We identified 64 potential trials. Thirteen reports were excluded: three were analyses of more than one study with no individual study data,w1–w3
five were in elderly patients with no evidence that they had overactive bladders,w4–w8
in one the placebo arm was not randomised,w9
two were not comparisons of suitable anticholinergic drugs with placebo,w10 w11
and two did not report suitable outcomes.w12 w13
We excluded 19 cross over trials as data were presented in a manner that was unsuitable for analysis.12 w14-w32
The remaining 32 trials were analysed.w33–w64
Some trials had multiple publications (maximum 13).w63
The 32 trials recruited a total of 6800 participants (1529 men and 3938 women; some trials did not report sex). Of these, 3866 (57%) participants were randomised to receive anticholinergic drugs and 1743 (26%) were randomised to receive placebo. Three trials did not report the numbers randomised to each group,w39 w40 w59
and three studies only reported the numbers after excluding the dropouts.w33 w41 w48
Sample sizes ranged from 20 to 1529, with a median of 155.w54 w63
Details of the 32 included trials appear on bmj.com. Inclusion and exclusion criteria were unclear for some of the trials. Most trials included people with overactive bladder, regardless of sex, but some were restricted to those with urge urinary incontinence, some to women only, and one to men with bladder outlet obstruction.w35
Trials compared the following active treatments with placebo: tolterodine (12 trials),w33–w35 w42 w46 w47 w49 w52-w55 w62 w63 oxybutynin chloride (10 trials),w34 w37 w42–w44 w50 w51 w58 w60 w64 trospium chloride (eight trials),w36 w38 w39 w44 w48 w49 w46 w61 propiverine (five trials),w41 w45 w51 w57 w59 emepronium bromide (one trial),w44 and propantheline (one trial).w60 Six trials compared two different anticholinergics with placebo (tolterodine and oxybutynin, oxybutynin and trospium chloride, tolterodine and trospium chloride, oxybutynin and propiverine, oxybutynin and propantheline).w34 w42 w44 w49 w51 w60 In four trials drugs were given by intravesical administration, and in all the remaining trials drugs were taken orally.w43 w44 w50 w54 In the trials of oral drugs, length of treatment ranged from 12 days to 12 weeks.w34 w42 w43 w53 w63 Outcome was measured at the end or shortly after the end of the treatment period in all trials.
The trials were more explanatory than pragmatic.13
Outcome was generally measured at the end of treatment, and there was a focus on urodynamic measures. Due to deficiencies in data reporting (for example, point estimate without measure of variation), many trials contributed limited data to the review.
Methodological quality of included studies
The method of group allocation was rarely described, although all trials but one were said to be double blind.w44
Although double blinding should adequately conceal group allocation, this is not guaranteed.14
Only one trial specifically stated that outcome assessors were blind to group allocation.w37
In some studies the code was broken on completion of the study, but only a few specified that this was after the analysis.
Baseline comparability of the groups was not mentioned in seven studies.w39 w44 w45 w48-w50 w64 The remaining 25 trials stated that the groups were comparable at baseline, although three studies did not provide supporting data.w35 w38 w59 In 13 trials the evaluation of treatment efficacy was conducted on intention to treat principles,w34 w37 w38 w44 w46 w48 w51-w54 w58 w63 w64 and seven trials specifically stated that a per protocol analysis was used to assess efficacy of treatment.w33 w36 w41 w42 w49 w61 w62
The description of withdrawals or dropouts was not adequate in eight trials.w33 w38 w39 w45 w48 w49 w59 w62 No dropouts occurred in the trials using single intravesical or oral doses of drug,w43 w44 w50 w54 w64 and in nine trials the dropout rate was 10% or less.w41 w46 w47 w50 w52 w53 w56 w57 w60 In the remainder, dropout rates ranged from 12% to 21%.w42 w43 w51 w58 w63
The primary, or only, reference for seven of the 32 included trials was a conference abstract, and none of these were interim reports of continuing studies.w33, w35 w39 w48-w50 w59 Eight trials reported the patient's perception of cure or improvement in symptoms (fig ).w34 w37 w41 w45 w53 w54 w58 w63 Those receiving active treatment were more likely to be subjectively improved (relative risk 1.41, 95% confidence interval 1.29 to 1.54). Nine trials reported on post-treatment leakage episodes in 24 hours (fig ).w34 w37 w41 w42 w46 w53 w55 w62 w63 Those taking an anticholinergic had about one leakage episode less in 48 hours than those taking placebo (estimated mean difference for reduction in number of leakage episodes in 24 hours 0.6, 0.4 to 0.8). Eight trials reported on the number of micturitions in 24 hours (fig ).w34 w41 w42 w46 w53 w55 w62 w63 Those taking an anticholinergic had about one less micturition in 48 hours than those taking placebo (estimated mean difference for reduction number of micturitions in 24 hours 0.6, 0.4 to 0.8). No significant heterogeneity was found in these results.
Effect of anticholinergics compared with placebo on patient perception of cure or improvement in symptoms of overactive bladder
Effect of anticholinergics compared with placebo on number of leakages in 24 hours and number of micturitions in 24 hours
Twelve trials reported either maximum cystometric capacity after treatment or change in maximum cystometric capacity after treatment (fig ).w33-w38 w44-w57 w60-w62 A larger increase in maximum cystometric capacity occurred in those receiving active treatment (estimated mean difference 54 ml, 43 ml to 66 ml). Significant heterogeneity was observed (P=0.027). When the data from Froehlich et al, in which participants received treatment by intravesical administration, was removed from the pooled analysis, there was an improvement in maximum cystometric capacity in favour of the drug group (49 ml, 38 ml to 61 ml), and the test for heterogeneity was no longer significant (P=0.51).w44
Effect of anticholinergics compared with placebo on maximum cystometric capacity and bladder volume at first contraction
Nine trials reported either volume at first contraction after treatment or change in volume at first contraction after treatment (fig ).w33 w36 w38 w47 w48,S55 w57 w60 w62 Volume at first contraction increased more in the drug group than in the placebo group (52 ml, 38 ml to 67 ml).
Residual volume after treatment or change in residual volume after treatment was reported by 11 trials (fig ).w33,S36 w38 w44 w47 w48 w51 w55 w57 w60 w62 On average, residual volume was 4 ml greater in the drug group (4 ml, 1 ml to 7 ml), but significant heterogeneity was observed (P<0.0001). When the data from Froehlich et al were excluded from the pooled analysis, those receiving active treatment had an increase in residual volume compared with those receiving placebo (3.5 ml, 0.1 ml to 6.8 ml), and the test for heterogeneity was no longer significant (P=0.14).w56
Effect of anticholinergics compared with placebo on residual urinary volume
Thirteen trials reported the number of people withdrawing owing to adverse events (fig ).w34-w36 w42 w46 w47 w51-w63 No significant difference was found in the number of withdrawals due to adverse events between drug and placebo groups (relative risk 1.01, 0.78 to 1.31), but there was significant heterogeneity (P=0.08). Excluding the data from Rentzhog et al, a dose ranging study of tolterodine, did not change the finding of the pooled analysis very much (1.05, 0.81 to 1.38), but the test for heterogeneity was no longer significant (P=0.24).w55
Effect of anticholinergics compared with placebo on withdrawal owing to side effects and dry mouth
Dry mouth was the most frequently reported side effect, and data on this were available from 20 trials (fig ).w33 w34 w37 w38 w42 w43 w46 w47 w49-w53 w55 w57 w58 w60-w63 The risk of dry mouth was two and a half times greater in the drug group (relative risk 2.56, 2.24 to 2.92). Significant heterogeneity was observed in this comparison (P<0.0001). Two trials, in elderly patients, had high rates of dry mouth in the placebo arm, perhaps as a consequence of polypharmacy.w37 w58When these two trials were excluded from the pooled analysis, the risk of dry mouth was nearly three times greater (2.88, 2.46 to 3.36), and the test of heterogeneity was no longer significant (P=0.11).
Despite clinical heterogeneity of the included trials (for example, sample populations and type of drug), sensitivity analyses did not show any differences in the results for age, sex, diagnosis (neurogenic or idiopathic detrusor overactivity), or type of drug. Changing to a random effects model to get the pooled estimate did not change the results much or change the interpretation. The quality of the trials was uniform in the elements that have been shown to be most related to bias—that is, hiding the group allocation and blinding, so sensitivity analysis on the basis of study quality was not done. Few trials reported quality of life measures, and none reported an economic analysis.