Discussion

Using the most recent Canadian nationally representative dietary intake survey, we modelled the impact of implementing a sugary drink tax – a widely implemented policy internationally that is under consideration in Canada – on energy intake from sugary drinks. We estimated a sugary drink tax would reduce both mean energy intake and the proportion of daily energy intake from sugary drinks, with the tiered tax associated with a potential small additional but meaningful reduction in energy intake compared with the 20 % flat-rate tax scenario. Additionally, we estimated both taxation structures would reduce inequities in mean energy intake from sugary drinks measured in the CCHS-N, through a greater reduction in energy intake in children/adolescents, males, lower education, food insecure and the low-income quintile groups. However, in each sugary drink taxation scenario, inequities in sugary drink consumption remained after modelled tax implementations. Moreover, 100 % juice contributed an additional 34 kcal to per capita or 1·9 % of daily energy intake, demonstrating the potential benefit of considering policies that reduce the consumption of beverages containing both added and natural sugars^{(Reference Popkin and Ng8,Reference Kao, Jones and Ohinmaa17)} . Overall, our study highlights the potential of sugary drink taxation, including the additional benefit of taxing 100 % juice, for reducing overall inequities in energy intake from sugary drinks in Canada.

Our findings from Canada add to the emerging international literature suggesting that tiered tax structures may be more effective than flat-rate taxes for reducing SSB consumption. For example, a US modelling study estimated a tiered tax to be more cost-effective than a flat-rate tax, with nearly double the health gains and savings^{(Reference Lee, Mozaffarian and Sy9)}. In our study, reductions in mean per-capita energy from sugary drinks were greater in the tiered compared with the 20 % flat-rate scenarios (40 v. 29 kcal, respectively), with greater per-capita mean energy intake from beverages with higher sugar concentration (tier 2 and 3 tax targets); however 95 % uncertainty intervals were overlapping after incorporating uncertainties related to OPE and CPE values. Additionally, our estimates are likely conservative, as shifts in consumer choice to lower sugar products and incentivized manufacturer product reformulation to reduce sugar content observed in real-world tiered sugary drink taxes were not modelled. For example, following the implementation of the two-tiered United Kingdom (UK) soft drinks industry levy [high tier: £0·24/l in the high tier (≥ 8 g/100 ml) and £0·18/l in the low tier (5–7·9 g/100 ml)], the purchased volume of drinks in the high levy tier decreased 44 % (95 % CI: 29–60), and the percentage of drinks with sugar in the high tier was reduced by 30 %^{(Reference Lee, Mozaffarian and Sy9,Reference Scarborough, Adhikari and Harrington29)} . Similarly, in Portugal, product reformulation following the introduction of a two-tiered SSB tax in 2017 (<8 g/100 ml: €0·08/l and ≥8 g/100 ml: €0·17/l) decreased the average energy density of SSB by 3·1 kcal/100 ml^{(Reference Goiana-da-Silva, Cruz-e-Silva and Gregório30,Reference Goiana-da-Silva, Severo and Cruz E Silva31)} . Tiered tax structures are consistent with the public health rationale for the implementation of sugary drink taxation, in that it can target both specific beverages and sugar content of these beverages^{(Reference Hattersley and Mandeville6)}. However, their limited implementation outside of high-income economies highlights potential implementation barriers, for example, the challenges in verifying and monitoring the sugar content of SSB^{(Reference Hattersley and Mandeville6)}.

Our findings suggest the introduction of a sugary drinks tax in Canada could reduce sugary drink consumption with the potential to reduce inequities in energy intake from sugary drinks. These findings are consistent with the international literature from empirical studies in the USA^{(Reference Bleich, Lawman and LeVasseur32)} and Mexico^{(Reference Ng, Rivera and Popkin33)} as well as modelling studies from Canada^{(Reference Kao, Jones and Ohinmaa17)} and Australia^{(Reference Lal, Mantilla-Herrera and Veerman18)} which suggest that SSB taxes reduce SSB purchases more among individuals with lower incomes and educational attainments. Further, sugary drink taxes have the potential to reduce social inequities in the prevalence and healthcare costs of associated health conditions. Modelling studies in Canada and elsewhere estimate larger potential health benefits among lower SEP groups with minimal additional financial burden concentrated among lower-income groups^{(Reference Kao, Jones and Ohinmaa17–Reference Briggs, Mytton and Kehlbacher19,Reference Andreyeva, Chaloupka and Brownell34–Reference Caro, Ng and Bonilla36)} . For example, a Canadian study modelling estimated the annual financial burden associated with the implementation of a 20 % flat-rate sugary drink tax would be larger in the lowest income quintiles ($43·52 CAD) compared with the highest income quintile ($38·74 CAD)^{(Reference Kao, Jones and Ohinmaa17)}. While this suggests the sugary drink tax was regressive, because price increases would represent a larger percentage of income from low compared with high-income earners, the tax difference was offset by the increase in averted DALYs (156 000 v. 125 000 DALYs) and increased lifetime healthcare savings ($2·27 v. $1·98 billion) in the lowest compared with highest income quintile, respectively^{(Reference Kao, Jones and Ohinmaa17)}.

Although our study estimated a sugary drink tax would reduce inequities in sugary drink consumption in Canada, it is critical that policymakers consider the wider impacts of the design and implementation of a sugary drink tax. While from a public health perspective implementing a sugary drink tax can have health benefits and generate revenue, critics have argued these taxes can be paternalistic and regressive, unfairly and disproportionately impacting lower-income consumers^{(Reference Grignon37,Reference Riediger and Bombak38)} . Appropriately balancing the ethical concerns raised by proponents and critics alike is important to the design of a sugary drink tax^{(Reference Mackay and Huber-Disla39)}. Transparency about the rationale for and anticipated benefits of the tax – health benefit, revenue generation, countering manipulation of consumers by sugary drink manufacturers^{(Reference Backholer, Vandevijvere and Blake7,Reference Mackay and Huber-Disla39)} – and how tax revenues will be equitably invested to balance the unfair tax burden for low SEP sugary drink consumers are critical considerations for assessing the degree to which a sugary drink tax is just ^{(3,Reference Petimar, Gibson and Roberto40)} . For example, the seven US cities with SSB taxes raise $134 million annually from SSB taxes and contribute 85 % of revenues towards supporting projects and programmes in impacted communities, for example, early childhood programmes ($58 million) and community improvements including recreation centres and libraries ($21 M), and increasing healthy food and beverage access ($17 million)^{(Reference Krieger, Magee and Hennings41)}. Considerations for the design, implementation and evaluation of a just sugary drink tax go beyond ‘to tax or not to tax’ and require insights from multiple disciplines and stakeholders, including constructive collaboration between finance and health policymakers^{(Reference Ng, Colchero and White42)}.

Sugary drink taxation policies are an important first step for reducing sugar consumption and related adverse health conditions. However, no single policy will reduce sugary drink consumption to healthy levels and additional interventions should be considered^{(Reference Valizadeh, Popkin and Ng43,Reference Andreyeva, Marple and Moore44)} . A recent Canadian study estimated that targeting SSB with taxation would be insufficient on its own to effectively reduce the health and economic burden of chronic diseases associated with excess sugar consumption from additional products^{(Reference Liu, Munasinghe and Maximova13)}. Moreover, it is imperative to consider taxation in the context of an embodied policy approach, with additional considerations for improving access (i.e. availability and affordability) to safe nutrient-dense items^{(Reference Valizadeh, Popkin and Ng43)} and reducing health inequities associated with material deprivation^{(Reference Kirkpatrick, Dodd and Parsons45)}. A recent systematic review found that food subsidy programmes are associated with increased purchasing of subsidised goods; however, there was uncertainty about associated shifts in consumption^{(Reference Andreyeva, Marple and Moore44)}. Moreover, food subsidy programmes^{(Reference Niebylski, Redburn and Duhaney46)} have existed for decades but are inadequate in alleviating material deprivation (food insecurity) in Canada^{(Reference Kirkpatrick, Dodd and Parsons45)}, wherein 2022, 6·9 million Canadians in the ten provinces lived in food-insecure households, with further inequities across geography, race/ethnicity and income⁽⁴⁷⁾. Recent economic shifts associated with reductions in food access (among other necessary goods and services)⁽⁴⁸⁾ reflect a timely need for broad policy action. Comprehensive monetary policies such as guaranteed annual income may provide autonomy in regard to accessing healthy and safe food options and reduce barriers associated with targeted food subsidy programmes, such as enrolment criteria and longevity^{(Reference Niebylski, Redburn and Duhaney46)}. Further research on this topic should be considered in Canada.

This study is not without limitations. Social desirability bias in self-reported data may have attenuated inequities towards the null as lower compared with higher SEP groups may be more likely to underreport the unfavourable consumption of sugary drinks^{(Reference Garriguet49)}. However, our previous analysis of sugary drink consumption in the CCHS-N found minimal impact on outcomes when adjusting for energy misreporting^{(Reference Warren, Hobin and Manuel12)}. We utilised single-day 24-h dietary recalls which are prone to random error associated with the within-person variation. However, single-day 24-h dietary recalls are sufficient for estimating mean intake on a given day^{(Reference Garriguet11)}. Our analyses required several assumptions for modelling inputs. Price elasticity estimates vary across modelling studies^{(Reference Andreyeva, Marple and Marinello5,Reference Kao, Jones and Ohinmaa17,Reference Teng, Jones and Mizdrak28)} . For example, a recent systematic review estimated average SSB price elasticity estimates of –1·00 (95 % CI: –1·47, –0·50)^{(Reference Teng, Jones and Mizdrak28)}, while income-specific price elasticities utilised in previous Canadian research ranged from –0·87 to –0·92^{(Reference Kao, Jones and Ohinmaa17)}. We leveraged the most recent meta-analysis which estimated a larger price elasticity of –1·39 from high-quality studies^{(Reference Andreyeva, Marple and Moore44)}. Despite the variability in price elasticities, our estimates align with other studies^{(Reference Kao, Jones and Ohinmaa17)}. We assumed consistent price elasticities for all sugary drinks, and the tax effect on different beverage types (e.g. soft drinks v. 100 % juice) may differ overall and across equity stratifiers. However, due to the heterogeneity of beverage options included in modelling price elasticity estimates, we assumed that the average impact represents an appropriate measure for our study. For the tiered taxation scenario, we assumed that cross-price elasticities would be reflected by a 20 % price increase which may not directly reflect the replacement related to low and high sugar beverage products.

Our study has several strengths. We leveraged the most recent population-representative nutrition survey, which contains the only data on consumption, to estimate the impact of sugary drink taxation on intake in the population and across equity stratifiers. Our comprehensive list of sugary drinks included traditional and novel types, including 100 % juice^{(Reference Czoli, Jones and Hammond50)}. We estimated the impact of two taxation structures in Canada, both of which have been implemented in real-world policies^{(Reference Backholer, Vandevijvere and Blake7)}. We tested the robustness of including both income-specific (primary analysis) and homogenous price elasticity estimates and found similar impacts of both taxation scenario across equity stratifiers. Finally, our socio-demographic-specific trends highlight the importance of population-level interventions to consider differential impacts across important equity stratifiers.