Article Text

Download PDFPDF

Efficacy of antiglaucoma fixed combination therapy versus unfixed components in reducing intraocular pressure: a systematic review
  1. J A Cox1,
  2. S P Mollan1,
  3. J Bankart2,
  4. R Robinson1
  1. 1
    University Hospitals of Coventry and Warwickshire NHS Trust, Coventry, UK
  2. 2
    Trent Research and Development Support Unit and Department of Health Sciences, University of Leicester, Leicester, UK
  1. Miss S P Mollan, University Hospitals of Coventry and Warwickshire NHS Trust, Clifford Bridge Road, Coventry CV2 2DX, UK; soozmollan{at}


Aim: To evaluate the efficacy of the fixed combination ocular hypotensive therapies compared with their non-fixed components used concomitantly for lowering intraocular pressure (IOP) in glaucoma and ocular hypertension.

Methods: A systematic review of the literature, up to May 2007, without limits on year or language of publication was performed. Seven randomised controlled trials (n = 2,083 eyes) were identified. Assessment of methodological quality was made using standardised criteria. Results were pooled quantitatively using meta-analysis methods, and statistical analysis was performed using STATA software. The difference in mean intraocular pressure (mm Hg) from baseline between the fixed combination and non-fixed component therapies was compared. Non-inferiority in terms of efficacy was set at an upper confidence limit of ⩽1.5 mm Hg for all time points (hour (Hr)0, Hr2 and Hr8) and evaluated at 12 weeks. Safety was evaluated from data on adverse events as reported in the included studies.

Results: Of the 679 abstracts identified, seven randomised controlled trials met the selection criteria. The quality scores of included studies were high (mean of 29.4, maximum score 30). The mean differences (95% CI) and p values at 12 weeks were as follows: 0.200 mm Hg, (CI −0.106 to 0.507), p = 0.20 for Hr0, 0.393 mm Hg (CI 0.038 to 0.747), p = 0.03 for Hr2 and 0.501 mm Hg (CI 0.156 to 0.846), p = 0.004 for Hr8. Although both Hr2 and Hr8 showed statistical significance favouring the non-fixed combinations, the non-inferiority measure ⩽1.5 mm Hg upper confidence limit was not exceeded.

Conclusions: Fixed combination therapies are equally safe and effective at lowering IOP as their non-fixed components administered concomitantly.

View Full Text

Statistics from

Realini and Fechtner1 calculated over 56 000 ways to reach maximal medical therapy for treatment of glaucoma in 2002. Today there are many more. If mono-therapy is not successful, a switch is advocated, and if partially responsive, adding in a second agent is recommended.2 However, the use of multiple medications, or complicated regimens, has been implicated in poor patient adherence3 and attempts to reduce frequency of dosing have been recommended.4 Additionally, the “washout” effect is a risk with multiple topical therapies,5 as is increasing possibility of local toxicity and side effects.6 7 The fixed-combination ocular hypotensive drugs provide a solution to some of these problems.

Many randomised controlled trials815 have addressed efficacy for mono-therapy and various fixed combination therapies as a first-line management of glaucoma. Of the systematic reviews, some have compared single therapies,16 17 Burr et al18 compared single therapies with surgery, and Webers et al19 included studies that investigated use of fixed combinations and concomitant use of single agents. As yet, however, there is no conclusion concerning the issue of the relative efficacy of using fixed or the non-fixed concomitant therapies. The aim of this systematic review was to investigate whether the fixed combination ocular hypotensive therapies are superior to their non-fixed concomitant components in reducing intraocular pressure (IOP) in glaucoma and ocular hypertension.


Literature search

Abstracts were identified through a computerised search of the Cochrane Controlled Trials Register, PubMed and Dialog Datastar for Medline and Embase, and references from relevant articles. The grey literature, such as conference abstracts, was also searched. Figure 1 details the sources that were searched to identify published studies from 1996 onwards. The search date was determined to include the pilot studies of the earlier fixed-combination therapies located in scoping searches in 1998. The search strategies used for Dialog Datastar Medline and Embase included an adaptation of the Cochrane phase 1 and phase 2 randomised controlled trial filters. Both keywords and medical subject heading (MeSH) terms were used. No limits were placed on the language of publication. If the abstract was relevant or contained insufficient information for exclusion, the article was obtained and reviewed.

Figure 1 Flow chart of the study selection process.

Of the 679 abstracts screened, 19 were from potentially relevant studies, of which 12 were excluded because they did not meet the selection criteria (fig 1).

Criteria for selection of studies

Only randomised, parallel, controlled trials or cross-over studies with an adult population (age ⩾18 years) were considered. Only trials reporting on glaucoma or ocular hypertension with a baseline intraocular pressure of ⩾22 mm Hg were included. Participants were administered one of the following fixed medications in comparison with their non-fixed components administered concomitantly: travoprost (Alcon Labaratories, Fortworth, TX) and a β-blocker; brimonidine tartrate (Allergan, Irivne, CA) and a β-blocker; dorzolamide (Merck & Co., Whitehouse Station, NJ) and a β-blocker; bimatoprost (Allergan) and a β-blocker; or latanoprost (Pfizer, New York) and a β-blocker. Other exclusion criteria are listed in fig 1, and if there was more than one reason for exclusion, only the first found is listed. Excluded publications were reassessed to make sure all eligible publications were included.

Data extraction and assessment of study quality

Data from included articles were extracted using a standardised extraction form. All were critically appraised with emphasis on selection bias, performance bias, attrition bias, detection bias and publication bias. An adapted methodology checklist combining the Critical Apprasial Skills Programme (CASP) assessment tool and the Cochrane Eyes and Vision Group quality assessment criteria was used. It comprised of 10 questions and was scored for quality inclusion as follows: 1 = inadequate, 2 = unclear and 3 = adequate.

Outcome measures

The primary outcome measure was the mean change in IOP (mm Hg) from baseline to 12 weeks. A cut-off point of ⩽1.5 mm Hg upper confidence limit was used to demonstrate equal efficacy of fixed and non-fixed therapies, as this is a frequently reported acceptance level of non-inferiority in glaucoma studies.9 10 15 IOP measurements taken were labelled hour 0 (Hr0) for the first measurement point, prior to the installation of the morning dosage, hour 2 (Hr2) for the second measurement taken 2 h later, and hour 8 (Hr8) taken 8 h following the first measurement. These measurement time frames were common among six studies for Hr0 and Hr2 and four studies for Hr8.

The secondary outcome was the number of patients experiencing adverse events during the study period; this is inclusive of the number of patients discontinuing with the study due to drug-related side effects.

Statistical methods

The Begg22 and Egger23 tests were used to assess for publication bias. STATA version 9.2 software was used to analyse the data. If heterogeneity tests were non-significant, fixed effects models were used, as they provide narrower 95% CIs than the equivalent random effects models, which are more appropriate where significant heterogeneity is detected. The issue of pooling both intention to treat (ITT) and per protocol (PP) data was re-examined by sensitivity analysis to determine whether the type of analyses used (from such a small number of studies) had an effect on the primary outcome.


Description of trials

Seven multi-centred randomised controlled trials were selected, most of which were multi-national. Although described as worldwide, the countries included were mainly Western: United States, Canada and Europe, with the exception of Hutzelmann et al, which also included centres based in South America (table 1).

Table 1 Characteristics and findings of trials included in the systematic review

They had a combined population of 2083 (range: 32–517), and the baseline characteristics are listed in table 2.

Table 2 Baseline characteristics of the randomised clinical trials evaluating the efficacy of fixed combination versus the non-fixed components for glaucoma and ocular hypertension

They incorporated four of the five glaucoma fixed combinations on the current market: Cosopt,12 15 Combigan,10 13 Duotrav11 14 and Xalacom.9 The only study for Ganfort 20 was excluded due to its short duration. Publication bias was tested using the Begg test22 and the Egger test,23 and both produced non-statistically significant results, providing no evidence of publication bias. However, with small sample sizes, these tests have a low power to detect publication bias.

The studies were all double-masked with fixed, vehicle and non-fixed eye-drops provided in kits of identical dropper bottles labelled with instillation times. A randomisation method using remote computer-generated randomisation schedules was used by six of the seven studies. All of the studies used a “run in” therapy to ensure a wash-out effect of previous medication prior to the onset of the use of the study medications. Six studies reported observer masking. Using calibrated Goldmann tonometry, the evaluation of the IOP measure was conducted at least twice by one or more individuals at each time point for each visit. Overall, the mean of two readings was recorded, or the mean of three if either of the two readings differed by more than 2 mm Hg10 and 4 mm Hg.11

There was a majority consensus to use a non-inferiority margin of 1.5 mm Hg, but not one of the studies used identical figures in their sample size calculations. The set statistical power ranged from 79 to 85%. Type 1 error levels were based on significance levels ranging between 0.05 and 0.001. The standard deviations used to determine sample size in relation to the variance of change in IOP from baseline ranged from 3.0 mm Hg to 4.2 mm Hg.

Difference in intraocular pressure between fixed and non-fixed preparations

The largest reported mean difference was 1.1 mm Hg at Hr 8.11 Two measures, at Hr014 and Hr2,15 showed a greater reduction in IOP using the fixed combination treatments (table 3).

Table 3 Primary outcome: mean difference (MD) and standard errors (SE) in mean intraocular pressure (mm Hg) change from baseline between the fixed and non-fixed combinations at week 12 for Hr0, Hr2 and Hr8

The peak diurnal reduction in mean IOP from baseline was achieved by both the fixed combination and non-fixed combinations of latanoprost9 followed by travoprost11 (table 4).

Table 4 Mean change in intraocular pressure from baseline as expressed in mm Hg at week 12


Results for heterogeneity were consistently non-significant with p values of 0.701 for Hr0, 0.312 for Hr2 and 0.188 for Hr8: indicating the results of the studies were relatively homogeneous. However, these tests have a low power with small sample sizes.

A graphical assessment of the non-inferiority of fixed versus non-fixed combination glaucoma therapies at week 12 is presented in fig 2 in the form of forest plots, for Hr0, Hr2 and Hr8. At all time periods (Hr0, Hr2 and Hr8) at week 12, the pooled results of the included studies fell below the non-inferiority level set at ⩽1.5 mm Hg of the upper confidence limit. Although, in each case, the dotted line of estimated treatment effect favours the non-fixed combinations, the fixed combinations were equally effective in reducing intraocular pressure at Hr0 (fig 2). At Hr2, (fig 2) the pooled analysis was statistically significant favouring the non-fixed combination 0.393 (CI 0.038 to 0.747), p = 0.03. Only four studies9 11 14 15 conducted IOP measurements at Hr8 (fig 2). The results show a statistically significant result in favour of the non-fixed combinations, 0.501 (CI 0.156 to 0.846), p = 0.004, but the pooled 95% confidence interval remains well within the non-inferiority limit.

Figure 2 Plots showing the weighted (study size) estimate in terms of difference in mean intraocular pressure mm Hg from baseline and associated 95% confidence intervals.

The results of the sensitivity analysis testing for pooling data from ITT and PP showed no difference in primary outcome compared with the overall results: 0.291 mm Hg (CI −0.09 to 0.67), p = 0.136 for the ITT studies alone, compared with 0.200 mm Hg (CI −0.106 to 0.507), p = 0.20 at week 12, Hr0, for the ITT and PP studies.

As a secondary outcome, this review examined the data presented by the included studies on adverse events and side effects (table 5).

Table 5 Overall incidence of adverse effects reported, likely drug-related adverse effects and proportion of patients who discontinued medication due to adverse effects

From the data provided, the three main ocular side effects reported were hyperaemia, ocular irritation and keratitis. All of the studies reported hyperaemia, and the three prostaglandin studies showed more cases reported than in the non-fixed groups. This difference was found to be significant in the Schuman (travoprost) study.14

The reporting of systemic adverse events was sporadic. The highest recorded was “bitter taste” with dorzolamide, a known side effect, and equally evident with either fixed or non-fixed therapies. Due to the nature of reporting differences for adverse events it is difficult to ascertain whether any of the adverse events recorded are specifically associated with the study medication. The number of patients reported to discontinue from the study due to intolerance or adverse events was found to be higher in the fixed combination groups in five versus two studies (table 5). However, the number of patients involved was extremely small in relation to the study population sizes (1–6%) with the exception of Konstas et al13 (12.5%).


The seven studies included in this review scored highly with respect to the quality of study methodology. The results of the studies were shown to be homogeneous and therefore valid for meta-analysis using a fixed effects model, although the heterogeneity test has a low power to reject homogeneity with small samples. Although publication bias could not be formally ruled out due to the small number of studies, the funnel plots showed no obvious deviations from symmetry (fig 2).

In assessing whether the fixed-combination therapies are equally effective compared with the non-fixed combinations in reducing intraocular pressure, the review demonstrated non-inferiority at the standard ⩽1.5 mm Hg upper confidence limit for all time points evaluated at 12 weeks. These results were consistent when accounting for differences in study analysis (ITT or PP) and for individual study variations in terms of size9 and quality.10 The magnitude of IOP reduction in the prostaglandin analogue therapies appeared to be greater than that of the alpha2-agonist and carbonic anhydrase inhibitor combinations (see table 3). However, no conclusion should be made with regard to which medication is the most potent, as this was not this review’s hypothesis and as such could not be fairly assessed statistically from the studies included.

It has been suggested that a difference of between 1.0 and 1.5 mm Hg in the upper confidence limit is acceptable in showing clinical non-inferiority between the fixed and non-fixed combination therapies.9 10 15 However, we note that this difference may be too small due to the fact that the error margin of Goldmann tonometry can be as much as 2 mm Hg.21

One of the main limitations in collating data from glaucoma studies is that there is no standard method of reporting IOP measures over a 24 h period. Although it is common practice to provide diurnal measurements, the actual time points measured are variable, meaning that in order to ascertain set time frames for review, it is necessary to extrapolate, which is not ideal. Study duration is also a confounding factor: 12 weeks was chosen as a minimum length of study in the inclusion criteria. It was reasoned that this was an adequate amount of time to assess changes in IOP and monitor any side-effects. However, this excluded the only study for bimatoprost/timolol (Ganfort),20 as its study duration was only 3 weeks. Longer follow-up periods are required to assess glaucoma progression, the occurrence of longer-term adverse events and patient adherence.

This review has highlighted that assessing adverse effects is variable, particular when side-effects are pooled in studies. As such, no statistical analyses could be used to determine differences between the use of fixed and non-fixed combinations. The results presented by the studies on safety were on the whole mixed. In terms of clinical importance, it may be more useful to ascertain the number of patients who had to discontinue from the study due to an intolerance or more serious adverse event associated directly with the study medication. In the seven studies reviewed, the non-fixed combinations discontinued less frequently than the fixed combinations, but the proportions of patients involved were minimal (1–6%) (table 5).

For those who require a multiple medication regime in order to maintain their intraocular pressure at a safe level, this systematic review of randomised clinical trials has indicated that the fixed combination therapies are as effective as their non-fixed components administered concomitantly.


We would like to thank Mr I Calder and Professor RAB Mollan for their critical review of the paper.


View Abstract


  • Competing interests: None.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.