Glaucoma case finding: a cluster-randomised intervention trial

Abstract

To assess the effect of an intervention comprising training in optic disc assessment, explicit referral criteria and ophthalmologist feedback on referred patients, on the number of optometrist referrals for suspected glaucoma seen at a referral site and the positive predictive value of those referrals. Optometric practices routinely referring to the Ealing Hospital Eye Clinic were randomly divided into two groups taking into consideration those practices, which shared an optometrist (a cluster) and the number of optometrist days worked per week. One group of practices acted as controls, while the other practices were invited to receive the intervention. Data on 397 new patients referred and presenting to Ealing Hospital with suspected glaucoma were collected over a 20-month period. The data on patients who had failed to attend their appointment were collected over 7 months of this period. The number of referrals seen, the positive predictive value of those referrals, and the attendance rate were calculated. Optometrist's opinions of the intervention were assessed qualitatively. Data relating to optometrist compliance with the intervention were also collected. The number of glaucoma referrals presenting to Ealing Hospital from the intervention practices was almost double that from the control practices (210 vs 119). When cluster randomisation, the number of optometrist days per cluster and the number of assessed referrals in the preintervention period are taken into consideration, it is estimated that the intervention is associated with a 52% increase in the number of referrals reaching Ealing Hospital. However, the design effect resulting from the cluster randomisation was unexpectedly high (of the order of 13-14)and so the confidence intervals around the estimate of 52% are very wide (95% c.i. 35% decrease to 253% increase, P = 0.34). There was no evidence of an association between optometrist compliance with the intervention and the number of referrals seen at Ealing Hospital. The positive predictive value (PPV)of referrals was similar for the intervention(0.49 (95% c.i. 0.42, 0.55)) and control groups(0.46 (95% c.i. 0.33, 0.60)). Optometrist opinions of the intervention were largely favourable. All expressed a willingness to participate in future programmes. A large difference in the number of referrals between the practice groups was observed. Since the PPV of referral was unchanged, the potential impact of the intervention in terms of numbers of new cases of glaucoma detected in the community is substantial. However, because of its large design effect, this trial does not provide conclusive evidence of an impact of the intervention on referral numbers. A considerably larger trial will be required to produce conclusive evidence of an effect.