Modelled health benefits of a sugar-sweetened beverage tax across different socioeconomic groups in Australia: A cost-effectiveness and equity analysis

Abstract

Background

A sugar-sweetened beverage (SSB) tax in Mexico has been effective in reducing consumption of SSBs, with larger decreases for low-income households. The health and financial effects across socioeconomic groups are important considerations for policy-makers. From a societal perspective, we assessed the potential cost-effectiveness, health gains, and financial impacts by socioeconomic position (SEP) of a 20% SSB tax for Australia.

Methods and findings

Australia-specific price elasticities were used to predict decreases in SSB consumption for each Socio-Economic Indexes for Areas (SEIFA) quintile. Changes in body mass index (BMI) were based on SSB consumption, BMI from the Australian Health Survey 2011–12, and energy balance equations. Markov cohort models were used to estimate the health impact for the Australian population, taking into account obesity-related diseases. Health-adjusted life years (HALYs) gained, healthcare costs saved, and out-of-pocket costs were estimated for each SEIFA quintile. Loss of economic welfare was calculated as the amount of deadweight loss in excess of taxation revenue. A 20% SSB tax would lead to HALY gains of 175,300 (95% CI: 68,700; 277,800) and healthcare cost savings of AU$1,733 million (m) (95% CI: $650m; $2,744m) over the lifetime of the population, with 49.5% of the total health gains accruing to the 2 lowest quintiles. We estimated the increase in annual expenditure on SSBs to be AU$35.40/capita (0.54% of expenditure on food and non-alcoholic drinks) in the lowest SEIFA quintile, a difference of AU$3.80/capita (0.32%) compared to the highest quintile. Annual tax revenue was estimated at AU$642.9m (95% CI: $348.2m; $1,117.2m). The main limitations of this study, as with all simulation models, is that the results represent only the best estimate of a potential effect in the absence of stronger direct evidence.

Conclusions

This study demonstrates that from a 20% tax on SSBs, the most HALYs gained and healthcare costs saved would accrue to the most disadvantaged quintiles in Australia. Whilst those in more disadvantaged areas would pay more SSB tax, the difference between areas is small. The equity of the tax could be further improved if the tax revenue were used to fund initiatives benefiting those with greater disadvantage.

Abstract

Anita Lal and colleagues model for and reveal the potential health benefits and cost savings from a sugar sweetened beverage tax in Australia.

Author summary

Why was this study done?

• Previous real-world evaluations of a sugar-sweetened beverage (SSB) tax showed that the SSB tax led to a reduction of SSB purchases for the total population, with larger effects for lower-income households.

• It was unknown what the healthcare cost savings, health gains, and financial impacts of an SSB tax would be for different income groups, in Australia or internationally.

What did the researchers do and find?

• We modelled the effect of a 20% SSB tax in Australia on life expectancy and health-adjusted life years before and after implementation of the tax, across quintiles of area-level socioeconomic deprivation.

• Our model predicts that the greatest health gains would accrue to the 2 lowest quintiles (most disadvantaged), leading to the highest healthcare cost savings in these quintiles.

• We estimate the increase in annual expenditure on SSBs to be AU$35 per capita in the lowest quintile, a difference of less than $5 compared to the highest quintile.

• Annual tax revenue was estimated at over AU$640 million.

What do these findings mean?

• A 20% SSB tax in Australia is likely to decrease SSB purchase and consumption, leading to significant health gains and healthcare expenditure savings across all quintiles of socioeconomic deprivation.

• The tax would generate considerable yearly revenue, which the government could use to further improve the health of the most disadvantaged.

• As with all simulation models, the model results represent the best estimate of a potential effect in the absence of stronger direct evidence.