Eight studies (n=2,301) met the criteria for inclusion in the meta-analysis.
Effect of dose on the mean change in outcome measure.
The raw data for each outcome measure was plotted against the dose of fluticasone. The graph showed the response beginning to plateau at a dose of 100 to 200 microg/day, with little improvement at higher doses.
Doses necessary to achieve 80 and 90% of the effect obtained with 1000 microg/day. A negative exponential line of best fit, derived from the weighted means of the effect at each dose, was modelled. From this, it was calculated that 80 and 90% of the benefit obtained with 1000 microg/day fluticasone was achieved at doses of 70 to 170 microg/day and 100 to 250 microg/day, respectively, depending on the outcome measure.
Dose necessary to achieve the maximum response. For four of the outcome measures (FEV1, morning PEF, evening PEF, beta-agonist use) it was possible to determine, by quadratic regression, the dose giving the peak effect. This ranged from 560 to 660 microg/day. In addition, it was possible to estimate the mean changes in the outcome measures.
Effect of lower versus higher doses on patients remaining in trials (5 trials). The ORs of patients remaining in a trial at a total dose of inhaled fluticasone of 200 microg/day, compared with higher doses, was 0.73 (95% confidence interval, CI: 0.49, 1.08). A test for homogeneity was not significant, with a value of 6.93 (d.f.=4, P<0.14). The random- effects pooled OR was 0.70 (95% CI: 0.38, 1.3).
Effect of lower versus higher doses on FEV1.
Four studies reported data on the comparison of 200 microg/day fluticasone with higher doses. The meta-analysis of the standardised difference in FEV1 showed a difference in FEV1 of 0.13 of a standard deviation, with a CI that included zero (95% CI: -0.02, +0.29). The pooled standard deviations for these 4 studies ranged from 0.43 to 0.76. A test for homogeneity was not significant. The random-effects pooled odds ratio was 0.13 (95% CI: -0.03, +0.30).