Evidence that a tax on sugar sweetened beverages reduces the obesity rate: a meta-analysis

BMC Public Health

Table 3 Own and cross price elasticity of SSBs

Author/year/country	Own-price elasticity		Cross-price elasticity
Author/year/country	Estimated	SE	Fruit juice	SE	Whole milk	SE	Diet drink	SE
Barquera S, et al. (2008)[44], Mexico	-1.085	0.195	-0.016	0.003	0.052	0.011
Bonnet C, et al. (2011)[45], France ¹	-2.206	0.133
Claro RS, et al. (2012)[46], Brazil ²	-0.85	0.434
Dharmasena S, et al. (2012)[42], USA ³	-2.255	0.550
Finkelstein EA, et al. (2010)[47], USA	-0.870	0.090
Finkelstein EA, et al. (2013)[18], USA	-1.320	0.005
Fletcher JM, et al. (2010b)[48], USA ⁴	-4.445	1.806	1.857	2.332	7.67	2.156
Lin BH, et al. (2011)[49], USA (Low-Income Population)	-0.949	0.082	0.473	0.127	0.242	0.129	-0.23	0.104
Lin BH, et al. (2011)[49], USA (High-Income Population)	-1.292	0.096	0.529	0.093	-0.054	0.13	-0.591	0.112
Smith TA, et al. (2010)[50], USA	-1.264	0.089	0.557	0.095	0.222	0.126	-0.457	0.103
Overall	-1.299		0.388		0.129		-0.423
(LCI – HCI)	(-1.089 - -1.509)		(0.0095 - 0.767)		(-0.085 - 0.342)		(-0.628 - -1.219)

In all the next four cases the authors were contacted by email and could not provide any additional information to explain the missing values needed. To enable inclusion of the studies, we estimated the following values:
¹Own-price elasticity: Consumer prices rose “by more than 3.4% on average” which led to a decrease in market share of 7.5%. Then -7.5/3.4 = -2.206. SE was inferred from Table VI: SE = (-0.07/1.16)*2.206 = 0.133.
²Information obtained from the paper: P < 0.05 at p(2-sides) = 0.05. Then SE = -[0.85]/1.96 = 0.434.
³The paper does not report SE, but only a p-value of 0.0000. It was used a website to derive an estimate of the Z-statistic (http://easycalculation.com/statistics/p-value-for-z-score.php). It was chose a value of 4.1, which is the lowest value that gives a two-sided p-value of 0.0000. Then SE = -[2.255]/4.1 = 0.550.
⁴SE inferred from mean and SE of grams of soft drink consumption (Table 4 in ref [48]): mean = -18.052, SE = 7.333, Ratio 7.33/-18.052 = -0.40622. Then SE = -4.445*-0.40622 = 1.806.
The numbers in parenthesis denote 95% confidence interval.

ISSN: 1471-2458