Study Design
- Daniel Amartya
- May 7, 2020
- 3 min read
Updated: May 11, 2020
Bradford Hill criteria
The Bradford Hill criteria, otherwise known as Hill's criteria for causation, are a group of 9 principles that can be useful in establishing epidemiologic evidence of a causal relationship between a presumed cause and an observed effect and have been widely used in public health research.
Plausibility: the association between cause and effect must make sense. However, Hill stated that mechanism on how can be limited due to lack of current knowledge.
Dose-response: this generally means that a higher dose incites a greater response. This is obviously not true in all cases, but can provide some sort of evidence of a causal relationship.
Temporality: the response occurs after the cause.
Strength of association: a small association does not necessarily mean no effect. However a large association does show that it is likely causal.
Consistency: Otherwise known as reproducibility, we expect consistent findings in different sample populations.
Reversibility: if the cause is removed, then the effect should disappear as well.
Levels of evidence
Systematic review / meta-analysis
What is a systematic review?
Summary of medical literature
Comprehensive literature search and critical appraisal of individual studies, and that uses appropriate techniques to combine these valid studies
Consist of pre-defined eligibility criteria and reproducible methodology, also known as a priori design.
Considered as the strongest type of evidence. Imagine the authors all attempt to answer the same question (topic), but combined into a single analysis.
Why are systematic reviews needed?
Healthcare decisions should be informed by the "best available research evidence"
There is way too many (unmanageable) studies for a person to read, hence a systematic review helps by summarising all available evidence into a single analysis
Helps practitioners to keep up to date with accumulating evidence and to help practice evidence-based medicine
Practice guidelines rely on previously published systematic reviews
Randomised controlled trials (RCT) (experimental study)
What is a randomised controlled trial?
Randomised controlled trial (RCT) is a type of scientific study aimed at reducing certain sources of bias achieved by randomly allocating subjects to two or more groups, treating them differently, and comparing the results to a measured response.
Elements in an RCT
Random allocation: refers to the process of assigning subjects (humans/animals) to different groups by means of randomisation (or simply "by chance")
Blinding: information that may affect or influence the participants is withheld until the experiment is completed.
Data collected prospectively: data collection is taken after intervention has been given. Data is taken forward in time.
Intervention/control groups: control group refers to a group of subjects that receive standard treatment or a placebo (nothing).
Exposure before outcomes
Standard protocol
Reporting in RCT
One of the guidelines that can help researchers report a RCT is CONSORT 2010. It is a checklist that ensures all the important pieces of information is presented in each section of the study.
Cohort studies (observational study)
What is a cohort study?
A cohort study is a particular form of longitudinal study that samples a cohort (group of people who share a defining characteristic, typically those who experienced a common event in a selected period).
There are two types of cohort studies:
Retrospective: looking back in time, thus using existing data such medical records or claims database [ based on my understanding, in general a cohort study spans over a period of time. A retrospective cohort is a study of certain group of time but the analysis uses data obtained in the past.
Prospective: collection of new data
But then, what is the difference between a prospective cohort study and a randomised controlled study?
Intervention in an RCT is investigator controlled, while the intervention in a cohort study is a naturally occurring phenomenon
In a cohort study, it is assumed that the subject at the beginning of the study is "disease free" of the outcome of interest.
Elements in cohort study
Data collected prospectively
Intervention/control group
Exposure before outcomes
Standard protocol
Reporting in cohort studies (observational study)
Similarly, a tool used to help report observational studies is called STROBE statement-checklist
Case control studies (observational study)
What is a case-control study?
A study that compares patients that have a disease or outcome of interest (cases) with patients that do not have the disease (controls), and looks back retrospectively to compare how frequently the exposure to a risk factor is present in each group to determine the risk factor of the disease. The difference with a cohort study, is that a cohort study will begin with the exposure first, and they will follow-up the outcome. A case-control study begins with the outcome of interest and looks back at what may have caused the outcome.
Elements in case-control study
Intervention/control group
Standard protocol
Cross-sectional studies (observational study)
A type of observational study that analyses data from a population, or a representative subset, at a specific point in time. The keyword here is a specific point in time. This is what differentiates a cross-sectional study from the other studies, the data is only taken from one time period.
Elements in cross-sectional study
Snapshot
Comments