A branch of observational epidemiology that goes beyond making simple observations on prevalence etc. by testing hypothesis about risk factors and the occurrence of certain diseases.
A branch of statistical analysis named after Thomas Bayes based on the concept of conditional probability. This means that prior information is taken into account in interpreting the result of hypothesis testing in a study.
This is a systematic error resulting in an incorrect interpretation between a possible cause (exposure) and an outcome (such as disease). The two main types of bias are selection bias, this means the two or more groups to be compared have not been selected in an even fashion, and information bias, this means that the collection of information about the groups to be compared has not been performed on an "equal playing field".
This means a variable with only 2 possible outcomes e.g. Yes/no or male/female.
A type of analytical epidemiology study which collects people with the disease of interest (cases) and compares them with similar people without the disease (controls) with respect to a previous history of things that may cause the disease (risk factors).
Another type of observational epidemiological study which follows a cohort of people forward in time (sometimes compared with a control group) in order to see if they develop a certain event of interests (e.g. disease).
A range of possible values around a sample summary estimate which will contain the true parameter value with a known probability. A 95% confidence interval will, when estimated repeatedly, be expected to include the true value of the parameter being estimated 95% of the time.
A variable which is independently related to the exposure (a risk factor for a disease) and the outcome (e.g. a disease). The inverse association between perinatal mortality and fathers wearing a silk tie may be confounded by social class!
A statement which requires certain essential features to be reported on clinical trials in the published literature. See http://www.consort-statement.org/
A variable which can take on an infinite number of numerical values within a certain range. Height and weight are examples of continuous variables.
Cross Over Study
A type of clinical trial whereby the participants act as their own control. If a treatment A is compared against treatment B, most typically the participants are undermised to either receiving A or B in the first half of the study. After a brief "washout" period they then swap around so that those receiving A in the first half of the study now receive treatment B and vice versa.
A type of simple epidemiological study which gathers information on possible association between causes and disease by correlating them at a population level rather than at the level of the individual. An example of this would be noticing that eczema prevalence is higher in hard water areas around Nottingham.
This is a way of standardising the magnitude of the treatment benefit in a randomised control study so that it can be combined with other studies. Whilst this is useful, the clinical interpretation of the effect size may be obscure.
Efficacy refers to the extent to which a drug can achieve its purpose under the strict conditions of a randomised controlled trial in ideal conditions. Effectiveness refers to what the drug can achieve when applied in real clinical practice.
These are studies aimed at showing that two treatments are roughly equivalent in their efficacy. They are often undertaken in order to show that a "need to" drug is just as effective as the existing drug in terms of efficacy, but it has some added advantage in terms of less side effects etc.
Evidence Based Medicine
A much abused term. Strictly speaking it should refer to "the conscientious, explicit and judicious use of current best evidence in making decisions about the care of individual patients".
This refers to how applicable the results of a study are to the entire population which ones wishes to generalise to. Internal validity refers to whether the results were true for the people within the study.
This is the effect which makes study participants behave differently because they know they are being observed.
Number of new cases per unit time e.g. the incidence of melanoma is 7 per 10,000 per year in Nottingham.
Intention to Treat
This refers to the principle of including all those participants who were initially randomised in the final analysis. This means that some of the people who have dropped out (for various reasons such as lack of efficacy, adverse effects or reasons not related to the medication) are all included in the analysis.
This is a technique used for statistical analysis when the outcome variable is binary (yes/no). It is particularly useful if there are many factors being explored at once.
A statistical technique for combining data from several clinical trials.
This is the number of people that one would theoretically need to treat in order to gain one additional positive outcome. Mathematically it is the reciprocal of the difference between the rate of events between two groups being compared.
When some variables (such as height and weight) are plotted against their frequency in a population. A bell-shaped curve emerges with certain statistical properties. This is the normal distribution.
This is a common method of expressing the "risk" of an exposure being associated with a disease in a case-control study. It is the ratio of the odds of an event occurring in one group to the other group being compared. You will have lots more about this in the course....
Parallel Group Study
A type of randomised controlled trial whereby participants are split into two or more groups and receive different treatments. They are then followed forward in time to see who gets better etc. Each participant only takes one particular medicine.
A term commonly used when evaluating diagnostic tests. It means the proportion of patients with a particular disease who have a positive test.
This is similar to the concept of sensitivity, except that it now refers to the proportion of people without the disease who have a negative test i.e. who are correctly classified as not having the disease of interest.
This is a measure of the spread of a distribution or values around a mean. If the distribution is a normal distribution, then the mean plus/minus 1.96 standard deviations will normally include 95% of the measurement.
This is a measure of the precision of the mean (rather than a spread of data around the mean like standard deviation). It is equal to the standard deviation of the individual measurements divided by the square root of the sample size.
This is a technique to analyse the study of time to certain events such as death or recovery. There are several methodologies used here, but the common factor is that of overcoming problems with single time point estimation because complete survival data may be missing.
An overview of a particular subject that is done in a systematic way in order to minimise the risk of bias. Typically a systematic review of randomised controlled trials starts with a clear question, a search strategy to identify possible studies, an appraisal of the content and quality of those studies, meta analysis (only if it makes sense), and then interpretation.
Type 1 Error
When conducting a statistical hypothesis test, type 1 error is the probability of rejecting the null hypothesis when it is in fact true. This is around 5% if indeed the "significance" value for the hypothesis testing was set at 0.05.
Type 2 Error
This is the probability of failing to detect an effect when in fact an effect was present. This typically occurs in very small studies which took little chance of demonstrating even sizeable effects. Please see Williams H C, Seed P. Inadequate size of "negative" clinical trials in Dermatology. Br J Dermatol 1993:128:317-326.