Absolute risk the probability of occurrence of a disease, death or other event over a period of time, i.e. incidence rate. (See Communicating risk, p.56.)
Absolute risk reduction the reduction in an outcome (e.g. death) comparing one group (e.g. a new drug) with a control group (placebo or standard treatment). (See How effective is the treatment, p.42 and Intervention studies and clinical trials, p.186.)
Attributable fraction the proportion of disease risk that would be prevented if the risk was changed to that of an unexposed person. (See Communicating risk, p.56).
Attributable risk a measure of the excess risk due to the factor concerned. It is calculated by subtracting the incidence rate in the unexposed (the background risk). (See Risk factors, p.12, Communicating risk, p.56, and Cohort studies, p.178.)
Audit see ‘Clinical audit’.
Basic reproductive number the number of new infections caused by one infected individual in an entirely susceptible population. (See Basic infectious disease epidemiology, p.106.)
Bias a systematic error in the design, the conduct, or the analysis (or a combination of these factors) of a study that can give rise to a false association. (See Basic concepts in epidemiological research, p.148 and Interpreting associations, p.150.)
Blinding (in a clinical trial) means that the patient does not know whether they are getting the experimental treatment or not. In a double blind trial neither the patient nor the investigator knows which treatment they are getting. (See Intervention studies and clinical trials, p.186).
Case a person with the condition of interest either for clinical purposes or study purposes. (See Is there anything wrong? p.10, and Basic concepts in epidemiological research, p.148.)
Case–control study a design of study where people with a condition (cases) are compared with those who do not have the condition (controls) with respect to one or more exposures of interest. (See Case–control studies, p.172.)
Case fatality rate the proportion of people with a disease who die from that disease. (See Decisions, decisions … the principles of clinical management, p.28 and Prognosis: mortality, p.32.)
Chi-squared test a statistical test for comparing the proportions of outcomes in different groups.
Clinical audit a quality improvement process that seeks to improve patient care and outcomes through systematic review of care against explicit criteria and the implementation of change. (See Clinical governance and ethics, p.122 and Chapter 6.)
Clinical guidelines systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances. (See Clinical guidelines, p.38.)
Cohort study a study in which individuals are selected on the basis of exposure status and are followed over a period of time to allow the frequency of occurrence of the outcome of interest in the exposed and non exposed groups to be compared. (See Cohort studies, p.178.)
Confidence interval given a sample of data, the sample value e.g. mean will be the best estimate of the true population value although there will be some uncertainty associated with this. A confidence interval quantifies this uncertainty and gives a range of values in which we are confident the true population value will lie. (See Confidence intervals, p.232.)
Confounding gives rise to a false association between an exposure and disease under investigation because it distorts the observed association through a third factor known as the confounder. (See Basic concepts in epidemiological research, p.148 and Confounding, p.154.)
Confounding variable a factor that is associated with both the exposure and outcome of interest. (See Basic concepts in epidemiological research, p.148.)
Control (as opposed to a case) a person without the outcome under study (in a case–control study), or a person not receiving the intervention (in a clinical trial). (See Case–control studies, p.172 and Intervention studies and clinical trials, p.186.)
Correlation a measure of association that is used for risk factors that are continuous variables or multiple categories. A correlation coefficient (r2) shows how much an increase in risk is explained by an increase in exposure. (See Correlation coefficients, p.262.)
Count the number of cases. The most basic measure of disease frequency. (See Measures of disease frequency, p.6.)
Critical appraisal the process of assessing the validity of research and deciding how applicable it is to the question you are seeking to answer. (See Critical appraisal, p.78.)
Critical proportion (pc) the proportion of the population that needs to be vaccinated to control the infection, and this is dependent on the basic reproductive number R0. (See Basic infectious disease epidemiology, p.106.)
Cross-sectional study a study that collects information on exposure and/or disease at one point in time. (See Surveys and cross-sectional studies, p.166.)
Disability-adjusted life years (DALYs) the sum of years of potential life lost due to premature mortality and the years of productive life lost due to disability. They are used in the Global Burden of Disease study. (See Chapter 11.)
Double blind see ‘Blinding’.
Ecological fallacy when group level information results are inappropriately referred to individuals. (See Time trend and geographical (ecological) studies, p.162.)
Ecological study these examine the association between exposures and outcomes by using aggregate data. (See Time trend and geographical (ecological) studies, p.162.)
Endemic when there is a constant presence (or incidence) of a disease in a particular community or geographic area. (See Basic infectious disease epidemiology, p.106.)
Epidemic when the incidence of a disease rises above the expected level in a particular community or geographic area. (See Basic infectious disease epidemiology, p.106.)
Epidemiology the study of the distribution and determinants of health and illness in populations. (See Types of evidence and study designs, p.146.)
Evidence-based medicine the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients. The practice of evidence-based medicine means integrating individual clinical expertise with the best available external clinical evidence from systematic research. (See Evidence-based management, p.36 and Section 2.)
Experimental event rate the incidence in the intervention arm of a study. (See Intervention studies and clinical trials, p.186.)
Exposure a factor that might affect an outcome. (See Basic concepts in epidemiological research, p.148.)
Forest plot a visual representation of the results of a meta-analysis. (See Meta-analysis, p.90.)
Frequentist statistics a method of analysis in which the data alone are used to make inferences. (See So what has my patient got? The Bayesian clinician, p.26.)
Funnel plot a visual representation to explore publication bias in a meta-analysis. Study size is plotted against the effect estimate (e.g. odds ratio). (See Meta-analysis, p.90.)
Health promotion the process of enabling people to increase control over, and to improve, their health. (See Health promotion in clinical practice, p.52.)
Herd (population) immunity the immunity of a group or population. High levels of immunity in the population protect those who are not immunized. (See Basic infectious disease epidemiology, p.106.)
Hyperendemic when there is a high constant presence (or incidence) of a disease in a particular community or geographic area. (See Basic infectious disease epidemiology, p.106.)
Hypothesis testing two alternative conclusions (hypotheses) are set up, and on the basis of the experiment one is accepted and the other rejected. This acceptance and rejection is always on the basis of some pre-specified levels of statistical confidence. (See Hypothesis tests, p.234.)
Incidence the number of new cases of the disease in a defined population over a defined period of time. (See Measures of disease frequency, p.6 and Cohort studies, p.178.)
Incubation period the time between initial contact with an infectious agent and the first appearance of symptoms of the disease associated with that infectious agent. (See Basic infectious disease epidemiology, p.106.)
Infant mortality rate the number of deaths of infants under age 1 per 1000 live births in a given year. (See Routine data, p.202.)
Infectious period the time during which a person is capable of transmitting the organism. (See Basic infectious disease epidemiology, p.106.)
Intention-to-treat analysis where the analysis of outcome in a clinical trial is based on the initial post-randomization treatment allocation, regardless of whether the participant ended up sticking to that protocol. (See Intervention studies and clinical trials, p.186.)
Lead time bias where survival appears to increase due to earlier detection of the disease as a result of screening. (See Evaluating screening programmes, p.116.)
Length bias occurs because cases detected through screening may be less aggressive. (See Evaluating screening programmes, p.116.)
Likelihood ratio the likelihood that a given test result would be expected in a patient with the disease compared to the likelihood that the same result would be expected in a patient without the disease. (See Likelihood ratios and post-test probabilities, p.22.)
Longitudinal study see ‘Cohort study’.
Matching a method for ‘controlling for’ (i.e. effectively removing) the effect of confounding at the design stage of a case-control study. (See Confounding, p.154.)
Maternal mortality rate the number of maternal deaths per 100,000 births. (See Routine data, p.202.)
Measurement bias see ‘Bias’.
Meta-analysis a statistical analysis of a collection of studies, often collected together through a systematic review. (See Meta-analysis, p.90.)
Morbidity the extent of disease and disability; in relation to prognosis it may include levels of disability, frequency of recurrence, etc. (See Decisions, decisions … the principles of clinical management, p.28 and Chapter 11.)
Natural history usually the course of the disease in the absence of treatment from the point of inception (or exposure to the causal agents) until recovery or death. (See Decisions, decisions … the principles of clinical management, p.28.)
Negative likelihood ratio (LR–) a measure of how much to decrease the probability of disease by if the test result is negative and is equal to 1 minus the sensitivity divided by the specificity. (See Likelihood ratios and post-test probabilities, p.22.)
Negative predictive value the probability that a person with a negative result does not have the condition. (See Validity of diagnostic tests, p.16 and Clinical uses of predictive value, p.18.)
Neonatal mortality rate the number of deaths of infants under 28 days per 1000 live births in a given year. (See Routine data, p.202.)
Normal distribution (also known as the Gaussian distribution or as a bell-shaped curve). This is the most commonly used mathematical distribution in statistics. It is defined for quantitative variables that can take any value from minus infinity to plus infinity, and is characterized by two parameters: its mean and standard deviation. (See Probability distributions, p.230.)
Null hypothesis see ‘Hypothesis testing’.
Number needed to treat (NNT) a measure of how many people with an illness would need to be treated to produce one desired outcome or prevent one adverse outcome (e.g. one heart attack prevented). (See How effective is the treatment, p.42 and Interpreting reports of clinical trials, p.44.)
Observer bias see ‘Bias’.
Odds another way to express probability. The mathematical relationship between odds and probability is: odds = probability / (1 – probability). (See Likelihood ratios and post-test probabilities, p.22.)
Odds ratio an estimate of relative risk obtained from case control studies where the incidence of disease is not known, but the frequency of exposure is measured in cases and controls. The odds ratio (of exposure) is the ratio between two odds, i.e. the odds of exposure in the cases divided by the odds of exposure in the controls. (See Risk factors, p.12, Communicating risk, p.56, and Case–control studies, p.172.)
Outbreak used interchangeably with epidemic, usually for a more localized increase in cases. (See Basic infectious disease epidemiology, p.106.)
Outcome the event or main quantity of interest in a particular study, e.g. death, contracting a disease, blood pressure. (See Basic concepts in epidemiological research, p.148.)
Pandemic when an epidemic spreads over several continents. (See Basic infectious disease epidemiology, p.106.)
Per protocol analysis see ‘Intention to treat’.
Population attributable risk (also known as the population excess risk) the proportion of the disease that is due to the risk factor in the population as a whole, and thus the proportion of disease in the population that should be prevented if the risk is removed (assuming there is a causal relationship). (See Communicating risk, p.56.)
Positive likelihood ratio (LR+) a measure of how much to increase the probability of disease by if the test result is positive and is equal to the sensitivity divided by (1-specificity). (See Likelihood ratios and post-test probabilities, p.22.)
Positive predictive value the probability that a person with a positive test result has the condition. (See Validity of diagnostic tests, p.16 and Clinical uses of predictive value, p.18.)
Pre-test probability the probability of the condition disorder before a diagnostic test result is known. (See Pre-test probability, p.20 and Likelihood ratios and post-test probabilities, p.22.)
Prevalence the proportion of people with a disease at any point (point prevalence) or period (period prevalence) in time. (See Measures of disease frequency, p.6.)
Prevention paradox the fact that many people exposed to a small risk may generate more disease than the few exposed to a large risk. (See Prevention strategies, p.98.)
Primary prevention the prevention of the onset of disease. (See Levels of prevention, p.100.)
Primordial prevention the prevention of factors promoting the emergence of risk factors. (See Levels of prevention, p.100.)
Probability the chance of an event occurring, ranges from 0 (no chance) to 1 (inevitable).
Prognosis the probable course and outcome of an illness, including duration of disease, morbidity, and mortality. (See Decisions, decisions … the principles of clinical management, p.28.)
Prognostic factors variables that influence outcome, such as age, sex, disease stage. (See Decisions, decisions … the principles of clinical management, p.28.)
Protective factor something that reduces a person’s chance of getting a disease. (See Risk factors, p.12.)
Publication bias the greater likelihood that research with statistically significant results will be published in the peer-reviewed literature in comparison to those with null or non-significant results. (See Analysis and interpretation of a meta-analysis, p.92.)
p-value the probability of obtaining the study result (relative risk, odds ratio, etc.) if the null hypothesis is true. The smaller the p-value, the easier it is for us to reject the null hypothesis and accept that the result was not just due to chance. (See Chapter 9.)
Randomization in clinical trials is done to remove bias in treatment allocation. (See Intervention studies and clinical trials, p.186.)
Recall bias see ‘Bias’.
Regression a method for quantifying the relationship between a dependent variable, or response and one or more independent or explanatory variables. Often used for controlling confounding in epidemiological analysis. (See Linear regression, p.260.)
Relative risk the increase (or decrease) in probability of disease given a particular exposure or risk factor. It is estimated in cohort or follow-up studies from the incidence in the exposed (those with the risk factor) divided by the incidence in the unexposed. (See Risk factors, p.12, Communicating risk, p.56, and Cohort studies, p.178.)
Relative risk reduction the percentage reduction in an outcome (e.g. death) comparing one group with a control group. (See How effective is the treatment, p.42 and Intervention and clinical trials, p.186.)
Restriction a method for controlling the effect of confounding at the design stage of a study.
Risk see ‘Absolute risk’.
Risk factor something that increases a person’s chance of getting a disease. This definition assumes that there is a causal relationship rather than simply a statistical association. Risk factors are often described as modifiable (for example environmental exposures and behaviours) or fixed (for example genetic predisposition, age, or ethnicity). (See Risk factors, p.12.)
Risk marker a factor that is associated with an increased risk of disease but has not been shown to be causal. (See Risk factors, p.12.)
Routine health data sources of information relating to the health of the population that are collected in an ongoing way rather than for any specific research project. (See Routine data, p.202.)
Sample the subset of the population that is included in a study. (See Chapter 9 Introduction, p.218.)
Screening the practice of investigating apparently healthy individuals with the object of detecting unrecognized disease or its precursors in order that measures can be taken to prevent or delay the development of disease or improve prognosis. (See Screening overview, p.112.)
Secondary prevention the halting of progression once the disease process is already established. (See Levels of prevention, p.100 and Screening overview, p.112.)
Selection bias occurs when people who participate in studies or screening programmes differ from those who do not. (See Evaluating screening programmes, p.116 and Basic concepts in epidemiological research, p.148.)
Sensitivity the ability of a diagnostic test to correctly identify people with the disease, i.e. the probability that a person with the disease will test positive. (See Validity of diagnostic tests, p.16.)
Specificity the ability of a diagnostic test to correctly identify people without the disease, i.e. the probability that a person without the disease will test negative. (See Validity of diagnostic tests, p.16.)
Standardization a method for controlling the effect of confounding at the analysis stage of a study. Used to produce a Standardized Mortality Ratio, a commonly used measure in epidemiology.
Stratification a method for controlling the effect of confounding at the analysis stage of a study. Risks are calculated separately for each category of confounding variable, e.g. each age group and each sex separately.
Survival the time from diagnosis to the outcome of interest (e.g. death). (See Decisions, decisions … the principles of clinical management, p.28, Prognosis mortality, p.32 and Measuring survival, p.274.)
Systematic review a systematic way of drawing together and synthesizing existing evidence and producing a summary appraisal in a narrative form. (See Systematic review, p.88.)
Tertiary prevention the rehabilitation of people with established disease to minimize residual disability and complications. (See Levels of prevention, p.100.)
Under-5 mortality rate the probability that a newborn baby will die before reaching age 5, as a number per 1000 live births.
Vaccination is the introduction of a substance (vaccine) into the body in order to stimulate the production of antibodies against an infection without causing the active infection. (See Basic infectious disease epidemiology, p.106 and Vaccination, p.108.)
Validity a measure of how well a test distinguishes diseased from non-diseased individuals. The concept can be equally well applied to symptoms and signs. A completely valid test/symptom/sign will be positive for all cases and negative for all non-cases. (See Validity of diagnostic tests, p.16 and Screening overview, p.112.)