Introduction

‘Sugary drinks’, defined as beverages with added sugars (‘sugar-sweetened beverages’ (SSB)) and beverages with natural intrinsic sugars (100% juice), contribute to excess dietary sugar intake (World Health Organization (WHO), 2015). Sugary drink consumption is associated with increased risk of obesity, type 2 diabetes, cardiovascular disease, cancer, disability-adjusted life years and mortality (Global Burden of Disease (GBD), 2018; Imamura et al., 2015; Makarem et al., 2018; Malik et al., 2013). Reducing consumption is a primary focus of global nutrition policies, including informational, financial, healthy default choices, and reduced availability interventions (Krieger et al., 2021). For example, the WHO recommends limiting free sugar to less than 10% of daily energy intake (WHO, 2015). Moreover, Canada’s updated 2019 Food Guide advises to ‘replace sugary drinks with water’, representing a change from the 2007 Canada’s Food Guide recommendations which included 100% juice as a fruit or vegetable serving (Health Canada, 2019b). In addition, Newfoundland and Labrador have included a 20-cent-per-litre sugary drink tax in the government’s proposed 2021–2022 budget, a first for Canada (Coady, 2021). There is a need to implement equitable interventions to reduce sugary drink consumption and associated health risks (Krieger et al., 2021).

In Canada, consumption of traditional sugary drinks (i.e., fruit drinks, regular soft drinks and 100% juice) has declined by one third between 2004 and 2015 (Czoli et al., 2019; Garriguet, 2019; Jones et al., 2019; Langlois et al., 2019). However, in 2015, sugary drinks contributed an average of 18% of total free sugar intake among Canadians (Liu et al., 2020) and were top dietary contributors of Canadians’ overall sugar intake across the lifespan (Kirkpatrick et al., 2019; Langlois et al., 2019). Of additional concern, purchases of novel types of SSB (e.g., energy and sport drinks) have increased among Canadians over time (Czoli et al., 2019), reflecting patterns observed in the United States (Drewnowski & Rehm, 2015; Han & Powell, 2013). Given the shifting consumption patterns, surveillance of sugary drinks overall and across beverage type is required to comprehensively assess population intake. Monitoring sugary drink consumption across socioeconomic position (SEP) is critical to understanding where inequities exist and their contribution to non-communicable disease risk. Low SEP groups have been found to consume more SSB and 100% juice in the USA (SEP: income and education) (Drewnowski & Rehm, 2015; Han & Powell, 2013), the United Kingdom (SEP: income) (Briggs et al., 2013) and Australia (SEP: income; SSB only) (Lal et al., 2017). In Canada, no differences in per capita SSB or 100% juice intake were observed across income (Jones et al., 2019), whereas intake across education and food security has yet to be established.

Comprehensive assessments of sugary drink consumption in Canada are needed. First, SEP indicators are not interchangeable proxies and descriptions of inequities associated with health knowledge (education), material resources (education, income and food security status) and material deprivation beyond accessing nutritious food (food security) are warranted (Braveman et al., 2005; Kirkpatrick et al., 2015). Moreover, age- and sex-specific patterns of sugary drink consumption across SEP should be considered. For example, compared to females, males have a higher prevalence of sugary drink consumption and consume more energy from these beverages in Canada (Garriguet, 2019; Jones et al., 2019; Langlois et al., 2019) and other high income countries (GBD, 2018). Finally, accounting for the episodic nature of beverage consumption is needed to better estimate social inequities in sugary drink consumption (Garriguet, 2019). Specifically, it is important to estimate the potential socioeconomic inequities in the prevalence of consumers, given that any consumption of these beverages is not recommended. Moreover, estimating socioeconomic inequities in mean energy intake from beverages allows us to assess the contribution of overall energy consumption among those who consumed (WHO, 2015).

Our objective was to estimate the sex-specific consumption of sugary drinks, SSB and 100% juice across SEP in a population-representative sample of Canadian children/adolescents and adults. This study fills a critical knowledge gap regarding the extent to which social inequities in sugary drink consumption exist in Canada.

Methods

Study population

We conducted a cross-sectional analysis of the nationally representative 2015 Canadian Community Health Survey–Nutrition (CCHS-N) Public Use Microdata File (Health Canada, 2017). The CCHS-N was conducted by Statistics Canada to assess dietary intakes of Canadians for the first time since 2004 using interviewer-administered 24-h dietary recalls using an adapted version of the Automated Multi-Pass Method from the United States Department of Agriculture (Statistics Canada, 2018). This survey used a multi-stage, cluster sampling approach to secure a sample of 20,487 Canadians aged 1 year and older living in private dwellings across the 10 provinces (61% response rate) (Health Canada, 2017).

We included single-day 24-h dietary recalls among Canadians 2 years and older (n = 20,115). We excluded respondents if they were breastfeeding or reported no energy intake (n = 200), or were missing information on income (n = 22), education (n = 41) or food security (n = 110). Our final analytic sample was 19,742 respondents. This study was approved by the Ethics Review Board at Public Health Ontario.

Beverage types: sugary drinks, sugar-sweetened beverages (SSB) and 100% juice

We assessed sugary drink, SSB and 100% juice consumption separately. We applied Health Canada’s definition of sugary drinks (Health Canada, 2019a) to derive beverage types using Nutrition Survey System (NSS) codes linked to Canadian Nutrient File descriptions (see Appendix 1, Table 3). Sugary drink codes (N = 249) represent the sum of SSB (N = 190) and 100% juice (N = 59) codes. Our beverage categorization aligned with a previous Canadian study (Jones et al., 2019), with additional exclusion of beverages with sugar added by consumer (e.g., tea and coffee) and inclusion of sugary drinks consumed with alcohol. Diet/low-calorie beverages, infant formulas, functional beverages (e.g., meal replacements) and alcoholic beverages (e.g., pure alcohol, beer, wine, pre-mixed seltzers) were not included as sugary drinks.

Beverage consumption categories are not mutually exclusive because sugary drink consumers include consumers of either SSB or 100% juice or both SSB and 100% juice. We estimated sugary drink, SSB and 100% juice intake using two measures: (1) prevalence of consumers on a given day, defined as the proportion that reported consuming each beverage type on the day prior to the 24-h recall interview; and (2) mean energy intake (kilocalories (kcal)) among those who consumed each beverage type on a given day, estimated among those who consumed the day prior to the 24-h recall interview. To further quantify energy intake, we estimated the relative contribution of energy from each beverage type. We report averages; therefore, a single dietary recall was sufficient considering mean intake on a given day reflects mean usual intake (Garriguet, 2019).

Socioeconomic position

The highest level of household education was categorized using four groups: ‘Less than high school’, ‘High school diploma’, ‘Certificate below bachelor’s degree’ (e.g., a trade, college or non-bachelor certificate) and ‘Bachelor degree or above’. Household food security status was assessed using eight questions for children/adolescents (aged 2–18) and ten for adults (aged 19 and over) with responses classified as either food secure (i.e., answered ‘yes’ to 0–1 questions about difficulty with income-related food access) or food insecure (moderate or severe, i.e., answered ‘yes’ to two or more questions about compromised quality/quantity or reduced food intake due to disrupted eating patterns) (Statistics Canada, 2018). Household income adequacy quintiles were derived based on the adjusted ratio of the respondents’ total household income reported in the previous 12 months to the low income cutoff corresponding to their household and community size (Statistics Canada, 2018).

Covariates

We stratified our sample by sex (female or male) and age (children/adolescents, 2–18 years or adults, 19 years and older).

Statistical analysis

We ran separate models for each outcome and SEP indicator. We applied modified Poisson regressions to estimate prevalence, prevalence differences and ratios (Appendix 1, Tables 6 and 7), and corresponding 95% confidence intervals (95% CI) across SEP using post-estimation marginal means analysis (Zou, 2004). Among those who consumed on a given day, we conducted linear regressions to estimate mean energy intake from each beverage type across SEP. For each beverage type, we estimated the relative (%) contribution of mean energy intake to the total energy intake from all sources. Primary analyses were unadjusted to describe consumption in a nationally representative sample (Conroy & Murray, 2020).

We conducted two sensitivity analyses. The residuals for the energy outcomes were right-skewed; therefore, we conducted square root-transformed regression and then back-transformed to original scale post-regression. Comparisons with the primary analysis revealed no differences in energy intake patterns (not shown); therefore, we reported untransformed energy outcomes to reflect reported energy intake from the CCHS-N (Lumley et al., 2002). Furthermore, we adjusted all models for dietary energy misreporting (under, over, plausible reporters, unclassified) to account for systematic error in self-reported dietary assessments (Appendix 2) (Garriguet, 2018).

We applied survey weights and bootstrap replicates (N = 500) provided by Statistics Canada in all models and used survey procedures ensuring results were representative of the Canadian population and appropriate variance estimation, respectively. Analyses were completed using SAS (v.9.4) and STATA (v.15).

Results

Prevalence of sugary drink, SSB and 100% juice consumers on a given day

Sex-specific prevalence of sugary drink, SSB and 100% juice consumers on a given day by SEP for children/adolescents is presented in Fig. 1 (estimates in Appendix 1, Table 4). Among female and male children/adolescents, the prevalence of consuming sugary drinks (females: 72% (95% CI: 68, 75); males: 78% (95% CI: 75, 80)), SSB (females: 52% (95% CI: 49, 56); males: 56% (95% CI: 53, 59)) and 100% juice (females: 37% (95% CI: 33, 40); males: 42% (95% CI: 38, 45)) on a given day was high. The absolute prevalence difference of sugary drink consumers ranged from 11 to 21 percentage-points higher among females in households with lower education (‘Less than high school’, ‘High school diploma’ and ‘Certificate below bachelor’s degree’) compared to ‘Bachelor’s degree or above’. Similarly, the prevalence of SSB consumers ranged from 8 to 27 percentage-points higher among females in households with lower education compared to those with ‘Bachelor’s degree or higher’. Among males, the prevalence of 100% juice consumers was 9 (95% CI: − 18, 0) percentage-points lower in food insecure compared to food secure households. In females, a lack of precision (i.e., uncertainty) was observed for the estimates related to higher prevalence of sugary drink and, separately, SSB consumers in food insecure compared to food secure and low compared to high income households (for sugary drinks only). In males, uncertainty surrounded higher prevalence of SSB consumers across education and comparing lowest to highest income. There was no clear pattern between the prevalence of 100% juice consumers and SEP in females and across education and income for males.

Fig. 1
figure 1

source: 2015 Canadian Community Health Survey–Nutrition. Children/adolescents include respondents aged 2–18 years (females: 3050; males: 3064). SEP indicators include highest household educational attainment, household food security status and household income quintiles. Dotted lines represent 95% confidence intervals (CI). Prevalence values for the point estimates are included along the right-hand side of each figure. Figure created in R-Studio

Sex-specific prevalence (%) of sugary drink consumption, on a given day, by beverage type and socioeconomic position in children/adolescents. Data

Full size image

Sex-specific prevalence of sugary drink, SSB and 100% juice consumers on a given day by SEP in adults is presented in Fig. 2 (estimates in Appendix 1, Table 5). Among female adults, the prevalence of consuming on a given day was 50% (95% CI: 47, 52) for sugary drinks, 35% (95% CI: 33, 38) for SSB and 22% (95% CI: 20, 24) for 100% juice. On the absolute scale, the prevalence of sugary drink and, separately, SSB female adult consumers in food insecure households was 10 (95% CI: 1, 19) and 14 (95% CI: 2, 27) percentage-points higher than food secure households, respectively. Moreover, among female adults, consumption of sugary drinks among the lowest (Q1) and fourth income quintile (Q4), respectively, was 11 (95% CI: 1, 20) and 10 (95% CI: 2, 19) percentage-points higher compared to the highest (Q5) quintile. Uncertainty surrounded the higher prevalence estimates of increased sugary drink consumption comparing Q2 and Q3 income groups to Q5 and the higher estimates of SSB consumption among lower income groups. There was no clear pattern across education for SSB consumption and across all SEP indicators for 100% juice consumption in females.

Fig. 2
figure 2

source: 2015 Canadian Community Health Survey–Nutrition. Adults include respondents aged 19 years and older (females: 7219; males: 6409). SEP indicators include highest household educational attainment, household food security status and household income quintiles. Dotted lines represent 95% confidence intervals (CI). Prevalence values for the point estimates are included along the right hand side of each figure. Figure created in R-Studio

Sex-specific prevalence (%) of sugary drink consumption, on a given day, by beverage type and socioeconomic position in adults. Data

Full size image

Among male adults, the prevalence of consuming on a given day was 58% (95% CI: 55, 60) for sugary drinks, 44% (95% CI: 41, 46) for SSB and 25% (95% CI: 23, 27) for 100% juice. No differences were observed in the prevalence of sugary drink consumers across SEP. The prevalence of SSB consumers was between 7 and 8 percentage-points higher among males with ‘High school diploma’ and ‘Credential below bachelor’s degree’ compared to ‘Bachelor’s degree or above’ and 9 (95% CI: − 1, 19) percentage-points higher among males from food insecure compared to food secure households. Conversely, the prevalence of 100% juice consumers was 9 (95% CI: − 18, 0) percentage-points lower among male adults from food insecure compared to food secure households. There is additional evidence that low education males are consuming less 100% juice than their high SEP counterparts. Given the uncertainty around the estimates, additional confirmation is required to support interpretations.

Sex-specific prevalence ratios of sugary drink, SSB and 100% juice consumers on a given day by SEP for children/adolescents (Appendix 1, Table 6) and adults (Appendix 1, Table 7) followed the same patterns described for the absolute differences.

Mean energy intake from sugary drinks, SSB and 100% juice, among consumers on a given day

Sex-specific mean energy intake from beverages among children/adolescents who consumed sugary drinks, SSB and 100% juice on a given day by SEP is presented in Table 1. Among female children/adolescents, mean daily energy intake was 211 kcal (95% CI: 200, 223) from sugary drinks, 197 kcal (184, 210) from SSB and 132 kcal (95% CI: 122, 143) from 100% juice, representing 12%, 11% and 8% of overall energy intake among consumers on a given day, respectively. No social inequities in mean energy intake were observed.

Table 1 Sex-specific mean energy intake (kcal) and mean differences from sugary drinks by socioeconomic position among children/adolescent consumers on a given day
Full size table

Among male children/adolescents, mean energy intake was 252 kcal (95% CI: 234, 270) from sugary drinks, 238 kcal (95% CI: 215, 260) from SSB and 149 kcal (95% CI: 136, 163) from 100% juice, representing 12%, 11% and 7% of overall energy intake among consumers, respectively. While the relative contribution of overall energy intake from sugary drinks and SSB was 3% higher among male children/adolescents with ‘Less than high school’ compared to ‘Bachelor’s degree or above’, uncertainty surrounded the mean differences. There were no clear patterns observed across SEP for 100% juice consumers.

Sex-specific mean energy intake among adult consumers of sugary drinks, SSB and 100% juice on a given day by SEP is presented in Table 2. Among female adults, mean energy intake was 186 kcal (95% CI: 175, 198) from sugary drinks, 193 kcal (95% CI: 180, 206) from SSB and 113 kcal (95% CI: 102, 124) from 100% juice, representing 11%, 11% and 6% of overall energy intake among consumers, respectively. Food insecure consumers, on average, consumed 27 kcal (95% CI: 3, 51) more energy from sugary drinks. Similar inequities were observed for SSB consumption across food security; however, additional verification is required due to uncertainty related to small sample size. One-hundred percent juice consumers with ‘Less than high school’ consumed, on average, 35 kcal (95% CI: 2, 67) more energy from 100% juice than female consumers with ‘Bachelor’s degree or above’.

Table 2 Sex-specific mean energy intake (kcal) and mean differences from sugary drinks by socioeconomic position among adult consumers on a given day
Full size table

Among male adults, mean energy intake from beverages was 243 kcal (95% CI: 230, 257) from sugary drinks, 246 kcal (95% CI: 230, 261) from SSB and 134 kcal (95% CI: 123, 145) from 100% juice, representing 10%, 10% and 6% of overall daily energy intake, respectively. Sugary drink consumers with ‘High school diploma’ and ‘Credential below bachelor’s degree’ consumed, on average, 68 and 39 kcal more energy from sugary drinks compared to those with ‘Bachelor’s degree or above’. Similar patterns were observed for SSB energy intake across education. While the relative contribution of energy intake among sugary drink and, separately, SSB consumers was 4% higher among food insecure compared to food secure male adults, the uncertainty surrounding the mean differences requires further verification. No patterns in energy intake were observed among male adult 100% juice consumers across SEP.

Accounting for systematic error related to energy misreporting

Adjusted intake patterns for dietary misreporting were consistent with the primary analyses (available upon request).

Discussion

This study identified sex-specific social inequities in sugary drink and SSB consumption in a nationally representative sample of Canadians, but patterns were inconsistent for 100% juice. Low education was associated with higher prevalence of sugary drink and SSB consumption on a given day among female children/adolescents, while food insecurity and low income were associated with a higher prevalence of sugary drink consumption in female adults. Among male children/adolescents and adults, a lower prevalence of 100% juice consumption on a given day was observed among food insecure compared to food secure groups. Among consumers on a given day, higher mean energy intake from sugary drinks and SSB was observed across food security in female adults and education in male adults. Among female adults who consumed 100% juice, low education was associated with higher energy intake. Across all SEP groups, Canadians reported a high prevalence of sugary drink, SSB and 100% juice consumption on a given day and have mean energy intake levels associated with poor health outcomes (WHO, 2015). Together our findings indicate the need for pro-equity interventions to reduce consumption among Canadians.

Population-level interventions such as informational and financial policies have potential to reduce unhealthy beverage consumption and may influence manufacturers to produce healthier products (Krieger et al., 2021). Informational interventions, including Canada’s ‘replace sugary drinks with water’ messaging (Health Canada, 2019b) or traffic-light labelling interventions (von Philipsborn et al., 2019), are important, but likely insufficient for achieving this goal. Moreover, informational interventions may impose literacy burdens on consumers, limiting access for certain groups (Krieger et al., 2021). Taxation policies are considered pro-equity interventions given the beneficial impact across subpopulations. Internationally, SSB taxation policies are associated with reduced SSB sales concentrated among lower SEP groups (Krieger et al., 2021; Popkin & Ng, 2021). Modelling studies have demonstrated the potential for beverage taxation policies to reduce social inequities in non-communicable diseases and increase healthcare cost savings concentrated among lower SEP groups (Kao et al., 2020; Lal et al., 2017). Finally, limiting SSB availability in subsidized food programs may improve diet quality of recipients, particularly when paired with incentives to increase fruit/vegetable purchases (Krieger et al., 2021).

Establishing baseline levels of beverage consumption by age, sex and SEP is essential for guiding the choice and design of interventions to equitably reduce sugary drink consumption and associated non-communicable disease in Canada (Popkin & Ng, 2021). Few studies have examined the association between SEP and sugary drink consumption, an important risk factor for health outcomes (Briggs et al., 2013; Han & Powell, 2013; Jones et al., 2019; Lal et al., 2017). A previous Canadian study using the CCHS-N reported no differences in per capita energy intake from SSB across income levels (Jones et al., 2019). Our study expanded upon this analysis by assessing total sugary drinks, incorporating the episodic nature of beverage consumption, disaggregating findings by sex, describing outcomes across income quintiles and examining additional household SEP measures (educational attainment and food security status). In our study, more patterns emerged for food security status and educational attainment compared to income for sugary drinks and SSB consumption, highlighting the importance of describing patterns across multiple SEP indicators in heath research (Braveman et al., 2005). To our knowledge, this is the first Canadian study to include food security status as an independent exposure of sugary drink intake incorporating work that identified food security status as a strong predictor of differential nutrient intake in Canada (Kirkpatrick et al., 2015). Our findings related to SSB consumption across education are consistent with associations found in the USA (Han & Powell, 2013). More income-related inequities in SSB consumption were observed in the USA, the UK and Australia compared to our study (Briggs et al., 2013; Han & Powell, 2013; Lal et al., 2017), which may be related to different subgroups and income measures analyzed across studies.

Our study found no clear pattern between SEP and 100% juice consumption consistent with a previous Canadian assessment of 100% juice energy intake across income quartiles (Jones et al., 2019). In our study, energy intake among female adult consumers from 100% juice was higher in the lowest education group. Conversely, males had a lower prevalence of 100% juice consumption in food insecure compared to secure groups. Despite the variability in our findings across 100% juice, monitoring social inequities in 100% juice consumption is important, as these have been observed in the USA (Drewnowski & Rehm, 2015) and the UK (Briggs et al., 2013). Differential uptake of Canada’s Food Guide recommendations to replace sugary drinks with water across SEP (Health Canada, 2019b), coupled with the lower cost of 100% juice compared to whole fruits and vegetables, could generate unintended inequities in 100% juice consumption (Darmon & Drewnowski, 2015).

This study has limitations. Our study relies on data collected in 2015 and may not reflect temporal changes to dietary intake patterns; however, we leveraged the most recent population-level nutrition data available. Social desirability bias in self-reported responses may have led to underestimates of consumption in some groups (e.g., females and higher SEP), which would bias observed social inequities towards the null (Garriguet, 2018). For example, self-reported beverage intakes in the CCHS-N were consistently lower than Canadian beverage sales over the same time period (Czoli et al., 2019). Therefore, we adjusted all models for energy misreporting status in our sensitivity analyses, which had minimal impact on our findings. Low sample sizes across SEP groups likely resulted in insufficient power to detect significant associations, particularly in children/adolescents; however, sole reliance on statistical significance for interpretations in epidemiology is incomplete (Wasserstein & Lazar, 2016). Analyses of single-day dietary recall estimates are prone to random error associated with within-person variability, which can be accounted for through analyses of multiple dietary recalls, such as the National Cancer Institute’s method for estimating usual dietary intake (Tooze et al., 2006). This bias may have contributed to the reduced precision of mean energy intake estimates across SEP due to smaller sample sizes in these analyses; however, for means a single 24-h recall is sufficient and our approach allowed us to assess our specific a priori research questions related to socioeconomic differences in the prevalence, per-consumer energy intake and relative contribution of total energy, on a given day. Twenty-four-hour recalls are prone to systematic error in energy reporting, which have been accounted for in the study design and sensitivity analysis (Health Canada, 2017). Reliance on the CNF descriptions to categorize beverages may have led to misclassification of certain codes; however, our categorizations are generally consistent with a previous study using the CCHS-N (Jones et al., 2019). Finally, inclusion of only pre-sweetened beverages likely underestimated consumption, though population-level interventions generally target pre-made items (Popkin & Ng, 2021; von Philipsborn et al., 2019).

This study’s strengths include analyzing three distinct primary exposures of household SEP to describe SEP-specific consumption patterns (Braveman et al., 2005; Kirkpatrick et al., 2015). We observed differential consumption patterns by sex, highlighting the importance of assessing sex-specific associations in health research (Johnson et al., 2009). We applied methods to analyze episodically consumed beverages (Han & Powell, 2013) and analyzed separate consumption patterns for sugary drinks, SSB and 100% juice associated with free, added and natural sugars, respectively (WHO, 2015). Finally, we included a comprehensive assessment of traditional and novel beverage types (Czoli et al., 2019; Jones et al., 2019).

Conclusion

To our knowledge, this is the first study to describe age- and sex-specific sugary drink, SSB and 100% juice consumption patterns across multiple indicators of SEP in a representative sample of Canadians. Future research on the effectiveness of interventions to equitably reduce sugary drink consumption in the Canadian context is warranted.

Contributions to knowledge

What does this study add to existing knowledge?

  • This study adds to a growing body of evidence highlighting population and socioeconomic patterns of beverage consumption by including multiple socioeconomic indicators (education, food security status and income adequacy quintile) to describe differential patterns of beverage intake. In addition, population-level intake on a given day was high across all SEP groups by age and sex.

    What are the key implications for public health interventions, practice or policy?

  • The design and implementation of pro-equity population-level interventions that reduce sugary drink consumption in the Canadian population are warranted.

  • Informational, financial, healthy defaults and limited availability interventions should be considered in reference to baseline consumption patterns for the design and implementation of sustainable strategies targeting food and social environments in Canada.