Duration of the common cold and similar continuous outcomes should be analyzed on the relative scale: a case study of two zinc lozenge trials

Table 1 Absolute and relative effects of zinc lozenges in the two randomized trials

	Mossad (1996) [9]		Petrus (1998) [10]
Trial groups:	Zinc	Placebo	Zinc	Placebo
N:	49	50	52	49
Duration of colds
Mean (days):	5.20	9.20	5.29	7.06
SD (days):	2.83	5.32	2.57	3.91
P, variance test to compare the SDs:	2*10⁻⁵		0.004
Effect of zinc on the absolute scale (days): (95% CI):	−4.00 (−5.7 to −2.3)		−1.77 (−3.1 to −0.47)
Quantiles of common cold duration (days)
1st quartile:	3	5	3	4
2nd quartile:	5	8	5	6
3rd quartile:	7	14	7	8
90th percentile:	9	17	8	14
Transformation to the %-scale*:
Mean (% of placebo group mean):	56.6%	100%	74.9%	100%
SD (% of placebo group mean):	30.7%	57.8%	36.4%	55.3%
Effect of zinc on the relative scale: (95% CI):	−43.4% (−62% to −24.9%)		−25.1% (−44% to −6.7%)

* Transformation to the %-scale indicates that the means and SDs of the Mossad [9] study were divided by 9.20 days, and the Petrus [10] study by 7.06 days. By this transformation the value of the zinc lozenge group mean becomes the RoM and the difference between the zinc and placebo groups gives the effect of zinc lozenges directly in percentages. Since this transformation is linear, the relative scale leads to a t-score identical with that obtained on the original values (i.e., on the absolute scale; the MD scale)

ISSN: 1471-2288