, caution should be taken in VF 91.1 92.2 27.4 99.7 generalizing the results to a community-based screening.39,44 Telemedicine-friendly Clearly, targeting screening programmes to populations at IOP 35.6 94.2 16.6 97.8 higher risk for glaucoma will identify a larger number of new VCDR 67.4 93.6 25.3 98.9 cases, and increase the screening tests PPV.45 In accordance VF 91.1 93.6 31.5 99.7 with the US Preventive Services Task Force report on conventional screening, the current study reports that screening IOP, intraocular pressure; NPV, negative predictive value; PPV, positive predictive value; Sen, sensitivity; Spec, specificity; VCDR, with telemedicine-friendly visual field testing and ophthal- vertical cup-to-disc ratio; VF, visual field test. moscopy identifies primary open-angle glaucoma patients with greater accuracy, whereas measurement of IOP is not a good screening tool for glaucoma. of IOP that provides an acceptable balance of sensitivity and In conclusion, the results of this study indicate that evalu- specificity for screening.18,32 IOP measurements above the ations of visual field and cup-to-disc ratio, using portable, usual cut-off point (greater than 21 mmHg) have a reported telemedicine-friendly digital devices, are most useful tool in sensitivity of 47% and specificity of 92% for diagnosing screening for glaucoma. When used together, these devices POAG, and IOP measurement does not appear to perform may become an alternative for conventional glaucoma screen- better in high-risk groups defined by age, race, sex or family ings. Country-specific guidelines for telemedicine-based history.33 Moreover, high IOP can also be seen in patients glaucoma screening by rural health workers need to be with ocular hypertension, the majority of whom may never established.

ACKNOWLEDGEMENTS 18. Quigley HA, Tielsch JM, Katz J, Sommer A. Rate of progression in open-angle glaucoma estimated from cross-sectional Financial support for this project was by Whitfeld Fellowship prevalence of visual field damage. Am J Ophthalmol 1996; 122: and Alvina King Medical Research Fellowship from Univer- 355–63. sity of Western Australia, Jack Hoffman Scholarship from 19. Bayer AU, Erb C. Short wavelength automated perimetry, fre- Lions Save Sight Foundation and Canadian Optometric Edu- quency doubling technology perimetry, and pattern electrore- cation Trust Fund from the Canadian Association of tinography for prediction of progressive glaucomatous Optometrists. The funding sources had no role in study standard visual field defects. Ophthalmology 2002; 109: 1009– design, data collection, data analysis, data interpretation or 17. in the writing of the report or the decision to submit the 20. Yamada N, Chen PP, Mills RP et al. Screening for glaucoma paper for publication. with frequency-doubling technology and Damato campimetry. Arch Ophthalmol 1999; 117: 1479–84. 21. Tribble JR, Schultz RO, Robinson JC, Rothe TL. Accuracy of REFERENCES glaucoma detection with frequency-doubling perimetry. Am J 1. Quigley HA, Broman AT. The number of people with glau- Ophthalmol 2000; 129: 740–5. coma worldwide in 2010 and 2020. Br J Ophthalmol 2006; 90: 22. Thomas R, Bhat S, Muliyil JP et al. Frequency doubling perim- 262–7. etry in glaucoma. J Glaucoma 2002; 11: 46–50. 2. Kumar S, Yogesan K, Constable IJ. Should telemedicine in eye 23. Tatemichi M, Nakano T, Tanaka K et al. Laterality of the per- care be funded in Australia? Med J Aust 2004; 181: 583. formance of glaucoma mass screening using frequency- 3. Kumar S, Yogesan K. Internet-based eye care: vision 2020. doubling technology. J Glaucoma 2003; 12: 221–5. Lancet 2005; 366: 1244–5. 24. Quaranta GM, Vetrugno M, Trabucco T, Cantatore F, Cardia 4. Yogesan K, Kumar S, Constable IJ. Tele-glaucoma: use of L. The role of frequency doubling technology (FDT) in glau- digital devices and web-based system for screening and detec- coma screening. Acta Ophthalmol Scand 2000; 78 (Suppl. 232): tion of glaucoma. Telemed J E Health 2004; 10: 45–7. 21–3. 5. Fleming C, Whitlock E, Beil T, Smit B, Harris R. Screening for 25. Medeiros FA, Sample PA, Weinreb RN. Frequency doubling primary open-angle glaucoma in the primary care setting: an technology perimetry abnormalities as predictors of glaucoma- update for the US Preventive Services Task Force. Ann Fam Med tous visual field loss. Am J Ophthalmol 2004; 137: 863–71. 2005; 3: 167–70. 26. Quigley HA, Dunkelberger GR, Green WR. Chronic human 6. Bland JM, Altman DG. Measuring agreement in method com- glaucoma causing selectively greater loss of large optic nerve parison studies. Stat Methods Med Res 1999; 8: 135–60. fibers. Ophthalmology 1988; 95: 357–63. 7. Bland JM, Altman DG. Statistical methods for assessing agree- 27. Sponsel WE, Arango S, Trigo Y, Mensah J. Clinical classifica- ment between two methods of clinical measurements. Lancet tion of glaucomatous visual field loss by frequency doubling 1986, 307–10. perimetry. Am J Ophthalmol 1998; 125: 830–6. 8. Mitchell P, Lee AJ. Open-angle glaucoma and systemic hyper- 28. Soliman MAE, De Jong LAMS, Ismaeil A et al. Standard achro- tension: the Blue Mountains eye study. J Glaucoma 2004; 13: matic perimetry, short wavelength automated perimetry, and 319–26. frequency doubling technology for detection of glaucoma 9. Weih LM, Nanjan M. Prevalence and predictors of open-angle damage. Ophthalmology 2002; 109: 444–54. glaucoma: results from the visual impairment project. Ophthal- 29. Maeda H, Tanaka Y, Nakamura M, Yamamoto M. Blue-on mology 2001; 108: 1966–72. yellow perimetry using an Armaly glaucoma screening 10. Kumar S, Mei-Ling TK, Constable IJ, Yogesan K. Internet program. Ophthalmologica 1999; 213: 71–5. based ophthalmology service: impact assessment. Br J Ophthal- 30. Mansberger SL, Johnson CA, Cioffi GA et al. Predictive value of mol 2005; 89: 1382–3. frequency doubling technology perimetry for detecting glau- 11. Tuulonen A, Ohinmaa A, Alanko HI et al. The application of coma in a developing country. J Glaucoma 2005; 14: 128–34. teleophthalmology in examining patients with glaucoma: a 31. Cioffi GA, Mansberger S, Spry P et al. Frequency doubling pilot study. J Glaucoma 1999; 8: 367–73. perimetry and the detection of eye disease in the community. 12. Li HK, Tang RA, Oschner K et al. Telemedicine screening of Trans Am Ophthalmol Soc 2000; 98: 195–202. glaucoma. Telemed J 1999; 5: 283–90. 32. American Academy of Ophthalmology Preferred Practice 13. Yogesan K, Constable IJ, Barry CJ et al. Evaluation of portable Pattern Committee Glaucoma Panel.Primary Open-Angle Glau- fundus camera for use in the teleophthalmologic diagnosis of coma, Limited Revision. 2003. Accessed 18 Feb 2007. Available glaucoma. J Glaucoma 1999; 8: 297–301. from: http://www.aao.org/aao/education/library/ppp/upload/ 14. Yogesan K, Constable IJ, Eikelboom RH, van Saarloos PP. Primary-Open-Angle-Glaucoma_.pdf Tele-ophthalmic screening using digital imaging devices. Aust 33. Jackson C, Bullock J, Pitt M et al. Screening for glaucoma in a NZJOphthalmol 1998; 26 (Suppl. 1): S9–11. Brisbane general practice – the role of tonometry. AustNZJ 15. Harper RA, Reeves BC. Glaucoma screening: the importance of Ophthalmol 1995; 23: 173–8. combining test data. Optom Vis Sci 1999; 76: 537–43. 34. Rezaie T, Child A, Hitchings R et al. Adult-onset primary open- 16. Michelson G, Groh MJ. Screening models for glaucoma. Curr angle glaucoma caused by mutations in optineurin. Science 2002; Opin Ophthalmol 2001; 12: 105–11. 295: 1077–9. 17. Paul H, Alvine KF, Demosthenes P. Screening for primary 35. Kass MA, Heuer DK, Higginbotham EJ et al. The Ocular open-angle glaucoma in the developed world: are we there yet? Hypertension Treatment Study: a randomized trial determines Can J Ophthalmol 2005; 40: 477–86. that topical ocular hypotensive medication delays or prevents

the onset of primary open-angle glaucoma. Arch Ophthalmol 41. Thomas R, Kumar RS, Chandrasekhar G, Parikh R. Applying 2002; 120: 701–13. the recent clinical trials on primary open angle glaucoma: the 36. Moseley MJ, Thompson JR, Deutsch J et al. Comparison of the developing world perspective. J Glaucoma 2005; 14: 324–7. Keeler Pulsair 2000 non-contact tonometer with Goldmann 42. Muller A, Vu HT, Ferraro JG, Keeffe JE, Taylor HR. Rapid and applanation. Eye 1993; 7: 127–30. cost-effective method to assess vision disorders in a population. 37. Mitchell P, Rochtchina E, Lee AJ, Wang JJ. Bias in self-reported Clin Experiment Ophthalmol 2006; 34: 521–5. family history and relationship to glaucoma: the Blue Moun- 43. Quigley HA. Current and future approaches to glaucoma tains Eye Study. Ophthalmic Epidemiol 2002; 9: 333–45. screening. J Glaucoma 1998; 7: 210–20. 38. Harris R. Screening for glaucoma. Waiting until our vision 44. Johannesson G, Gudmundsdottir GJ, Linden C. Can the preva- clears. BMJ 2005; 331: E376–7. lence of open-angle glaucoma be estimated from a retrospec- 39. Maier PC, Funk J, Schwarzer G et al. Treatment of ocular hyper- tive clinical material? A study on the west coast of Iceland. Acta tension and open angle glaucoma: meta-analysis of randomised Ophthalmol Scand 2005; 83: 549–53. controlled trials. BMJ 2005; 331: 134. 45. Grodum K, Heijl A, Bengtsson B. A comparison of glaucoma 40. Guirguis J. Rationale for the USPSTF recommendation on patients identiﬁed through mass screening and in routine clini- screening for glaucoma. Am Fam Physician 2005; 72: 1184, 1187. cal practice. Acta Ophthalmol Scand 2002; 80: 627–31.