Table 3. Changes in Taxed and Untaxed Beverage Volume Sold in Cook County, Illinois, Relative to St Louis County and City, Missouri^a.

Beverage type	Posttax period				Postrepeal period				Overall change
	Change in level		Change in slope		Change in level		Change in slope
	Coefficient estimates (95% CI)	Exponentiated coefficients	Coefficient estimates (95% CI)	Exponentiated coefficients	Coefficient estimates (95% CI)	Exponentiated coefficients	Coefficient estimates (95% CI)	Exponentiated coefficients	Coefficient estimates (95% CI)	Exponentiated coefficients
Taxed	−0.297 (−0.415 to −0.179)	0.743	−0.002 (−0.014 to 0.011)	0.998	0.266 (0.124 to 0.408)	1.305	0.005 (−0.008 to 0.018)	1.005	0.066 (−0.039 to 0.172)	1.068
SSB	−0.262 (−0.383 to −0.141)	0.770	−0.003 (−0.015 to 0.010)	0.997	0.257 (0.111 to 0.403)	1.293	0.006 (−0.008 to 0.019)	1.006	0.059 (−0.049 to 0.167)	1.061
ASB	−0.461 (−0.581 to −0.341)	0.631	0.005 (−0.008 to 0.018)	1.005	0.328 (0.184 to 0.473)	1.388	−0.003 (−0.016 to 0.010)	0.997	0.013 (−0.093 to 0.120)	1.013
Untaxed	0.036 (−0.034 to 0.106)	1.037	−0.003 (−0.011 to 0.004)	0.997	−0.012 (−0.097 to 0.073)	0.988	0.005 (−0.003 to 0.013)	1.005	0.034 (−0.029 to 0.096)	1.035

Abbreviations: ASB, artificially sweetened beverage; SSB, sugar-sweetened beverage.

^aExponentiated coefficients and overall change were computed to show changes in percentage terms. Estimates are from interrupted time series models for the change in intercept and slope in log volume in Cook County relative to St Louis after the Cook County sweetened beverage tax was imposed and after it was repealed, controlling for month. Posttax changes are relative to the end of the pretax period, and postrepeal changes are relative to the end of the posttax period. The overall change in Cook County relative to St Louis at the end of the postrepeal period relative to the end of the pretax period was computed as the sum of the 2 change-in-intercept terms and the 2 change-in-slope terms multiplied by the number of weeks to which they applied. Models were fit by feasible generalized least squares, allowing for first-order autoregressive autocorrelation within each site with site-specific coefficients, and a heteroskedastic error structure assuming no correlation between sites. The volume sold measure used for this analysis is the total across UPCs within each site and week. The analysis included 122 pretax weeks (from March 29, 2015, to July 29, 2017), 16 tax weeks (from August 6, 2017, to November 25, 2017), and 35 postrepeal weeks (from December 3, 2017, to August 4, 2018).