3. Animal studies and expert opinion
Considered the lowest level of evidence
Animal studies provide mechanistic data but may not apply to humans
Expert opinion can be biased unless based on higher levels of
evidence, i.e. systematic reviews or meta-analyses
4. Observational studies
Studies where a population sample or groups are observed, but there is
no intervention
Includes cross-sectional, case-control, and cohort studies
Subject to confounding
However, may be the most appropriate study design for many studies
involving diet and nutrition
Important for study design to address confounders
Selection of representative population samples
Statistical adjustment for confounders
5. Confounding
When the dependent variable (diet or nutrient) is associated with another
confounding variable than can affect the outcome variable (e.g.
cardiovascular disease risk, cancer risk etc.)
Smoking
Alcohol
Socioeconomic
status
BMI
Age
Sex
6. Cross-sectional studies
Data is collected from a sample of a population group at a single point
in time
Small cross-sectional studies determining associations
Can be used to determine prevalence in populations
Includes data collected from national surveys
NANS - National Adult Nutrition Survey, Ireland 2008-10
NDNS - National Diet and Nutrition Survey, UK rolling survey
NHANES - National Health and Nutrition Examination Survey, US rolling
survey
8. Case-control studies
Typically reserved for studies involving diseases or outcomes that are
very rare
Control participants are selected from the general population and
matched to cases through age, sex, and other potentially confounding
variables
Retrospective analysis of exposure variable
Odds ratio is calculated to determine disease risk
Odds ratio is an estimate of relative risk when the total population is
unknown
10. Cohort studies
Prospective - take a group and define the exposure then monitor the
disease/outcome over time
Prospective cohort studies may last decades
Retrospective - take a group and define by disease/outcome and
retrospectively assess the exposure (i.e. look back in time)
Retrospective cohort studies are subject to poor recall and recall bias
Confounding variables require adjustment in statistical analysis
11. Cohort studies
EPIC
The European Prospective Investigation into Cancer and Nutrition (EPIC) is a
prospective cohort with more than 521 000 study participants enrolled from
23 centres in 10 western European countries.
Nurses' Health Study
Starting with the original Nurses’ Health Study in 1976, the studies are now in
their third generation with Nurses’ Health Study 3 (which is still enrolling
male and female nurses) and count more than 275,000 participants.
12. Randomised-controlled trials (RCT)
Considered the gold-standard research study design
Chances of bias are reduced because any confounding variables
should be randomly distributed in both the treatment and placebo
groups
Many nutritional studies cannot be answered by RCT design studies
What are some reasons that nutritional studies cannot be
conducted using RCT designs?
13. Meta-analyses
Statistical analysis of data from multiple studies
Help to identify heterogeneity in results (i.e. when conflicting results
are reported by different studies)
Increase sample size and therefore statistical power to detect
associations
Give more precise estimates of an effect
15. Systematicreviews
Review of scientific literature that is carried out in a specific and
methodical way
Narrow research question
Search strategy is defined, systematic, and often performed in duplicate
Methods and specific outcomes are defined
Involves some form of quality assessment of studies or sensitivity analysis
of exclusion of potentially biased reports
Results presented categorically by study type/characteristics
Conclusion is evidence-based
17. Systematicreviews
Researchers conducting systematic reviews use explicit, systematic
methods that are selected with a view aimed at minimizing bias, to
produce more reliable findings to inform decision-making.
Cochrane Reviews are updated to reflect the findings of new evidence
when it becomes available
https://www.cochranelibrary.com/
18. Reporting guidelines for studies
www.equator-network.org
STROBE-nut
PRISMACONSORTSTROBE
A collection of reporting guidelines for studies of different designs
Aim to improve the reliability and value of published health research literature
Promoting transparent and accurate reporting and wider use of robust reporting
guidelines
19. Reporting guidelines for studies
STROBE-nut
PRISMA
CONSORT
STROBE
Strengthening The Reporting of OBservational Studies in Epidemiology – nutritional
epidemiology
CONsolidated Standards Of Reporting Trials
Preferred Reporting Items for Systematic Reviews and Meta-Analyses
20. Assessing the quality of studies
Is the study sample representative of the population of interest?
What are the eligibility criteria?
What is the sample size?
Is the calculation of sample size explained in the methods?
What measures have been taken to address potential sources of bias?
In the methods? In the discussion?
Is bias addressed in the methods for dietary assessment?
Is heterogeneity addressed?
21. Canyouthinkofimportantconfoundingvariablesthathaven'tbeenaddressed?
Does the information within the introduction and
discussion reflect the scientific literature?
Are the exposure and outcome variables well defined?
What are the sources of funding?
Is information included on non-respondents and dropouts from the study?
Assessing the quality of studies