Basic principles
Study design
There are two major classes of the study designs:
- Observational, without intervention (the lower below, the better the design) (can only establish
correlation or association, but no causality)
o Ecological study: look at the association between behaviour and something else of
groups; not of individuals so it cannot be extrapolated to an individual. It can give a false
cause-effect correlation.
o Cross-sectional study: measure the outcome and the exposure of individuals at the
same time. You don’t know which one comes first, so the main limitation is the inability
to assign the temporal relationship between exposure and outcome.
o Case-control study: compares cases and controls with a questionnaire. Useful to study
rare diseases (particularly used for diet and cancer). You look back in time. Problems:
Bias like selection bias (choose proper control group)
Confounding: a wrong correlation can be found, intake of one thing can
correlate with other things
Recall bias: the cases can recall better what they have eaten
o Cohort study: individuals register over time what they have eaten. The illness and death
is registered in this period. The studies are large. The advantage is that you look
forward. The cause and effect can be obtained. The limitation is the food frequency
questionnaire (people don’t report accurate) and confounding.
- Experimental, with intervention (most powerful research design, can establish causality)
o Intervention study: change people diets. People are randomly assigned to a group with
experimental or control (placebo) treatment. Relevant parameters are compared before
and after exposure. No intervention study can be done with things that probably harm
people. A casual association can be obtained (not with other observational studies),
there is no possible interference, confounders or bias. Another advantage is that there
are all kinds of parameters you can look at. The problem is effect on long term: people
don’t want to do things for a long time, lot of people are needed and it is expensive.
Food intake can be determined by:
- Food frequency questionnaires (how often and how much food is consumed)
- 24 h dietary recall (all foods and amounts is consumed in past 24 h, performed several times)
- Food diary/record (for weak or longer, major concern is that people change their food
consumption, it is expensive, time consuming and people may become sloppy in recording)
- Taking pictures of food (it is difficult to convert images into quantitative food intake)
The advantage of the approaches is that it is efficient and can be used to collect data of a lot of
individuals. A disadvantage is that people are not 100% truthful.
Converting information from food intake to nutrient intake is done with food composition tables,
however they are incomplete.
1
Study design
There are two major classes of the study designs:
- Observational, without intervention (the lower below, the better the design) (can only establish
correlation or association, but no causality)
o Ecological study: look at the association between behaviour and something else of
groups; not of individuals so it cannot be extrapolated to an individual. It can give a false
cause-effect correlation.
o Cross-sectional study: measure the outcome and the exposure of individuals at the
same time. You don’t know which one comes first, so the main limitation is the inability
to assign the temporal relationship between exposure and outcome.
o Case-control study: compares cases and controls with a questionnaire. Useful to study
rare diseases (particularly used for diet and cancer). You look back in time. Problems:
Bias like selection bias (choose proper control group)
Confounding: a wrong correlation can be found, intake of one thing can
correlate with other things
Recall bias: the cases can recall better what they have eaten
o Cohort study: individuals register over time what they have eaten. The illness and death
is registered in this period. The studies are large. The advantage is that you look
forward. The cause and effect can be obtained. The limitation is the food frequency
questionnaire (people don’t report accurate) and confounding.
- Experimental, with intervention (most powerful research design, can establish causality)
o Intervention study: change people diets. People are randomly assigned to a group with
experimental or control (placebo) treatment. Relevant parameters are compared before
and after exposure. No intervention study can be done with things that probably harm
people. A casual association can be obtained (not with other observational studies),
there is no possible interference, confounders or bias. Another advantage is that there
are all kinds of parameters you can look at. The problem is effect on long term: people
don’t want to do things for a long time, lot of people are needed and it is expensive.
Food intake can be determined by:
- Food frequency questionnaires (how often and how much food is consumed)
- 24 h dietary recall (all foods and amounts is consumed in past 24 h, performed several times)
- Food diary/record (for weak or longer, major concern is that people change their food
consumption, it is expensive, time consuming and people may become sloppy in recording)
- Taking pictures of food (it is difficult to convert images into quantitative food intake)
The advantage of the approaches is that it is efficient and can be used to collect data of a lot of
individuals. A disadvantage is that people are not 100% truthful.
Converting information from food intake to nutrient intake is done with food composition tables,
however they are incomplete.
1
