Study reveals current asssessment methods can be imprecise
"It is not enough to assume that comparing a single pair of intraocular pressure (IOP) measurements before and after starting treatment gives an accurate assessment of effectiveness," emphasized Dr Alan P. Rotchford (Tennent Institute of Ophthalmology, Gartnavel Hospital, Great Western Road, Glasgow, UK) when discussing the results of a recent study on the day-to-day variability in IOP in glaucoma and ocular hypertension (OHT),1 and on the repeatability of measurements of effectiveness of glaucoma medication.2
Assessing individuals
To solve this issue, Dr Rotchford explained that it is possible to average the effect from multiple pre and post-treatment visits, which would give a true assesment of the therapeutic effect of the medication. However, additional visits to a clinic are not only an inconvenience for patients but are also more costly.
"So, in this study we wanted to determine for the first time the degree to which estimates of the effect of glaucoma medication are repeatable," he added. "We also wanted to know whether taking multiple IOP measurements before and after starting treatment does really give a significantly more accurate estimate of therapeutic effect and, if so, how many measurements are necessary."
Examining IOP measurement variability
To examine the day-to-day variability of IOP measurements, Dr Rotchford and colleagues enrolled patients presenting with primary open-angle glaucoma (POAG) or OHT with an IOP of greater than 21 mmHg who needed bilateral antihypertensive medication. "Following recruitment each participant had seven further visits at weekly intervals and at each of these visits IOP was measured at 8 am, 11 am and 4 pm," Dr Rotchford said.
During the first three visits each patient was completely untreated for IOP and then after the third visit, travoprost was introduced monocularly into the eye with the highest IOP value. After the fourth visit, medication was introduced into the patient's fellow eye and they were treated binocularly for the rest of the study.
"We analysed the three weekly visits before treatment and the three after treatment to measure the repeatability of the effect of travoprost," continued Dr Rotchford. "The measurements were taken in a masked fashion and the median of the three readings was taken as representative each time IOP was measured. The IOP at the recruitment visit was discarded for the analysis to avoid the problem of regression to the mean inherent in studies where recruitment is based on finding an abnormal pressure."
Large range of effect
"Travoprost is on average a highly effective first line treatment but with a very large range of effect in individual eyes," revealed Dr Rotchford. "The measured effect is poorly repeatable and so assessing the therapeutic effect from a single visit is very inaccurate." The precision for a single visit was only ±5 mmHg or ±70% of the mean effect he noted.
"This clearly represents a significant level of uncertainty which would likely result in incorrect clinical decision-making, that is to assume incorrectly that the medication is working or, conversely, to switch medications unnecessarily," added Dr Rotchford. This poor repeatability was a surprise finding for the team.
"Using a measure of the variability of the therapeutic effect, we modelled the benefit of additional measurements on the level of precision," he continued. "As expected, we found that increasing the number of measurements of therapeutic effect before deciding whether or not the medication is working effectively improves precision. However, the improvement is relatively modest for each additional visit and is subject to the 'law' of diminishing returns. While a single paired visit only yields a precision of ±5 mmHg, as many as 8 baseline and 8 post-treatment visits would be needed to improve this to ±2 mmHg."
Multiple measurements required
Assuming that the comparison of a single pair of IOP measurements, those before treatment and those after, gives an accurate assessment of effectiveness of treatment is not valid, according to Dr Rotchford. "Multiple pre- and post-treatment measurements are required with precision increasing with the square root of the number of visits. Although, this has obvious cost implications," he said.
As an alternative option, Dr Rotchford noted that it is possible to use a monocular treatment trial in which treatment is begun in one eye at the first visit and the change in IOP in the fellow eye used to adjust the effect seen at the follow-up visit. "We have shown that this gives a substantially more precise estimate of effect in this group of patients in a previous paper,"3 he added. "A monocular treatment trial works so long as the variability of IOP between fellow eyes is less than the day-to-day variability in one eye. It can never give as precise an estimate as using a large number of measurements but for medications with no cross-over effect it is more precise than using a single pair of IOP measurements with no additional visits needed."
Special contributor
Dr Alan P. Rotchford is a consultant ophthalmologist based at the Tennent Institute of Ophthalmology, Glasgow, Scotland. He specialises in glaucoma and may be reached by Email: rotchford@doctors.org.uk
Dr Rotchford has indicated no relevant financial disclosures
References
1. A.P. Rotchford et al., Br. J. Ophthalmol., 2012;96:967–970.
2. A.P. Rotchford et al., Br. J. Ophthalmol., 2012, Sep 18 (Epub ahead of print).
3. A.J. King et al., Ophthalmology, 2011;118(11):2190–2195.