Validity of the Monocular Trial of Intraocular Pressure–Lowering at Different Time Points in Patients Starting Topical Glaucoma Medication

Importance  Establishing the true therapeutic effect of eyedrops when initiating glaucoma therapy is important. Accurate prediction of the intraocular pressure (IOP)–lowering response in the fellow eye when using a monocular trial eliminates the need for additional office visits to confirm the therapeutic effect.

Objective  To investigate the validity of the monocular trial in patients commencing topical glaucoma treatment at different time points.

Design, Setting, and Participants  Prospective cohort study of untreated patients with open-angle glaucoma or ocular hypertension at a hospital-based glaucoma service among treatment-naive individuals. The study dates were October 1, 2008, to November 30, 2009.

Interventions  Participants had 8 visits. After the recruitment visit, IOP was measured in both eyes by masked applanation tonometry at 8 am, 11 am, and 4 pm for 7 consecutive weeks. Treatment with travoprost, 0.004%, was commenced at week 3 in the trial eye and at week 4 in the fellow eye.

Main Outcomes and Measures  Three IOP outcomes were measured for the trial eye, including unadjusted IOP-lowering effect, adjusted IOP-lowering effect, and true therapeutic effect.

Results  Of 30 topical glaucoma treatment–naive individuals (11 male and 19 female), 16 had ocular hypertension and 14 had primary open-angle glaucoma. Their mean (SD) age was 64.4 (12.6) years (age range, 42-88 years). The unadjusted IOP-lowering effect overestimated the true therapeutic effect by mean (SD) 2.5 (4.8), 3.1 (3.8), and 4.9 (4.4) mm Hg at 8 am, 11 am, and 4 pm, respectively, and the mean (SD) adjusted IOP-lowering effect was almost identical to the true therapeutic effect at each of the 3 time points (0.43 [3.87], 0.02 [2.82], and −0.40 [3.90]), respectively. The correlation between the unadjusted effect of treatment and the true therapeutic effect was 0.55 (95% CI, 0.23-0.76), and the effect when adjusted by the monocular trial was 0.72 (95% CI, 0.49-0.86). Fellow eye responses to treatment were correlated at all time points (r range, 0.78-0.86). Treatment did not demonstrate any effect on the diurnal pattern of IOP.

Conclusions and Relevance  The monocular trial of therapy is effective in accurately predicting the response of an untreated eye to monotherapy with a prostaglandin analogue at all daytime time points measured. There is no requirement for patients to be seen at the same time of day after treatment has commenced. The effect in the first eye predicts both the likelihood and magnitude of an effect in the second eye at all time points during office hours and negates the requirement for an additional visit to check the therapeutic effect when commencing therapy in the second eye.

Quiz Ref IDThe monocular trial of topical therapy when initiating glaucoma eyedrops was first suggested by Drance and colleagues^1,2 as a way of measuring the therapeutic effectiveness of a medication by using the untreated fellow eye as a control. This approach has remained in glaucoma guidelines,^3,4 but the validity of the trial has been challenged in recent years.^5-9

In a previous publication,¹⁰ our group demonstrated the validity of the monocular trial for measurements at a single time point (11 am). However, one of the concerns of the monocular trial has been the question of diurnal variation,⁸ and no investigation, to our knowledge, has explored the validity of the monocular trial at different time points. In clinical practice, it is often not practicable to ensure that all intraocular pressure (IOP) measurements before and after starting treatment can be obtained at the same time of day, so it is important to know if the validity of the monocular trial remains when the pretreatment and posttreatment times are different. In this study, we investigated the validity of the monocular trial at the 8 am, 11 am, and 4 pm time points and determined its effectiveness when the pretreatment and posttreatment times are different.

Box Section Ref ID

The study was approved by the Nottingham Ethics Committee, Nottingham, England. All patients completed written informed consent before participation in the study. The study dates were October 1, 2008, to November 30, 2009. The methods are described in detail elsewhere.¹⁰ In summary, patients with primary open-angle glaucoma or ocular hypertension found to have an IOP exceeding 21.0 mm Hg in both eyes at their initial clinic visit were invited to participate. All patients were naive to topical glaucoma treatment.

Inclusion criteria were open angle on gonioscopy, IOP exceeding 21.0 mm Hg in both eyes at the diagnostic visit, and no previous treatment with topical antiglaucoma medication. Exclusion criteria were previous intraocular surgery, unwillingness to participate, and inability to provide written informed consent.

Goldmann applanation tonometry was used to measure IOP. After the recruitment visit, 3 masked measurements were obtained on each occasion, and the median of these 3 was taken as representative.

After the recruitment visit (V0), each participant had further IOP measurements obtained for 7 consecutive weeks. The IOP was measured at 8 am, 11 am, and 4 pm. The visit regimen is summarized in Table 1. The first 3 postrecruitment visits were baseline assessments before the administration of travoprost, 0.004% (visits 1-3). Medication was started in the eye with the higher recruitment IOP at the third visit. The patient was examined 1 week later, and travoprost was then commenced in the second eye (visit 4). The patients attended 3 more visits during treatment in both eyes (visits 5-7). Therefore, the patients were recruited on an intent-to-treat basis for both eyes. A standardized evening regimen (8 pm) for eyedrop instillation was adopted.

Several different therapeutic IOP outcomes were defined, including unadjusted IOP-lowering effect, adjusted IOP-lowering effect, and true therapeutic effect. Unadjusted IOP-lowering effect is the difference in IOP between the visit at which the decision to treat was made (V0) and the first posttreatment visit (V4) (ie, [V0 − V4]^{first eye}), which represents the effect that would be seen in normal clinical practice. Adjusted IOP-lowering effect is the difference in IOP of the treated eye between V0 and V4 minus the change in IOP of the untreated fellow eye between the same 2 visits, which represents the monocular drug trial (ie, [V0 − V4]^{first eye} − [V0 − V4]^{second eye}). True therapeutic effect is the difference in IOP between the 3 baseline visits and 3 posttreatment visits) (ie, (mean [V1, V2, and V3]^{first eye} − mean [V5, V6, and V7]^{first eye}), which corrects for regression to the mean by ignoring the recruitment IOP and represents the real therapeutic effect of treatment controlled for day-to-day variation.^11,12

A paired t test was used to test the significance of differences in IOP outcomes. Correlations for IOP reduction in each eye were made using the Pearson linear correlation coefficient. Comparison of variance was performed using the F test. Analysis was performed using statistical software (SPSS, version 15.0; SPSS Inc).

In total, 32 patients agreed to participate in the study. Two patients failed to complete all visits. Of the 30 patients who completed the study, 11 were male and 19 were female. The mean (SD) age for the group was 64.4 (12.6) years (age range, 42-88 years). There were 16 with ocular hypertension and 14 with primary open-angle glaucoma. Two of the patients were of Afro-Caribbean origin, and the others were of white race/ethnicity. The mean (SD) recruitment IOPs (V0) were 28.2 (3.7) and 26.0 (2.1) mm Hg in the trial and fellow eyes, respectively.

The mean (SD) central corneal thickness measurements of the treated and fellow eye groups were 565 (42) and 568 (38) µm, respectively (P = .33). The mean (SD) spherical equivalents of the treated and fellow eye groups, respectively, were 0.9 (2.3) and 0.8 (2.7) diopters (P = .49). In total, 7 patients had systemic hypertension, and 2 were receiving treatment for type 2 diabetes mellitus.

Table 2 lists the adjusted and unadjusted IOPs at the 3 different time points. A decrease in IOP as a result of regression to the mean was observed between recruitment and baseline measurements in both eyes at all time points, ranging from 2.1 to 4.0 mm Hg for trial eyes and 2.8 to 4.6 mm Hg for fellow eyes. The mean baseline IOPs were 26.1 and 23.2, 25.8 and 22.7, and 24.2 and 21.4 mm Hg for the 8 am, 11 am, and 4 pm time points in the trial and fellow eyes, respectively.

The mean true effect of treatment varied according to the time of day between 7.7 and 8.6 mm Hg in the trial eye and between 6.0 and 6.7 mm Hg in the fellow eye. These values are considered to be the real short-term therapeutic effects of the medication in this group. They represent reductions from baseline IOP of 31.0% to 33.3% for first eyes and 27.5% to 29.5% for second eyes. Treatment effects between fellow eyes were strongly correlated (Table 2).

Quiz Ref IDFor the 11 am time point, the mean unadjusted effect of treatment in the first eye was 11.7 mm Hg, whereas the mean adjusted effect of treatment was 8.6 mm Hg, almost identical to the true mean effect of treatment. Therefore, the normal clinical practice of starting treatment at V0 without adjustment with a monocular trial overestimated the effectiveness of treatment by a mean of 3.1 mm Hg (36.0%). Adjusting for the results using a monocular trial reduced this error to a mean of 0.02 mm Hg (0.2%) (P < .001).

Similar results were seen at the other time points. For the 8 am time point, the unadjusted treatment effect (10.3 mm Hg) overestimated the true effect (8.1 mm Hg) by a mean of 1.9 mm Hg, whereas adjusting with the monocular trial reduced this difference to 0.43 mm Hg. For the 4 pm time point, the unadjusted treatment effect (12.2 mm Hg) overestimated the true effect (7.7 mm Hg) by a mean of 4.5 mm Hg, although adjustment reduced this difference to −0.40 mm Hg.

The mean absolute value (modulus) of the error was 3.8 mm Hg (44.2% of the true effect) for the unadjusted effect and 2.1 mm Hg (24.4%) for the adjusted effect (P < .02) at the 11 am time point. By comparison, these values were 4.3 mm Hg (53.0%) vs 3.0 mm Hg (37.0%) at 8 am (P = .09) and 5.4 mm Hg (70.1%) vs 3.2 mm Hg (41.6%) at 4 pm (P = .007).

The correlation between the unadjusted effect of treatment and the true therapeutic effect was 0.55 (95% CI, 0.23-0.76), whereas the effect when adjusted by the monocular trial was 0.72 (95% CI, 0.49-0.86). The results were not materially different at the 8 am and 4 pm time points.

The Figure shows cumulative plots for the adjusted and unadjusted estimates of the therapeutic effect, demonstrating the percentage of values within a given range of the true effectiveness of treatment for 8 am, 11 am, and 4 pm. At each time point, the area under the curve was greater for the adjusted effect, confirming the benefit of the monocular trial. The proportion of individuals in whom the measured treatment effect was within 2 SDs of the true effect was 96.6% using adjustment with monocular treatment effect compared with 75.9% for the standard unadjusted approach at the 11 am time point (P = .004). The equivalent values were 89.7% vs 68.9% (P = .11) for 8 am and 96.4% vs 60.7% (P = .02) for 4 pm (McNemar χ² test for paired data).

Whenever a patient is given a new treatment to reduce IOP, an assumption is made that the measured reduction in pressure is a reasonably accurate reflection of the true therapeutic effect. However, this assumption is not always the case. Previous work on this group of patients showed that the repeatability of measurements of the therapeutic effect was in fact poor.¹² At the 3 time points, the 95% CIs ranged between ±5.4 and ±5.9 mm Hg (±67.2% to ±88.0%) of the true mean effect, and a measured reduction of less than 7.0 mm Hg could not be distinguished from a chance finding, which is clearly of significant clinical relevance.

One solution to this inherent imprecision is to increase the number of IOP measurements before and after starting treatment. However, because the SE only declines in proportion to the square root of the number of measurements, the improvement is modest for each additional visit and is subject to the law of diminishing returns. While a single paired visit only yields a precision of ±5.0 mm Hg, as many as 8 baseline and 8 posttreatment visits would be required to improve this precision to ±2.0 mm Hg,¹² which is clearly not a desirable solution in a busy ophthalmology practice.

An alternative is the monocular (or uniocular) trial, initially proposed by Drance¹ and cited in the American Academy of Ophthalmology³ and European Glaucoma Society⁴ guidelines. Authors have recently raised concerns about the validity of this approach.^5,6,9 Bhorade⁸ highlighted that few studies had addressed the classic monocular trial that adjusts IOP in the trial eye based on IOP changes in the fellow eye and recommended a prospective study to evaluate its validity. The present study was designed to address this issue.

This patient cohort demonstrates that using a monocular trial significantly improves the precision of the estimation of the treatment effect without any additional clinic visits. Both the modulus of the mean error and the cumulative size of the error for each individual favored the monocular trial over the standard approach at all 3 time points.

Furthermore, because all the patients were recruited in the morning and yet the monocular trial is shown to be an effective tool for estimating the therapeutic effect at 4 pm, it can be used even when the pretreatment and posttreatment measurement times do not correspond. In the standard approach, all pretreatment and posttreatment measurements must be obtained at the same time of day to allow for diurnal variation.

This article is not to imply that the monocular trial gives a perfect estimate of the effect of treatment. However, it will provide a more precise measure than the standard approach as long as the correlation between the IOP in fellow eyes is greater than the day-to-day variability within the treated eye. Diurnal variation is well reported, but the degree of day-to-day variation in IOP is not. In this patient group, we found that even when measured at the same time of day this precision (95% CI) ranged between ±3.7 and ±4.3 mm Hg (±15.3% to ±18.2%).¹¹ It is this within-eye variability that underlies the demonstrated effectiveness of the monocular trial.

Our findings show a general reduction in pretreatment IOP throughout the day, as would be expected. This result corresponds to a reduction of approximately 2.0 mm Hg between the 8 am and 4 pm time points for the trial (higher IOP) eye, with a consistent difference of approximately 3.0 mm Hg between fellow eyes at all times (r range, 0.78-0.86). This finding demonstrates the pattern of diurnal variation that would be expected and confirms consistent IOP differences between the trial eye and fellow eye throughout the day. These results are consistent with the findings of other authors for IOP variation in individuals with ocular hypertension and glaucoma.^13,14

All the recruitment IOP values were morning measurements (9 am to 1 pm). It can be seen that for all 3 time points there is a reduction in IOP between the recruitment IOP (V0) and the baseline IOP (V1-V3). For the 11 am time point, this reduction is attributable to regression to the mean, which is only partly the case for the 8 am and 4 pm time points. To obtain true regression to the mean for these points, the recruitment IOP would have to have been obtained at the corresponding times. It can be seen that there is a mean decrease in IOP of 4.0 mm Hg between recruitment and the 4 pm baseline before therapy is even commenced, which could in some cases be mistaken for a therapeutic effect in itself. Unless a monocular trial is performed, the pretreatment and posttreatment measurements must be obtained at the same time of day to avoid this mistake. With a monocular trial, the assessment of the therapeutic effect of treatment is shown to be equally accurate even when the pretreatment and posttreatment measurements do not correspond.

It has been suggested previously that the response of second eyes to therapy was unpredictable, thus negating the value of the monocular trial,^5,15 although our results suggest that this conclusion is not the case. The mean reduction in IOP varies between 31% and 33% for treatment of the first eye and between 27.5% and 29.5% for treatment of the second eye, with correlations for these proportional reductions between 0.78 and 0.86. So it can be seen that if travoprost, 0.004%, has a therapeutic effect on the first eye, then it is almost certain that it will have a therapeutic effect on the second eye, and the magnitude of this response will be similar in both eyes. This finding argues against the requirement for additional clinic visits to assess the response of treatment in the second eye.

There are 2 main sources of variability that could potentially lead to misinterpretation of the IOP-reducing effect of a glaucoma eyedrop, including inherent day-to-day variability (of which regression to the mean is a specific example) and diurnal variation in untreated IOP throughout the day. However, by using the monocular trial, this variability can be corrected for to a significant degree to allow a more accurate assessment of the therapeutic effect.

Quiz Ref IDFrom a practical point of view, the use of the monocular trial provides physicians with confidence to assume that if an effect is observed in the first eye then it will occur in the second eye. The true treatment effect in the second eye was 3.8% less than in the first eye (mean, 29.5% vs 33.3%), and this difference had an SD of 6.0%. When treatment is started in the second eye, it can be predicted that the effect will be between 17.6% and 41.3% reduction in IOP with 95% certainty (29% ± 2 SDs). As we have shown previously in these patients using analysis of within-individual variation, such a degree of precision would require 3 pretreatment and 3 posttreatment IOP measurements.¹²

For glaucoma eyedrops, the therapeutic effect (if it occurs) will be achieved within a few days, and in our practice it is usual to assess the therapeutic effect approximately 4 weeks after treatment with eyedrops is commenced. If at this stage (assuming adherence) there is no therapeutic effect, we would consider the eyedrop to be ineffective and consider an alternative agent.

The limitations of this study relate to the generalizability of the findings. We have only assessed 3 time points during office hours. However, we believe that this practice provides a representative sample of IOP measurement, and it is reasonable to extrapolate that the time points between these measurements will behave in a similar way, considering the consistency we have demonstrated between our measurements. However, we do not have any nighttime measurements; therefore, it is not possible for us to comment on the validity of the monocular trial for assessing the therapeutic effect in the supine sleeping posture. We have only tested our study participants with a single prostaglandin analogue, so we cannot speculate whether these findings would necessarily be replicated with other topical glaucoma medications. Clearly, it would not be useful for treatment with β-blockers owing to the well-known crossover effect between eyes. Similarly, we have only tested treatment-naive individuals, so we do not know whether these observations will apply to those already using topical medication.

Quiz Ref IDDespite these caveats, it is reasonable to conclude that the monocular trial of therapy is more effective than using single pretreatment and posttreatment measurements in estimating the response of an untreated eye to monotherapy with a prostaglandin analogue during office hours. There is no requirement for patients to be seen at the same time of day after treatment has been commenced. The effect in the first eye predicts both the likelihood and magnitude of an effect in the second eye at all time points during office hours and negates the requirement for an additional visit to check the therapeutic effect when commencing therapy in the second eye. We endorse the adoption of this technique in busy ophthalmology practices, where multiple pretreatment and posttreatment measurements are not practicable.

Submitted for Publication: October 31, 2015; final revision received January 11, 2016; accepted March 15, 2016.

Corresponding Author: Anthony J. King, MD, FRCOphth, MMedSci, Department of Ophthalmology, Nottingham University Hospital, Derby Rd, Nottingham NG7 2UH, England (anthony.king@nottingham.ac.uk).

Published Online: May 5, 2016. doi:10.1001/jamaophthalmol.2016.0994.

Author Contributions: Dr King had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Both authors.

Acquisition, analysis, or interpretation of data: King.

Statistical analysis: Rotchford.

Administrative, technical, or material support: Both authors.

Conflict of Interest Disclosures: Both authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr King reported being an advisory board member for Alcon and Santen for which he received payment. No other disclosures were reported.

Funding/Support: This study was supported by an unrestricted grant from Alcon UK (Dr King).

Role of the Funder/Sponsor: The funding organization had no input into the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.