The aim of a programme screening for asymptomatic field defects must be two-fold: to detect true positives (the association between asymptomatic field defects and genuine pathology) and to reduce false positives (such as with a field screening protocol for glaucoma). When screening fields we also need to be aware of the differences between glaucomatous and non-glaucomatous field defects, since this will influence referral practice.
Audit Protocol
Between December 2000 and April 2001 we audited the results of consultations in the Bournemouth Eye Unit based on referrals for asymptomatic field defects via GOS18 forms.
Twenty-three referrals were received during this period for asymptomatic field defects, of which 20 sets of notes were located.
The patients' age range was from 27-90. We compared the screening field performed by the optometrist with the Humphrey threshold field performed in the eye department.
With regard to reducing false positives, we concentrated on those patients referred as glaucoma suspects. We took as our gold standard the Bury St Edmunds' audit1 that demonstrated 60 per cent of patients referred as glaucoma suspects received a final positive diagnosis of glaucoma.
For detecting true positives, we concentrated on previous studies reviewing the association between asymptomatic field defects and genuine pathology. We selected a paper that reported 52 per cent of patients referred to one department with asymptomatic field defects turned out to have identifiable pathology (both glaucomatous and non-glaucomatous).2 This was, therefore, our second gold standard.
Method
We used a standardised system for grading the field defects (Table 1). Our objectives were to:
Compare the consistency of field defects between screening fields and threshold fields and possibly associate consistency with diagnosis. Each patient had both monocular field defects for screening and threshold fields compared. Only one eye had to demonstrate a defect classified the same on screening and threshold fields for that patient to be labelled as consistent.
With glaucoma referrals, review the corroborative evidence provided with an asymptomatic field defect to support the referral and the strength of association between types of evidence and a final positive glaucoma diagnosis.
With regard to non-glaucoma referrals, the characteristics of defects more likely to be associated with this group.
Review if the referral patterns in our region achieved our audit standards.
Results
Glaucoma referrals
Eleven of 20 patients were referred as glaucoma suspects with asymptomatic field defects.
One was referred with field loss only, two with fields and raised intraocular pressure, two with fields and suspicious discs, six with fields and a family history of glaucoma, and one with field changes and a previous diagnosis of glaucoma but since discharged to the optometrist glaucoma follow-up scheme.
Of these, 63 per cent (seven patients) had a positive diagnosis of glaucoma or glaucoma progression made. Two patients had non-glaucomatous pathology found which explained the field defects (one dense cataract and one tilted disc) and two had no pathology found.
On threshold field-testing, positive glaucoma diagnosis patients demonstrated either arcuate, central or ungradable defects. Three patients' defects were consistent with their screening fields and four were inconsistent.
Non-glaucoma referrals
Nine of 20 patients were referred with asymptomatic defects of unknown aetiology. Thirty-three per cent (three patients) had an identifiable pathology; two had chorioretinal scars and a previous optic neuritis. Six had no pathology found.
If we consider all the patients with identifiable non-glaucomatous pathology, both from the glaucoma suspect and non-glaucoma referral groups (five in total), we find that all their field defects were consistent between screening and threshold fields. On threshold testing they demonstrated either temporal wedges or paracentral scotomas.
Conclusions
Glaucoma referrals
We found a 63 per cent true positive referral rate for glaucoma patients. We have therefore achieved our audit standard. The problem of improving the true positive rate is derived from the fact that glaucoma is a condition of relatively low prevalence. A test with 99 per cent specificity and 99 per cent sensitivity would still generate a 49.5 per cent false positive rate (see box below for explanation). Our audit suggests that optometrists in the Bournemouth and Poole region are therefore theoretically doing better than this.
The patient referred with field loss alone had a positive glaucoma diagnosis. In fact Newman et al1 showed that referral with isolated field loss was most weakly associated with a positive diagnosis of glaucoma. Isolated change in cup-to-disc ratios was more strongly associated and isolated raised intraocular pressure was most strongly associated. When considering more than one finding in combination, the results become more complicated. Newman's table is shown in Table 2.
The most often cited corroborative factor on GOS18 forms to support referral as a glaucoma suspect with asymptomatic field loss was a positive family history of glaucoma. Strictly speaking a family history should principally be used to identify the group that need to be screened rather than to support a referral. Interestingly, the patients classified with either non-glaucomatous pathology or no identifiable pathology in the glaucoma suspects referral group all had a family history of glaucoma as the corroborative factor supporting glaucoma referral.
Four eyes of seven patients with positive glaucoma diagnoses demonstrated 'ungradable defects'. In fact generalised field depression has been shown to be highly sensitive in its association with early glaucoma. We included generalised depression as ungradable since it is screened out by threshold field algorithms. It is also of no use in screening since predictably generalised depression is not specific for early glaucoma.
Non-glaucoma referrals
Our positive diagnosis rate for all significant pathology (both glaucomatous and non-glaucomatous) was 60 per cent. Thus, we have achieved our audit standard.
Considering all patients, only 50 per cent had consistent defects when comparing screening to threshold fields. It is pertinent therefore that all the patients who were found to have significant non-glaucomatous pathology had highly consistent defects. This is not an age influence, since the average age of these patients was 62, whereas the average age for those with a positive glaucoma diagnosis was 66. These findings do agree with previous studies demonstrating marked consistency between screening and threshold field defects for well defined retinal lesions.3
The ophthalmologist presented with a well-defined retinal lesion and a consistent field defect on a screening field should consider if any extra information would be provided by performing a threshold field.
The field defects associated with non-glaucomatous pathology tended to be paracentral or temporal. This is again in agreement with previous studies.4
Local optometry practices achieved our audit standards. We therefore hesitate to circulate guidelines based on our findings.
We can learn from Vernon et al who distributed a protocol for optometric glaucoma referral and managed to reduce the true positive referral rate from 40 per cent to 32.3 per cent.5 To close the audit loop we have therefore merely circulated our results and will be reauditing later this year. We hope that our audit provides some evidence basis for everyday practice.
References
1 Newman DK, Anwar S, Jordan K. Glaucoma screening by optometrists: positive predictive value of visual field testing. Eye, 1998; 12: 921-24.
2 Gutteridge IF. Clinical significance of detecting visual field loss. Am J Optom Physiol Opt, 1985; 62: 275-81.
3 Bass SJ, Feldman J. Visual field defects in well-defined retinal lesions using Humphrey and Dicon perimeters. Optometry, 2000; 71: 643-52.
4 Anton A, Maquet JA, Mayo A, Tapia J, Pastor JC. Value of logistic discriminant analysis for interpreting initial visual field defects. Ophthalmology, 1997; 104: 525-31.
5 Vernon SA, Ghosh G. Do locally agreed guidelines for optometrists concerning the referral of glaucoma suspects influence referral practice? Eye, 2001; 15: 458-63.
6 Rumney NJ, Henson DB. Glaucoma screening by optometrists: positive predictive value of visual field testing (letter). Eye, 1999; 13: 605-6.
Steve Madill is a specialist registrar at Queen Alexandra Hospital, Portsmouth, Steve Lash is an SHO at Southampton Eye Unit and Anne Denning is consultant ophthamologist at Bournemouth Eye UnitThe aim of a programme screening for asymptomatic field defects must be two-fold: to detect true positives (the association between asymptomatic field defects and genuine pathology) and to reduce false positives (such as with a field screening protocol for glaucoma). When screening fields we also need to be aware of the differences between glaucomatous and non-glaucomatous field defects, since this will influence referral practice.
Audit Protocol
Between December 2000 and April 2001 we audited the results of consultations in the Bournemouth Eye Unit based on referrals for asymptomatic field defects via GOS18 forms.
Twenty-three referrals were received during this period for asymptomatic field defects, of which 20 sets of notes were located.
The patients' age range was from 27-90. We compared the screening field performed by the optometrist with the Humphrey threshold field performed in the eye department.
With regard to reducing false positives, we concentrated on those patients referred as glaucoma suspects. We took as our gold standard the Bury St Edmunds' audit1 that demonstrated 60 per cent of patients referred as glaucoma suspects received a final positive diagnosis of glaucoma.
For detecting true positives, we concentrated on previous studies reviewing the association between asymptomatic field defects and genuine pathology. We selected a paper that reported 52 per cent of patients referred to one department with asymptomatic field defects turned out to have identifiable pathology (both glaucomatous and non-glaucomatous).2 This was, therefore, our second gold standard.
Method
We used a standardised system for grading the field defects (Table 1). Our objectives were to:
Compare the consistency of field defects between screening fields and threshold fields and possibly associate consistency with diagnosis. Each patient had both monocular field defects for screening and threshold fields compared. Only one eye had to demonstrate a defect classified the same on screening and threshold fields for that patient to be labelled as consistent.
With glaucoma referrals, review the corroborative evidence provided with an asymptomatic field defect to support the referral and the strength of association between types of evidence and a final positive glaucoma diagnosis.
With regard to non-glaucoma referrals, the characteristics of defects more likely to be associated with this group.
Review if the referral patterns in our region achieved our audit standards.
Results
Glaucoma referrals
Eleven of 20 patients were referred as glaucoma suspects with asymptomatic field defects.
One was referred with field loss only, two with fields and raised intraocular pressure, two with fields and suspicious discs, six with fields and a family history of glaucoma, and one with field changes and a previous diagnosis of glaucoma but since discharged to the optometrist glaucoma follow-up scheme.
Of these, 63 per cent (seven patients) had a positive diagnosis of glaucoma or glaucoma progression made. Two patients had non-glaucomatous pathology found which explained the field defects (one dense cataract and one tilted disc) and two had no pathology found.
On threshold field-testing, positive glaucoma diagnosis patients demonstrated either arcuate, central or ungradable defects. Three patients' defects were consistent with their screening fields and four were inconsistent.
Non-glaucoma referrals
Nine of 20 patients were referred with asymptomatic defects of unknown aetiology. Thirty-three per cent (three patients) had an identifiable pathology; two had chorioretinal scars and a previous optic neuritis. Six had no pathology found.
If we consider all the patients with identifiable non-glaucomatous pathology, both from the glaucoma suspect and non-glaucoma referral groups (five in total), we find that all their field defects were consistent between screening and threshold fields. On threshold testing they demonstrated either temporal wedges or paracentral scotomas.
Conclusions
Glaucoma referrals
We found a 63 per cent true positive referral rate for glaucoma patients. We have therefore achieved our audit standard. The problem of improving the true positive rate is derived from the fact that glaucoma is a condition of relatively low prevalence. A test with 99 per cent specificity and 99 per cent sensitivity would still generate a 49.5 per cent false positive rate (see box below for explanation). Our audit suggests that optometrists in the Bournemouth and Poole region are therefore theoretically doing better than this.
The patient referred with field loss alone had a positive glaucoma diagnosis. In fact Newman et al1 showed that referral with isolated field loss was most weakly associated with a positive diagnosis of glaucoma. Isolated change in cup-to-disc ratios was more strongly associated and isolated raised intraocular pressure was most strongly associated. When considering more than one finding in combination, the results become more complicated. Newman's table is shown in Table 2.
The most often cited corroborative factor on GOS18 forms to support referral as a glaucoma suspect with asymptomatic field loss was a positive family history of glaucoma. Strictly speaking a family history should principally be used to identify the group that need to be screened rather than to support a referral. Interestingly, the patients classified with either non-glaucomatous pathology or no identifiable pathology in the glaucoma suspects referral group all had a family history of glaucoma as the corroborative factor supporting glaucoma referral.
Four eyes of seven patients with positive glaucoma diagnoses demonstrated 'ungradable defects'. In fact generalised field depression has been shown to be highly sensitive in its association with early glaucoma. We included generalised depression as ungradable since it is screened out by threshold field algorithms. It is also of no use in screening since predictably generalised depression is not specific for early glaucoma.
Non-glaucoma referrals
Our positive diagnosis rate for all significant pathology (both glaucomatous and non-glaucomatous) was 60 per cent. Thus, we have achieved our audit standard.
Considering all patients, only 50 per cent had consistent defects when comparing screening to threshold fields. It is pertinent therefore that all the patients who were found to have significant non-glaucomatous pathology had highly consistent defects. This is not an age influence, since the average age of these patients was 62, whereas the average age for those with a positive glaucoma diagnosis was 66. These findings do agree with previous studies demonstrating marked consistency between screening and threshold field defects for well defined retinal lesions.3
The ophthalmologist presented with a well-defined retinal lesion and a consistent field defect on a screening field should consider if any extra information would be provided by performing a threshold field.
The field defects associated with non-glaucomatous pathology tended to be paracentral or temporal. This is again in agreement with previous studies.4
Local optometry practices achieved our audit standards. We therefore hesitate to circulate guidelines based on our findings.
We can learn from Vernon et al who distributed a protocol for optometric glaucoma referral and managed to reduce the true positive referral rate from 40 per cent to 32.3 per cent.5 To close the audit loop we have therefore merely circulated our results and will be reauditing later this year. We hope that our audit provides some evidence basis for everyday practice.
References
1 Newman DK, Anwar S, Jordan K. Glaucoma screening by optometrists: positive predictive value of visual field testing. Eye, 1998; 12: 921-24.
2 Gutteridge IF. Clinical significance of detecting visual field loss. Am J Optom Physiol Opt, 1985; 62: 275-81.
3 Bass SJ, Feldman J. Visual field defects in well-defined retinal lesions using Humphrey and Dicon perimeters. Optometry, 2000; 71: 643-52.
4 Anton A, Maquet JA, Mayo A, Tapia J, Pastor JC. Value of logistic discriminant analysis for interpreting initial visual field defects. Ophthalmology, 1997; 104: 525-31.
5 Vernon SA, Ghosh G. Do locally agreed guidelines for optometrists concerning the referral of glaucoma suspects influence referral practice? Eye, 2001; 15: 458-63.
6 Rumney NJ, Henson DB. Glaucoma screening by optometrists: positive predictive value of visual field testing (letter). Eye, 1999; 13: 605-6.
Steve Madill is a specialist registrar at Queen Alexandra Hospital, Portsmouth, Steve Lash is an SHO at Southampton Eye Unit and Anne Denning is consultant ophthamologist at Bournemouth Eye UnitThe aim of a programme screening for asymptomatic field defects must be two-fold: to detect true positives (the association between asymptomatic field defects and genuine pathology) and to reduce false positives (such as with a field screening protocol for glaucoma). When screening fields we also need to be aware of the differences between glaucomatous and non-glaucomatous field defects, since this will influence referral practice.
Audit Protocol
Between December 2000 and April 2001 we audited the results of consultations in the Bournemouth Eye Unit based on referrals for asymptomatic field defects via GOS18 forms.
Twenty-three referrals were received during this period for asymptomatic field defects, of which 20 sets of notes were located.
The patients' age range was from 27-90. We compared the screening field performed by the optometrist with the Humphrey threshold field performed in the eye department.
With regard to reducing false positives, we concentrated on those patients referred as glaucoma suspects. We took as our gold standard the Bury St Edmunds' audit1 that demonstrated 60 per cent of patients referred as glaucoma suspects received a final positive diagnosis of glaucoma.
For detecting true positives, we concentrated on previous studies reviewing the association between asymptomatic field defects and genuine pathology. We selected a paper that reported 52 per cent of patients referred to one department with asymptomatic field defects turned out to have identifiable pathology (both glaucomatous and non-glaucomatous).2 This was, therefore, our second gold standard.
Method
We used a standardised system for grading the field defects (Table 1). Our objectives were to:
Compare the consistency of field defects between screening fields and threshold fields and possibly associate consistency with diagnosis. Each patient had both monocular field defects for screening and threshold fields compared. Only one eye had to demonstrate a defect classified the same on screening and threshold fields for that patient to be labelled as consistent.
With glaucoma referrals, review the corroborative evidence provided with an asymptomatic field defect to support the referral and the strength of association between types of evidence and a final positive glaucoma diagnosis.
With regard to non-glaucoma referrals, the characteristics of defects more likely to be associated with this group.
Review if the referral patterns in our region achieved our audit standards.
Results
Glaucoma referrals
Eleven of 20 patients were referred as glaucoma suspects with asymptomatic field defects.
One was referred with field loss only, two with fields and raised intraocular pressure, two with fields and suspicious discs, six with fields and a family history of glaucoma, and one with field changes and a previous diagnosis of glaucoma but since discharged to the optometrist glaucoma follow-up scheme.
Of these, 63 per cent (seven patients) had a positive diagnosis of glaucoma or glaucoma progression made. Two patients had non-glaucomatous pathology found which explained the field defects (one dense cataract and one tilted disc) and two had no pathology found.
On threshold field-testing, positive glaucoma diagnosis patients demonstrated either arcuate, central or ungradable defects. Three patients' defects were consistent with their screening fields and four were inconsistent.
Non-glaucoma referrals
Nine of 20 patients were referred with asymptomatic defects of unknown aetiology. Thirty-three per cent (three patients) had an identifiable pathology; two had chorioretinal scars and a previous optic neuritis. Six had no pathology found.
If we consider all the patients with identifiable non-glaucomatous pathology, both from the glaucoma suspect and non-glaucoma referral groups (five in total), we find that all their field defects were consistent between screening and threshold fields. On threshold testing they demonstrated either temporal wedges or paracentral scotomas.
Conclusions
Glaucoma referrals
We found a 63 per cent true positive referral rate for glaucoma patients. We have therefore achieved our audit standard. The problem of improving the true positive rate is derived from the fact that glaucoma is a condition of relatively low prevalence. A test with 99 per cent specificity and 99 per cent sensitivity would still generate a 49.5 per cent false positive rate (see box below for explanation). Our audit suggests that optometrists in the Bournemouth and Poole region are therefore theoretically doing better than this.
The patient referred with field loss alone had a positive glaucoma diagnosis. In fact Newman et al1 showed that referral with isolated field loss was most weakly associated with a positive diagnosis of glaucoma. Isolated change in cup-to-disc ratios was more strongly associated and isolated raised intraocular pressure was most strongly associated. When considering more than one finding in combination, the results become more complicated. Newman's table is shown in Table 2.
The most often cited corroborative factor on GOS18 forms to support referral as a glaucoma suspect with asymptomatic field loss was a positive family history of glaucoma. Strictly speaking a family history should principally be used to identify the group that need to be screened rather than to support a referral. Interestingly, the patients classified with either non-glaucomatous pathology or no identifiable pathology in the glaucoma suspects referral group all had a family history of glaucoma as the corroborative factor supporting glaucoma referral.
Four eyes of seven patients with positive glaucoma diagnoses demonstrated 'ungradable defects'. In fact generalised field depression has been shown to be highly sensitive in its association with early glaucoma. We included generalised depression as ungradable since it is screened out by threshold field algorithms. It is also of no use in screening since predictably generalised depression is not specific for early glaucoma.
Non-glaucoma referrals
Our positive diagnosis rate for all significant pathology (both glaucomatous and non-glaucomatous) was 60 per cent. Thus, we have achieved our audit standard.
Considering all patients, only 50 per cent had consistent defects when comparing screening to threshold fields. It is pertinent therefore that all the patients who were found to have significant non-glaucomatous pathology had highly consistent defects. This is not an age influence, since the average age of these patients was 62, whereas the average age for those with a positive glaucoma diagnosis was 66. These findings do agree with previous studies demonstrating marked consistency between screening and threshold field defects for well defined retinal lesions.3
The ophthalmologist presented with a well-defined retinal lesion and a consistent field defect on a screening field should consider if any extra information would be provided by performing a threshold field.
The field defects associated with non-glaucomatous pathology tended to be paracentral or temporal. This is again in agreement with previous studies.4
Local optometry practices achieved our audit standards. We therefore hesitate to circulate guidelines based on our findings.
We can learn from Vernon et al who distributed a protocol for optometric glaucoma referral and managed to reduce the true positive referral rate from 40 per cent to 32.3 per cent.5 To close the audit loop we have therefore merely circulated our results and will be reauditing later this year. We hope that our audit provides some evidence basis for everyday practice.
References
1 Newman DK, Anwar S, Jordan K. Glaucoma screening by optometrists: positive predictive value of visual field testing. Eye, 1998; 12: 921-24.
2 Gutteridge IF. Clinical significance of detecting visual field loss. Am J Optom Physiol Opt, 1985; 62: 275-81.
3 Bass SJ, Feldman J. Visual field defects in well-defined retinal lesions using Humphrey and Dicon perimeters. Optometry, 2000; 71: 643-52.
4 Anton A, Maquet JA, Mayo A, Tapia J, Pastor JC. Value of logistic discriminant analysis for interpreting initial visual field defects. Ophthalmology, 1997; 104: 525-31.
5 Vernon SA, Ghosh G. Do locally agreed guidelines for optometrists concerning the referral of glaucoma suspects influence referral practice? Eye, 2001; 15: 458-63.
6 Rumney NJ, Henson DB. Glaucoma screening by optometrists: positive predictive value of visual field testing (letter). Eye, 1999; 13: 605-6.
Steve Madill is a specialist registrar at Queen Alexandra Hospital, Portsmouth, Steve Lash is an SHO at Southampton Eye Unit and Anne Denning is consultant ophthamologist at Bournemouth Eye Unit
Register now to continue reading
Thank you for visiting Optician Online. Register now to access up to 10 news and opinion articles a month.
Register
Already have an account? Sign in here