The authors concluded that patient characteristics such as baseline severity and duration of erectile dysfunction can significantly affect the findings of efficacy trials of sildenafil. Given the small number and size of the trials and limitations in the study methodology, the authors' findings are suggestive rather than definitive.

To evaluate treatment effect modifiers in trials of oral sildenafil for erectile dysfunction (ED).

MEDLINE and the Cochrane CENTRAL Register were searched to October 2005; the search terms were reported. The drug manufacturer was contacted for further information. The reference lists of identified articles and reviews were checked for other relevant publications.

Study designs of evaluations included in the review

Double-blind, randomised placebo-controlled trials (RCTs) of either a parallel-group or crossover design were eligible for inclusion. Studies had to report the sample size.

Specific interventions included in the review

Studies of flexible-dose oral sildenafil were eligible for inclusion. The included studies had a mean treatment duration ranging from 6 to 26 weeks (10 studies had a treatment duration of 12 weeks).

Participants included in the review

Studies in patients being treated for ED using oral sildenafil were eligible for inclusion. In the included studies, themean age of the patients ranged from 50.9 to 60.5 years. A minority of studies reported the proportion of white patients (range: 9 to 41%). The participants included European, Asian, Latin American and US American men. The mean duration of ED ranged from 3 to 5.3 years and the baseline mean erectile function (EF) score ranged from 9.5 to 14.5. ED aetiology was stated as 'psychogenic' for between 4% and 44.3% of patients.

Outcomes assessed in the review

Studies reporting the mean EF domain of the International Index of Erectile Functioning were eligible for inclusion.

How were decisions on the relevance of primary studies made?

Two reviewers independently applied the inclusion criteria. Any disagreements were resolved with the involvement of a third reviewer.

One reviewer assessed the methodological quality of the studies using the Jadad scale. The criteria assessed included randomisation, blinding and drop-outs.

One reviewer extracted the data on study characteristics, while two reviewers independently extracted the data on outcomes. Any disagreements were resolved by discussion. For each study, the mean difference in the EF domain score and standard deviation were extracted or calculated from reported statistics. For crossover studies, only data from the pre-crossover phase were used.

How were the studies combined?

The studies were not combined to produce a pooled estimate of the effectiveness of sildenafil. Instead, inverse variance-weighted meta-regression models were used to test the impact of treatment effect modifiers on the mean post-treatment EF domain score for both placebo and intervention groups, together with the mean treatment effect. The authors stated that univariate regression models were fitted because of power considerations. Pearson's 'r' was used to test possible relationships between the variables. Publication bias was assessed using a funnel plot and Egger's test.

How were differences between studies investigated?

Regression analyses were conducted to investigate the impact on treatment effect of various trial characteristics: mean trial duration, mean age, mean baseline EF score, proportion of men with psychogenic aetiology, mean ED duration, sample size, methodological quality, year of publication and impact factor of publishing journal in year of publication.

Thirteen RCTs (n=2,996) were included in the review (the number of participants ranged from 111 to 443).

Eleven of the 13 studies scored 3 points out of a possible 5 for quality; given this low variation, the impact of methodological quality on treatment effect was not investigated.

Modifiers of the post-treatment EF domain score in the placebo group: patients who had a higher baseline EF domain score (erectile dysfunction of minor severity) showed a greater placebo response (p<0.001). No other confounders such as treatment duration, age of patients, aetiology of ED or ED duration had a statistically significant effect.

Modifiers of the post-treatment EF domain score in the sildenafil group: no confounders were shown to have a statistically significant effect.

Modifiers of treatment effect: trials where EF was less severe in patients at baseline showed a significantly lower effect (p<0.001); trials in patients with a longer history of ED showed a significantly greater effect (p=0.0018). Further analysis showed that treatment effect was strongly associated with the post-treatment EF domain score of the placebo group (p<0.001) but not with the post-treatment EF domain score of the sildenafil group.

Distorting effect modifiers: older trials showed a greater effectiveness of the sildenafil treatment than trials published more recently (p=0.006).

Publication bias: the funnel plot and Egger's test suggested there was no evidence of publication bias (p=0.858).

Patient characteristics, such as baseline severity and duration of disease, can significantly affect the findings of efficacy trials using flexible-dose oral sildenafil for ED in broad-spectrum populations. However, sildenafil has a reliable effect almost irrespective of these factors. Variation in trial findings is largely caused by varying responses to placebos.

The inclusion criteria were clearly defined for the intervention,patients, study design and outcomes of interest. Only two databases were searched and no efforts were made to identify unpublished studies. It is therefore possible that relevant studies have not been included. The study selection and data extraction processes were conducted in duplicate, thereby reducing the potential for error and bias. Validity was assessed using an established checklist, although only the composite score was presented; this makes it difficult for the reader to judge the validity of the studies for themselves. Only one reviewer was involved in the validity assessment, therefore the possibility of reviewer error or bias cannot be discounted. The analysis focused on the effects of differences between trials on estimates of the absolute and relative effectiveness of sildenafil. The authors discussed possible limitations of the methodology, including the potential for multiple associations between variables to increase the possibility of statistically significant results arising by chance. Given the limitations in the study methodology, the authors' findings are suggestive rather than definitive.

Practice: The authors stated that clinicians need to be aware that patient characteristics may significantly affect the findings of efficacy trials of flexible-dose oral sildenafil when applying the findings to their patients.

Research: The authors stated that investigators, clinical guideline developers, policy makers and economic evaluators should consider the effect of patient characteristics on efficacy when performing or interpreting the results of trials of sildenafil.

1. Burls A, Gold L, Clark W. Systematic review of randomised controlled trials of sildenafil (Viagra) in the treatment of male erectile dysfunction. Br J Gen Pract 2001;51:1004-12. 2. Fink HA, MacDonald R, Rutks IR, Nelson DB, Wilt TJ. Sildenafilfor male erectile dysfunction: a systematic review and meta-analysis. Arch Intern Med 2002;162:1349-60. 3. Markou S, Perimenis P, Gyftopoulos K, Athanasopoulos A,Barbalia G. Vardenafil (Levitra) for erectile dysfunction: a systematic review and meta-analysis of clinical trial reports. Int J Impot Res 2004;16:470-8.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.