Hoorcolleges Blok Algemene Wetenschappelijke Vorming Jaar 2
CLIP 1 – Research Questions :
The goal of medical research is to improve the medical practice. Medical research is performed for
different reasons, described below. This is the classification of research questions.
- To prevent a disease or to make sure about a risk factor. This is called etiology. For example,
is a high caloric diet a risk factor for cardiovascular disease?
- To figure out the diagnosis of a patient. What is the probability of having a hip fracture if the
affected leg is shorter and in exorotation?
- To determine the best treatment. Does chloroquine treatment reduce risk of mortality
among COVID-19 patients admitted to the ICU?
- To determine the prognosis of a disease. What is the probability of dying within 5 years after
breast cancer diagnosis?
Research questions are questions that arise in practise of when reading a paper. They are then the
starting point for designing a study to answer the question. So, research questions must be
answerable. The study to answer the research question is composed of different components. This
could be done via two ways.
- PICO :
o Patient (or population)
o Intervention (most easy describable in a treatment research but still possible to
describe in other researches)
o Comparator
o Outcome
- DDO :
o Domain (this is the population or patient)
o Determinant (intervention + comparator)
o Outcome
“what is the predictive value of a second ultrasound, in case a first ultrasound was inconclusive, in
patients suspected of having an acute appendicitis”?
- Classification : diagnosis
- P : patients with a suspected acute appendicitis, in whom the first ultrasound was
inconclusive.
- I : a second ultrasound
- C:-
- O : appendicitis or not appendicitis
“Is smoking a risk factor for lung cancer?”
- Classification = etiology
- P = humans
- I = smoking
- C = non smoking
- O = lung cancer
,Hoorcolleges Blok Algemene Wetenschappelijke Vorming Jaar 2
CLIP 2 – randomised control trial :
RCT = randomised control trial. For example, if there is a new diuretic drug for patients with
hypertension and you want to study the effect of the drug, you set up a randomised control trial.
- Research question = what is the effect of the new diuretic drug on blood pressure compared
to no diuretic treatment in patients with hypertension.
- Classification = treatment
- P = patients with hypertension
- I = the new diuretic drug
- C = no diuretic treatment
- O = blood pressure reduction.
If you measure a high blood pressure, and give a diuretic treatment. If after a couple weeks you
measure a lower blood pressure, this does not necessarily mean that the drug is effective. You have
to take into account a lot of other factors that may have changed between measure moment one
and two.
Regression to the mean is an effect that can influence the outcome of a study. Some values fluctuate
over time and have a mean value. If you select patients with hypertension and only measure once,
you could have measured during the peak in fluctuation. The blood pressure will then lower anyway
after a period of time, this is the regression to the mean. So the dropping of the blood pressure is
then not because of the drug that was admitted.
The outcome is a combination of different things. The interest in a study of the effect of a drug, the
interest is only in the outcome of the treatment.
- Treatment (T)
- Natural course, like regression to the mean. (NC)
- Extraneous factor. These are mostly lifestyle factors like smoking or going to the gym. But
also other treatments or other drugs. (EF)
- Error processes. The natural variation because of the measurement devices that you use. (V)
Outcome with treatment = T + NC + EF + V
Outcome without treatment = NC + EF + V
To determine the effect of the treatment, you have to compare these two outcomes. So the other
factors have to be the same between the two groups of patients. So the groups have to be
comparable with respect to NC, EF and V and only differ with respect to the treatment. In that case,
an observed difference in the outcome between the groups can be attributed to the only aspects that
the two groups differ on, the treatment.
So in order to achieve the comparability, the RCT has design elements, these are
- Randomisation and concealment of allocation. The population is randomly divided between
the two control groups, this is randomisation. The patients don’t know in which group they
are, this is concealment of the allocation. To be sure, the doctors who treat the patients,
don’t know either in which group the patient is. As a result, treatment allocation is
independent of patient characteristics.
- Blinding. This means that someone doesn’t know what type of treatment the patients
receive. Participants should not know which treatment they receive because that could
influence their behaviour. This also applies to treating physicians, nurses and relatives.
, Hoorcolleges Blok Algemene Wetenschappelijke Vorming Jaar 2
Blinding aims to keep the groups comparable during the follow-up. There are different ways
to achieve blinding. So concealment of allocation is that the patient doesn’t know if he is in
the placebo or treatment group. Blinding is that the patient doesn’t find out in which group
he is. So making the non-treatment comparable to the actual treatment. Via :
o Placebo. The treatment that tastes, looks and smells like the active treatment but
doesn’t contain the active compound. This is sometimes difficult in surgery of
physiotherapy.
o Active comparator. This is a different treatment with the same goals.
Blinded outcome assessment is that the one who measures the outcome should not know
about the treatment status or patients, because that could influence the measurements. One
exception is an all-cause mortality.
- Standardisation is the fourth design element. This is that certain elements are standardised.
This is to minimize the error processes and to improve interpretability of the treatment
effect. So there is a standardisation of intervention, so it is very clearly stated how, when and
how much treatment is admitted. Standardisation of concomitant care, is what doctors are
and are not allowed to do during the trial to prevent interference with the treatment. And
standardisation of outcome assessment, so how the measurement are obtained. In the
example of blood pressure it is clearly stated how the blood pressure should be measured,
how many times and during which period of the day.
So comparability is key.
Going back to the example, the research question
is still vague. A comparator could be a placebo
drug or different diuretic drug. Note that this will
answer a different question. If the comparator is
hydrochlorothiazide (= different drug), the study
will answer the question whether the new drug will have more of less effect that
hydrochlorothiazide, not if the new drug is effective at all.
With every study, you have to take into account the fact if the study is ethical and if there is a
relevant comparison. Like if there is a known working drug for hypertension, is it ethical to treat
patients with a placebo?
CLIP 3 – Sample size calculation :
The sample size calculation is the calculation you use to determine how many patients (= sample size)
are needed in a study design. If the patients are divided into two groups, and there are too few
patients, no differences are detectable. And if they are detected, they could have happened because
of chance and not of the study. If you have too many participants, the study may not be ethical and
too expensive. There are different factors that are important for deciding the sample size. Practical
factors are ;
- Number of eligible patients treated at a center.
- Number of patients willing to participate.
- Time
- Money
Statistical factors are ;
