Glaucoma screening during regular optician visits: the feasibility and specificity of screening in real life

Authors

  • Margriet M. de Vries,

    1. Department of Ophthalmology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Remco Stoutenbeek,

    1. Department of Ophthalmology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Rogier P. H. M. Müskens,

    1. Department of Ophthalmology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands
    Search for more papers by this author
  • Nomdo M. Jansonius

    1. Department of Ophthalmology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands
    2. Department of Epidemiology, Erasmus Medical Center, Rotterdam, The Netherlands
    Search for more papers by this author

Dr Nomdo M. Jansonius
Department of Ophthalmology
University Medical Center Groningen
PO Box 30.001
9700 RB Groningen
The Netherlands
Tel: + 31 50 3612510
Fax: + 31 50 3611709
Email: n.m.jansonius@umcg.nl

Abstract.

Purpose:  To determine the feasibility and specificity of glaucoma screening during regular optician visits.

Methods:  In four optician shops, glaucoma screening was offered to 400 consecutive visitors aged 45 years or above. If the visitor agreed to participate, an intraocular pressure measurement and – in those with a pressure below 25 mmHg – a frequency-doubling perimeter (FDT) C20-1 visual field screening test were performed. Those with an elevated pressure or at least one reproducibly abnormal test location on FDT were referred to our hospital.

Results:  Three-hundred and fifty-two of 400 consecutive visitors (88%) were screened. Fifteen of the unscreened visitors were not screened because they were already regularly checked by an ophthalmologist related to glaucoma. Forty-two of 352 screened participants (12%) were referred. Of these 42 referrals, seven were diagnosed with glaucoma, 10 were diagnosed with ocular hypertension (OHT), 12 did not have any eye disease, seven had an eye disease other than glaucoma or OHT that was diagnosed previously and six were newly diagnosed with an eye disease other than glaucoma or OHT. The specificity of the screening protocol was 91% (95% confidence interval 88–94%).

Conclusions:  Glaucoma screening at the optician shop was feasible, but the specificity of the screening protocol was rather low. With more stringent cut-off points (30 mmHg; at least two reproducibly abnormal test locations), the specificity could be improved to 96% (94–98%), apparently without a significant loss of sensitivity. This suggests that screening during regular optician visits might be a viable approach.

Introduction

Glaucoma is a chronic eye disease, which eventually can lead to irreversible blindness. The prevalence of open-angle glaucoma in the 40+ population is approximately 2% and the 10-year incidence in the Dutch 55+ population 3% (Burr et al. 2007; Czudowska et al. 2010). Symptoms often go unnoticed because of the insidious nature; half of the patients are undetected at a certain moment (Hollows & Graham 1966; Wolfs et al. 2000). Because treatment can slow down the progression of glaucoma (Heijl et al. 2002; Maier et al. 2005), screening could be useful. At this moment, there is no nationwide periodic population-based screening programme for glaucoma in the Netherlands nor in any other country.

The low prevalence of glaucoma is one reason why glaucoma screening is not obviously cost-effective (Vaahtoranta-Lehtonen et al. 2007; Burr et al. 2007; Hernández et al. 2008). The prevalence (prior probability) may be increased by limiting the screening to high-risk groups. Although this might work in some populations, we showed – in a Caucasian population in the Western world with an easily accessible healthcare system – that preselection on the basis of known glaucoma risk factors (family history and myopia) would not improve the performance of a screening programme. The reason was that patients with these risk factors were already more likely to be picked up in regular care (Stoutenbeek et al. 2008). A second reason why glaucoma screening is not obviously cost-effective is the lack of an ideal screening test (Mowatt et al. 2008). An ideal screening test is simple, cheap and fast, and combines the required very high specificity (above 95%; Vaahtoranta-Lehtonen et al. 2007) with an acceptably high sensitivity. A third reason is length bias (Johnson et al. 2003). This bias arises if undetected patients differ from the already detected ones. For example, the former group might have a less aggressive disease course. In a recent study, we showed that indeed patients detected during a population-based screening differed from patients picked up in regular care (Stoutenbeek et al. 2008).

Given all the arguments against the introduction of a nationwide glaucoma screening programme, the question is left if current case-finding strategies could be improved. Current case-finding strategies consist mainly of (i) ophthalmologists paying attention to glaucoma during visits for any reason, (ii) ophthalmologists warning and checking family members of patients with glaucoma and (iii) opticians screening their shop visitors by measuring the intraocular pressure (IOP). Paying attention to glaucoma during visits for any reason has been shown to be a cost-effective approach (Peeters et al. 2008); the warning and checking of family members seem to be effective as well (Stoutenbeek et al. 2008). An IOP measurement alone, however, has a poor screening performance (Mowatt et al., 2008). This does not disqualify the optician shop as a potential screening site: after all, 83% of the population at risk for glaucoma visits an optician at least once in 5 years (Stoutenbeek & Jansonius 2006). Hence, some kind of screening with a better performance than an IOP measurement alone, performed at the optician shop, would very likely improve glaucoma case finding – at little additional costs.

The aim of this study was to determine the feasibility and specificity of a two-stage glaucoma screening (in strict terminology still to be called case finding) during regular optician visits. For this purpose, selected optician shops screened consecutive visitors aged 45 years or above. Following a strict protocol, they performed an IOP measurement and – in those with a pressure below 25 mmHg – a frequency-doubling perimetry (FDT; Carl Zeiss Meditec Inc., Dublin, CA, USA) screening test. Those with an elevated IOP or a reproducibly abnormal FDT test result were referred to our hospital and assessed in more detail.

Material and Methods

Study population

Optician shops were eligible if they (i) had both a tonometer and an FDT device and (ii) were located within 30 km of our hospital. Four optician shops fulfilled both criteria and were invited to participate; all agreed. Glaucoma screening was offered to 400 consecutive visitors (100 visitors per optician shop) of 45 years and above. Self-declared glaucoma patients who regularly visited an ophthalmologist were not screened. The ethics board of the University Medical Center Groningen approved the study protocol. All participants provided written informed consent.

Screening and ophthalmic assessment

Figure 1 shows the screening protocol as followed by the opticians. Screening was performed between March and July 2010. The IOP was measured with non-contact tonometry (NCT), as this was the only device available in the shops. If an IOP of 25 mmHg or higher was measured in at least one eye, the visitor was referred; else, an FDT test using the C20-1 screening mode was performed on both eyes. If at least one abnormal test location was found, and confirmed in at least one same test location, the visitor was referred. It was shown earlier that this confirmation roughly halves the number of false-positive FDT test results without a significant loss of sensitivity (Heeg et al. 2005b).

Figure 1.

 Flow diagram for glaucoma screening as used by the opticians.

The ophthalmic exam was performed on average 2 weeks after the screening, with a maximum of 2 months. The referral information included the IOP, the FDT test results, and refraction and visual acuity data. The additional investigations were performed by one physician investigator (MMV) and included three IOP measurements with NCT (NT-2000; Nidek, Gamagori, Japan), an FDT C20-1 screening test, perimetry with a Humphrey field analyzer (HFA 30-2 SITA; Carl Zeiss Meditec), a measurement of the central corneal thickness (Tomey SP-3000 pachymeter; Tomey, Nagoya, Japan) and a 35° fundus photograph of the posterior pole taken in mydriasis. One of three experienced ophthalmologists (RS, RPHMM or NMJ) measured the IOP with Goldmann applanation tonometry (Haag-Streit, Bern, Switzerland), performed a dilated ophthalmic exam and classified the referred participant in one of five categories using all available information. These five categories were (i) glaucoma, (ii) ocular hypertension (OHT), (iii) no evidence for any eye disease, (iv) previously diagnosed eye disease not being glaucoma or OHT and (v) newly diagnosed eye disease not being glaucoma or OHT. For perimetric glaucoma, two consecutive, reliable visual field tests had to be abnormal in at least one eye, according to previously published criteria (Heeg et al. 2005a; Wesselink et al. 2009). Defects had to be in the same hemifield, and at least one depressed test point of these defects had to have exactly the same location on both fields. Moreover, defects had to be compatible with glaucoma and without any other explanation (altitudinal defect, arcuate scotoma, nasal step, central island or temporal island, or combinations thereof, were considered compatible; homonymous or bitemporal anopia or isolated central scotoma were not). The first test was discarded because of a learning effect. Therefore, at least three tests had to be performed before glaucoma could be diagnosed. For glaucomatous optic neuropathy, at least one of the following was required: (i) a vertical cup-disc ratio of more than 0.7, (ii) a cup-disc ratio asymmetry of more than 0.2, (iii) a vertical elongation of the cup or (iv) a disc haemorrhage. This is the definition of the ISGEO glaucoma classification working group for use in ophthalmic epidemiology (Foster et al. 2002); the cut-off points for the recommended 97.5th percentiles were taken from the Rotterdam Study (Wolfs et al. 2000; Ramdas et al. 2011). For OHT, an IOP above 20 mmHg had to be measured at least twice, and no glaucomatous visual field defect was allowed.

All referred visitors completed a questionnaire concerning the reason for visiting the optician. We also made an inventory of possible risk factors for open-angle glaucoma, family history of glaucoma, myopia and corticosteroid use, and of ethnicity, ophthalmic surgery in the past and diabetes.

Data analysis

Descriptive statistics was used to describe the characteristics of the visitors and patients (referred visitors). A t-test was used to compare groups. A p-value of 0.05 or less was considered statistically significant.

The specificity was calculated on an ‘intent-to-screen’ basis. In this approach, to be compared with an ‘intent-to-treat’ basis in trials, the number of screened persons equals the number of visitors offered screening (Lachin 2000). Furthermore, the number of normal test results equals the number of non-referred visitors, the number of true-positive cases equals the number of referred visitors in which glaucoma was diagnosed, and the number of false positives equals the number of referred visitors minus the number of true-positive cases. We assumed the number of false negatives to be small in comparison with the number of true negatives (see Discussion section). Hence, the specificity was approximated as the number of non-referred visitors divided by the number of visitors offered screening minus the number of true-positive cases. The corresponding 95% confidence interval was calculated using the binomial distribution. True-positive cases were cases with glaucoma; we also made calculations with OHT cases with a confirmed IOP above 25 mmHg and cases with a newly detected eye disease not being glaucoma or OHT that required treatment added to the true-positive cases.

Results

Table 1 shows the results for all optician shop visitors offered screening. Of the 400 consecutive visitors, two (0.5%) were excluded because of missing data, leaving 398 visitors. Three-hundred and fifty-two visitors (88%) were screened according to the protocol. Fifteen visitors were not screened because they either had glaucoma or were regularly checked for glaucoma by an ophthalmologist (self-declared glaucoma patients); nine because of lack of time at the optician shop. Twenty-two visitors did not want to or could not undergo screening. Reasons mentioned were as follows: incapable to execute the test(s) (7), not willing because already known with an ocular disorder not being glaucoma (7; no reason to refrain from screening, though) and unspecified reasons (8). Table 1 shows that most of the visitors referred by opticians B and D had an abnormal FDT test result, while most of the visitors referred by optician C had an IOP of 25 mmHg or above. Repeating an abnormal FDT test result roughly halved the number of (false positive) referrals. Interestingly, eight of 50 referred visitors did not show up after being invited for additional investigations. The average age of these visitors was 72 years, that is, significantly higher than that of the remaining of the group (p < 0.001; t-test). In the specificity calculations, they were considered non-referred.

Table 1.   Characteristics of the screened population (N = 398).
 Optician AOptician BOptician COptician DTotal
  1. IOP = intraocular pressure; FDT = frequency-doubling perimeter.

Gender (male/female)41/57 32/68 39/61 34/66146/252
Age [years; mean (range)]60 (45–90) 64 (45–93) 63 (46–89) 69 (45–91) 64 (45–93)
Number of consecutive visitors98100100100398 (100%)
Screened by the optician
 Yes89 91 97 75352 (88%)
 No, because
  Self-declared glaucoma patient 4  4  2  5 15 (4%)
  Lack of time 0  0  0  9  9 (2%)
  Did not want to or could 5  5  1 11 22 (6%)
Referred to the hospital
 Yes, because
  IOP ≥ 25 mmHg 5  1 15  1 22 (6%)
  Reproducibly abnormal FDT 4  6  8 10 28 (8%)
 No, because
  IOP < 25 mmHg and normal 1st FDT76 77 63 58274 (78%)
  IOP < 25 mmHg and normal 2nd FDT 4  7 11  6 28 (8%)
Referred but not further assessed 1  0  4  3  8 (16%)
Referred and assessed 8  7 19  8 42 (12%)

Forty-two visitors were investigated at the hospital. Table 2 gives their characteristics. Nineteen of 42 were referred because of an elevated IOP and 23 of 42 because of a reproducibly abnormal FDT test result. The most frequently reported reason for visiting the optician was related to glasses. Only a minority of the referred visitors had myopia and/or a positive family history of glaucoma as risk factor.

Table 2.   Characteristics of the referred population (N = 42).
  1. IOP = intraocular pressure; FDT = frequency-doubling perimeter; VA = visual acuity; CE = cataract extraction.

Gender (male/female)16/26
Age [years; median (range); male/female]63 (45–78)/71 (47–87)
IOP as measured by the optician (highest value of both eyes) in those referred because of IOP ≥ 25 mmHg [n = 19; mmHg; median (range)]27 (25–43)
IOP as measured by the optician (highest value of both eyes) in those referred because of a reproducibly abnormal FDT [n = 23; mmHg; median (range)]17 (12–24)
Reason(s) for optician visit
 Glasses25
 Problems with reading4
 Decreased vision5
 IOP measurement1
 Family history1
 General control1
 Driving test1
 Fundus photograph3
 Unclear1
VA [worse/better eye; median (range)]0.90 (0.10–1.40)/1.00 (0.35–1.40)
Myopia (spherical equivalent; most negative value of both eyes)
 ≤−4.00 dioptres4
 −3.99 to −0.01 dioptres6
 No myopia32
Family member(s) with glaucoma
 Yes4
 No19
 Do not know19
Ethnicity
 Caucasian39
 Other3 (Indonesians)
Diabetes mellitus
 Yes7 (1 type I, 6 type II)
 No35
Eye operation (any eye)
 Yes10 (CE 8; blepharoplasty 1; trauma 1)
 No32
Any corticosteroid use
 Yes5
 No, sporadic3
 No, never19
 Don’t know15

Table 3 presents the final classification of the referred population. Seven (17%) were diagnosed with glaucoma. Ten (24%) were diagnosed with OHT, of which one had already been diagnosed before the screening. Twelve (28%) did not have any eye disease, and seven (17%) had an eye disease other than glaucoma or OHT that was diagnosed previously. Finally, six (14%) were newly diagnosed with an eye disease other than glaucoma or OHT. Five of seven patients with glaucoma were referred because of an abnormal FDT test result, two because of a high IOP.

Table 3.   Diagnoses of the referred population (N = 42).
 Referral reasonTotal
Elevated IOPAbnormal FDT
  1. IOP = intraocular pressure; FDT = frequency-doubling perimeter; DRP = diabetic retinopathy; CVA = cerebrovascular accident.

  2. * One patient was previously diagnosed with ocular hypertension.

  3. Although captured by IOP, this visitor had FDT abnormalities as well (as assessed at the hospital).

Glaucoma257
Ocular hypertension*10010
No evidence for glaucoma or any other eye disease5712
Eye disease other than glaucoma; previously diagnosed:077
 Cataract022
 Opticopathy by Lyme disease011
 Old eye trauma011
 Status after pan-laser for DRP011
 Quadrant anopsia after CVA022
Eye diseases other than glaucoma; newly diagnosed246
 Papil hypoplasia011
 Chronic central serous retinopathy011
 Quadrant anopsia caused  by a tumour101
 Congruent partial homonymous hemianopsia caused by CVA011
 Inconclusive visual field test112

The specificity of the screening protocol was calculated to be 91% (95% confidence interval 88–94%). If we would consider OHT with a confirmed IOP above 25 mmHg and newly detected eye diseases that required treatment as being worthwhile to be captured, the number of true positives would increase from 7 to 16, yielding a specificity of 93%. In the 12 referrals without glaucoma or any other eye disease, on average, one visit (maximum two visits) was sufficient to come to this conclusion and discharge the subject after reassurance.

Table 4 shows the characteristics of the patients diagnosed with glaucoma or OHT. The median age of the group with glaucoma was 78 years (range: 62–87 years) and of the group with OHT 60 years (range: 48–74 years). Risk factors for glaucoma were not clearly present in these groups. The median (range) standard automated perimetry mean deviation of patients diagnosed with glaucoma was −6.0 dB (−4.2 to −15.7 dB) for their worse eye and −4.2 dB (−1.9 to −12.3 dB) for their better eye. Six of seven patients with glaucoma and one of ten OHT patients had a glaucomatous optic neuropathy according to the criteria as stated in the Methods Section.

Table 4.   Characteristics of the patients diagnosed with OHT or glaucoma.
 Glaucoma (n = 7)OHT (n = 10)
  1. OHT = ocular hypertension; VA = visual acuity; DM = diabetes mellitus; NCT = non-contact tonometry; AT = applanation tonometry.

  2. * Blepharoplasty.

Gender (male/female) 3/4  5/5
Age [years; median (range)]78 (62–87) 60 (48–74)
VA [worse/better eye; median (range)] 0.7 (0.1–1.0)/0.8 (0.6–1.0)  1.0 (0.4–1.2)/1.0 (0.9–1.2)
EthnicityAll CaucasianAll Caucasian
Family history of glaucomaYes 1; no 6Yes 1; no 9
DM I/IIYes 0; no 7Yes 3; no 7
Eye operationYes 2; no 5Yes 1*; no 9
Any corticosteroid useYes 2; no 4; don’t know 1No 7; do not know 3
Myopia (spherical equivalent; most negative value of both eyes)
 ≤−4.00 dioptres 0  1
 −3.99 to −0.01 dioptres 1  0
 No myopia 6  9
NCT as measured in the hospital/AT [mmHg; highest value of both eyes; median (range)]18 (15–26)/27 (15–32) 24 (22–30)/23 (20–30)
Corneal thickness [μm; mean value of both eyes; median (range)]550 (484–582)578 (517–613)
Open angleYes 7; no 0Yes 10; no 0
Lens: phakic/pseudophakic at least one eye 5/2 10/0
Glaucomatous optic neuropathy at least one eyeYes 6; no 1Yes 1; no 9
Pseudo exfoliation at least one eyeYes 1; no 6Yes 1; no 9
Pigment dispersion syndrome at least one eyeYes 1; no 6Yes 0; no 10

Discussion

The specificity of this two-stage screening protocol was 91%. Glaucoma was newly diagnosed in 17% of the referred visitors, that is, in 2% of the screened population. The ophthalmic consultation was considered worthwhile in about one-fifth of the remaining referred visitors.

As mentioned in the Materials and Methods Section, we calculated the specificity on an ‘intent-to-screen’ basis. If we would only count the actually screened visitors (n = 352), the specificity would be 90%. If we would consider our screening a screening for any eye disease rather than for glaucoma, the specificity would increase to 93% (see Results section). In our specificity calculation, we assumed the number of false negatives to be small in comparison with the number of true negatives. This is a logical assumption in any disease with a low prevalence (such as glaucoma), but it is not a critical assumption. For example, even with 10% of the non-referred population being false negatives (the real percentage is more likely to be typically 1%), the calculated specificity would change by no more than 1% (from 91% to 90%).

The power of this study was targeted at the determination of the specificity, which is, unlike the sensitivity, pivotal from the point of view of cost-effectiveness. A tentative comparison of the number of newly detected glaucoma cases (2% of the screened population) with the estimated prevalence of undetected glaucoma (typically 1%) suggests an at least reasonable sensitivity (Burr et al. 2007; Hollows & Graham 1966; Wolfs et al. 2000).

It is unlikely that the FDT has a higher sensitivity for the earliest stages of glaucoma than imaging devices or standard automated perimetry (Cello et al. 2000; Trible et al. 2000; Jansonius & Heeg 2009). Reasons to select, nonetheless, the FDT for screening at the optician shop were, amongst others, the results of a recent systematic review and meta-analysis dealing with glaucoma screening tests (Mowatt et al. 2008), its ease of use, the short testing time and the relatively low initial and maintenance costs. Its very high sensitivity for all, but the earliest stages of glaucoma (Stoutenbeek et al. 2004), should prevent missing patients with more advanced disease stages. The earliest stages are more likely to be picked up if an imaging device would be added for those with a normal FDT test result, but this approach has been shown to have a substantially lower specificity (Heeg et al. 2005b). In this respect, it is important to realize that a widely performed screening with its need for a high specificity will never be able to pick up the earliest stages (as a result of cost-effectiveness issues). Therefore, our approach should not be considered as a replacement for currently performed case-finding activities in clinical settings, but should rather function as an additional safety net for undetected glaucoma patients who are currently reaching the ophthalmologist too late. Also, our approach allows for the referral of those with an obviously elevated IOP irrespective of the presence of FDT abnormalities. On average, those with an obviously elevated IOP progress faster than those with an IOP within normal limits (Anderson et al. 2001;Heijl et al. 2002). Finally, it is important to keep in mind that the majority of the optician shop visitors will revisit the optician within 5 years (Stoutenbeek & Jansonius 2006). This provides the overlooked early cases a second chance to be picked up. A periodic population-based screening at 5-year intervals seems to yield a timely detection of the majority of the patients with glaucoma (Stoutenbeek et al. 2008).

There are countless studies published concerning glaucoma screening, but the vast majority of them focused on a single device applied to a single eye and were conducted in subjects that were not allowed to have any concurrent eye disease. In real-life screening, however, both eyes are tested in unselected subjects and several devices may be applied in a parallel or serial fashion. Therefore, the performance of a screening protocol cannot be predicted from ‘clean’ single-eye, single-test sensitivity and specificity values. Hence, the first step in weighing the costs and benefits of glaucoma screening is the assessment of the performance of entire screening protocols. Several studies evaluated glaucoma referral schemes or guidelines that were designed to reduce the amount of false-positive glaucoma referrals and to improve the quality of referrals from optometrists (Vernon & Ghosh 2001; Henson et al. 2003; Ang et al. 2009; Bourne et al. 2010; Davey et al. 2010; Devarajan et al. 2011). Of these studies, only one study presented a specificity value, being 87% (Devarajan et al. 2011). None of the studies mentioned the use of the FDT device. On average, three (range three to five) tests were included, and different strategies were used including assessments at more than one site before an ophthalmologist was consulted (for example, community optometrists referring to specialized optometrists). This suggests that our protocol has a favourable specificity/complexity ratio, with only two tests (NCT and FDT) and one site (optician shop).

The Finnish cost-effectiveness study mentioned a required specificity of more than 95% for a cost-effective glaucoma screening (Vaahtoranta-Lehtonen et al. 2007). Although this depends to some extent on model assumptions and may be different in other countries because of demographic and healthcare differences, the specificity of our protocol of 91% does not meet this requirement. To reduce the amount of false-positive referrals, aiming to increase the specificity, several strategies may be employed. The most obvious strategies are (i) the change in the cut-off points of the tests and (ii) preselection with risk factors for glaucoma. Preselection with risk factors would obviously reduce the yield of our screening protocol because there were no clear risk factors present in the group of patients diagnosed with glaucoma (Table 4). This is in line with the results of an earlier study as mentioned in the Introduction (Stoutenbeek et al. 2008). If we would increase the FDT cut-off point to at least two reproducibly abnormal test locations, the specificity would increase from 91% to 93%. If we would increase the IOP cut-off point to 30 mmHg, the specificity would increase to 94%. If we would employ both measures simultaneously, the specificity would increase to 96% (95% confidence interval 94–98%). In all three scenarios, the number of detected glaucoma patients would remain unaffected. Obviously this is a secondary analysis and the number of identified glaucoma patients is too small to allow for a thorough study of the effects of alternative strategies on the sensitivity. Nevertheless, these findings suggest that the use of more stringent cut-off points might be a more viable approach than preselection based on risk factors.

A potential source of bias is the selection of opticians who participated in this study. Only a minority of the opticians in the Netherlands currently has both a tonometer and an FDT device. Hence, the included opticians may differ in aptitude for screening from their colleagues. This may hamper the performance if such a screening protocol would be introduced nationwide. On the other hand, the current practice (in the Netherlands but also in many other countries) in optician shops is an IOP measurement alone. As mentioned in the Introduction Section, an IOP measurement alone has a rather poor screening performance (Mowatt et al. 2008). Hence, given the results of the present study, the replacement of an IOP measurement alone by our approach would very likely improve glaucoma case finding – at little additional costs.

In conclusion, glaucoma screening at the optician shop was feasible, but the specificity of the original screening protocol was rather low. However, with more stringent cut-off points for IOP and FDT, a specificity of 96% could be reached, apparently without loss of sensitivity. This suggests that a cost-effective glaucoma screening during regular optician shop visits is not unrealistic.

Acknowledgements

The authors wish to thank Moritsz Schelwald Optiek in Groningen, Madlener Optiek in Vries, Woldringh Optiek in Groningen and Het Huis Opticiens in Drachten, all in the Netherlands. The authors do not have a conflict of interest. This work was presented at the ARVO meeting 2011, Fort Lauderdale, FL, USA. Financial support was provided by Stichting Oogfonds Nederland, Glaucoomfonds Nederland. The funders had no role in the design and the conduct of the study, in the data collection, analysis and interpretation, and in the preparation, review or approval of the manuscript.

Ancillary