Figure 1.

 Pooled relative risk of hospital admission comparing addition of single and multiple doses of ipratropium bromide to β₂ agonists in trials reporting data relative to hospital admission: one trial (Watson et al) not included in overall estimate of multiple dose protocol as absence of event in both groups prevented estimation of relative risk. Trials stratified according to concomitant administration of systemic corticosteroids. Width of horizontal line represents 95% confidence interval around point estimate (black square). Size of point estimate represents relative weight (% weight) of each trial in the pooled summary estimate (diamond). Vertical line is line of no effect (relative risk 1.0)