Taxed and untaxed beverage intake by South African young adults after a national sugar-sweetened beverage tax: A before-and-after study

Michael Essman; Lindsey Smith Taillie; Tamryn Frank; Shu Wen Ng; Barry M Popkin; Elizabeth C Swart

doi:10.1371/journal.pmed.1003574

. 2021 May 25;18(5):e1003574. doi: 10.1371/journal.pmed.1003574

Taxed and untaxed beverage intake by South African young adults after a national sugar-sweetened beverage tax: A before-and-after study

Michael Essman ^1,², Lindsey Smith Taillie ^1,², Tamryn Frank ³, Shu Wen Ng ^1,², Barry M Popkin ^1,², Elizabeth C Swart ^3,^*

Editor: Sanjay Basu⁴

PMCID: PMC8148332 PMID: 34032809

Abstract

Background

In an effort to prevent and reduce the prevalence rate of people with obesity and diabetes, South Africa implemented a sugar-content-based tax called the Health Promotion Levy in April 2018, one of the first sugar-sweetened beverage (SSB) taxes to be based on each gram of sugar (beyond 4 g/100 ml). This before-and-after study estimated changes in taxed and untaxed beverage intake 1 year after the tax, examining separately, to our knowledge for the first time, the role of reformulation distinct from behavioral changes in SSB intake.

Methods and findings

We collected single-day 24-hour dietary recalls from repeat cross-sectional surveys of adults aged 18–39 years in Langa, South Africa. Participants were recruited in February–March 2018 (pre-tax, n = 2,459) and February–March 2019 (post-tax, n = 2,489) using door-to-door sampling. We developed time-specific food composition tables (FCTs) for South African beverages before and after the tax, linked with the diet recalls. By linking pre-tax FCTs only to dietary intake data collected in the pre-tax and post-tax periods, we calculated changes in beverage intake due to behavioral change, assuming no reformulation. Next, we repeated the analysis using an updated FCT in the post-tax period to capture the marginal effect of reformulation. We estimated beverage intake using a 2-part model that takes into consideration the biases in using ordinary least squares or other continuous variable approaches with many individuals with zero intake. First, a probit model was used to estimate the probability of consuming the specific beverage category. Then, conditional on a positive outcome, a generalized linear model with a log-link was used to estimate the continuous amount of beverage consumed. Among taxed beverages, sugar intake decreased significantly (p < 0.0001) from 28.8 g/capita/day (95% CI 27.3–30.4) pre-tax to 19.8 (95% CI 18.5–21.1) post-tax. Energy intake decreased (p < 0.0001) from 121 kcal/capita/day (95% CI 114–127) pre-tax to 82 (95% CI 76–87) post-tax. Volume intake decreased (p < 0.0001) from 315 ml/capita/day (95% CI 297–332) pre-tax to 198 (95% CI 185–211) post-tax. Among untaxed beverages, sugar intake increased (p < 0.0001) by 5.3 g/capita/day (95% CI 3.7 to 6.9), and energy intake increased (p < 0.0001) by 29 kcal/capita/day (95% CI 19 to 39). Among total beverages, sugar intake decreased significantly (p = 0.004) by 3.7 (95% CI −6.2 to −1.2) g/capita/day. Behavioral change accounted for reductions of 24% in energy, 22% in sugar, and 23% in volume, while reformulation accounted for additional reductions of 8% in energy, 9% in sugar, and 14% in volume from taxed beverages. The key limitations of this study are an inability to make causal claims due to repeat cross-sectional data collection, and that the magnitude of reduction in taxed beverage intake may not be generalizable to higher income populations.

Conclusions

Using a large sample of a high-consuming, low-income population, we found large reductions in taxed beverage intake, separating the components of behavioral change from reformulation. This reduction was partially compensated by an increase in sugar and energy from untaxed beverages. Because policies such as taxes can incentivize reformulation, our use of an up-to-date FCT that reflects a rapidly changing food supply is novel and important for evaluating policy effects on intake.

Rina Swart and co-authors use real world data to demonstrate that sugar-sweetened beverage tax was associated with reduced sugary drink intake in a low-income population within a middle-income country.

Author summary

Why was this study done?

In 2018, South Africa became the first sub-Saharan African country to implement a sugary beverage tax to reduce consumption of added sugars.
Sugar-sweetened beverage (SSB) taxes have been shown to reduce purchases, but few studies have included dietary intake, and they are focused in higher income countries like the United States. To our knowledge, this is the first study to examine the effects of an SSB tax using detailed 24-hour dietary recall data focused on a low-income population from the Global South.
To our knowledge, this is the first study to evaluate real-world changes from an SSB tax policy by empirically quantifying how much of the overall change in sugar intake from taxed beverages came from consumers’ behavioral changes versus reformulation of beverages.

What did the researchers do and find?

Using dietary intake data collected from a low-income South African township before and after the SSB tax was implemented, we examined changes in sugar, calorie, and volume intake from taxed and untaxed beverages.
We used detailed brand-product-specific nutrition facts panel data collected at baseline and a year after tax implementation to allow us to examine both behavioral change and reformulation.
We found the intake of sugar, calories, and volume of taxed beverages decreased by 9.0 g (31%), 39 kcal (33%), and 117 ml (37%) per capita per day, respectively. The intake of sugar, calories, and volume of untaxed beverages increased by 5.3 g (36%), 30 kcal (29%), and 339 ml (58%) per capita per day, respectively. Water accounted for the majority (52%) of the pre–post difference in volume of untaxed beverage intake.
Of the 9.0 g per capita per day (31%) total reduction in sugar from taxed beverages, 6.4 g per capita per day (22%) was due to behavioral differences. When accounting for reformulation, there was an additional 2.6 g per capita per day (9%) reduction in the post-tax period.

What do these findings mean?

SSB taxes were associated with reduced sugary drink intake in a low-income population within a middle-income country.
We quantified the observed changes in beverage intake due to reformulation and behavior change, thereby examining the 2 primary goals of an SSB tax based on sugar concentration.
Future research will be needed to understand how responses to sugar-based beverage taxes may vary by socioeconomic status.

Introduction

Consumption of sugar-sweetened beverages (SSBs) is increasing particularly rapidly in low- and middle-income countries [1]. Given the major impact of SSBs on obesity and many key noncommunicable diseases, and building upon evidence that national SSB taxes reduced SSB purchases in Mexico and Chile in 2014, over 40 countries have now implemented a national SSB tax or increased an existing tax in the last decade [2–9]. South Africa, with one of the highest SSB consumption rates in Africa and a growing prevalence of type 2 diabetes, is the first sub-Saharan African country to institute a sugary beverage tax [10], implemented in April 2018.

The South African tax, called the Health Promotion Levy (HPL), is one of the first SSB taxes to be based on sugar content, applying a fixed 2.1-cent tax rate for every gram of sugar (both intrinsic and added) above a 4 g/100 ml threshold [11]. Early calculations suggest that the average tax rate is approximately 10%. A similar threshold-based multi-tiered SSB levy has been passed in the United Kingdom, but the HPL structure is potentially stronger as each additional gram per 100 ml imposes a greater tax [12]. Taxes based on sugar concentration may lead to greater impact on health outcomes than volume-based taxes, both by reducing the purchases of SSBs and by encouraging product reformulation to lessen the tax burden, thereby reducing excessive added sugars [13,14].

Heretofore, no major national tax evaluation has examined changes in dietary intake of taxed beverages as a key outcome. This study is unique in using detailed dietary intake data on a sample of low-income young adults in South Africa. This is important because two-thirds of cardiovascular deaths occur in middle- and high-income countries, and within those countries it is the lowest income communities with the highest risk [15]. To our knowledge, this study is also the first SSB tax evaluation of a national tax to use detailed 24-hour dietary recall data. Previous work estimated crude store intercepts and used SSB frequency questionnaires for small city evaluations, but these methods miss other sources of consumption and do not represent key segments of a country [16–20]. Other studies have used high-quality household purchase data or cruder aggregate sales data [6,7,21,22]. While these data are important, they exclude many sources of SSBs that dietary measures overcome. For example, collecting purchase information through exit interviews at stores could miss other locations where participants shopped. Food purchase studies can also miss incidental beverage purchases and underestimate the total consumption. Another source of consumption data, aggregate sales from Euromonitor International, can miss the sales from vending and smaller nonchain stores. In contrast, individual dietary intake data record what each individual actually consumes, regardless of where it was purchased. Although 24-hour recall data are susceptible to both random and systematic errors, they are considered by the National Cancer Institute the least biased self-report dietary intake instruments and the most appropriate for evaluating the effects of an intervention [23].

The objectives of our study are to estimate differences in total sugar, energy, and volume from taxed (>4 g sugar/100 ml ready-to-drink), untaxed (≤4 g sugar/100 ml ready-to-drink), and total beverages using 24-hour dietary recall data from the Langa township of South Africa before and 1 year after the HPL. A novel contribution of our study to the SSB tax literature involves linking updated food composition tables (FCTs) with dietary recall data. We developed time-specific FCTs for South African beverages before and after the HPL, which are linked with the dietary assessment instruments. This allows us to examine separately for the first time the role of reformulation as distinct from behavioral changes in SSB intake.

Methods

This study was part of a student dissertation and not registered in a public repository. However, the original proposal of the study design and analytical plan from the dissertation proposal can be found in S1 Protocol. This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 STROBE Checklist).

Data sources and measures

To evaluate the HPL, we analyzed single-day 24-hour dietary recalls from repeat cross-sectional surveys of young township adults aged 18–39 years living in the lower income Langa township near Cape Town, South Africa (S1 Survey). Our study population in the Langa township was selected due to the stability of the community for repeated data collections across time and because it contains a large number of young adults who are heavy consumers of SSBs. At last count, Langa had 17,402 households and 52,401 inhabitants (50.4% female), of whom 99.1% were of Black African race [24].

Data were collected in February–March 2018 (pre-tax, 2 months before the SSB tax implementation; n = 2,481) and a post-tax survey 12 months later in February–March 2019 (n = 2,507) to measure differences in beverage consumption following the HPL (Fig 1). Twenty-two diet records in the pre-tax group (0.9%) and 18 diet records in the post-tax group (0.7%) were dropped for reporting less than 400 daily kcal. Thus, 2,459 and 2,489 diet recalls were included in the final analysis for pre-tax and post-tax, respectively (Table 1). Because the survey was designed to capture young adults’ SSB intake, the only eligibility criterion was being 18–39 years of age. Participants were recruited using a door-to-door sampling method of all identifiable households in Langa until the target sample size of approximately 2,500 households was achieved at each collection period. Fieldworkers approached every household in Langa to ascertain if a household member met the entry requirement (aged 18–39 years) and if that individual was willing to participate. Only 1 diet assessment was completed for 1 individual within a given household. Where 2 qualifying participants were present in the household, the first qualifying participant was selected if the household number in the survey was an uneven number, and the second participant was selected if the household number was an even number. If 3 or more qualifying participants were present in the household, a random numbers list was used to select the respondent. The household questionnaire was conducted digitally and included a geolocation. The geolocations provided us with maps to ensure that we covered all areas. Participants received a supermarket voucher worth R30 (US$2.19) after participating. At the post-tax data collection, participants were asked whether they had been previously surveyed. However, the data are not longitudinal because individuals’ diet surveys cannot be linked across time.

Table 1. Sociodemographic information for Langa survey of adults aged 18–39 years.

Variable	Pre-tax (n = 2,459)	Post-tax (n = 2,489)
Male	34.8%	34.8%
Female	65.2%	65.2%
LSM category¹
LSM 3	1.2%	1.6%
LSM 4	13.1%	19.1%^*
LSM 5	34.4%	49.7%^*
LSM 6	34.9%	27.2%^*
Missing/incomplete data	16.4%	2.5
Age (years)	27.9 (6.0)	27.8 (6.2)
Male BMI (kg/m²)	22.8 (4.4)	22.6 (4.1)
Female BMI (kg/m²)	29.6 (7.3)	30.3 (7.5)

Open in a new tab

Data given as percent or mean (SD).

¹South African Living Standards Measure (LSM) [30].

*Indicates statistically significant difference (p < 0.001) from pre-tax using Fisher exact test.

To record anthropometry, fieldworkers used standardized scales and stadiometers to record the weight and height of each participant after the diet recall was completed, measuring each twice. Body mass index (BMI) was calculated by dividing weight in kilograms by the square of height in meters, using the average of the 2 measurements.

Measuring dietary intake

For the diet assessment, 24-hour diet recalls were conducted by interviewers with nutrition training. Participants reported what foods and drinks were eaten, how foods and beverages were prepared, whether anything was added, and the quantity consumed. A multiple-pass approach, including detailed prompting, was used to enhance completeness.

Linking FCTs to beverage categories

We created composite nutritional records for beverages based on the current food supply and consumer purchases. First, nutrition facts panel (NFP) data were collected from South African grocery stores in February and March 2018. This was repeated exactly a year later in February and March 2019. The stores were sampled in 2 ways. First, we obtained permission and collected all NFP data from beverages at all major chains functioning in the Cape Town region. Second, we visited small stores in the sample township area of Langa to ensure we collected any beverages missed in the chain stores that may have been consumed by our sample population. Products from each round of NFP data collection were linked to a database of 114 beverage codes for creating a South Africa FCT. Fieldworkers who coded the 24-hour recalls created codes by brand name for each SSB, allowing linkages between NFP data and dietary intake data. Fieldworkers always prompted whether beverages were diet or regular beverage types to ensure reported beverages were correctly placed in the beverage code that was above or below the 4 g/100 ml taxation threshold. Each beverage code was given an average nutrient profile, weighted by household purchase data from Kantar Worldpanel, a panel dataset of household packaged food and beverage purchases. The pre-tax beverage FCT was linked to 2018 NFP–Kantar data, and the updated post-tax beverage FCT was linked to 2019 NFP–Kantar data. Each FCT beverage code was also categorized by taxation status. Beverage taxation status was determined by a 2-step process: (1) whether the product category is taxable, as 100% fruit juice and unsweetened milks are exempt from the HPL, and (2) among all other beverages that are taxable, those with a total sugar concentration greater than 4 g/100 ml are classified as taxed and those with 4 g/100 ml or less are untaxed. We considered all tax-exempt beverages and beverages with total sugar ≤ 4 g/100 ml as untaxed. Beverage categories were ultimately analyzed as either taxed or untaxed according to the beverage grouping system (S1 Table).

Analytical approach

All analyses were conducted in Stata, version 16 [25]. Our key outcome variables include intake of total sugar (grams), energy (kilocalories), and volume (milliliters) for total beverages, taxed beverages, untaxed beverages, and subcategories of taxed and untaxed beverages. Intake estimates for these beverage categories were made for the pre-tax and post-tax collection periods. There were few deviations from the initial analysis plan outlined in S1 Protocol. For this study, we added total beverage intake as an outcome to demonstrate the total effect of the policy on beverage intake, not just taxed and untaxed beverage intake alone. We do not report results for beverage subcategories, to maintain focus and clarity on changes in taxed, untaxed, and total beverage intake after the HPL.

Estimation of beverage intake

We estimated beverage intake from 24-hour recalls using a 2-part model implemented with the Stata twopm command to account for beverage groups that have a high percentage of non-consumers, meaning study participants who did not report a particular beverage category consumed [26]. We used a probit model for the first part (likelihood of consumption), and conditional on consumption, we used a generalized linear model with log-link, which gives unbiased estimates of amount consumed, for the second part [27]. The gamma distribution with a log-link is appropriate for data that are continuous, positive, and right-skewed, and these assumptions fit our intake data. Primary outcomes are reported with 95% confidence intervals, and statistically significant differences between groups were calculated using a Wald test [28]. The main comparisons were whether the predicted mean intakes of sugar, energy, and volume were different in the post-tax period compared to the pre-tax period. Models were adjusted for age (continuous, range 18–39), sex, weekday versus weekend of intake (binary), average daily temperature (obtained from National Centers for Environmental Information [29]), and socioeconomic status using the Living Standards Measure (LSM) of the South African Audience Research Foundation (previously the South African Advertising Research Foundation) [30]. LSM is a standardized composite measure commonly used to segment based on socioeconomic status in South Africa and has been widely used by industry since 1989 [31]. It is based on 29 items, including mostly household assets [30]. Other measures such as income are considered less reliable in South Africa for categorizing socioeconomic status. With a total range from 1 to 10, the range of LSM in our study, from 3 to 6, includes low to middle socioeconomic status in South Africa. LSM 3 and LSM 4 were combined to increase power for comparisons with the lowest group. Models are adjusted for the same covariates in both steps. We did not adjust for intake from foods as we were interested in capturing the effects of the policy on beverage intake, which could include substitution effects.

In our main analyses, pre–post beverage intake comparisons were made using the pre-tax beverage FCT linked to both the pre-tax beverage intake data and the post-tax beverage intake data. Thus, any changes in beverage intake would be due to behavioral change alone, since we effectively assume no reformulation. Next, we analyzed post-tax beverage intake linked to the updated beverage FCT to reflect the nutritional composition of beverages at each time point. This analysis reflects the combined effects of reformulation and behavioral change.

Beverage intake by socioeconomic status

To determine whether our results differed by socioeconomic category (LSM), an interaction term between time of data collection and LSM category was added to adjusted models, and tested for statistical significance using a Wald test.

Sensitivity analyses

We conducted a series of sensitivity analyses to test modeling decisions. We tested whether the 2-part model was less appropriate for modeling total beverage intake, with a percentage of consumers over 90%, comparing results from the 2-part model to a generalized linear model with a log-link alone. We investigated adding BMI as a covariate in our models, in case there were differences in reporting of beverage intake by body mass. We also excluded participants who were present both at pre-tax and post-tax data collection (12.4% of post-tax sample), in case participants who repeated the survey responded differently from those taking it for the first time. A part of our sample data in the post-tax period was collected when a drought was in function and strict water controls were enforced. We tested whether our results were confounded by water shortages by controlling for a variable indicating whether participants changed their beverage consumption due to drought. Finally, to test for the absence of bias due to missing LSM data, we tested whether missingness on LSM depended on the outcome (beverage intake), conditional on the covariates.

Ethical approval

The study protocol, questionnaires, procedures, and informed consent forms were approved by the Biomedical Research Ethics Committee of the University of the Western Cape (#BM17/8/20 and #BM18/6/2) as well as by University of North Carolina at Chapel Hill Institutional Review Board (#18–2028). All participants gave written informed consent to enroll in the study.

Patient and public involvement

Participants and the public were not involved in the design, conduct, reporting, or dissemination plans of our research.

Results

Study population characteristics are presented in Table 1. From pre-tax to post-tax, the percent of respondents in LSM 4 and LSM 5 increased (p < 0.001), and the percent of respondents in the highest category, LSM 6, decreased (p < 0.001). There were no significant differences between the 2 time points for any other sociodemographic characteristics.

Adjusted results

Total effects for sugar, energy, and volume intakes for taxed beverages

Sugar intake from taxed beverages decreased (p < 0.0001) from 28.8 g/capita/day (95% CI 27.3 to 30.4) pre-tax to 19.8 (95% CI 18.5 to 21.1) post-tax, a 31.4% reduction (Table 2). Energy intake from taxed beverages decreased (p < 0.0001) from 121 kcal/capita/day (95% CI 114 to 127) pre-tax to 82 (95% CI 76 to 87) post-tax, a 32.5% reduction (Table 2). Volume intake from taxed beverages decreased (p < 0.0001) from 315 ml/capita/day (95% CI 297 to 332) pre-tax to 198 (95% CI 185 to 211) post-tax, a 37.1% reduction (Table 2). Confidence intervals for absolute differences are reported in Table 2. Sugar, energy, and volume intakes of taxed beverage subcategories are reported in S2–S4 Tables.

Table 2. Model adjusted predicted values for sugar, energy, and volume intakes for taxed, untaxed, and total beverages.

Category	Sugar intake, g/capita/day (95% CI)	Energy intake, kcal/capita/day (95% CI)	Volume intake, ml/capita/day (95% CI)
Taxed beverages
Pre-tax	28.8 (27.3 to 30.4)	121 (114 to 127)	315 (297 to 332)
Post-tax	19.8 (18.5 to 21.1)	82 (76 to 87)	198 (185 to 211)
Absolute difference	−9.1 (−11.2 to −6.9)^**	−39 (−48 to −30)^**	−117 (−139 to −94)^**
Untaxed beverages
Pre-tax	15.0 (13.9 to 16.0)	105 (99 to 112)	587 (563 to 610)
Post-tax	20.3 (19.2 to 21.4)	135 (128 to 141)	926 (899 to 953)
Absolute difference	5.3 (3.7 to 6.9)^**	29 (19 to 39)^**	340 (303 to 376)^**
Total beverages
Pre-tax	43.8 (41.9 to 45.7)	226 (217 to 235)	901 (876 to 927)
Post-tax	40.1 (38.5 to 41.6)	216 (208 to 224)	1,124 (1097 to 1151)
Absolute difference	−3.7 (−6.2 to −1.2)^* p = 0.004	−10 (−23 to 2)	223 (184 to 261)^**

Open in a new tab

From models adjusting for age, sex, weekday versus weekend, average daily temperature, and socioeconomic status (Living Standards Measure). Exact p-values are reported unless p < 0.001.

*Statistically significant difference at the p < 0.01 level.

**Statistically significant difference at the p < 0.0001 level.

Total effects for sugar, energy, and volume intakes for untaxed beverages

Sugar intake from untaxed beverages increased (p < 0.0001) from 15.0 g/capita/day (95% CI 13.9 to 16.0) pre-tax to 20.3 (95% CI 18.2 to 21.4) post-tax, a 35.5% increase. Energy intake from untaxed beverages increased (p < 0.0001) from 105 kcal/capita/day (95% CI 99 to 112) pre-tax to 135 (95% CI 128 to 141) post-tax, a 28.6% increase. Volume intake from untaxed beverages increased (p < 0.0001) from 587 ml/capita/day (95% CI 563 to 610) pre-tax to 926 (95% CI 899 to 953) post-tax (Table 2). Water accounted for the majority (52%) of this pre–post difference in untaxed beverage intake (S4 Table). Sugar, energy, and volume intakes of untaxed beverage subcategories are reported in S2–S4 Tables.

Total effects for sugar, energy, and volume intakes for total beverages

Sugar intake from total beverages significantly decreased (p = 0.004) from 43.8 g/capita/day (95% CI 41.9 to 45.7) pre-tax to 40.1 (95% CI 38.5 to 41.6) post-tax, a 8.4% reduction (Table 2). However, there was no significant change in energy intake comparing pre-tax (226 kcal/capita/day; 95% CI 217 to 235) with post-tax (216 kcal/capita/day; 95% CI 208 to 224; p = 0.1) (Table 2). Volume intake was 223 ml/capita/day (95% CI 184 to 261) greater post-tax, partially driven by the large increase in water intake (S4 Table).

Reformulation effects for sugar, energy, and volume intakes

For taxed beverages, we found greater pre–post differences when accounting for reformulation. Sugar intake from beverages was 28.8 g/capita/day (95% CI 27.3 to 30.4) in the pre-tax period and 22.4 (95% CI 21.1 to 23.8) in the post-tax period (−22.2%). When accounting for reformulation, there was a 9.2% additional reduction compared to the pre-tax period, to 19.8 g/capita/day (95% CI 18.5 to 21.1) in the post-tax period (Fig 2). Energy intake was 121 kcal/capita/day (95% CI 114 to 127) in the pre-tax period and 92 (95% CI 86 to 97) in the post-tax period (−24.1%). When accounting for reformulation, there was an 8.4% additional reduction compared to the pre-tax period, to 82 kcal/capita/day (95% CI 76 to 87) in the post-tax period (Fig 3). Volume intake was 315 ml/capita/day (95% CI 297 to 352) in the pre-tax period and 241 (95% CI 227 to 256) in the post-tax period (−23.3%). When accounting for reformulation, there was a 13.7% additional reduction compared to the pre-tax period, to 198 ml/capita/day (95% CI 185 to 211) in the post-tax period (Fig 4).

Fig 2 — “Post-Tax: Behavior Only” refers to estimates of beverage intake using only the pre-tax food composition table (FCT), which keeps nutrient content constant. “Post-Tax: Reformulation Included” refers to estimates of beverage intake using the post-tax FCT, which accounts for reformulation changes in the nutrient content of beverages. Error bars represent 95% confidence intervals.

Fig 3 — “Post-Tax: Behavior Only” refers to estimates of beverage intake using only the pre-tax food composition table (FCT), which keeps nutrient content constant. “Post-Tax: Reformulation Included” refers to estimates of beverage intake using the post-tax FCT, which accounts for reformulation changes in the nutrient content of beverages. Error bars represent 95% confidence intervals.

Fig 4 — “Post-Tax: Behavior Only” refers to estimates of beverage intake using only the pre-tax food composition table (FCT), which keeps nutrient content constant. “Post-Tax: Reformulation Included” refers to estimates of beverage intake using the post-tax FCT, which accounts for reformulation changes in the nutrient content of beverages. Error bars represent 95% confidence intervals.

For untaxed beverages, energy intake was 105 kcal/capita/day (95% CI 99 to 112) in the pre-tax period and 141 (95% CI 134 to 148) in the post-tax period (+33.9%). When accounting for reformulation, there was less of an increase, to 135 (+27.5%) kcal/capita/day (95% CI 76 to 87) in the post-tax period (Fig 3).

For total beverages, sugar intake was 43.8 g/capita/day (95% CI 41.9 to 45.7) in the pre-tax period and 42.4 (95% CI 40.9 to 44.0) in the post-tax period (−3.2%). When accounting for reformulation, there was a 5.4% additional reduction to 40.1 g/capita/day (95% CI 38.5 to 41.6) in the post-tax period, making the total difference from pre-tax statistically significant (p = 0.004) (Fig 2). Energy from total beverages was 226 kcal/capita/day (95% CI 217 to 235) in the pre-tax period and 233 (95% CI 224 to 241) in the post-tax period. When accounting for reformulation, there was a 4.4% total reduction to 216 kcal/capita/day (95% CI 208 to 224) in the post-tax period, although this reduction was not statistically significant (Fig 3).

Beverage intake by LSM

All socioeconomic groups consumed significantly less sugar, energy, and volume from taxed beverages post-tax compared to pre-tax (S5 Table). For untaxed beverages, all LSM groups significantly increased their intake of sugar, energy, and volume post-tax compared to pre-tax (S5 Table). For total beverages, only LSM 6 had a statistically significant reduction in sugar intake post-tax (S5 Table). There were no differences in the magnitude of change between LSM groups for taxed, untaxed, or total beverage intakes.

Sensitivity analyses

We tested whether the 2-part model was less appropriate for modeling total beverage intake, with a percentage of consumers over 90%. We found a less than 1-kcal difference between results using the 2-part model and solely a generalized linear model with a log-link. The sensitivity analysis including BMI in the beverage intake model estimated an additional reduction of 0.5 g of sugar (2 kcal/capita/day) post-tax compared to pre-tax for taxed beverages. Therefore, to be more conservative in our conclusions, we did not include BMI in the model for our main results. We also tested whether our results were affected by excluding participants who were present at both pre-tax and post-tax data collection (12.4%) and estimated an additional 2 kcal/capita/day reduction when excluding any repeats. Therefore, to be more conservative in our conclusions, we kept the entire sample. When we controlled for whether participants changed their beverage consumption due to drought, the results shifted by less than 1 kcal when estimating energy intake from taxed beverages. We found missingness on LSM did not depend on the outcome (beverage intake). This suggests our intake estimates are not biased due to the missing LSM data.

Discussion

This study evaluated the effects of the South African HPL on taxed, untaxed, and total beverage intake by young adults in a Cape Town township 1 year after the tax was implemented. Our novel contribution to the SSB tax literature is the use of updated FCTs linked to dietary intake data to more closely track changes in the food supply after the tax, thus allowing us to separate the effects of changing consumer behavior and industry reformulation. We found that the HPL was followed by a 9.1 g/capita/day (31.4%) reduction in sugar intake, a 39 kcal/capita/day (32.5%) reduction in energy intake, and a 117 ml/capita per day (37.1%) reduction in volume intake from taxed beverages at 1 year post-implementation. We found a 5.3 g/capita/day increase in sugar intake and a 29 kcal/capita/day increase in energy intake from untaxed beverages that partially compensated for this large reduction in taxed beverage intake. Across all beverages, we found a total reduction of 3.7 g/capita/day (8.4%) in sugar consumption (p = 0.004). Our results indicate overall that the behavior change of these adults was associated with reductions of 22% in taxed beverage sugar intake and 3.2% in total beverage sugar intake compared to pre-tax levels, and our estimate of reformulation was the remainder of the post-tax difference.

To our knowledge, this is the first detailed dietary intake survey to find significant results of a national tax effort on dietary intake [32]. The present study collected 24-hour recalls, which are more suitable for estimating mean intakes in a population than frequency questionnaires, and had a larger sample size than earlier studies [33]. For example, Silver et al. found a 19.8% reduction in volume intake and a 13.3% reduction in caloric intake of SSBs following the SSB tax in Berkeley, CA, US, but these findings lacked precision and did not reach statistical significance, potentially due to low baseline intakes in Berkeley (45 kcal/capita/day) compared to our larger high-consuming population pre-tax (121 kcal/capita/day) [21]. Our study population was also a low-income community, and greater reductions in taxed beverage intake could be due to greater price sensitivity. Results from Mexico have also shown the greatest reductions in SSB purchases among the lowest income groups following a tax [34].

Our key methodological contribution is the ability to separate behavioral change from reformulation effects using time-specific FCTs linked to each dietary intake collection period. Accurate FCTs are necessary to calculate nutrient intakes and total energy from 24-hour recalls and frequency questionnaires [35,36]. A study on the threshold-based SSB tax in the United Kingdom noted the combined effects of behavioral change and reformulation in reducing sugar consumption from beverages [37], but to our knowledge, this is the first study to separately quantify the contribution of each in the context of a real-world policy evaluation of dietary intake. Behavioral change accounted for reductions of 24% in energy, 22% in sugar, and 23% in volume compared to pre-tax, while reformulation accounted for additional reductions of 8% in energy, 9% in sugar, and 14% in volume from taxed beverages. Changes in volume intake due to reformulation should be interpreted as effects of updating sugar concentration that led to changed taxed beverage classification. For example, a beverage from the pre-tax period could become categorized as untaxed if it was reformulated below the 4 g/100 ml threshold.

It is important to note that despite the utility of these new methods, we may still be underestimating the effects of the HPL on taxed beverage intake in our study sample. Recent analyses of the impact of the UK soft drink levy found significant reductions in sugar levels in SSBs between the announcement and implementation of the levy [37,38]. The initial announcement of a plan to tax SSBs was made by Finance Minister Pravin Gordhan during a budget speech in February 2016, over 2 years before the SSB tax was ultimately implemented. Our study did not capture any preemptive reductions in beverage sugar content such as have been suggested in other countries [37,38]. Other data sources such as purchase data will be important for capturing changes in the nutrition content of beverages in the time period between the announcement and implementation of the tax.

Other work used a different decomposition method to capture reformulation of the food supply, although for purchase data [39,40]. Griffith and colleagues used a decomposition method that was useful for measuring changes in the salt density of food purchases [39]. Both Griffith et al. [39] and Spiteri and Soler [40] focus on disaggregating 3 components: reformulation of existing food products by manufacturers, the net effect of the launching/removal of products, and consumer switching between products [39,40]. Because our study is focused on the changes in total beverage intake after the policy, we attempt to disaggregate only the effects of consumer behavior change from the total effects of reformulation. Our study provides a straightforward translation of results in grams of sugar or kilocalories per capita per day.

Our results indicate an increase in sugar, energy, and volume intake of untaxed beverages after the tax directionally consistent with but greater in magnitude than increases in a study in Berkeley that measured changes in both taxed and untaxed beverage consumption [21]. Water was a major component, accounting for 177 ml/capita/day (52%) of the increase in untaxed beverage volume (S3 Table). Part of this effect could be seen as a shift away from taxed to untaxed beverages, as found in other studies [19–21,34]. However, Cape Town, South Africa, experienced a drought and severe water use restrictions from March to September 2018 [41]. Therefore, we cannot disentangle the tax-related effects from the effects of the drought on water consumption.

Analyzing results by socioeconomic status, sugar, energy, and volume intake decreased from taxed beverages and increased from untaxed beverages for all groups (S5 Table). However, there were no differences in absolute changes between groups, likely because the sample as a whole is relatively low income. For comparison, a recent pre–post study using purchase data from a nationally representative South Africa sample found greater reductions in sugar consumption (−32.7%) in LSM 4–6—nearly the identical range to our sample—compared to the higher socioeconomic group of LSM 7–10 (−20.4%) [42].

This study has several limitations. Given our data are cross-sectional, we are not able to follow all individuals over time, only measure differences in population means. Social desirability bias could have affected reporting and caused us to underestimate SSB intake. It is also possible that after the tax, social norms may have shifted so that the effect of social desirability bias is even greater after SSBs are subject to tax, causing an overestimation of reductions in SSB intake in this population. The magnitude of changes in taxed beverage consumption after the HPL in this study may not be generalizable to higher income populations, given that other studies have found the largest changes in sugary beverage intake among the lowest income groups analyzed [34]. However, the methods used in this study are useful for future work as they can separate the 2 crucial components of behavioral change and reformulation, regardless of the total effect of a tax on income groups. Finally, although we examined pre–post differences in beverage intakes under 2 scenarios, first accounting for differences in intake only, followed by accounting for differences in intake plus reformulation effects, we cannot isolate the specific types of reformulation that may have occurred, nor identify any causal effects. For example, changes may be due to reduced sugar content within taxation categories, or products may have switched categorization from taxed to untaxed if they were reformulated below the 4 g/100 ml threshold. To the extent that we did not carefully capture individual beverage heterogeneity within a wider brand-product-specific categorization weighting approach using Kantar data, we might misrepresent some of the nuanced industry changes.

Our study has several strengths, including a large sample of high-consuming young adults to increase study power and the ability to detect changes; the use of dietary intake data, which are a more suitable measure for mean population intakes than frequency questionnaires [33]; the use of a 2-part model for beverage intake [43]; and the development of time-varying FCTs linked with these dietary data to estimate changes in sugar, energy, and volume intake after the HPL.

Conclusion

Using a large sample of a high-consuming, low-income population, we found large reductions in taxed beverage intake, separating the components of behavioral change from reformulation. This reduction was partially compensated by an increase in sugar and energy intake from untaxed beverages. Because policies such as taxes can incentivize reformulation, our use of time-specific beverage FCTs that reflect a rapidly changing food supply is novel and important for evaluating future taxation policies’ impact on dietary intake.

Supporting information

S1 STROBE Checklist

(DOC)

Click here for additional data file.^{(107.5KB, doc)}

S1 Protocol. Study protocol.

(DOCX)

Click here for additional data file.^{(752.1KB, docx)}

S1 Survey. Diet survey.

(PDF)

Click here for additional data file.^{(237.3KB, pdf)}

S1 Table. Beverage classification system.

(DOCX)

Click here for additional data file.^{(23KB, docx)}

S2 Table. Model adjusted predicted intake of total sugar for taxed and untaxed beverage subcategories, Langa adults aged 18–39 years.

(DOCX)

Click here for additional data file.^{(18.6KB, docx)}

S3 Table. Model adjusted predicted intake of energy (kilocalories) for taxed and untaxed beverage subcategories, Langa adults aged 18–39 years.

(DOCX)

Click here for additional data file.^{(18.5KB, docx)}

S4 Table. Model adjusted predicted intake of volume for taxed and untaxed beverage subcategories, Langa adults aged 18–39 years.

(DOCX)

Click here for additional data file.^{(18.4KB, docx)}

S5 Table. Model adjusted predicted intakes of total sugar, energy, and volume from taxed beverages for each LSM category.

(DOCX)

Click here for additional data file.^{(16.3KB, docx)}

Acknowledgments

We wish to thank Karen Ritter for exceptional assistance with the data management, and Jessica Ostrowski and Bridget Hollingsworth for research assistance, Safura Abdool Karim for legal review of the legislation and regulation, and Emily Yoon for administrative assistance. We want to thank Chuma Maqina, Kelly Leman Scott, Sarona Munyai, Sikhumbule Joni, Lamla Sajini, Sisa Honesi Martins, Olwethu Mateta, Siyabulela Mditshwa, Andiswa Magwaza, Abongwe Benge, Ayakha Sigwabe, Masingita Nyalungu, Namhla Baskiti, Andisiwe Siramza, Siphesihle Thotsho, Nomhle Poni, and Ntombizodwa Nako for dietary intake and anthropometry collection, as well as Mawande Nelani, Pamela Magubane, Emmanuel Obasa, Zukiswa Mfinyongo, Ziyanda Lufele, Zikhona Lobola, Asisipho Majongile, Nqabakazi Majeke, Yanga Gadu, Busiswa Mqhayi, Priscilla Nkhensani, Thandokazi Magqaza, Onele Maqambayi, Nonelwa Ngcingane, Tembakazi Msutwana, Sisipho Mabaso, Chumani Matiwane, Akhona Rasmeni, and Sisipho Mngqnge for household questionnaires. We also want to thank Luchrechia Afrika, Vashika Chibba, Thandokazi Mahuzi, Kholiswa Manisi, Tyler Coats, Valentine Khumalo, Aneeqah Latief, Zintle Phekana, Stephanie Röhrs, Sharna Solomons, and Morongoa Tlhako, who conducted nutrition label data collection in 2018 and 2019. For the software used to collect and enter the nutrition label data, we thank the National Center for Advancing Translational Sciences (NIH UL1TR001111). We also thank The George Institute and Discovery Vitality for sharing early nutrition label data. For the use of the South African Food Data System FCTs, we thank the South African Medical Research Council.

Abbreviations

FCT: food composition table
HPL: Health Promotion Levy
LSM: Living Standards Measure
NFP: nutrition facts panel
SSB: sugar-sweetened beverage

Data Availability

Data analyzed for this paper form part of a primary project which is currently being written up in other publications. Data used for this paper will therefore be available upon request and granted for replication purposes. Data are available from the UNC Carolina Digital Repository (https://cdr.lib.unc.edu/). For data inquiries, please contact Donna Miles (drmiles@email.unc.edu) and Jessica Ostrowski (jessica.ostrowski@unc.edu).

Funding Statement

Funding for this study comes from Bloomberg Philanthropies (https://www.bloomberg.org/; received by BP), including Subcontract #5108311 with the University of the Western Cape. This research also received support from the Population Research Infrastructure Program awarded to the Carolina Population Center (P2C HD050924) at The University of North Carolina at Chapel Hill by the Eunice Kennedy Shriver National Institute of Child Health and Human Development. ME was supported by a Predoctoral Fellowship from NIH Training Grant T32 HL129969. Other support includes student scholarships from International Development Research Center (IDRC) Project number 108425-001 and the DST/NRF Centre of Excellence in Food Security project number 180401 (https://www.uwc.ac.za/Faculties/CHS/soph/Pages/The%20Centre-of-Excellence-in-Food-Security.aspx). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1.Popkin BM, Hawkes C. Sweetening of the global diet, particularly beverages: patterns, trends, and policy responses. Lancet Diabetes Endocrinol. 2016;4(2):174–86. 10.1016/S2213-8587(15)00419-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
2.Malik VS, Schulze MB, Hu FB. Intake of sugar-sweetened beverages and weight gain: a systematic review. Am J Clin Nutr. 2006;84(2):274–88. 10.1093/ajcn/84.1.274 [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Te Morenga L, Mallard S, Mann J. Dietary sugars and body weight: Systematic review and meta-analyses of randomised controlled trials and cohort studies. BMJ. 2013;345(7891):e7492. 10.1136/bmj.e7492 [DOI] [PubMed] [Google Scholar]
4.Malik VS, Popkin BM, Bray GA, Despres J-P, Willett WC, Hu FB. Sugar-sweetened beverages and risk of metabolic syndrome and type 2 diabetes: a meta-analysis. Diabetes Care. 2010;33(11):2477–81. 10.2337/dc10-1079 [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Malik VS, Popkin BM, Bray GA, Despres J-P, Hu FB. Sugar-sweetened beverages, obesity, type 2 diabetes mellitus, and cardiovascular disease risk. Circulation. 2010;121(11):1356–64. 10.1161/CIRCULATIONAHA.109.876185 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Arantxa Colchero M, Rivera-Dommarco J, Popkin BM, Ng SW. In Mexico, evidence of sustained consumer response two years after implementing a sugar-sweetened beverage tax. Health Aff (Millwood). 2017;36(3):564–71. 10.1377/hlthaff.2016.1231 [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Caro JC, Corvalán C, Reyes M, Silva A, Popkin B, Taillie LS. Chile’s 2014 sugar-sweetened beverage tax and changes in prices and purchases of sugar-sweetened beverages: an observational study in an urban environment. PLoS Med. 2018;15(7):e1002597. 10.1371/journal.pmed.1002597 [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Global Food Research Program. Resources. Chapel Hill (NC): Global Food Research Program; 2020. [cited 2020 Apr 1]. Available from: http://globalfoodresearchprogram.web.unc.edu/resources/. [Google Scholar]
9.Backholer K, Blake M, Vandevijvere S. Sugar-sweetened beverage taxation: an update on the year that was 2017. Public Health Nutr. 2017;20(18):3219–24. 10.1017/S1368980017003329 [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Statistics South Africa. Mortality and causes of death in South Africa, 2016: findings from death notification. Pretoria: Statistics South Africa; 2018. [cited 2021 Feb 8]. Available from: https://www.statssa.gov.za/publications/P03093/P030932016.pdf. [Google Scholar]
11.South African Revenue Service. Health promotion levy on sugary beverages. Pretoria: South African Revenue Service; 2021. [cited 2021 April 26]. Available from: http://www.sars.gov.za/customs-and-excise/excise/health-promotion-levy-on-sugary-beverages/. [Google Scholar]
12.HM Revenue and Customs. Soft drinks industry levy. London: HM Revenue and Customs; 2016 [cited 2021 Feb 8]. Available from: https://www.gov.uk/government/publications/soft-drinks-industry-levy/soft-drinks-industry-levy.
13.Francis N, Marron DB, Rueben KS. The pros and cons of taxing sweetened beverages based on sugar content. Washington (DC): Urban Institute; 2016. [cited 2021 Apr 22]. Available from: https://www.heart.org/idc/groups/ahaecc-public/@wcm/@adv/documents/downloadable/ucm_490771.pdf. [Google Scholar]
14.Veerman L. The impact of sugared drink taxation and industry response. Lancet Public Health. 2017;2(1):e2–3. 10.1016/S2468-2667(16)30039-1 [DOI] [PubMed] [Google Scholar]
15.World Health Organization. Global status report on noncommunicable diseases 2014. Geneva: World Health Organization; 2014 [cited 2021 Apr 22]. Available from: https://apps.who.int/iris/bitstream/handle/10665/148114/9789241564854_eng.pdf;jsessionid=7DD0AC09B03E54923DCFB3C640D71713?sequence=1. [Google Scholar]
16.Cawley J, Frisvold D, Hill A, Jones D. The impact of the Philadelphia beverage tax on purchases and consumption by adults and children. J Health Econ. 2019;67:102225. 10.1016/j.jhealeco.2019.102225 [DOI] [PubMed] [Google Scholar]
17.Fletcher JM, Frisvold D, Tefft N. Taxing soft drinks and restricting access to vending machines to curb child obesity. Health Aff (Millwood). 2010;29(5):1059–66. 10.1377/hlthaff.2009.0725 [DOI] [PubMed] [Google Scholar]
18.Lee MM, Falbe J, Schillinger D, Basu S, McCulloch CE, Madsen KA. Sugar-sweetened beverage consumption 3 years after the Berkeley, California, sugar-sweetened beverage tax. Am J Public Health. 2019;109(4):637–9. 10.2105/AJPH.2019.304971 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Falbe J, Thompson HR, Becker CM, Rojas N, McCulloch CE, Madsen KA. Impact of the Berkeley excise tax on sugar-sweetened beverage consumption. Am J Public Health. 2016;106(10):1865–71. 10.2105/AJPH.2016.303362 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Zhong Y, Auchincloss AH, Lee BK, Kanter GP. The short-term impacts of the Philadelphia beverage tax on beverage consumption. Am J Prev Med. 2018;55(1):26–34. 10.1016/j.amepre.2018.02.017 [DOI] [PubMed] [Google Scholar]
21.Silver LD, Ng SW, Ryan-Ibarra S, Taillie LS, Induni M, Miles DR, et al. Changes in prices, sales, consumer spending, and beverage consumption one year after a tax on sugar-sweetened beverages in Berkeley, California, US: a before-and-after study. PLoS Med. 2017;14:e1002283. 10.1371/journal.pmed.1002283 [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Fletcher JM, Frisvold DE, Tefft N. The effects of soft drink taxes on child and adolescent consumption and weight outcomes. J Public Econ. 2010;94(11–12):967–74. 10.1016/j.jpubeco.2010.09.005 [DOI] [Google Scholar]
23.Thompson FE, Kirkpatrick SI, Krebs-Smith SM, Reedy J, Schap TE, Subar AF, et al. The National Cancer Institute’s dietary assessment primer: a resource for diet research. J Acad Nutr Diet. 2015;115(12):1986–95. 10.1016/j.jand.2015.08.016 [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Statistics South Africa. Langa. Pretoria: Statistics South Africa; 2020 [cited 2020 May 14]. Available from: http://www.statssa.gov.za/?page_id=4286&id=318.
25.StataCorp. Stata statistical software: release 16. College Station (TX): StataCorp; 2019. [Google Scholar]
26.Belotti F, Deb P, Manning WG, Norton EC. twopm: two-part models. Stata J. 2015;15(1):3–20. [Google Scholar]
27.Manning WG, Mullahy J. Estimating log models: to transform or not to transform? J Health Econ. 2001;20(4):461–94. 10.1016/s0167-6296(01)00086-8 [DOI] [PubMed] [Google Scholar]
28.Amrhein V, Greenland S, McShane B. Scientists rise up against statistical significance. Nature. 2019;567:305–7. 10.1038/d41586-019-00857-9 [DOI] [PubMed] [Google Scholar]
29.National Centers for Environmental Information. Data tools: find a station. Asheville (NC): National Centers for Environmental Information; 2020. [cited 2020 Feb 13]. Available from: https://www.ncdc.noaa.gov/cdo-web/datatools/findstation. [Google Scholar]
30.South African Advertising Research Foundation. The SAARF AMPS Living Standards Measure (LSM). Johannesburg: South African Advertising Research Foundation; 2011. [cited 2019 May 1]. Available from: http://www.saarf.co.za/amps-technicalreport/technicalreport-Jan2011—Dec 2011/data files/Technical/21—Tech 2011B~Pages92-97.pdf. [Google Scholar]
31.South African Audience Research Foundation. Living Standards Measure. Johannesburg: South African Audience Research Foundation; 2017. [cited 2019 Apr 29]. Available from: http://www.saarf.co.za/lsm/lsms.asp. [Google Scholar]
32.Teng AM, Jones AC, Mizdrak A, Signal L, Genç M, Wilson N. Impact of sugar-sweetened beverage taxes on purchases and dietary intake: systematic review and meta-analysis. Obes Rev. 2019;20(9):1187–204. 10.1111/obr.12868 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.National Cancer Institute. Dietary assessment primer: evaluating the effect of an intervention on diet. Bethesda (MD): National Cancer Institute; 2019. [cited 2019 Jan 16]. Available from: https://dietassessmentprimer.cancer.gov/approach/intervention.html. [Google Scholar]
34.Colchero MA, Molina M, Guerrero-López CM. After mexico implemented a tax, purchases of sugar-sweetened beverages decreased and water increased: difference by place of residence, household composition, and income level. J Nutr. 2017;147(8):1552–7. 10.3945/jn.117.251892 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Satija A, Yu E, Willett WC, Hu FB. Understanding nutritional epidemiology and its role in policy. Adv Nutr. 2015;(6):5–18. 10.3945/an.114.007492 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.National Cancer Institute. Dietary assessment primer: learn more about food composition databases for 24-hour dietary recalls and food records. Bethesda (MD): National Cancer Institute; 2019. [cited 2019 Apr 30]. Available from: https://dietassessmentprimer.cancer.gov/learn/recall-record.html. [Google Scholar]
37.Bandy LK, Scarborough P, Harrington RA, Rayner M, Jebb SA. Reductions in sugar sales from soft drinks in the UK from 2015 to 2018. BMC Med. 2020;18(1):20. 10.1186/s12916-019-1477-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Scarborough P, Adhikari V, Richard A. Harrington R, Elhussein A, Briggs A, Rayner M, et al. Impact of the announcement and implementation of the UK Soft Drinks Industry Levy on sugar content, price, product size and number of available soft drinks in the UK, 2015–19: a controlled interrupted time series analysis. PLoS Med. 2020;17(2):e1003025. 10.1371/journal.pmed.1003025 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Griffith R, O’Connell M, Smith K. The importance of product reformulation versus consumer choice in improving diet quality. Economica. 2017;84(333):34–53. [Google Scholar]
40.Spiteri M, Soler LG. Food reformulation and nutritional quality of food consumption: an analysis based on households panel data in France. Eur J Clin Nutr. 2018;72(2):228–35. 10.1038/s41430-017-0044-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Pitt C. City of Cape Town relaxes water restrictions, tariffs to level 5. news24. 2018 Sep 10 [cited 2020 Apr 30]. Available from: https://www.news24.com/SouthAfrica/News/city-of-cape-town-relaxes-water-restrictions-tariffs-to-level-5-20180910.
42.Stacey N, Edoka I, Hofman K, Swart R, Popkin BM, Ng S. Changes in beverage purchases following the announcement and implementation of South Africa’s Health Promotion Levy: an observational study. Lancet Planet Health. 2021;5(4):e200–8. 10.1016/S2542-5196(20)30304-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Haines P, Guilkey D, Popkin B. Modeling food consumption decisions as a two-step process. Am J Agr Econ. 1988;70(3):543–52. [Google Scholar]

PLoS Med. doi: 10.1371/journal.pmed.1003574.r001

Decision Letter 0

Adya Misra

12 Aug 2020

Dear Dr Swart,

Thank you for submitting your manuscript entitled "Taxed and untaxed beverage consumption by young adults in Langa, South Africa before and one year after a national sugar-sweetened beverage tax" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff [as well as by an academic editor with relevant expertise] and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by .

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Adya Misra, PhD,

Senior Editor

PLOS Medicine

PLoS Med. doi: 10.1371/journal.pmed.1003574.r002

Decision Letter 1

Adya Misra

2 Nov 2020

Dear Dr. Swart,

Thank you very much for submitting your manuscript "Taxed and untaxed beverage consumption by young adults in Langa, South Africa before and one year after a national sugar-sweetened beverage tax" (PMEDICINE-D-20-03807R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by Nov 23 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Adya Misra, PhD

Senior Editor

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Please revise your title according to PLOS Medicine's style. Your title must be nondeclarative and not a question. It should begin with main concept if possible. "Effect of" should be used only if causality can be inferred, i.e., for an RCT. Please place the study design ("A randomized controlled trial," "A retrospective study," "A modelling study," etc.) in the subtitle (ie, after a colon).

Abstract: could we perhaps rephrase “growing burdens of obesity and diabetes” I wonder if this is a bit stigmatising and could be revised

Please clarify what you mean here “We estimated beverage intake using a two-part model, with a probit model for the first part and a generalized linear model with log-link for the second part”. Please also outline briefly how you planned to test behavioural versus reformulation effects

The Data Availability Statement (DAS) requires revision. For each data source used in your study:

a) If the data are freely or publicly available, note this and state the location of the data: within the paper, in Supporting Information files, or in a public repository (include the DOI or accession number).

b) If the data are owned by a third party but freely available upon request, please note this and state the owner of the data set and contact information for data requests (web or email address). Note that a study author cannot be the contact person for the data.

c) If the data are not freely available, please describe briefly the ethical, legal, or contractual restriction that prevents you from sharing it. Please also include an appropriate contact (web or email address) for inquiries (again, this cannot be a study author).

At this stage, we ask that you include a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary

Please use square brackets for references

Please add p-values of up to three decimal places, providing exact values throughout unless p<0.001

Please provide copies of questionnaires used or citations if they have been previously published

Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

Table 1- I suggest you add the units for “Age”

Discussion- please start this section with a brief description of what was done and the main findings

I suggest the discussion is tempered, by adding “our results indicate” rather than “we show” due to the observational nature of this study

Please also temper all assertions of primacy by adding “ to our knowledge”

Please remove the funding information from the main text and add this information to the financial disclosure section to the article meta-data

Please ensure that the study is reported according to the STROBE guideline, and include the completed STROBE checklist as Supporting Information. Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

The STROBE guideline can be found here: http://www.equator-network.org/reporting-guidelines/strobe/

When completing the checklist, please use section and paragraph numbers, rather than page numbers.

Comments from the reviewers:

Reviewer #1: This is an excellent manuscript that provides a thorough evaluation of an important population health intervention. The methods that are reported are novel, and the focus on the low income group is important. This paper makes an excellent contribution to the literature.

I have some minor comments:

1. If the pre-tax data were collected two months before the implementation of the tax then some reformulation of soft drinks could have taken place already, which would result in an underestimate of the impact of the tax on sugar consumption. Our analysis of the impact of the UK sugar drink levy on sugar levels in drinks suggested that two months before the implementation of the levy the proportion of drinks with sugar levels over the 5g per 100mL levy threshold had already fallen by about 20 percentage points (Scarborough et al. PLoS Med, 2020, https://doi.org/10.1371/journal.pmed.1003025).

2. Lines 15—167: The description of how food diary data are linked with the nutrient composition tables is not very clear, and would be helped by an example. From my understanding, in the food diary data there was sometimes ambiguity at a brand-level (e.g. "Coca-Cola" could cover regular, cherry, mango, vanilla, etc.), and these multiple options were dealt with by weighting using Kantar World Panel data on consumption. If that is the case, then it would be useful to know whether this ambiguity extended to inclusion of both diet (i.e. zero calorie variants) and regular drinks amongst the same brand, which would be a big limitation.

3. The percentage split of men and women at each timepoint in table 1 are identical. Just checking whether this is a typo!

4. Lines 282-290: This should start "For total beverages" rather than "For untaxed beverages". However, the next bit of text does refer to 'untaxed beverages, so this paragraph needs a little bit of editing.

5. It is not clear to me how reformulation can have affected the volume of drink consumed. The only mechanism that I can think is that the respondents recorded how many units of drink they consumed (e.g. a bottle, a can) and the size of these units changed as a result of the tax. If so, then that should be made clear in the methods. Also, how were these units linked to specific serving sizes (e.g. in the UK, Coca-Cola is available in 250mL and 330mL cans. If a respondent reports consumption of 'a can of Coca-Cola' how do you choose which serving size to assign?)

6. I would like more information about how the nutrition data were collected. The manuscript simply says "First, nutrition facts panel (NFP) data were collected from South African grocery stores in February and March 2018. This was repeated exactly a year later in February and March 2019." How were the stores sampled? How were the products sampled? Was a complete audit conducted, or were only drinks that appeared in the food diary included?

Reviewer: Peter Scarborough

Reviewer #2: Thank you for the opportunity to review "Taxed and untaxed beverage consumption by young adults in Langa, South Africa before and one year after a national sugar-sweetened beverage tax." This manuscript uses a before and after study design to estimate the effect of the South African SSB tax on the consumption of sugar, energy and volume of both taxed and untaxed beverages, in a low-income adult community. The main strength of the paper is the use of detailed 24-hour dietary recalls linked with updated food composition databases for beverage. The paper is very clear, well structured and well written.

In my read, the main limitation of the paper is the pre/post study design, as differences observed could be due to all sorts of time-varying factors, such as differences in the population sampled (either due to sampling process, change within the population, or change of the population, such as rising income), pre-existing decreasing trends in SSB purchases, changes in accessibility of beverages. In addition, the authors mention that there was a drought and severe water use restriction during part of the baseline data collection period. To account for some of these confounding effects, the authors adjust their results for a limited number of variables available, incl. age, day of the week, temperature, and a measure of socio-economic status, yet this does not rule out the presence of residual bias, such that the strength of the policy effect may remain uncertain.

My specific comments and suggestions are as follows

1. Results summary in the abstract and conclusion: in my read, there seems to be a mismatch between what the paper says it intends to do and the results highlighted in the abstract. Both the introduction and the abstract indicate that the study aims to estimate changes in taxed and untaxed beverage intake. However, the only results presented in the abstract highlight the decrease in taxed beverage intake, showing that the tax might have had a positive effect on sugar and energy reduction. However, more than half of the reduction observed is compensated by what appears to be substitution towards untaxed beverages, which have dramatically increased in terms of overall volume consumption. This means that all in all, the SSB tax resulted in a relatively small reduction in sugar intake and an uncertain reduction in total energy from beverages in that population.

2. I would remove the emphasis on the fact that looking at differences between reformulation and behavioural change is a novelty of this paper. Rachel Griffith et al. proposed a more sophisticated decomposition method in a 2014 IFS working paper "The importance of product reformulation versus consumer choice in improving diet quality". The paper makes the distinction between changes due to reformulation, behavioural change and introduction of new products. The method was applied to beverages by Marine Spiteri & Louis-Georges Soler in paper published in the European Journal of Clinical Nutrition in 2017 ("Food reformulation and nutritional quality of food consumption: an analysis based on households panel data in France").

3. Related the above point, how were new beverages introduced after the baseline treated in the analyses?

4. It wasn't very clear to me how the nutritional values were attributed to beverages, in particular to what extent approximations had to be made using weights derived from purchase data. Could the authors provide more information on how detailed the beverage lists in the 24-hour recall was (i.e. how many beverage codes)? And how did the list of beverage codes in the 24-hour recall differ from the product level information available in Kantar? There is potentially some variability within beverage codes, which could affect the results.

5. Could you indicate when the tax was announced in the text and in the timeline. In the UK, there was for example an announcement effect which triggered reformulation before the introduction of the tax. As a result, the total effect of the tax might not be fully captured by the analyses presented in the paper if reformulation started before the baseline data collection. This might deserve to be discussed.

6. Could you explain a bit more on how the door to door sampling was conducted. Was there an element of randomisation for selecting the households to survey? Some important difference in socio-economic status were observed in the pre and post data collection waves. I was wondering whether the sampling method could have been responsible for part of those differences. You might have useful geographical location on the respondents at each wave, which could indicate systematic differences for example. Given the repeated cross-sectional study design, and the fact that few covariates are used in the models, I feel that I would need to be further reassured that the pre and post data collection do not systematically differ.

7. I was wondering why other dietary information were not taken into account in the analyses. Could these for example provide an indication of the extent which the 2 samples systematically differ, which could then bias the results? Could these serve to adjust the results?

8. Following the previous point, I was very curious to see a very large increase in the total volume consumed in the post-tax data collection in the adjusted model (+25%). How could this be explained? Was this also observed in the Kantar data analysed (paper under review cited)? Could this reflect a difference in socio-economic level between the 2 samples? Or rather the drought of 2018, as only briefly mentioned in the discussion. If the latter, this is potentially a massive confounding factor which should be further highlighted

9. Table 1 - could you indicate what the LSM category mean in the table

10. Why LSM was the only socioeconomic indicator used? Were other indicators available?

11. It would be interesting to have an idea of the extent to which adjusted and unadjusted results differ. Maybe not necessarily in the main text.

12. Page 3 line 94: "while these data are important, they exclude many sources of SSBs which dietary measures overcome". Could you give an example of those sources of SSBs.

13. What distributional assumptions are used for the second part of the model (for which a 'log-link' function is specified)? Do the data meet the assumptions?

14. About the 2-part model: the type of model seems to make sense for the analysis of some food groups, which are rarely consumed. However, this might be less the case for total beverage volumes or sugar, or even reformulated products which contain a mixture of sugar and artificial sweeteners. I was wondering if the authors could reassure the reader that the results are not affected by this methodological choice, by reproducing some of the more general results using an alternative modelling strategy, where the 2 part-model does not feel as much justified.

15. Could the authors provide all the coefficients of the fitted model for peer review purposes, and not just the adjusted predicted mean of the outcome.

16. Sensitivity analysis. No change in energy intake when moving LSM from the analysis is no indication of an absence of bias, as both intake values could be biased indeed. To rule out whether results could be biased, the authors should assess whether missingness on LSM depends on the outcome (intake), conditional on the covariates (see e.g. Carpenter and Kenward, Multiple Imputation and its applications)

17. Page 8 line 325-326: could the authors avoid the word 'statistically significant', this only makes sense with reference to a certain level of significance. Also, the sentence seems to indicate that there was strong evidence of energy reduction overall, whereas the 95%CI includes 0, and the p-value (which is not reported) appears to be higher than 0.01.

18. Page 5 line 170 "our key outcome variables include mean adjusted" - I am not sure what the authors mean by this. Outcomes (i.e. variables) are intake, not "mean adjusted" intake (i.e. summary statistic)?

19. Figures 2-4. From the caption and figures it is not very clear what behavioural, and reformulation mean. I would be more explicit so that the figure can be read independently.

20. Why didn't you run the analysis on households who were interviewed at each wave as a sensitivity analysis, as opposed to running it on those who were not. Focusing on the 'longitudinal sample' would ensure that the population is the same, which would rule out the concern that the populations are different between the 2 waves.

Reviewer #3: This is a well-conducted study on beverage consumptions by young adults in Langa, South Africa before and after a national sugar-sweetened beverage tax. The study design, datasets, statistical methods and analyses, and presentation of results (tables and figures) are mostly adequate. However, there are a few issues especially interpretation of results needing attention.

1) The title and the abstract. The title is "Taxed and untaxed beverage consumption by...", however in the abstract, it's all about taxed beverage consumption in the findings and conclusion and nothing about untaxed and total beverage consumption. It's an incomplete summary/finding which doesn't match the title.

2) Interpretation of study results. Table 2 is a key table summarising the main findings of the study. We can see clearly that after introducing the SSB tax there were decreases in sugar, energy and volume consumption in taxed beverage but increases in untaxed beverage, and overall slight or no change in total beverage consumption. So far the paper is mainly focused on the results of taxed beverage consumption changes but largely ignored the untaxed and total beverage results, which is inadequate. To evaluate the effect of SSB tax fairly a holistic approach of presenting and interpreting all the results including taxed, untaxed and total beverage consumptions is needed in order to provide a complete picture of the effectiveness and impact of SSB tax policy. Balanced presentation and interpretation of results are needed in discussion, conclusion and in abstract.

3) The two-part model is key in estimation of beverage intake. Authors said the reason why the model was used is in the first part to account for beverage groups that have a high percentage of non-consumers (likelihood of consumption). Can authors make it explicit on what 'non-consumers' means? Were these young people consumer or non-consumer? What's the percentage and how it was modelled? Unless all these are made clear, readers will have doubts on the validity of using this two-part model.

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. 2021 May 25;18(5):e1003574. doi: 10.1371/journal.pmed.1003574.r003

Author response to Decision Letter 1

24 Nov 2020

Attachment

Submitted filename: PMEDICINE-D-20-03807R1 responses to reviewers final.docx

Click here for additional data file.^{(17.4MB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003574.r004

Decision Letter 2

Raffaella Bosurgi

4 Feb 2021

Dear Dr. Swart,

Thank you very much for re-submitting your manuscript "Changes in taxed and untaxed beverage intake by South African young adults after a national sugar-sweetened beverage tax: a before-and-after study" (PMEDICINE-D-20-03807R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by reviewers below. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

Editorial points:

Thank you for addressing the previously stated issues. Before we proceed, please address the following issues:

• Please revise your title according to PLOS Medicine's style. Your title must be nondeclarative. For example, “Taxed and untaxed beverage intake by South African young adults after a national sugar-sweetened beverage tax: a before-and-after study.”

• Under the Competing Interests statement, please declare that Barry Popkin is an Academic Editor with PLOS Med.

• In the Author Summary, please avoid assertions of primacy. Specifically, the second bullet point under “Why was this study done?”, use the phrase, “To our knowledge, this is the first…”

• Your study is observational and therefore causality cannot be inferred. Please remove language that implies causality throughout the manuscript, such as “The majority of this increase was due to ...” Refer to associations instead. For example, in the third and forth bullet point under “What did the researchers do and find?”, line 359, line 378 to 390, line 393, Discussion section, and so forth.

• Please attenuate you statement in line 539 to 540.

• Change Methods, line 3 to rename Appendix A as Supporting Information file A or 1 and so forth.

• In Table 2 and 5, please provide exact p-values unless p <0.001

• In the tables, please explain, in the legend, hat the error bars represent

• Do not underline the text in line 445-446

• Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

• Under references, please indicate the access date for all websites consistently.

• Please update reference 42 – it is no longer under review.

• Indicate that Supporting Information file “Diet final” is a survey. Consider renaming it as such.

• In the third bullet point of 'What did the researchers find?' there is a typo, where 'non-taxed beverages' are referred to as 'taxed beverages.’ Please revise this.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Feb 11 2021 11:59PM.

Sincerely,

Beryne

PLOS | OPEN FOR DISCOVERY

Beryne Odeny MD, MPH, PhD(c)

Associate Editor | PLOS Medicine

1265 Battery Street, Suite 200, San Francisco, CA 94111

bodeny@plos.org

plos.org | Facebook | Twitter | Blog

------------------------------------------------------------

Requests from Editors:

Comments from Reviewers:

Reviewer #1: The authors have fully addressed all of my concerns at the first review round and I have only one very small further comment to make. In the third bullet point of 'What did the researchers find?' there is a typo, where 'non-taxed beverages' are referred to as 'taxed beverages'.

Thank you for the opportunity to review this paper.

Peter Scarborough

Reviewer #2: It was a pleasure to read the revised version of the manuscript. The authors have done an excellent job in systematically addressing all my initial comments and concerns, and I have no further comments.

The paper looks stronger in my view and will be a nice contribution to the journal.

Reviewer #3: Thanks authors for their great effort to improve the manuscript. I am satisfied with the response and revision. No further issues needing attention.

Any attachments provided with reviews can be seen via the following link:

[LINK]

PLoS Med. 2021 May 25;18(5):e1003574. doi: 10.1371/journal.pmed.1003574.r005

Author response to Decision Letter 2

14 Feb 2021

Attachment

Submitted filename: PMEDICINE-D-20-03807R2 Response document_final.docx

Click here for additional data file.^{(30.4KB, docx)}

PLoS Med. doi: 10.1371/journal.pmed.1003574.r006

Decision Letter 3

Raffaella Bosurgi

24 Feb 2021

Dear Dr Swart,

On behalf of my colleagues and the Academic Editor, Dr. Sanjay Basu, I am pleased to inform you that we have agreed to publish your manuscript "Taxed and untaxed beverage intake by South African young adults after a national sugar-sweetened beverage tax: a before-and-after study" (PMEDICINE-D-20-03807R3) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Beryne Odeny

Associate Editor

PLOS Medicine

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S1 STROBE Checklist

(DOC)

Click here for additional data file.^{(107.5KB, doc)}

S1 Protocol. Study protocol.

(DOCX)

Click here for additional data file.^{(752.1KB, docx)}

S1 Survey. Diet survey.

(PDF)

Click here for additional data file.^{(237.3KB, pdf)}

S1 Table. Beverage classification system.

(DOCX)

Click here for additional data file.^{(23KB, docx)}

S2 Table. Model adjusted predicted intake of total sugar for taxed and untaxed beverage subcategories, Langa adults aged 18–39 years.

(DOCX)

Click here for additional data file.^{(18.6KB, docx)}

S3 Table. Model adjusted predicted intake of energy (kilocalories) for taxed and untaxed beverage subcategories, Langa adults aged 18–39 years.

(DOCX)

Click here for additional data file.^{(18.5KB, docx)}

S4 Table. Model adjusted predicted intake of volume for taxed and untaxed beverage subcategories, Langa adults aged 18–39 years.

(DOCX)

Click here for additional data file.^{(18.4KB, docx)}

S5 Table. Model adjusted predicted intakes of total sugar, energy, and volume from taxed beverages for each LSM category.

(DOCX)

Click here for additional data file.^{(16.3KB, docx)}

Attachment

Submitted filename: PMEDICINE-D-20-03807R1 responses to reviewers final.docx

Click here for additional data file.^{(17.4MB, docx)}

Attachment

Submitted filename: PMEDICINE-D-20-03807R2 Response document_final.docx

Click here for additional data file.^{(30.4KB, docx)}

Data Availability Statement

[pmed.1003574.ref001] 1.Popkin BM, Hawkes C. Sweetening of the global diet, particularly beverages: patterns, trends, and policy responses. Lancet Diabetes Endocrinol. 2016;4(2):174–86. 10.1016/S2213-8587(15)00419-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref002] 2.Malik VS, Schulze MB, Hu FB. Intake of sugar-sweetened beverages and weight gain: a systematic review. Am J Clin Nutr. 2006;84(2):274–88. 10.1093/ajcn/84.1.274 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref003] 3.Te Morenga L, Mallard S, Mann J. Dietary sugars and body weight: Systematic review and meta-analyses of randomised controlled trials and cohort studies. BMJ. 2013;345(7891):e7492. 10.1136/bmj.e7492 [DOI] [PubMed] [Google Scholar]

[pmed.1003574.ref004] 4.Malik VS, Popkin BM, Bray GA, Despres J-P, Willett WC, Hu FB. Sugar-sweetened beverages and risk of metabolic syndrome and type 2 diabetes: a meta-analysis. Diabetes Care. 2010;33(11):2477–81. 10.2337/dc10-1079 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref005] 5.Malik VS, Popkin BM, Bray GA, Despres J-P, Hu FB. Sugar-sweetened beverages, obesity, type 2 diabetes mellitus, and cardiovascular disease risk. Circulation. 2010;121(11):1356–64. 10.1161/CIRCULATIONAHA.109.876185 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref006] 6.Arantxa Colchero M, Rivera-Dommarco J, Popkin BM, Ng SW. In Mexico, evidence of sustained consumer response two years after implementing a sugar-sweetened beverage tax. Health Aff (Millwood). 2017;36(3):564–71. 10.1377/hlthaff.2016.1231 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref007] 7.Caro JC, Corvalán C, Reyes M, Silva A, Popkin B, Taillie LS. Chile’s 2014 sugar-sweetened beverage tax and changes in prices and purchases of sugar-sweetened beverages: an observational study in an urban environment. PLoS Med. 2018;15(7):e1002597. 10.1371/journal.pmed.1002597 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref008] 8.Global Food Research Program. Resources. Chapel Hill (NC): Global Food Research Program; 2020. [cited 2020 Apr 1]. Available from: http://globalfoodresearchprogram.web.unc.edu/resources/. [Google Scholar]

[pmed.1003574.ref009] 9.Backholer K, Blake M, Vandevijvere S. Sugar-sweetened beverage taxation: an update on the year that was 2017. Public Health Nutr. 2017;20(18):3219–24. 10.1017/S1368980017003329 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref010] 10.Statistics South Africa. Mortality and causes of death in South Africa, 2016: findings from death notification. Pretoria: Statistics South Africa; 2018. [cited 2021 Feb 8]. Available from: https://www.statssa.gov.za/publications/P03093/P030932016.pdf. [Google Scholar]

[pmed.1003574.ref011] 11.South African Revenue Service. Health promotion levy on sugary beverages. Pretoria: South African Revenue Service; 2021. [cited 2021 April 26]. Available from: http://www.sars.gov.za/customs-and-excise/excise/health-promotion-levy-on-sugary-beverages/. [Google Scholar]

[pmed.1003574.ref012] 12.HM Revenue and Customs. Soft drinks industry levy. London: HM Revenue and Customs; 2016 [cited 2021 Feb 8]. Available from: https://www.gov.uk/government/publications/soft-drinks-industry-levy/soft-drinks-industry-levy.

[pmed.1003574.ref013] 13.Francis N, Marron DB, Rueben KS. The pros and cons of taxing sweetened beverages based on sugar content. Washington (DC): Urban Institute; 2016. [cited 2021 Apr 22]. Available from: https://www.heart.org/idc/groups/ahaecc-public/@wcm/@adv/documents/downloadable/ucm_490771.pdf. [Google Scholar]

[pmed.1003574.ref014] 14.Veerman L. The impact of sugared drink taxation and industry response. Lancet Public Health. 2017;2(1):e2–3. 10.1016/S2468-2667(16)30039-1 [DOI] [PubMed] [Google Scholar]

[pmed.1003574.ref015] 15.World Health Organization. Global status report on noncommunicable diseases 2014. Geneva: World Health Organization; 2014 [cited 2021 Apr 22]. Available from: https://apps.who.int/iris/bitstream/handle/10665/148114/9789241564854_eng.pdf;jsessionid=7DD0AC09B03E54923DCFB3C640D71713?sequence=1. [Google Scholar]

[pmed.1003574.ref016] 16.Cawley J, Frisvold D, Hill A, Jones D. The impact of the Philadelphia beverage tax on purchases and consumption by adults and children. J Health Econ. 2019;67:102225. 10.1016/j.jhealeco.2019.102225 [DOI] [PubMed] [Google Scholar]

[pmed.1003574.ref017] 17.Fletcher JM, Frisvold D, Tefft N. Taxing soft drinks and restricting access to vending machines to curb child obesity. Health Aff (Millwood). 2010;29(5):1059–66. 10.1377/hlthaff.2009.0725 [DOI] [PubMed] [Google Scholar]

[pmed.1003574.ref018] 18.Lee MM, Falbe J, Schillinger D, Basu S, McCulloch CE, Madsen KA. Sugar-sweetened beverage consumption 3 years after the Berkeley, California, sugar-sweetened beverage tax. Am J Public Health. 2019;109(4):637–9. 10.2105/AJPH.2019.304971 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref019] 19.Falbe J, Thompson HR, Becker CM, Rojas N, McCulloch CE, Madsen KA. Impact of the Berkeley excise tax on sugar-sweetened beverage consumption. Am J Public Health. 2016;106(10):1865–71. 10.2105/AJPH.2016.303362 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref020] 20.Zhong Y, Auchincloss AH, Lee BK, Kanter GP. The short-term impacts of the Philadelphia beverage tax on beverage consumption. Am J Prev Med. 2018;55(1):26–34. 10.1016/j.amepre.2018.02.017 [DOI] [PubMed] [Google Scholar]

[pmed.1003574.ref021] 21.Silver LD, Ng SW, Ryan-Ibarra S, Taillie LS, Induni M, Miles DR, et al. Changes in prices, sales, consumer spending, and beverage consumption one year after a tax on sugar-sweetened beverages in Berkeley, California, US: a before-and-after study. PLoS Med. 2017;14:e1002283. 10.1371/journal.pmed.1002283 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref022] 22.Fletcher JM, Frisvold DE, Tefft N. The effects of soft drink taxes on child and adolescent consumption and weight outcomes. J Public Econ. 2010;94(11–12):967–74. 10.1016/j.jpubeco.2010.09.005 [DOI] [Google Scholar]

[pmed.1003574.ref023] 23.Thompson FE, Kirkpatrick SI, Krebs-Smith SM, Reedy J, Schap TE, Subar AF, et al. The National Cancer Institute’s dietary assessment primer: a resource for diet research. J Acad Nutr Diet. 2015;115(12):1986–95. 10.1016/j.jand.2015.08.016 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref024] 24.Statistics South Africa. Langa. Pretoria: Statistics South Africa; 2020 [cited 2020 May 14]. Available from: http://www.statssa.gov.za/?page_id=4286&id=318.

[pmed.1003574.ref025] 25.StataCorp. Stata statistical software: release 16. College Station (TX): StataCorp; 2019. [Google Scholar]

[pmed.1003574.ref026] 26.Belotti F, Deb P, Manning WG, Norton EC. twopm: two-part models. Stata J. 2015;15(1):3–20. [Google Scholar]

[pmed.1003574.ref027] 27.Manning WG, Mullahy J. Estimating log models: to transform or not to transform? J Health Econ. 2001;20(4):461–94. 10.1016/s0167-6296(01)00086-8 [DOI] [PubMed] [Google Scholar]

[pmed.1003574.ref028] 28.Amrhein V, Greenland S, McShane B. Scientists rise up against statistical significance. Nature. 2019;567:305–7. 10.1038/d41586-019-00857-9 [DOI] [PubMed] [Google Scholar]

[pmed.1003574.ref029] 29.National Centers for Environmental Information. Data tools: find a station. Asheville (NC): National Centers for Environmental Information; 2020. [cited 2020 Feb 13]. Available from: https://www.ncdc.noaa.gov/cdo-web/datatools/findstation. [Google Scholar]

[pmed.1003574.ref030] 30.South African Advertising Research Foundation. The SAARF AMPS Living Standards Measure (LSM). Johannesburg: South African Advertising Research Foundation; 2011. [cited 2019 May 1]. Available from: http://www.saarf.co.za/amps-technicalreport/technicalreport-Jan2011—Dec 2011/data files/Technical/21—Tech 2011B~Pages92-97.pdf. [Google Scholar]

[pmed.1003574.ref031] 31.South African Audience Research Foundation. Living Standards Measure. Johannesburg: South African Audience Research Foundation; 2017. [cited 2019 Apr 29]. Available from: http://www.saarf.co.za/lsm/lsms.asp. [Google Scholar]

[pmed.1003574.ref032] 32.Teng AM, Jones AC, Mizdrak A, Signal L, Genç M, Wilson N. Impact of sugar-sweetened beverage taxes on purchases and dietary intake: systematic review and meta-analysis. Obes Rev. 2019;20(9):1187–204. 10.1111/obr.12868 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref033] 33.National Cancer Institute. Dietary assessment primer: evaluating the effect of an intervention on diet. Bethesda (MD): National Cancer Institute; 2019. [cited 2019 Jan 16]. Available from: https://dietassessmentprimer.cancer.gov/approach/intervention.html. [Google Scholar]

[pmed.1003574.ref034] 34.Colchero MA, Molina M, Guerrero-López CM. After mexico implemented a tax, purchases of sugar-sweetened beverages decreased and water increased: difference by place of residence, household composition, and income level. J Nutr. 2017;147(8):1552–7. 10.3945/jn.117.251892 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref035] 35.Satija A, Yu E, Willett WC, Hu FB. Understanding nutritional epidemiology and its role in policy. Adv Nutr. 2015;(6):5–18. 10.3945/an.114.007492 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref036] 36.National Cancer Institute. Dietary assessment primer: learn more about food composition databases for 24-hour dietary recalls and food records. Bethesda (MD): National Cancer Institute; 2019. [cited 2019 Apr 30]. Available from: https://dietassessmentprimer.cancer.gov/learn/recall-record.html. [Google Scholar]

[pmed.1003574.ref037] 37.Bandy LK, Scarborough P, Harrington RA, Rayner M, Jebb SA. Reductions in sugar sales from soft drinks in the UK from 2015 to 2018. BMC Med. 2020;18(1):20. 10.1186/s12916-019-1477-4 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref038] 38.Scarborough P, Adhikari V, Richard A. Harrington R, Elhussein A, Briggs A, Rayner M, et al. Impact of the announcement and implementation of the UK Soft Drinks Industry Levy on sugar content, price, product size and number of available soft drinks in the UK, 2015–19: a controlled interrupted time series analysis. PLoS Med. 2020;17(2):e1003025. 10.1371/journal.pmed.1003025 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref039] 39.Griffith R, O’Connell M, Smith K. The importance of product reformulation versus consumer choice in improving diet quality. Economica. 2017;84(333):34–53. [Google Scholar]

[pmed.1003574.ref040] 40.Spiteri M, Soler LG. Food reformulation and nutritional quality of food consumption: an analysis based on households panel data in France. Eur J Clin Nutr. 2018;72(2):228–35. 10.1038/s41430-017-0044-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref041] 41.Pitt C. City of Cape Town relaxes water restrictions, tariffs to level 5. news24. 2018 Sep 10 [cited 2020 Apr 30]. Available from: https://www.news24.com/SouthAfrica/News/city-of-cape-town-relaxes-water-restrictions-tariffs-to-level-5-20180910.

[pmed.1003574.ref042] 42.Stacey N, Edoka I, Hofman K, Swart R, Popkin BM, Ng S. Changes in beverage purchases following the announcement and implementation of South Africa’s Health Promotion Levy: an observational study. Lancet Planet Health. 2021;5(4):e200–8. 10.1016/S2542-5196(20)30304-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[pmed.1003574.ref043] 43.Haines P, Guilkey D, Popkin B. Modeling food consumption decisions as a two-step process. Am J Agr Econ. 1988;70(3):543–52. [Google Scholar]

PERMALINK

Taxed and untaxed beverage intake by South African young adults after a national sugar-sweetened beverage tax: A before-and-after study

Michael Essman

Lindsey Smith Taillie

Tamryn Frank

Shu Wen Ng

Barry M Popkin

Elizabeth C Swart

Roles

Abstract

Background

Methods and findings

Conclusions

Author summary

Why was this study done?

What did the researchers do and find?

What do these findings mean?

Introduction

Methods

Data sources and measures

Fig 1. Timeline for data collection.

Table 1. Sociodemographic information for Langa survey of adults aged 18–39 years.

Measuring dietary intake

Linking FCTs to beverage categories

Analytical approach

Estimation of beverage intake

Beverage intake by socioeconomic status

Sensitivity analyses

Ethical approval

Patient and public involvement

Results

Adjusted results

Total effects for sugar, energy, and volume intakes for taxed beverages

Table 2. Model adjusted predicted values for sugar, energy, and volume intakes for taxed, untaxed, and total beverages.

Total effects for sugar, energy, and volume intakes for untaxed beverages

Total effects for sugar, energy, and volume intakes for total beverages

Reformulation effects for sugar, energy, and volume intakes

Fig 2. Estimated daily intake in grams of sugar per capita from taxed, untaxed, and total beverages.

Fig 3. Estimated daily intake in energy (kilocalories) per capita from taxed, untaxed, and total beverages.

Fig 4. Estimated daily intake in milliliters per capita from taxed, untaxed, and total beverages.

Beverage intake by LSM

Sensitivity analyses

Discussion

Conclusion

Supporting information

Acknowledgments

Abbreviations

Data Availability

Funding Statement

References

Decision Letter 0

Adya Misra

Roles

Decision Letter 1

Adya Misra

Roles

Author response to Decision Letter 1

Decision Letter 2

Raffaella Bosurgi

Roles

Author response to Decision Letter 2

Decision Letter 3

Raffaella Bosurgi

Roles

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases