Introduction
- Poor
quality of diet is a major cause of mortality and disability worldwide
- Diet-related
health burdens due to non-communicable chronic diseases (NCDs) are now
surpassing those due to undernutrition in nearly every region of the world
- However,
the differences in dietary patterns across the world, and how such dietary
patterns are changing with time, are not well established.
- Little
is known about dietary patterns across the world based on consumption of
healthier foods and nutrients versus consumption of unhealthy foods and
nutrients.
Aim
- To
characterise global changes (or trends) in dietary patterns nationally and
regionally and to assess heterogeneity by age, sex, national income, and type
of dietary pattern.
- We
analysed global dietary information derived from individual-based national
surveys
Methods
Methods
1)
Global dietary consumption by country, age, sex, and time
-
Focused on 20 foods & nutrients having at least probable or
convincing evidence of effects on major NCDs, including cardiovascular
diseases, diabetes, and diet-related cancers
- Systematically
searched, identified, and compiled data from nationally representative
dietary surveys, large subnational surveys (when national surveys were not
available), and UN FAO food balance sheets; for sodium intake, surveys
assessing urinary sodium were additionally identified
- Total
information compiled: 325 dietary surveys (233 nationally representative;
154 were undertaken before 2000), covering 88.7% of global adult population;
142 surveys on urinary sodium, covering 71.9% of global adult population.
- Also
compiled year-specific data for national availability of 10 foods and 10
nutrients
-
Focused on data for adults (aged ≥ 20 years) only.
- Dietary
intakes adjusted for 2000 kcal/day was evaluated to assess diet quality
independently of diet quantity and to reduce measurement error within and
across surveys (because energy intake is related to under-reporting or
overreporting of dietary consumption and adjustment for total energy intake
partly corrects the error).
- For
all dietary factors, an age-integrating Bayesian hierarchical model was
developed, that estimated the mean intake levels and its statistical
uncertainty for each age-sex-country-year stratum, accounting for
differences in dietary data, survey methods, representativeness, and sampling
and modelling uncertainty
- The
dataset included estimates of dietary consumption for 26 subgroups (men and
women and 13 age categories from 20-24.9 years to ≥80 years) within all 187
countries with a year 2000 population greater than 50 000 in 1990 and 2010, covering
4.42 billion adults across 21 world regions.
2) Characterisation
of dietary patterns
- For analysis, 17 of the 20 dietary factors
were compiled, excluding 3 factors (calcium [assessed milk instead]; seafood
omega-3s [assessed fish instead]; and fruit juice, with its equivocal evidence
for effects on major health outcomes)
- 2
different dietary patterns were modelled:
i) Based
on relatively high consumption of 10 healthy items (fruits, vegetables,
beans and legumes, nuts and seeds, whole grains, milk, total polyunsaturated
fatty acids, fish, plant omega-3s, and dietary fibre);
ii) Based
on relatively low consumption of 7 unhealthy items (unprocessed red
meats, processed meats, sugar-sweetened beverages, saturated fat, trans-fat,
dietary cholesterol, and sodium).
- For comparison,
overall dietary pattern that incorporated all 17 dietary factors together
were also modelled.
To derive a score for each pattern, the mean age-specific, sex-specific, and nation-specific intakes of each dietary factor in 2010 were divided into quintiles, based on all 4862 age-specific, sex-specific, and country-specific strata.
- Each
quintile was assigned an ordinal score
- Higher
scores were given to quintiles with higher mean intakes of healthier foods
(1 to 5 points).
- Higher
scores were given to quintiles with lower mean intakes of unhealthier foods
(5 to 1 points)
- For
each population stratum, scores across different dietary items were summed to
obtain the total score for each of three dietary patterns: healthy items, unhealthy
items, and all items combined
- For comparability,
every score was standardised to a 100-point scale (higher scores equivalent
to healthier diets).
- To optimise
comparability of trends over time, the quintile cutpoints for every dietary
factor in 2010 were used to generate quintile cutpoints for every dietary item
in 1990
Statistical Analysis
- Each
dietary pattern was assessed by country, sex, age, and national income.
- Hierarchical
linear regression was modelled in which age and sex strata were nested within
every nation and random intercepts were estimated
- To
estimate national, regional, and global means, each age and sex stratum was weighted
by the proportion of adults within each contributing country.
- Every
model included age, sex, and national income simultaneously to assess whether
any of these key sociodemographic factors were independently associated with
the dietary pattern score.
- To
test linear trends in dietary patterns across age and national income, ordinal categories
of age and income were assessed as continuous variables.
- Similar
models were used to test trends in dietary patterns from 1990 to 2010, after
standardisation of age and sex distributions to 2010 to assess changes independent
of varying demographics with time
Statistical uncertainty was quantified with Monte
Carlo simulations
- We
simultaneously propagated the uncertainty in the estimated dietary intake of
all items in every age, sex, country, and time stratum by randomly drawing from
the 95% uncertainty interval (UI) of intake and combining results across 1000
iterations. The 95% UIs were derived from estimated SEs based on within-iteration
and between-iteration variances. Using the median and SE, we evaluated Wald
statistics (square of β/SE) to test the null hypothesis for each result from the
regression analyses.
Results
- In
2010, consumption levels of key foods and nutrients related to NCDs varied
across their quintile categories by between two-fold to more than 50-fold (Table
1)
- The largest
variation was noted for mean
- wholegrain consumption (10th to 90th percentiles: 12-157 g/day),
- fruit juice (1.4-86 g/day),
- nuts and seeds (1.5-19.4 g per day),
- beans and legumes (1.6-147 g/day),
- milk (33-230 g/day),
- seafood omega-3 fats (22-553 mg/day),
- plant omega-3 fats (0.2-1.5 g/day),
- sugar-sweetened beverages (33-293 g/day), and
- processed meats (3.9-34 g/day).
- Between
the 17 dietary factors contributing to dietary patterns, correlations across countries
were moderate or weak (r=-0.44 to 0.48).
The mean
(SD, range) global dietary pattern scores out of a maximum
(healthiest) of 100 were:
- 44.0 (10.5,
13.8-64.5) on the basis of ten healthy foods and nutrients,
- 52.1 (18.6,
15.2-93.4) on the basis of seven unhealthy foods and nutrients, and
- 51.9
(9.3, 27.5-75.3) on the basis of all 17 foods and nutrients (Table 2)
- Both
the healthy and unhealthy pattern score were moderately associated with overall
score (Spearman r=0.63 for healthy pattern score, r=0.70 for unhealthy pattern
score). In contrast, the healthy and unhealthy pattern had very little
intercorrelation across countries (r=-0.08, p=0.14)
By Age
and Gender
- For
all three patterns, older adults had better dietary patterns than
did younger adults (table 2). On average, women also had better
dietary patterns than did men.
By
National Income
- Higher
national income was associated with better quality for the healthy dietary
pattern, accounting for 15.7% of between-country variability of the score
(p=0.0005), and
with much worse quality for the unhealthy dietary pattern, accounting for
46.9% of between-country variability (p<0.0001).
- Compared
with low-income countries, high-income countries had higher healthy dietary
pattern scores (adjusted mean difference 2.5, 95% UI 0.3-4.1), but substantially
lower unhealthy dietary pattern scores.
- Unhealthy
dietary pattern scores were also substantially lower in upper middle-income
countries (-25.2; 95% UI -30.2 to -20.2) and lower middle-income (-18.5, -23.7 to -13.2)
countries than in low-income countries
- In
post-hoc analysis, high-income nations showed a nonsignificant positive
correlation between the two types of pattern scores (r=0.27), whereas
low-income nations showed an inverse correlation (r=-0.24;
p>0.05 each; pinteraction>0.1 by
national income). These differences between healthy and unhealthy foods were
largely masked when only one overall dietary pattern score was assessed
By Years
- Between
1990 and 2010, global dietary patterns based on more healthy items improved
modestly (by 2.2 points, 95% UI 0.9-3.5; figure 4, appendix pp 47–51),
indicating greater consumption of these more healthy foods and nutrients.
- By
contrast, global dietary patterns based on fewer unhealthy items worsened (-2.5;
95% UI -3.3 to -1.7), indicating concomitant increased consumption of these
unhealthy foods and nutrients.
-These trends were weakly
correlated across countries (r= -0.08 overall, range -0.15 to 0.09 in
the four national-income categories; p>0.05 each).
- These
trends did not significantly vary by age or sex (p>0.4
each), but significantly varied by national income (p<0.02 each; appendix p
47 figure S24).
- Nations
with higher incomes had larger improvements in diet patterns based on healthy
items than did nations with lower incomes; for example, by 2.5 points (95% UI 0.5-4.6)
comparing high-income to low-income countries.
- By contrast,
middle-income nations showed the largest worsening in diet patterns based on
unhealthy items: compared with high-income nations, greater worsening by 2.5
points (95% UI 0.5-4.5) and by and 2.8 points (95% UI 0.9-4.8) was noted in
upper-middle nations and lower-middle income nations, respectively
- Although
most world regions showed modest improvements in dietary patterns
between 1990 and 2010 on the basis of more healthy items, such
improvements were generally not noted in the poorest regions, including
in sub-Saharan Africa and the Andean states of Latin America
- Conversely,
most regions of the world showed substantial declines in diet quality based
on increased consumption of unhealthy items.
-Despite
some improvement by 2010, dietary scores for unhealthy items in wealthy countries
remained among the worst in the world.
- As
seen for absolute scores, most of these differences in national and regional
trends were far less apparent when examining the dietary pattern aggregating
both healthy and unhealthy dietary items (figure 4).
Discussion
- Consumption
of healthier foods and nutrients has modestly increased during the past two decades;
however, consumption of unhealthy foods and nutrients has increased to a
greater extent
- Improvements
in healthier foods were seen in high-income and middle income countries; by
contrast, no improvements were seen in the poorest regions
- Associations
between socioeconomic status and diet quality might vary substantially for diet
patterns based on healthy versus unhealthy items, and also that such diet
patterns are only weakly correlated.
- Although
a monotonic relation between wealth and diet quality has been frequently
proposed, we noted high-income nations at both extremes of healthy dietary patterns
Limitations
- Did
not assess within-country variations of diets and socioeconomic characteristics,
- Individual-based
data are subject to measurement errors, and were incomplete for some regions,
dietary factors, and years. These limitations were incorporated into
uncertainty in the analysis, but could cause sampling bias, information bias,
or both.
Conclusion
- Global
diet quality varies substantially by age, sex, and national income, and fairly
independent heterogeneity is evident for diet patterns based on eating more
healthy versus fewer unhealthy foods and nutrients
- Increases
in unhealthy patterns are outpacing increases in healthy patterns in most world
regions
- These
findings emphasise the need to better elucidate the societal, policy, and food
industry determinants of these differences and trends, and to implement
policies to address these inequities and improve diet quality globally
*Note: only
Table 1, Table 2 and Figure 1 are shown here.
No comments:
Post a Comment