Читать книгу Statistics in Nutrition and Dietetics - Michael Nelson - Страница 31

The Dodecahedron: ‘If you hadn’t done this one properly, you might have gone the wrong way’.

When designing an experiment, you should do it in such a way that allows the null hypothesis to be disproved. The key is to introduce and protect a random element in the design.

Consider some research options for the study to test whether anti‐oxidant vitamins supplements improve walking distance in peripheral arterial disease (PAD). In the sequence below, each of the designs has a weakness, which can be improved upon by introducing and protecting further elements of randomization.

1 Choose the first 100 patients with PAD coming into the clinic, give them the treatment, observe the outcome.Patients may naturally improve with time, without any intervention at all. Alternatively, there may be aplacebo effect (patients show improvement simply as a result of having taken part in the study because they believe they are taking something that is beneficial and alter their behaviour accordingly), even if the treatment itself is ineffective.

This is a weak observational study. Introduce a control group which receives a placebo.

1 Allocate the first 50 patients to the treatment group, the next 50 to the placebo group.If the two groups differ by age, sex, or disease severity, this could account for apparent differences in improvement between the groups.

This is a weak experimental study. Introduce matching.

1 Match patients in pairs for age, sex, disease severity; assign the first patient in each pair to receive treatment, the second patient to receive a placebo.

The person assigning patients may have a subconscious preference for putting one patient first in each pair. Does the patient know which treatment they are getting?

This is a weak placebo‐controlled intervention trial. Introduce randomization and blinding.

1 Allocate patients to treatment or placebo randomly within pairs. Make sure that the researcher does not know which patient is to receive which treatment (the researcher is then said to be ‘blind’ to the allocation of treatment). Make sure that the patient does not know which treatment they are receiving (keep the patient ‘blind’ as well). This makes the study ‘double blind’.

2 Conduct a placebo‐controlled randomized double‐blind intervention trial.

‘Randomization properly carried out is the key to success.’

(Sir Ronald Fisher)

Intervention studies of this type are often regarded as the most robust for testing a hypothesis. But sometimes randomized controlled trials are not ethical, especially if the exposure may be harmful (e.g. smoking, or increasing someone’s saturated fatty acid intake), or it is not possible to blind either the subjects or the researchers because the treatment being given is so obviously different from the placebo. In these cases, it is possible to mimic intervention studies in samples that are measured at baseline and then followed up over months or years. These types of studies raise issues about how to deal with factors that cannot be controlled for but which might affect the outcome (e.g. in a study looking at the impact of diet on risk of cardiovascular disease, the influence of smoking and diabetes) and changes in the general environment (e.g. indoor smoking bans) that have the potential to affect all the subjects in the study. Complex statistical analysis can be used to cope with some of these design issues. The ability to test the hypothesis in a robust way remains.