Did the Philadelphia Beverage Tax Reduce Childhood Obesity?

This study examines whether the 2017 Philadelphia beverage tax affected pediatric weight outcomes. Using electronic health records from a pediatric health system, researchers analyzed data from 136,078 youth in a panel study and 258,584 youth in a cross-sectional study, comparing weight changes in Philadelphia to nearby control regions. The tax led to no significant changes in standardized BMI (zBMI) or obesity prevalence after two years. The observed zBMI difference was -0.004 (95% CI, -0.009 to 0.001), and the odds ratio for obesity was 1.02 (95% CI, 0.97-1.08), indicating minimal and clinically insignificant effects.

Full Citation and Link to Article

Gregory, E. F., Roberto, C. A., Mitra, N., Edmondson, E. K., Petimar, J., Block, J. P., Hettinger, G., & Gibson, L. A. (2024). The Philadelphia beverage tax and pediatric weight outcomes. JAMA Pediatrics. https://doi.org/10.1001/jamapediatrics.2024.4782

Extended Summary

Central Research Question

The study investigates whether the 2017 Philadelphia beverage tax led to changes in pediatric weight outcomes. Specifically, it asks: Did the implementation of the beverage tax reduce standardized body mass index (zBMI) and obesity prevalence among children and adolescents in Philadelphia compared to youth in nearby regions without the tax? While prior research has examined the effects of sweetened beverage taxes on sales and consumption, the impact on actual weight outcomes in children remains unclear. This study aims to fill that gap by evaluating whether Philadelphia’s tax, which raised beverage prices and reduced sales, had measurable health effects.

Previous Literature

Consumption of sugar-sweetened beverages (SSBs) is strongly linked to childhood obesity and other adverse health outcomes, such as metabolic disorders and dental issues. Studies indicate that reducing SSB intake can lower obesity rates, particularly among youth, who often consume excessive added sugars. Policies targeting SSBs, such as taxes, have been implemented in various countries and cities, with mixed results.

Research on SSB taxes suggests they lead to reduced purchases. For instance, previous studies have found that Philadelphia’s beverage tax caused a 30% price increase and a 25-35% decline in taxed beverage sales. However, whether these changes translate to reduced BMI is unclear.

Prior studies on beverage taxes and weight outcomes have produced inconsistent results. A Seattle-based study found modest reductions in BMI for children, while a Mexican study observed a decline in overweight prevalence among adolescent girls. In contrast, studies in Mauritius and other U.S. cities found no effect on BMI. The mixed findings suggest that factors such as socioeconomic status, age, race, and beverage substitution behaviors may influence the effectiveness of SSB taxes.

This study builds on prior research by using a robust dataset and statistical approach to analyze whether Philadelphia’s beverage tax produced measurable improvements in pediatric weight outcomes.

Data

The study uses electronic health records (EHRs) from a pediatric health system that operates 28 primary care offices in the Philadelphia region. The dataset includes weight and height measurements of children aged 2 to 18 years before and after the tax.

Two datasets were analyzed:

Panel Analysis (136,078 youth): Includes children with at least one BMI measurement before the tax (2014-2016) and at least one after the tax (2018-2019), allowing for longitudinal comparisons.
Cross-Sectional Analysis (258,584 youth): Includes all children with at least one BMI measurement between 2014 and 2019, enabling broader population analysis.

Participants were categorized into two groups: children living in Philadelphia (subject to the tax) and children in nearby Pennsylvania and New Jersey counties (control group). The study excluded children with complex medical conditions, those who moved between regions, and those with missing demographic data.

Key variables include zBMI (standardized BMI adjusted for age and sex) and obesity prevalence (BMI at or above the 95th percentile for age and sex). Covariates such as age, sex, race, ethnicity, Medicaid insurance status, healthcare usage, and socioeconomic status were included to account for potential confounders.

Methods

The study employs a difference-in-differences (DiD) approach, a statistical method commonly used to assess policy impacts by comparing changes over time between a treatment group (Philadelphia) and a control group (nearby counties). To improve comparability, the study applies inverse probability of treatment weights (IPTW), a technique that balances covariate distributions between groups.

Two types of regression models were used:

Mixed-Effects Linear Models: To estimate changes in zBMI before and after the tax, accounting for individual and neighborhood-level factors.
Logistic Regression Models: To assess changes in obesity prevalence (percentage of children with BMI ≥95th percentile).

The study also conducted subgroup analyses by age, race, sex, Medicaid status, and baseline weight status to determine whether the tax had differential effects across demographic groups. Sensitivity analyses examined alternative model specifications and excluded certain time periods to ensure robustness.

Findings and Size Effects

The study found no significant overall association between the Philadelphia beverage tax and pediatric weight outcomes two years post-implementation.

Primary Results

zBMI Change: The difference in zBMI change between Philadelphia and the control group was -0.004 (95% CI, -0.009 to 0.001) in both panel and cross-sectional analyses, indicating no meaningful reduction in standardized BMI.
Obesity Prevalence: The odds ratio for having a BMI ≥95th percentile was 1.02 (95% CI, 0.97-1.08) in the panel analysis and 1.01 (95% CI, 0.95-1.07) in the cross-sectional analysis, showing no significant effect on obesity rates.

Subgroup Findings

Race: White children in Philadelphia experienced a small but statistically significant reduction in zBMI compared to White children in the control group (panel: -0.034; cross-sectional: -0.018). No significant differences were observed for Black children.
Age: In the cross-sectional analysis, the tax was associated with a small decrease in zBMI for adolescents aged 13-18 (-0.030) but a small increase in zBMI for children aged 6-12 (0.027).
Baseline Weight Status: Children who were already overweight or obese saw a slight reduction in zBMI in the panel analysis (-0.020).
Medicaid Status: Medicaid-insured children had a small reduction in zBMI (-0.012), while those not insured by Medicaid had a slight increase (0.006).

Overall, while some small effects were observed in subgroups, they were inconsistent across analyses and of low clinical significance. The largest statistically significant effect (a zBMI change of -0.034 for White youth) corresponds to a weight difference of less than one pound over two years, highlighting the minimal impact of the tax on individual weight trajectories.

Conclusion

The Philadelphia beverage tax successfully reduced taxed beverage sales, but this study finds no meaningful impact on pediatric weight outcomes within two years. These results align with prior research suggesting that while SSB taxes reduce consumption, their effect on BMI is minimal or highly context-dependent.

Several factors may explain the lack of impact:

Substitution Effects: Children may have replaced taxed beverages with other high-calorie drinks, such as 100% fruit juice or untaxed powdered drinks.
Cross-Border Shopping: Philadelphia residents may have purchased SSBs in nearby untaxed areas, diluting the effect of the tax.
Time Frame: Two years may not be enough to observe significant weight changes, especially if children maintained overall caloric intake despite reduced SSB consumption.
Multiple Policy Approaches Needed: Given the complexity of pediatric obesity, SSB taxes alone may be insufficient; comprehensive policies combining taxation with education, access to healthier foods, and other interventions may be necessary.

Despite the lack of observed weight effects, the tax’s broader health and economic implications remain important. The revenue from the tax funds prekindergarten programs and public infrastructure, which may have long-term benefits for children’s well-being. Future research should examine the cumulative effects of beverage taxes over longer periods and in combination with other public health initiatives.

This study underscores the need for a multi-faceted approach to tackling childhood obesity and suggests that while beverage taxes can reduce sugar consumption, their direct impact on weight outcomes may be limited.