Table 3.

Sensitivity analyses of the effect of ondansetron on the number of retches + vomits (R+V) induced by cisplatin (5 or 10 mg kg⁻¹)

	Effect estimate (proportion of R+V difference)	95% CI	Number of comparisons	Overall effect (Z tests)	Heterogeneity (χ2)
Cisplatin dose and duration of the observation period
5 mg/kg (24 h)	−0.61	−0.87 to −0.35	5	P < 0.001***	P = 0.800
10 mg/kg (24 h)	−0.41	−0.95 to 0.13	4	P = 0.13	P = 0.004**
10 mg/kg (6 h)	−0.68	−1.09 to −0.26	3	P = 0.001**	P = 0.140
10 mg/kg (4 h)	−0.70	−0.94 to −0.46	8	P < 0.001***	P = 0.100
10 mg/kg (2 h)	−0.93	−1.15 to −0.70	4	P < 0.001***	P = 0.750
Ondansetron dose
0.01 mg kg−1	−0.71	−1.16 to −0.26	1	P = 0.002**	N/A
0.1–0.5 mg kg−1	−0.40	−0.69 to −0.10	8	P = 0.009**	P = 0.030*
1–10 mg kg−1	−0.83	−0.95 to −0.70	15	P < 0.001***	P = 0.600
Mode of delivery of ondansetron
i.v.	−0.77	−0.95 to −0.60	11	P < 0.001***	P = 0.130
i.p.	−0.59	−0.81 to −0.37	11	P < 0.001***	P = 0.030*
s.c.	−0.74	−1.45 to −0.03	2	P = 0.040*	P = 0.710
Treatment time
With cisplatin	−0.67	−0.86 to −0.48	15	P < 0.001***	P = 0.010*
30 min before	−0.68	−1.09 to −0.26	3	P = 0.001**	P = 0.140
Twice daily	−0.74	−1.45 to −0.03	4	P = 0.04*	P = 0.710
Three times daily	−0.72	−0.98 to −0.48	2	P < 0.001***	P = 0.26
Mode of administration of cisplatin
i.v.	−0.77	−0.97 to −0.57	9	P < 0.001***	P = 0.100
i.p.	−0.62	−0.81 to −0.44	15	P < 0.001***	P = 0.080
Animal origin
UK	−0.70	−0.88 to −0.53	15	P < 0.001***	P = 0.020*
USA	−0.67	−0.91 to −0.44	7	P < 0.001***	P = 0.330
NZ	−1.00	−0.83 to −0.17	1	P = 0.020*	N/A
Quality score
Less that 5/9	−0.65	−0.95 to −0.34	5	P < 0.001***	P = 0.060
5 to 7/9	−0.76	−0.97 to −0.56	9	P < 0.001***	P = 0.260
7.5 to 9/9	−0.65	−0.88 to −0.42	10	P < 0.001***	P = 0.050

The effect estimate was computed as the weighted mean difference (WMD) and expressed as the proportion of retches and vomits in the control group. An effect estimate of −1 indicates that emesis was abolished in the treatment group, 0 indicates that the treatment had no effect on the R+V response and an effect estimate >0 indicates that the treatment increased the number of R+V. The variables examined were the variant of the cisplatin model (5 or 10 mg kg⁻¹ cisplatin and the duration of the observation period), the mode of administration of cisplatin, the dose of ondansetron, the regimen of ondansetron administration (mode of delivery and timing relative to cisplatin administration), the animal origin (country animals were bred) and the quality score assigned to the study where comparisons were extracted