Literature search and study selection

In accordance with the Meta-analysis Of Observational Studies in Epidemiology (MOOSE) guidelines , we conducted a literature search regarding the effect of SSB and ASB intake on hypertension, using PubMed, Embase and Web of Science databases up to May 2015. The search terms were as follows: “sweetened beverage, sweetened drink, sugary drink, soft drink, carbonated drink, soda, fruit drink, lemonade, squashes or punch” combined with “hypertension, hypertensive, blood pressure, systolic OR diastolic”. In addition, we reviewed the reference lists of relevant articles to search for more eligible studies. Only human studies published in English as full-length articles were considered. Additionally, studies were included in the meta-analysis if they met the following criteria: a prospective cohort design; the exposure of interest was the consumption of SSBs or ASBs; the outcome of interest was defined as incident hypertension or high blood pressure; relative risks (RRs) with 95% confidence intervals (CIs) were reported. Studies focused on patients with specific diseases were excluded. Among the eligible studies, two articles reported results from the same cohort , so we included the results from the more recent article in the meta-analysis .

Data extraction

Two investigators (Y.K. and Y.J.) independently extracted data from selected articles, and discrepancies were resolved by reviewing the original articles and discussion. We considered one article as three prospective studies, because it reported separated RRs from three large cohort studies, including the Nurses’ Health Study (NHS) I , the NHS II and the Health Professionals Follow-up Study (HPFS) . The following data were extracted for each study: study design; duration of follow-up; publication year; first author’s last name; cohort name; geographic region; sample size; beverage category; adjustment factors; and RR and 95% CI for the association between various categories of SSB or ASB intake and hypertension. The most fully adjusted RRs were extracted when the study provided several RRs.

Quality assessment

We used the Newcastle-Ottawa quality assessment scale to assess the quality of studies included in the meta-analysis. Studies were evaluated on three aspects, as follows: the selection of study groups (0–4 points); the comparability of groups (0–3 points); and ascertainment of outcome (0–4 points). Studies with a total score of ≥ 8 points were considered to be of high quality.

Statistical analysis

The natural logarithm of the RR of each study was combined using the DerSimonian and Laird random-effects models, which incorporate both within-study and between-study variations , to estimate a pooled RR of hypertension for the highest versus lowest category of beverage consumption. The RR from each study as well as a pooled RR are presented as a forest plot, where the size of the data markers (squares) corresponds to the inverse of the variance of the natural logarithm of the RR from each study, and the diamond indicates a pooled RR. To test statistical heterogeneity among the studies, the Cochran Q test was used, and inconsistency was quantified by I ² statistics [100% × (Q– df )/Q] . We conducted subgroup analyses by sex, duration of follow-up and adjustment for body mass index (BMI), and used a meta-regression model to assess the variations in risk estimates by study characteristics. Furthermore, we performed a sensitivity analysis in which one study at a time was excluded; the pooled estimate was recalculated for the other studies to evaluate the effect of each individual study.

The dose-response association among different categories of beverage intake was estimated using generalized least-squares trend estimation analysis, based on the method developed by Greenland and Longnecker . When the study did not report the median or mean dose for categories of beverage consumption, we used the midpoint of the category. The highest open-ended category was assumed to have the same interval of intake as the previous category. After unifying the unit to servings/day, we estimated the pooled RR for a 1 serving/day increase in beverage consumption. One study included in the highest versus lowest meta-analysis was left out because it had only two categories of beverage intake . In addition, we examined a potential non-linear dose-response relationship between beverage intake and hypertension, by adding a quadratic term of beverage intake into the model.

Finally, publication bias was assessed through the tests of Begg and Mazumdar and Egger et al. . A two-sided P < 0.05 was considered statistically significant. All the statistical analyses were carried out using Stata/SE software, version 12.0 (Stata Corporation, College Station, TX, USA).

Study characteristics

The detailed literature search process is presented in Fig. 1 . We identified seven prospective cohort studies with a total of 250,185 subjects and 81,471 incident cases of hypertension that met the inclusion criteria of the meta-analysis . The characteristics of the prospective cohort studies included in the meta-analysis are presented in Table 1 . Six studies were conducted in the USA and one study was performed in Spain. The age at baseline ranged from 18 to 84 years, and duration of follow-up ranged from 4 to 38 years. All studies used food frequency questionnaires for dietary assessment. Four studies provided sex-specific RRs . The systolic and diastolic blood pressure criteria for defining hypertension were > 140 and 90 mmHg in four studies , ≥ 130 and 85 mmHg in two studies and ≥ 135 and 85 mmHg in one study . The ascertainment of hypertension was done through the measurement of blood pressure , medical record review and self-report . All studies adjusted for age, smoking, physical activity and energy intake in multivariable models. Four studies provided RRs that were adjusted for BMI . Study quality scores ranged from 6 to 9 out of a possible score of 11 points. Of the seven studies, five received a score of > 8 points, indicating high quality .

Process used to select prospective cohort studies for a meta-analysis. ASB: artificially sweetened beverage; SSB: sugar-sweetened beverage.

Table 1

Characteristics of prospective cohort studies included in the meta-analysis of consumption of sugar-sweetened and artificially sweetened beverages and the risk of hypertension.

Source	Country; cohort name	Duration of follow-up (years)	Age at baseline (years)	Cases/subjects ( n / n )	Sex	Consumption of SSBs/ASBs (serving)	RR (95% CI)	Adjustment for covariates
Dhingra et al., 2007	USA; FHS	4	52.9	1377/6449	M and F	SSBs		Age, sex, baseline level of the metabolic syndrome component, physical activity index, smoking, dietary consumption of saturated fat, trans fat, fibre, magnesium, total calories and glycaemic index
						None (ref)	1.0 (ref)
						1/day	1.16 (0.92, 1.47)
						≥ 1/day	1.18 (0.96, 1.44)
						≥ 2/day	1.20 (0.90, 1.60)
Nettleton et al., 2009	USA; MESA	7	45–84	843/3363	M and F	ASBs		Age, study site, sex, race/ethnicity, energy intake education, physical activity, smoking status, pack-years and weekly supplement use or more
						Rare/never	1.0 (ref)
						< 1 serving/week	1.07 (0.83, 1.37)
						<1 serving/day	1.11 (0.91, 1.34)
						≥ 1 serving/day	1.17 (0.95, 1.45)
Duffey et al., 2010	USA; CARDIA	10	18–30	609/2639	M and F	SSBs		Age, sex, race, CARDIA exam centre, weight, smoking status, energy from food, total physical activity, energy from the three other beverages (low-fat milk, whole-fat milk, fruit juice) and energy from alcohol
						Quartile 1 (ref)
						Quartile 4	1.06 (1.01, 1.12)
Cohen et al., 2012	USA; NHS I	38	30–55	42,022/88,540	F	SSBs		Age, race, family history of hypertension, physical activity, calcium, magnesium and vitamin D intake, cereal fibre and trans fat intake, carbohydrate consumption, DASH-style diet, total fructose consumption, daily calories, alcohol, whether or not trying to lose weight, smoking status, oral contraceptive use, non-narcotic analgesic use, mutually controlled for SSB and ASB intake, BMI, weight change between surveys
						< 1/month (ref)	1.0 (ref)
						1–4/month	1.02 (0.99, 1.04)
						2–6/week	1.04 (1.01, 1.07)
						≥ 1/day	1.12 (1.08, 1.17)
						ASBs
						< 1/month (ref)	1.0 (ref)
						1–4/month	1.03 (1.00, 1.06)
						2–6/week	1.07 (1.04, 1.10)
						≥ 1/day	1.11 (1.08, 1.14)
Cohen et al., 2012	USA; NHS II	16	25–42	21,873/97,991	F	SSBs		As above
						< 1/month (ref)	1.0 (ref)
						1–4/month	1.00 (0.96, 1.04)
						2–6/week	1.07 (1.03, 1.11)
						≥ 1/day	1.17 (1.11, 1.23)
						ASBs
						< 1/month (ref)	1.0 (ref)
						1–4/month	1.01 (0.97, 1.06)
						2–6/week	1.06 (1.01-1.10)
						≥ 1/day	1.12 (1.08, 1.16)
Cohen et al., 2012	USA; HPFS	22	40–75	13,439/37,360	M	SSBs		Age, race, family history of hypertension, physical activity, calcium, magnesium and vitamin D intake, cereal fibre and trans fat intake, carbohydrate consumption, DASH-style diet, total fructose consumption, daily calories, alcohol, whether or not trying to lose weight, smoking status, non-narcotic analgesic use, mutually controlled for SSB and ASB intake, BMI, weight change between surveys
						< 1/month (ref)	1.0 (ref)
						1–4/month	0.97 (0.93, 1.02)
						2–6/week	1.04 (1.00, 1.10)
						≥ 1/day	1.06 (0.99, 1.14)
						ASBs
						< 1/month (ref)	1.0 (ref)
						1–4/month	1.08 (1.02, 1.13)
						2–6/week	1.09 (1.04, 1.14)
						≥ 1/day	1.20 (1.14, 1.26)
Sayon-Orea et al., 2014	Spain; SUN	8.1	36.4	1308/13,843	M and F	SSBs		Age, sex, baseline BMI, family history of hypertension, self-reported hypercholesterolaemia, physical activity, years of university education, smoking status, total energy intake, energy-adjusted sodium, potassium, low-fat dairy, olive oil, fruit, vegetables, cereals, legumes, meat, whole-fat dairy and fish consumption, alcohol
						None (ref)	1.0 (ref)
						< 7/week	1.07 (0.94, 1.22)
						≥ 7/week	1.34 (1.09, 1.65)

Only gold members can continue reading. Log In or Register to continue

Share this:

• X
• Facebook

Related

Related posts:

1. Long-term prognosis of left-sided native-valve Staphylococcus aureusendocarditis
2. Prediction of new onset of resting pulmonary arterial hypertension in systemic sclerosis
3. Novel materials and devices in the transcatheter management of congenital heart diseases – the future comes slowly (part 1)
4. A French patient cohort with low prevalence of diabetes mellitus and Takotsubo syndrome

Stay updated, free articles. Join our Telegram channel

Join