Association Between Soft Drink Consumption and Mortality in 10 European Countries

CONCLUSIONS AND RELEVANCE This study found that consumption of total, sugar-sweetened, and artificially sweetened soft drinks was positively associated with all-cause deaths in this large European cohort; the results are supportive of public health campaigns aimed at limiting the consumption of soft drinks. study, the high level of consumption of total, sugar-sweetened, and artificially sweetened soft drinks was associated with elevated risks of death from all causes. Positive as-sociationswereobservedbetweensugar-sweetenedsoftdrinks and digestive disease deaths as well as between artificially sweetened soft drinks and circulatory disease deaths. Further studies are needed to investigate the possible adverse healtheffectsofartificialsweeteners.Theresultsofthisstudy are supportive of ongoing public health campaigns aimed at reducing the consumption of soft drinks.

T he frequent consumption of sugar-sweetened soft drinks increases energy intake, which can lead to weight gain and obesity. [1][2][3] In 2010, the worldwide burden of adiposity-associated cardiovascular diseases, cancers, and type 2 diabetes associated with consumption of sugar-sweetened soft drinks was estimated to be 184 000 deaths. 4 Reformulation of sugar-sweetened soft drinks, in which sugar is replaced with low-or no-calorie sweeteners, is being driven by consumer awareness and fiscal instruments, such as taxes. 5 Artificially sweetened soft drinks have few or no calories; however, their long-term physiological and health implications are largely unknown. [6][7][8] Whether regular consumption of soft drinks (total, sugarsweetened, or artificially sweetened) is associated with greater all-cause and cause-specific mortality is uncertain, given that inconsistent findings were reported from previous prospective studies. 9,10 Recently, a joint analysis of the Health Professionals Follow-up Study (HPFS) and Nurses' Health Study (NHS) reported that a higher level of consumption of sugarsweetened and artificially sweetened beverages was associated with greater all-cause mortality in the United States. 11 Similarly, a positive association between artificially sweetened beverage consumption and all-cause mortality among USbased women was also reported by the Women's Health Initiative Observational Study. 12 Large-scale studies are required to examine the association between soft drink consumption and health outcomes, in particular research based in populations outside of the United States. To our knowledge, a large-scale European-based analysis of soft drink consumption and mortality outcomes has not been undertaken.
For cause-specific mortality, previous studies on soft drink consumption and cardiovascular disease mortality have found positive associations. 11,13 However, whether these associations differed by type of circulatory diseases (eg, ischemic heart disease and cerebrovascular diseases) is uncertain. Similarly, scant data are available to date on mortality from other major chronic conditions such as cancers, digestive diseases, and neurodegenerative diseases.
We investigated the association of consuming total, sugarsweetened, and artificially sweetened soft drinks with total and cause-specific mortality among participants in the European Prospective Investigation into Cancer and Nutrition (EPIC), an ongoing, large multinational cohort with more than 41 000 recorded deaths reported during the follow-up period.

Study Population
EPIC is a multicenter cohort of 521 330 participants who were recruited between January 1, 1992, and December 31, 2000, predominantly from the general populations of 10 European countries (Denmark, France, Germany, Greece, Italy, the Netherlands, Norway, Spain, Sweden, and the United Kingdom). 14,15 Written informed consent was provided by all study participants. Ethical approval for this study was provided by the International Agency for Research on Cancer and the institutional review boards of the local participating EPIC centers.
For the current analysis, we excluded participants who at baseline reported cancer (n = 22 537), heart disease (n = 12 619), stroke (n = 3683), or diabetes (n = 12 461); participants in the highest or lowest 1% of the distribution of the ratio between energy intake to estimated energy requirement (ie, those with implausible dietary intake data; n = 8828); and participants with missing soft drink consumption or missing follow-up information (n = 9459

Assessment of Exposure
Dietary intake was assessed during the baseline enrollment visit (1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) by country-specific instruments that were developed and validated within the various source populations in EPIC. 14,15 Self-administered questionnaires were used in all centers, except in Greece, Spain, and Ragusa (Italy), where data were collected during personal interviews. In Malmö (Sweden), a combined semiquantitative food frequency questionnaire and 7-day dietary diary and diet interview was used. For soft drink consumption, participants recorded the number of glasses per month, week, or day; the structure of the questions varied somewhat by country and questionnaire. The dietary questionnaires for most countries collected information on the frequency of consumption (per glass) of "low calorie or diet fizzy soft drinks," "fizzy soft drinks, eg cola, lemonade," and "fruit squash or cordial." Soft drink consumption (grams per day, which is roughly equivalent to the amount in milliliters; 1 glass was equal to approximately 250 mL) was calculated using typical glass sizes in each center.
Total soft drinks referred to a combination of soft drinks, carbonated and isotonic drinks, and diluted syrups. Total soft drink consumption was subdivided into sugar-sweetened and artificially sweetened soft drink consumption for all countries except Italy, Spain, and Sweden, where types of soft drinks were unmeasured. The reproducibility and validity of the dietary questionnaires were assessed in some countries, [14][15][16][17][18][19][20][21] with correlations between repeated dietary questionnaires and with 24-hour dietary records ranging from 0.46 to 0.77 for soft or nonalcoholic drinks in the Netherlands, France, Germany, and Spain. Lifestyle questionnaires, administered at recruitment, were used as a source of information on educational attainment, smoking habits, alcohol intake, physical activity, reproductive and menstrual characteristics, and other variables.

Ascertainment of Deaths
Data on vital status as well as the cause and date of death were collected by EPIC centers through record linkages with cancer registries, boards of health, and death indices in Denmark, Italy, the Netherlands, Norway, Spain, Sweden, and the United Kingdom or through active follow-up (inquiries by mail or telephone to municipal registries or regional health departments or to physicians or hospitals) in Germany, Greece, and France. For the current study, follow-up of participants from baseline (1992-2000) occurred from December 2009 to December 2013 for countries with record linkage. The end of follow-up was considered to be the last known contact with participants in France (June 2008), Germany (December 2009), and Greece (December 2012). Loss to follow-up was relatively low at 1.5%.

Statistical Analysis
Hazard ratios (HRs) and 95% CIs for mortality were estimated using Cox proportional hazards regression models with age as the primary time metric. Time at study entry was age at recruitment, and exit time was age at death or the last date at which follow-up was considered complete in each EPIC center. Models were stratified by age at recruitment in 1-year categories, sex, and EPIC center. Soft drink consumption was categorized by the frequency of glasses consumed (<1 glass per month, 1 to 4 glasses per month, >1 to 6 glasses per week, 1 to <2 glasses per day, or ≥2 glasses per day, with 1 glass being equal to 250 mL). Linear trend tests across exposure groups were evaluated using the median category variables as continuous terms. Multivariable models were adjusted for alcohol consumption; smoking status, intensity, and duration; body mass index (BMI) (calculated as weight in kilograms divided by height in meters squared); physical activity; educational status; menopausal status; ever use of menopausal hormone therapy; and dietary intakes of total energy, red and processed meats, coffee, fruit and vegetable juices, and fruits and vegetables. Further adjustment for dietary fiber intake resulted in virtually unchanged risk estimates, so this variable was not included in the final multivariable models. Sugarsweetened and artificially sweetened soft drinks were also mutually adjusted.
The association between soft drink consumption and mortality was also assessed across subgroups of smoking status, BMI, physical activity, and alcohol consumption. Tests for interaction were performed with the likelihood ratio test of models with and without interaction terms. Heterogeneity across countries was explored using a meta-analysis approach. 22 We further investigated the shape of the association between soft drink consumption and all-cause mortality using restricted cubic splines with knots defined by the midpoints of aforementioned categories. The proportional hazard assumption was satisfied using Schoenfeld residuals 23 analyses.
In sensitivity analyses, we excluded BMI from the multivariable models to assess the potential mediating role of adiposity for the association between soft drinks and mortality. To investigate reverse causality, analyses were conducted excluding deaths within the first 8 years of follow-up. We examined the associations of sole consumption of sugarsweetened and artificially sweetened soft drinks with mortality. In addition, we examined the associations between soft drink consumption and mortality according to the death ascertainment method (linkage or active follow-up), with nonconsumers of soft drinks as the reference group, and with adjustment of the multivariable models for the World Cancer Research Fund dietary score 24 (rather than individual dietary covariates). As a negative control analysis, we also examined the associations between soft drink consumption and deaths from external causes. All statistical tests were 2-sided, and P < .05 was considered statistically significant. Data analyses were performed from February 1, 2018, to October 1, 2018.
The soft drink consumption and all-cause mortality associations were generally consistent across subgroups of other mortality risk factors. For sugar-sweetened soft drinks, a positive association was found among participants with a BMI of 30 or higher (obese) but not among those with a BMI between 25 and under 30 (overweight) (HR, 1.23; 95% CI, 1.10-1.39 vs HR, 0.98; 95% CI, 0.90-1.06; P = .002; Figure). Results remained similar when country-specific estimates were pooled in a random-effects meta-analysis (eTable 1 in the Supplement). No heterogeneity across countries was found for artificially sweetened soft drinks and all-cause mortality (I 2 = 0%; P = .86); however, heterogeneity was detected for sugarsweetened soft drinks (I 2 = 63%; P = .01), although positive associations were found for all but 1 country.

Cancer
Total, sugar-sweetened, and artificially sweetened soft drink consumption was not associated with risk of deaths from overall cancer (Table 3), breast cancer, or prostate cancer (Table 4). Total soft drink consumption was positively associated with colorectal cancer deaths (≥1 glass per day vs <1 glass per month; HR, 1.25; 95% CI, 1.07-1.47; P = .004) (Table 4), with statistically nonsignificant associations found for sugar-sweetened and artificially sweetened soft drinks.

Digestive Diseases
Higher level of consumption of total soft drinks and sugarsweetened soft drinks (≥1 glass per day vs <1 glass per month; HR, 1.59; 95% CI, 1.24-2.05; P < .001), but not artificially sweetened soft drinks, was associated with digestive disease mortality. Similar associations were found for men and women (Table 3).

Neurodegenerative Diseases
Total soft drink consumption was positively associated with risk of Parkinson disease mortality (≥1 glass per day vs <1 glass per month; HR, 1.59; 95% CI, 1.07-2.36; P = .02), with similar magnitude nonsignificant associations found for artificially sweetened and sugar-sweetened soft drinks (Table 4). Soft drinks were not associated with Alzheimer disease mortality.

Sensitivity Analyses
Sugar-sweetened and artificially sweetened soft drink consumption was positively associated with circulatory disease and digestive disease mortality among participants with a healthy weight (eTable 2 in the Supplement). Similar positive associations between soft drink consumption and mortality outcomes were found when the multivariable models excluded BMI adjustment (eTable 3 in the Supplement), former: quit Ն20 y; current: pipe, cigar, occasional; current or former: missing; or unknown); smoking duration (<10, 10 to <20, 20 to <30, 30 to <40, Ն40 y, or smoking duration unknown); ever use of contraceptive pill (yes, no, or unknown); menopausal status (premenopausal, postmenopausal, perimenopausal or unknown menopausal status, or surgical postmenopausal); ever use of menopausal hormone therapy (yes, no, or unknown); and intakes of total energy (kcal per day), red and processed meat (g per day), fruits and vegetables (g per day), coffee (g per day), and fruit and vegetable juice (g per day) (all continuous); and stratified by age (1-year categories), EPIC center, and sex. d Sugar-sweetened and artificially sweetened soft drinks were mutually adjusted. Italy, Spain, and Sweden were excluded from these analyses because information on type of soft drink consumption was not collected.

Discussion
In this large multinational European study, higher level of consumption of total, sugar-sweetened, and artificially sweetened soft drinks was associated with increased risk of death from all causes. The positive association between soft drink consumption and mortality was evident for both men and women. Only artificially sweetened, and not sugarsweetened, soft drinks were associated with deaths from circulatory diseases, whereas for digestive disease deaths, only sugar-sweetened soft drinks were associated with higher risk. The high level of consumption of sugar-sweetened and artificially sweetened soft drinks has previously been linked to elevated risks of obesity, type 2 diabetes, and cardiovascular

Sugar-Sweetened Soft Drinks
The hazard ratios (HRs) are for comparing participants who drank Ն1 glass per day with those who had <1 glass per month. Multivariable Cox regression model adjusted for body mass index (BMI), calculated as weight in kilograms divided by height in meters squared (<22, 22 to <25, 25 to <30, 30 to <35, or Ն35); physical activity index (inactive, moderately inactive, moderately active, or active); educational status (none; primary school completed; technical or professional school; secondary school; longer education, including university; or not specified); alcohol consumption (nonconsumer, <5, 5 to <15, 15 to <30, or Ն30 g per day); smoking status and intensity (never; current: 1-15 cigarettes per day; current: 16-25 cigarettes per day; current: Ն16 cigarettes per day; former: quit Յ10 y; former: quit 11-20 y; former: quit Ն20 y; current: pipe, cigar, occasional; current or former: missing; or unknown); smoking duration (<10, 10 to <20, 20 to <30, 30 to <40, Ն40 y, or smoking duration unknown); ever use of contraceptive pill (yes, no, or unknown); menopausal status (premenopausal, postmenopausal, perimenopausal or unknown menopausal status, or surgical postmenopausal); ever use of menopausal hormone therapy (yes, no, or unknown); and intakes of total energy (kcal per day), red and processed meat (g per day), fruits and vegetables (g per day), coffee (g per day), and fruit and vegetable juice (g per day) (all continuous); and stratified by age (1-year categories), EPIC center, and sex. Sugar-sweetened and artificially sweetened soft drinks were mutually adjusted. Italy, Spain, and Sweden were excluded from the sugar-sweetened and artificially sweetened soft drinks analyses because information on type of soft drink consumption was not collected. Median alcohol consumption was 5.4 g per day. EPIC indicates European Prospective Investigation into Cancer and Nutrition. disease. 1,[25][26][27] We found that sugar-sweetened soft drinks were positively associated with all-cause mortality, a result consistent with findings from an HPFS/NHS analysis 11 but inconsistent with findings from smaller Singapore-and US-based studies. 10,28 We also found positive associations between consumption of artificially sweetened soft drinks and all-cause mortality, a finding consistent with results of the HPFS/NHS and Women's Health Initiative studies. 11,12 Overall, to our knowledge, this current study was the largest to date to investigate the associations between soft drink consumption and mortality outcomes as well as the first comprehensive European-based analysis. We found nonlinear J-shaped associations for soft drink consumption and all-cause mortality, with higher risks observed at consumption levels of more than 125 mL per day (half a glass) of artificially sweetened soft drinks and more than 225 mL per day (approximately 1 glass) of sugar-sweetened soft drinks, and the lowest risks found from drinking 50 mL per day. The reduced risk found at low consumption levels may be the result of reverse causality (analogous to what has been observed for alcohol consumption and all-cause mortality 29-31 ), driven by participants with disease symptoms reporting nonconsumption of soft drinks.
The role of BMI in the soft drink consumption and mortality outcomes association is complex, with adiposity likely a mediating and confounding factor that varies by cause of death. In our analyses, results were unchanged between the multivariable models with or without BMI adjustment, suggesting that the observed associations may be independent of adiposity. In support of this hypothesis, positive associations were found between total, sugar-sweetened, and artificially sweetened soft drinks with deaths from all causes, circulatory diseases, and digestive diseases among participants with healthy weight. These results may suggest that soft drinks al-ter mortality risk independently of adiposity, possibly because of the high glycemic index of sugar-sweetened soft drinks, 32 which elevates blood glucose levels and may in turn lead to insulin resistance and inflammation. 33,34 For sugar-sweetened soft drinks, we found a positive association with all-cause mortality for participants who were obese but no association for participants who were overweight. The lack of association among overweight participants is inconsistent with the result of an analysis of the HPFS/ NHS study, which found a positive association with all-cause mortality among overweight adults. 11 It is unclear why we observed a positive association for sugar-sweetened soft drinks among obese but not among overweight participants. We cannot exclude the role of chance in these subgroup analyses, and additional large-scale prospective studies are required to examine these associations further.
In cause-specific analyses, we found positive associations between artificially sweetened soft drink consumption and deaths from circulatory diseases; these results are consistent with those in the HPFS/NHS and Women's Health Initiative analyses. 11,13 These results were largely based on the positive association between ischemic heart disease deaths and consumption of artificially sweetened soft drinks. Reverse causality is a possible explanation for this positive association, with unhealthy individuals at the study baseline (eg, those who were overweight or obese; those with prediabetes) switching from drinking sugar-sweetened to artificially sweetened soft drinks to control their body weight. However, this association persisted when deaths recorded in the first 8 years of follow-up were excluded. In addition, positive associations between artificially sweetened soft drinks and all-cause and circulatory diseases mortality were found among participants with healthy weight. Possible biological mechanisms that may explain the positive associations between artificially sweetened soft drinks and mortality outcomes are unclear. Limited experimental evidence suggests that artificial sweeteners may induce glucose intolerance, 35 but further studies are needed into the possible adverse effects of the long-term consumption of artificial sweeteners commonly used in soft drinks, such as aspartame and acesulfame potassium. 36 A higher level of soft drink consumption was associated with greater risk of death from digestive diseases, with a positive association only found for sugar-sweetened soft drinks. Hyperglycemia resulting from consumption of sugarsweetened soft drinks may alter gut-barrier function and in-crease the risk of enteric infection. 37 Furthermore, fructose, a sugar commonly used in soft drinks, promotes liver lipogenesis, which can lead to nonalcoholic fatty liver disease and lower insulin sensitivity. [38][39][40] We observed no association between soft drink consumption and overall cancer mortality. This result is consistent with findings in most previous studies, which found little evidence of a direct association between soft drink consumption and cancer risk, 2 but it is inconsistent with that in the HPFS/NHS analysis, which reported a positive association between sugarsweetened beverages and cancer mortality. 11 We did observe former: quit Ն20 y; current: pipe, cigar, occasional; current or former: missing; or unknown); smoking duration (<10, 10 to <20, 20 to <30, 30 to <40, Ն40 y, or smoking duration unknown); ever use of contraceptive pill (yes, no, or unknown); menopausal status (premenopausal, postmenopausal, perimenopausal or unknown menopausal status, or surgical postmenopausal); ever use of menopausal hormone therapy (yes, no, or unknown); and intakes of total energy (kcal per day), red and processed meat (g per day), fruits and vegetables (g per day), coffee (g per day), and fruit and vegetable juice (g per day) (all continuous); and stratified by age (1-year categories), EPIC (European Prospective Investigation into Cancer and Nutrition) center, and sex. b One glass is equal to approximately 250 mL. c Number of deaths based on total soft drink consumption models. d Sugar-sweetened and artificially sweetened soft drinks were mutually adjusted. Italy, Spain, and Sweden were excluded from these analyses because information on type of soft drink consumption was not collected. a positive association between total soft drink consumption and risk of colorectal cancer mortality, but our analyses could not ascertain whether sugar-sweetened or artificially sweetened soft drinks were factors in this association. In the HPFS/NHS analysis, a borderline positive association was observed between sugar-sweetened beverages and colon cancer mortality. 11 Further studies into soft drinks and cancer are warranted to identify whether direct or indirect (through weight gain or overweight or obesity status, which are strong risk factors for multiple cancers 41 ) associations exist for various cancer types.
In addition, we observed that a higher level of total soft drink consumption was associated with greater risk of Parkinson disease mortality, with positive nonsignificant associations found for sugar-sweetened and artificially sweetened soft drinks. To our knowledge, this study is the first to link soft drink consumption with Parkinson disease, and additional studies are required to examine this association.

Limitations
To our knowledge, this study is the largest to date to investigate the association between soft drink consumption and mortality. However, it has several limitations. Given the observational design of the study, it is not possible to establish causality between soft drink consumption and mortality, and we recog-nize that the observed associations may be biased because of residual confounding. However, the large number of participants and recorded deaths (approximately 42 000) allowed us to conduct analyses by subgroups of other mortality risk factors, and we generally observed similar associations across subgroups of considered risk factors. Furthermore, the negative control analysis found no association between consumption of soft drinks and deaths from external causes. This study was also limited by a single assessment of soft drink consumption at baseline.

Conclusions
In this study, the high level of consumption of total, sugarsweetened, and artificially sweetened soft drinks was associated with elevated risks of death from all causes. Positive associations were observed between sugar-sweetened soft drinks and digestive disease deaths as well as between artificially sweetened soft drinks and circulatory disease deaths. Further studies are needed to investigate the possible adverse health effects of artificial sweeteners. The results of this study are supportive of ongoing public health campaigns aimed at reducing the consumption of soft drinks.