Editor,
In their article Foster et al have addressed an issue of relevance to developing countries that have a high prevalence of angle closure. The likelihood ratio of a positive test using results that we have reported (sensitivity 61.9% and specificity 89.3%) is approximately 6 which is similar to the LR for the authors' 15% cutoff. Invocation of pupillary block to explain the discrepancy seems reasonable. Twenty-seven percent of our cases do not open following a laser PI2. Technically these patients had plateau iris that could be missed by the limbal chamber depth (LCD). If we were to exclude such cases our LCD sensitivity would be more similar to the authors'.

While our results are similar, we would apply them differently. The authors consider LCD to have two major applications: opportunistic screening in ill-equipped clinics and population based screening. In both situations they emphasize the need for a high sensitivity, accepting a "certain number" of false positives.

In a developing country, screening in an ill-equipped clinic is quite similar to population based screening. Persons who are test positive will have to travel long distances to seek expert help at centers that are "better." While the goal of screening is identification of pathology (or eyes at risk), false positives can be a major logistical problem. Six percent of Mongolian eyes were occludable on gonioscopy but only 5-10% of these might actually go on to closure. The positive predictive value of LCD for the detection of occludable angles is reasonably high, but referral based on LCD alone (which includes false positives) would account for 20% of the population. As denial of treatment, in this case to the majority and that too after travelling long distances, is undesirable, one option might be to treat all referred occludable angles. In that case, based on LCD alone, one in five Mongolians would need to travel to obtain a laser iridotomy that is not necessarily innocuous and one that is probably not needed in the first place. The same logic applies to population based screening but involves even more false positives.

In both situations we would argue for a test with a high specificity and settle for reasonable sensitivity. The sensitivity and specificity of LCD, especially for established ACG is indeed high, but is the specificity high enough? As table 4 shows the positive predictive value is actually lower for angle closure and angle closure glaucoma as compared to occludable angles. This of course is a function of prevalence, and further emphasizes the need for a high specificity. More so in the case of occludable angles than established ACG. On the other hand the negative predictive value of LCD is good for most grades (especially grade I). Hence, at least in a busy clinic situation, a negative LCD can be used to rule out occludable angles and established angle closure.

A positive LCD on its own doesn't mean very much. However, LCD (LCD greater than or equal to 5%) can be used in conjunction with raised IOP. This would increase the specificity to about 99% (Specificity for both tests positive = [1-(1-specificity of the first test)(1- specificity of the second test)]. The sensitivity would decrease as the product of the two test sensitivities but will mainly affect detection of occludable angles, not established angle closure glaucoma.

If an appropriate test is not available, perhaps we should not screen at all. Certainly with the long-term goal of improving the quality of ophthalmology, teaching gonioscopy to ophthalmologists in developing countries might be a better option. While gonioscopy itself cannot predict which eyes will go into closure, it is the current gold standard; ophthalmologists in developing countries could be taught this technique with as much ease (or difficulty) as the registrars who used it for the study.

Editor,
We congratulate Hugh Taylor on his editorial in which he discusses the volume of cataract surgery which needs to be done. We were particularly interested by his calculations drawing attention to the effect of changing the visual threshold at which the decision to undertake cataract surgery is made.

We have reviewed a series of eight audits of cataract surgery to determine the visual acuities at which patients were put on the waiting list for cataract extraction. The audits were designed to show surgical outcomes but they also list the visions at the point when the decision to operate was made. They include all patients during short periods between 1982 when intracapsular extraction without lens insertion was the norm and the first six months of 2000 when practically all cases were phacoemulsification through a clear corneal incision with a foldable lens. They are all based on the throughput of a single firm of a teaching hospital. The table shows visual acuity at listing for cataract extraction.

These results are heartening in that they show that the pool of dense cataracts resulting in visions of 6/60 or worse has decreased (from 77% in 1982 to 19% in 2000). They confirm that as surgical techniques have improved the demand for surgery at an early stage has dramatically increased. In the 1982 audit only 2.6% of cases saw 6/12 or better whereas by 2000 this figure had risen to 47.7%.

We accept that visual acuity is far from being a comprehensive measure of visual disability but none the less it is useful as an indicator of trends. The trend is clearly towards earlier cataract surgery and it is likely to be maintained resulting in increasing surgical volumes. The answer to Hugh Taylor's question, "How much surgery do we have to do?" is more and still more. The problem then becomes partly political in that governments decide on maximum waiting times for surgery with the effect that it becomes difficult to prioritise those patients with a serious degree of cataract over those who are simply suffering inconvenience.

Editor,
We read Frau et al's report with interest and noted that our article in Ophthalmology was not cited as reference (Larkin G, Flaxel C, Leaver P. Phacoemulsification and silicone oil removal through a single corneal incision. Ophthalmology 1998;105:2023-7). In this article, we reported our experience at Moorfields Eye Hospital with 34 eyes prospectively evaluated to look at the efficacy and potential complications of combined cataract extraction and silicone oil removal with posterior chamber lens implantation. We also reported the method of Ms Maria Restori, BSc, MSc, MIPSM, Ophthalmic Ultrasound Specialist at Moorfields Eye Hospital for calculating the IOL power in an oil filled eye with correction for the specific gravity of silicone oil taken into consideration.

Our findings were that the procedure was safe and effective for these eyes that had often had many previous surgeries. The visual outcome in these eyes was generally good with improvement in visual acuity, even with recurrent retinal detachment or pre-existing macular pathology. We also concluded that it was safer to place a rigid posterior chamber implant after silicone oil removal due to potential contraction of the anterior capsule limiting the view of the retina post-operatively. Our technique was a passive technique but might easily be done with the I/A handpiece as this group reported.

We feel that it would have been appropriate for them to make reference to our study since it presents a much larger series with more detailed follow-up.