BMS1042 Lecture Notes - Lecture 7: Confidence Interval, Statistic, Cohort Study
20 Jun 2018
School
Department
Course
Professor
STUDY DESIGN
The methods behind the results are the key to good research. Poor choice of study design is the
downfall of many research studies. Why? Think about all the ways that choosing the wrong
study design administered in the wrong way can impact on your outcomes?
Are you introducing error or confounding? Is the project feasible?
Over the next two weeks you will learn about the pros and cons of each study design. This
theme will equip you with the knowledge to be able to start thinking about which research
questions is match with correct study design.
7. CHOOSING THE RIGHT STUDY DESIGN
First, we need to define a few terms.
Case: Someone who has the disease
Control: Some who does not have the disease
Exposure: A factor that changes the likelihood of the development of the disease
Outcome: The outcome of the study, often the development of the disease.
The point is, we use different study designs to examine the relationship between “exposure” and
an “outcome”.
Identify, describe and differentiate between descriptive and analytical study designs
Both descriptive and analytical study designs are observational trials. Unlike experimental trials,
there is no intervention and we let nature take its course.
In descriptive studies, there is no comparison group, and so is often seen as the lowest level of
evidence. It’s mostly used to generate hypotheses, and can act as the first observation in a rare
occurrence. In analytical studies, we have a comparison group.
Identify, describe and differentiate between case-study, ecological, cross-sectional,
case-control, cohort studies and RCTs (with advantages and limitations)
DESCRIPTIVE:
Case studies
are detailed reports of a single patient that has experienced a particularly adverse
and rare event. A case series
is a report on a series of patients who have the particularly
adverse and rare event. These are descriptive studies as there is no control group.
These are usually used to just fix the problem, so you don’t want to compare anyway.
Advantage: Quick, cheap, and identifies rare diseases/ exposures, which lead to more rigorous
studies.
Limitations: Selection bias, confounders.
Ecological studies
are population-based studies. Whatever results we find, we cannot infer on
an individual. It basically compares different populations at the same time, or the same
population at different time periods.
Advantages: Quick, easy, cheap, and hypothesis generating
Limitations: Highly susceptible to bias and confounding, can only infer associations and not
causations, and can easily confuse characteristics of groups for characteristics of individuals.
ANALYTICAL
Cross-sectional
studies gather information about the exposure and outcome at the same time.
This is really just a snapshot in time scenario. Here, we’re only looking at the prevalence of a
disease.
Advantages: Cheap, easy, able to estimate prevalence, and useful for understanding the
etiology of a disease by generating a hypothesis.
Limitations: We cannot infer a causal inference because it is a “snapshot” in time, therefore we
cannot infer what is the cause and what is the effect. The situation may also provide differing
results if another time-frame had been chosen.
Case controls
are a simple method of investigating rare diseases. They compare the
occurrence of possible causes in cases and controls. Here, data is collected in a point in time,
whilst exposures are collected at a previous point in time. Thus, case-controls have a
retrospective nature as we are first looking at cases and controls as defined by the
absence/presence of a disease. Then, we look backwards in time to assess the presence of the
exposure of interest. The key is, you know the disease, you know the outcome, but you do not
know the exposure.
The direction of inquiry is backwards.
The exposure can be determined by patient interviews or medical records. It’s our job to identify
the start and duration of exposure, which is only determined after the development of the
disease. In other words, the development of the disease focuses us to find a common exposure.
Advantage: Easy, fast, cheap, and good for rare outcomes and long-term diseases.
Limitations: The selection of the control group is difficult as unknown exposure beings
uncertainty in risk factors, therefore uncertainty in “matching” participants. Since it is
retrospective in nature, there is a high probability of recall bias, selection bias, and
measurement error. The way it is dealt is by increasing the sample of the control group to give a
more accurate population.
In case controls, we’re interest in odds ratio which is the odds of exposure in cases / odds of
exposure in controls.
Cohort studies
are observational and analytical studies. They allow nature to take its course,
whilst also having a comparison group. A cohort study involves following patients up over time.
It’s about comparing the incidence of an outcome between the exposed and not exposed. In
every cohort study, participants are selected on the basis of exposure, and then they are
followed up to identify whether or not they have the outcome. The direction of inquiry is
forwards, regardless of a prospective or retrospective study.
Prospective: Start with the population of interest and select a sample. Then separate the group
according to exposed and unexposed. Last, follow up to see if they have the disease.
Retrospective: This is the same except we look back to past documents to search for
suspected exposures.
The cohort needs to be free of the outcome prior to the exposure, but also be potentially
exposed and therefore at risk. The biggest problem is that the longer the study goes on, the
more likely participants are to drop out of the study.
Advantages: Identifies the natural history of a disorder, it identifies the temporal sequence
between cause and outcome, it is good for rare exposures and common outcomes, and involves
a rigorous epidemiological design.
Limitations: Selection bias, control and experimental group may differ on important predictors
of outcome, it is insufficient to study rare diseases, loss of follow-up, and it is expensive and
time-consuming.
Cohort studies uses relative risk (risk ratio).
Document Summary
The methods behind the results are the key to good research. Poor choice of study design is the downfall of many research studies. Over the next two weeks you will learn about the pros and cons of each study design. This theme will equip you with the knowledge to be able to start thinking about which research questions is match with correct study design: choosing the right study design. First, we need to define a few terms. Control: some who does not have the disease. Exposure: a factor that changes the likelihood of the development of the disease. Outcome: the outcome of the study, often the development of the disease. The point is, we use different study designs to examine the relationship between exposure and an outcome . Identify, describe and differentiate between descriptive and analytical study designs. Both descriptive and analytical study designs are observational trials.
