Critical appraisal

Question 1

Accuracy

Answer 1

The degree to which a measurement or an estimate based on measurement represents the true value of the attribute being measured. Not to be confused with PRECISION or VALIDITY.

Question 2

Adjustment

Answer 2

A summarizing procedure for comparing two or more sets of data (such as mortality rates in different populations) that minimizes the effects of factors, such as age composition. Adjustment for age and other differences is usually performed with procedures of direct or indirect STANDARDIZATION.

Question 3

Association

Answer 3

Statistical dependence between two or more events, characteristics, or other variables. Associations can be broadly grouped under two headings, causal (or protective) and non-causal, resulting from CHANCE, BIAS or CONFOUNDING.

Question 4

Audit

Answer 4

An examination or review of a process, by implication often a searching review, especially when carried out by an officially constituted authority.

Question 5

Berkson’s bias

Answer 5

Syn: REFERRAL BIAS. A form of SELECTION BIAS that causes, for example, hospital cases and community controls in a CASE CONTROL STUDY to be systematically different from one another because the combination of exposure to risk and occurrence of disease increases the likelihood of being admitted to the hospital. This produces a systematically higher exposure rate among hospital patients, so it distorts the ODDS RATIO.

Question 6

Bias

Answer 6

Deviation of results or inferences from the truth or processes leading to such deviation. Sometimes classified into SELECTION BIAS (e.g. REFERRAL BIAS, LOSS TO FOLLOW-UP BIAS, PUBLICATION BIAS) and INFORMATION BIAS (e.g. recall bias).

Question 7

Case control study

Answer 7

A study which starts with the identification of people with the disease (or the health-related state) of interest and a suitable control (comparison or reference) group without the disease. It then compares the (previous) exposure of cases and controls to the risk factor or factors of interest. It can be PROSPECTIVE or RETROSPECTIVE.

Question 8

Case fatality rate

Answer 8

The proportion of the cases of a condition of interest who die of the condition within a specified period of follow-up after diagnosis and/or medical intervention, often expressed as percentages of deaths in a 1-year or 5-year period.

Question 9

Case report

Answer 9

A descriptive account of a case of a condition of interest, providing details of all the relevant circumstances, occasionally providing a basis for inferences about causes and possible methods of control that could be applied in the future.

Question 10

Case series

Answer 10

A descriptive account of a series of cases of a condition of interest, often highlighting similarities in their circumstances and clinical characteristics.

Question 11

Causal inference

Answer 11

The method of logical reasoning that seeks to use all available facts to arrive at a conclusion about the relationship of particular causes to their effects. This is the main method of reasoning in the public health sciences, especially epidemiology. See also HILL’S CRITERIA.

Question 12

Clinical trial

Answer 12

An EXPERIMENTAL STUDY that involves the administration of a test regimen to humans to evaluate its EFFICACY, EFFECTIVENESS and safety. The term is subject to wide variation in usage, from the fist use in humans without any control treatment to a rigorously designed and executed experiment involving test and control treatments and randomization. Several phases of clinical trials are distinguished:

Question 13

phase I clinical trial

Answer 13

Concentrates on safety and pharmacological and physiological effects. For example, the first introduction of a candidate vaccine or a drug into a human population to determine its safety and mode of action. Phase I trials usually involve fewer than 100 healthy volunteers.

Question 14

phase II clinical trial

Answer 14

Pilot studies of EFFICACY. These usually involve 200 to 500 volunteers; with vaccines, the focus is on immunogenicity, and with drugs, on demonstration of safety and EFFICACY in comparison to other existing regimens. Usually but not always, subjects are randomly allocated to study and control groups.

Question 15

phase III clinical trial

Answer 15

Extensive clinical trials. This phase is intended for complete assessment of safety and EFFECTIVENESS. It involves larger numbers, perhaps thousands, of volunteers, usually with random allocation to study and control groups, and may be a multicentre trial.

Question 16

phase IV clinical trial

Answer 16

With drugs, this phase is conducted after the national drug restriction authority (e.g., the Food and Drug Administration in the United States or the UK Committee on Safety of Medicines) has approved the drug for distribution. Phase IV trials may include research designed to explore a specific pharmacological effect, to establish the incidence of adverse reactions (many of which are rare), or to determine the effects of long-term use. Sometimes termed post-marketing surveillance.

Question 17

Clinical trials of complex interventions

Answer 17

Step 0 – explore relevant theory to ensure best choice of intervention and predict confounders and design issues
Step I – identify components of intervention and evidence to show how they influence outcomes
Step II – define components of replicable intervention and feasible protocol for comparing this with appropriate alternative
Step III – compare intervention and control (both well defined) in many centres with adequate statistical power
Step IV – test whether others can replicate intervention and findings in uncontrolled settings over the long term

Question 18

Cochrane Collaboration

Answer 18

An international organization of clinicians, epidemiologists, patients, and others that aims to help health professionals make well-informed decisions about health care by preparing, maintaining, disseminating, and promoting the accessibility of systematic reviews of the effects of health care interventions. Cochrane Reviews are prepared and updated by collaborating authors, working in Cochrane Collaborative Review Group, using explicitly defined methods to minimize the effects of BIAS, and, where appropriate and readable, META-ANALYSIS to reduce imprecision.

Question 19

Coherence

Answer 19

The quality of evidence that fits most neatly with other facts and existing theories about the situation under investigation. See also HILL’S CRITERIA.

Question 20

Cohort study

Answer 20

A study which starts with the identification of people who are exposed to the risk factor of interest and those who are not. It then compares the incidence of the development of the disease or health-related state in these different groups. The measure of association is the RELATIVE RISK, RATE RATIO or HAZARD RATIO (each of these terms mean essentially the same thing)..

Question 21

Confounding

Answer 21

The distortion of the relationship between two factors brought about by their mutual relationship with a third factor.

Question 22

Confounding factor (or variable), also known as a confounder

Answer 22

A variable that is associated with the outcome of the investigation (such as the disease in a case control study) but which is also associated with the factor (exposure) under study.

Question 23

CONSORT

Answer 23

An acronym meaning consolidated standards for reporting of trials, an initiative of the International Committee of Medical Journal Editors to ensure consistency and quality of the conduct of clinical trials by requiring conformity to a predetermined protocol, consistent use of methods, procedures, participant assignment, masking (blinding) procedures, data analysis, etc

Question 24

Construct validity

Answer 24

The extent to which an experiment, laboratory test, or set of measurements conforms to or correlates with the underlying theoretical concepts. See also VALIDITY.

Question 25

Content validity

Answer 25

In the terminology of survey methods, content validity is the extent to which the test items, e.g., questions, opinion statements offered to participants in a survey, are representative of the domain or topic that is the target of the survey. See also VALIDITY.

Question 26

Critical appraisal

Answer 26

Systematic evaluation of research reports, consisting of a detailed scrutiny and logical analysis of all phases of the process with the aim of assessing whether it conforms to acceptable standards, or, if it does not, identifying its shortcomings.

Question 27

Cross-over design

Answer 27

A research design sometimes used in clinical trials, in which the study and comparison regimens are switched after observing the effects on the study group, in order to determine whether the PLACEBO effect has influenced the outcome.

Question 28

Cross-sectional study

Answer 28

A study that examines the relationship of health to other variables of interest at a particular time. For example a PREVALENCE study of, say hypertension in a defined population.

Question 29

Descriptive study

Answer 29

A study concerned with and designed only to describe the existing distribution of variables, without regard to any hypothesis.

Question 30

Distribution-free method

Answer 30

A method of statistical analysis that does not depend on the mathematical form of the underlying distribution that is being analyzed.

Question 31

Dose-response relationship

Answer 31

The relationship of observed outcomes in a population to varying levels of a protective or harmful agent such as a form of medication or an environmental contaminant. For instance, a linear dose-response relationship reflects a direct increase in risk as exposure increases. The dose-response relationship has central importance in toxicology and environmental health. It is one of HILL’S CRITERIA of causality.

Question 32

Double-blind trial

Answer 32

A procedure of blind assignment to study and control groups and blind assessment of outcome, designed to ensure that ascertainment of outcome is not biased by knowledge of the group to which an individual was assigned. Double refers to both parties i.e., the observer(s) in contact with the subjects and the participants in the study and control groups. Blinding is sometimes referred to as “MASKING”

Question 33

Ecological fallacy

Answer 33

The BIAS that may occur because an association observed between variables on an aggregate level does not necessarily represent the association that exists at an individual level. An example is “particularizing from the general” by assuming that a particular individual faces the same risk of dying from coronary heart disease as the whole population in the “risk category” to which the individual belongs.

Question 34

Ecological study

Answer 34

A study in which the units of analysis are populations or groups of people, rather than individuals (e.g. a study of average body mass index in a number of countries compared with the cancer mortality rates in these same countries). As a research method, ecological studies are fraught with difficult problems, notably the inability to allow for the effects of CONFOUNDING variables. See also ECOLOGICAL FALLACY

Question 35

effect modifier

Answer 35

A factor that modifies the action of a causal factor. For example, age modifies the effect for many conditions, and immunization status modifies the action of pathogenic organisms as causal factors in specific communicable diseases. See also CONFOUNDING.

Question 36

Face validity

Answer 36

The apparent relationship of findings or observations to the topic that is under investigation or observation (“on the face of it”). See also VALIDITY.

Question 37

False negative

Answer 37

A negative test result occurring in a person who actually has the disease or possesses the attribute for which the test is done. Rates of false-negative and false-positive test results are indicators, respectively, of the SENSITIVITY and SPECIFICITY of SCREENING TESTS.

Question 38

False positive

Answer 38

A positive test result in a person who actually does not have the condition for which the test is being done.

Question 39

Forest plot

Answer 39

A diagrammatic representation of the components of a META-ANALYSIS showing the conclusion (odds ratio, relative risk or standardised mean differences) of each individual study, together with its confidence interval and, usually, a pooled estimate of the parameter combining all of the studies identified.

Question 40

Funnel plot

Answer 40

A plotting device used in meta-analysis to detect PUBLICATION BIAS. The estimate of risk is plotted against sample size. If there is no publication bias, the plot is funnel-shaped; but if studies showing significant results are more likely to be published, there is a “hole in the lower left corner” of the funnel.

Question 41

Generalisability (Syn. external validity)

Answer 41

The degree to which the inference can validly be drawn from a study and extrapolated to the wider population from which the study subjects are drawn.

Question 42

Healthy worker effect

Answer 42

The effect of prior selection, observed in the early stages of studies of workers in industries exposed to occupational hazards. Because unhealthy people are more likely to be rejected and healthy people are more likely to be recruited, the sickness and death rates initially may be lower in the exposed workforce than in the general population.

Question 43

Henle-Koch postulates

Answer 43

Syn: KOCH’S POSTULATES. These are four criteria that usually suffice to confirm the causal relationship of a pathogenic organism to a specific infectious disease. The postulates are: 1. The agent must be demonstrable in every case of the diseases; 2. it must produce the disease when a pure culture is given to a susceptible animal 3. After isolation in culture, the agent must be able to produce the disease in experimental animals; and (added by Koch) 4. The agent can be recovered from the experimental animal.

Question 44

Hill’s (or Bradford Hill’s) criteria of causation

Answer 44

The first complete statement of the criteria suggesting that an observed association may be causal:

1. Consistency: The association is consistent when results are replicated in studies in different settings using different methods.
2. Strength: This is defined by the size of the risk as measured by appropriate statistical tests.
3. Specificity: This is established when a single putative cause produces a specific effect.
4. Dose-response relationship: An increasing level of exposure (in amount and/or time) increases the risk.
5. Temporal relationship: Exposure always precedes the outcome. This is the only absolutely essential criterion.
6. Biological plausibility: The association agrees with currently accepted understanding of pathobiological processes. Coherence: The association should be compatible with existing theory and knowledge.
7. Reversibility: The condition can be altered (e.g., prevented or ameliorated) by removing or changing the exposure.

Question 45

Historical cohort study

Answer 45

Syn: RETROSPECTIVE COHORT STUDY. A COHORT STUDY that makes use of pre-existing health-related facts about members of a defined population who can be classified into exposed and unexposed groups on the basis of information that may have been collected years or even decades earlier.

Question 46

Information bias

Answer 46

Systematic error due to differences in the quality of information obtained from different groups in a study. For example, RECALL BIAS.

Question 47

Intention to treat analysis

Answer 47

A strategy used in RANDOMIZED CONTROLLED TRIALS that aims to prevent bias due to loss of participants by including all participants in data analysis, whether or not they completed the intervention given to the group of which they were a member.

Question 48

Inter-observer variation

Answer 48

Sometimes called observer error, but “error” is an inappropriate word for this phenomenon. A pervasive problem in all OBSERVATIONAL studies is that when two or more people examine a patient, an x-ray, an electrocardiogram, a microscopic slide of tissue, blood cells, or other natural phenomena, they do not always agree on what they see and on its significance.

Question 49

Intra-observer variation

Answer 49

Syn: intraobserver error. Differing readings of the same phenomenon by the same observer on second or other subsequent review. With chest x-rays, the same observer reports different findings on re-examination of the same set of x-rays about once in 20 films, i.e., the intraobserver variation rate is about 5%

Question 50

Lead time bias

Answer 50

A systematic error introduced when follow-up does not begin at comparable stages in the natural history of the disease. Lead time bias can occur in comparisons of survival rates for conditions such as cancer when allowance is not made for the stage of the disease when diagnosis was made and treatment begun. Thus, early diagnosis of case in a SCREENING programme and early treatment may or may not influence the survival rate. It may merely mean that the presence of the disease is known for a larger proportion of the person’s lifetime.

Question 51

Length bias

Answer 51

A systematic error due to selection, by screening, of disproportionate numbers of long-duration cases (cases who survive longest) in one group but not in another.

Question 52

Medical audit

Answer 52

A systematic evaluation of the process of medical care based on statistical analyses of length of stay, procedures, outcomes, complication rates, etc., in relation to diagnoses, service providers, etc., all aimed at finding ways to enhance efficiency and EFFECTIVENESS of care.

Question 53

Meta-analysis

Answer 53

A statistical synthesis of the data from separate but similar (i.e. not significantly heterogeneous) group of studies leading to a quantitative summary of the pooled result.

Question 54

Negative predictive value

Answer 54

Syn: PREDICTIVE VALUE OF A NEGATIVE TEST. The probability that a person with a negative test is a true negative (i.e. does not have the disease) (or true negatives as a proportion of all those with a negative test).

Question 55

Nested case control study

Answer 55

A CASE CONTROL STUDY “nested” (i.e. included) within a COHORT STUDY.

Question 56

Observational epidemiological studies

Answer 56

Studies in which exposure is not determined as part of the study design. In other words, already existing exposures and their consequences are studied. The most common examples of observational epidemiological studies are CASE CONTROL and COHORT studies

Question 57

Observer bias

Answer 57

The tendency of an observer to “see what is expected or wanted” rather than what is actually there.

Question 58

Observer error

Answer 58

An error of observation or measurement due to failure of the observer to identify, measure accurately, or interpret some aspect of the phenomena that are being observed.

Question 59

Observer variation

Answer 59

This common phenomenon is not necessarily either OBSERVER ERROR or BIAS. There are two varieties. INTEROBSERVER VARIATION occurs when two or more observers observe the same phenomenon or situation and arrive at different conclusions. INTRA-OBSERVER VARIATION occurs when the same observer repeats an observation, measurement, or assessment, especially of a complex phenomenon that is observed again after an interval during which other unrelated observations have been made or other tasks performed. The subsequent observation may lead to a conclusion that differs from the original one.

Question 60

Per protocol analysis

Answer 60

In a RANDOMISED CONTROLLED TRIAL (particularly an EXPLANATORY RCT), the inclusion in the analysis of only those subjects or participants who followed the protocol i.e. took the drug under study or participated in the intervention concerned. Contrasted with INTENTION-TO-TREAT ANALYSIS.

Question 61

Positive predictive value

Answer 61

Syn: PREDICTIVE VALUE OF A POSITIVE TEST. The probability that a person with a positive test is a true positive (i.e. does have the disease) (or true positives as a proportion of all those with a positive test).

Question 62

Predictive value

Answer 62

In SCREENING and diagnostic tests, the probability that a person with a positive test is a true positive (i.e., does have the disease) is referred to as the “PREDICTIVE VALUE OF A POSITIVE TEST.” The PREDICTIVE VALUE OF A NEGATIVE TEST is the probability that a person with a negative test does not have the disease.

Question 63

Probability

Answer 63

A measure that ranges from zero to 1 of the likelihood that a random event will occur or the degree to which a statement or assumption is true. This must lie between 0 and 1: the probability is 0 if the event is certain not to occur, 1 if it is certain to occur. See also NULL HYPOTHESIS.

Question 64

Proportion

Answer 64

A RATIO in which the numerator is included in the denominator. It may be expressed as a fraction, a decimal fraction, or a percentage, and it must, by definition, lie between zero and 1 (or zero and 100 in the case of a percentage. An example is PREVALENCE.

Question 65

Prospective study

Answer 65

A study dealing with disease or health related events that have not yet occurred at the study’s point of commencement (can apply to case control studies, cohort studies and nested case control studies).

Question 66

Publication bias

Answer 66

Tendency of editors (and authors) to publish articles containing positive findings, especially “new” results, in contrast to reports that do not yield “significant” results.

Question 67

Randomised Controlled Trial (RCT) – Explanatory

Answer 67

As above but the “pragmatic” term signifies a trial which aims to establish the EFFECTIVENESS of an intervention in the “real world”. For this reason, the subjects recruited are often chosen to provide as “realistic” a group as possible (in terms of the likely characteristics of the patient population in which the intervention will be used) and, in all cases, the analysis is carried out on an “INTENTION TO TREAT” basis.

Question 68

Randomised Controlled Trial (RCT)

Answer 68

Syn: randomised clinical trial. An EXPERIMENTAL STUDY in which subjects in a population are randomly allocated into groups, usually called “intervention” and “control” groups, to receive or not to receive an experimental preventive or therapeutic procedure. The results are assessed by rigorous comparison of rates of disease onset, death, recovery, or other appropriate outcome in the intervention and control groups, respectively. Randomised controlled trials are generally regarded as the most scientifically rigorous method of hypothesis testing available in epidemiology.

Question 69

Recall bias

Answer 69

A systematic error that is caused by differing degrees of accuracy in remembering past events (including, for example, previous medical history) between different groups in a study. This bias often arises when people’s past history includes events that they have forgotten or not mentioned because they consider them irrelevant, embarrassing or shameful, e.g., details of premarital or extramarital sexual activity, or because their memory of unpleasant experiences has been suppressed. A type of INFORMATION BIAS

Question 70

Receiver operating characteristic (ROC) curve

Answer 70

A graphic means for comparing two or more screening tests in terms of the trade-off between sensitivity and specificity.

Question 71

Referral Bias

Answer 71

A type of SELECTION BIAS caused by the fact that participants in, for example a CASE CONTROL STUDY based in a hospital clinic may be materially different from the population of cases from which they are drawn.

Question 72

Reliability

Answer 72

Syn. reproducibility. The characteristic by which a measuring instrument or procedure yields a series of identical or closely similar measurements of the same variable that is repeatedly measured by the same observer using the same instrument or procedure.

Question 73

Repeatability

Answer 73

The quality of a test whereby repetition of the same protocol and procedures yields the same or closely similar results or responses each time - an important criterion of tests used in clinical diagnosis. The distinction between repeatability and RELIABILITY is that the latter is a property of the measuring instrument, whereas repeatability is determined by interaction of the observer, the subject, and the instrument.

Question 74

Reproducibility

Answer 74

The quality of an experiment or test wherein repetition under identical or similar circumstances gives the same or a closely similar result to that achieved on the first occasion. See also RELIABILITY and REPEATABILITY.

Question 75

Retrospective

Answer 75

A study dealing with disease or health related events that have already occurred at the study’s point of commencement (can apply to CASE CONTROL STUDIES, COHORT STUDIES and NESTED CASE CONTROL STUDIES).

Question 76

Self controlled case series

Answer 76

A study in which that the incidence of a given outcome when subjects are exposed (adverse events during the taking of a drug, for example) is compared with the incidence when the same subject is not exposed.

Question 77

Selection bias

Answer 77

Systematic error due to differences between those selected for study and those not selected. For instance, if a study relies on volunteers, these tend to be qualitatively different from those who do not volunteer for whatever reason, such as being able to attend a clinic because they are mobile. Hospital patients are a selected sample of persons with a condition, such as coronary heart disease, from which those at either end of the continuum from sudden death to silent infarct are excluded. Cf. INFORMATION BIAS

Question 78

sensitivity(Of a screening or diagnostic test)

Answer 78

The proportion of those who have the disease that the test correctly identifies as having it or the probability of identifying a case using the test.

Question 79

Specificity

Of a screening or diagnostic test

Answer 79

The proportion of those who do not have the disease that the test correctly identifies as not having it or the probability of correctly identifying the fact that a person does not have the disease using the test.

Question 80

Statistical power

Answer 80

The ability of a study (usually as a result of its sample size) to avoid ALPHA or BETA (TYPE 1 or TYPE 2) errors.

Question 81

Statistical significance

Answer 81

Statistical methods allow an estimate to be made of the probability of the observed difference or association between variables being due to chance. Usually the level of statistical significance is sated by the P VALUE which is the probability that the observed difference or association is due to chance.

Question 82

Systematic error

Answer 82

A consistent bias due to faulty instruments that yield incorrect readings or results, e.g., because of calibration, wrong dilution of reagents, etc.

Question 83

Systematic review

Answer 83

A method of analysis based on pooled data from as many sources as possible, preferably, but not necessarily, using exactly comparable research studies. In this respect, a systematic, review may not be quite the same as a META-ANALYSIS, which relies on comparable studies. The aim is to make maximum use of the best available evidence from all available sources in EVIDENCE-BASED DECISION MAKING. See also COCHRANE COLLABOARATION.

Question 84

Temporal relationship

Answer 84

An essential feature in interpretation of events is their relationship to one another in time. In all causal relationships, the cause must precede the effect.

Question 85

Test-retest reliability

Answer 85

A method of calibrating equipment, verifying the VALIDITY of findings, and assessing the dimensions of instrumental error conducted by repeating the same test several or many times under conditions as nearly identical as feasible. See REPRODUCIBILITY.

Question 86

Tripe-blind study

Answer 86

A DOUBLE-BLIND STUDY in which, as well as withholding information from study subjects and investigators, the statistician performing statistical analysis of the results is not informed about which subjects were in the control group, or the expected results of the analysis.

Question 87

Type 1 error

Answer 87

Syn: alpha error. The error of wrongly rejecting a true null hypothesis, i.e., declaring that there is a significant difference when in reality there is not.

Question 88

Type 2 error

Answer 88

Syn: beta error. The error of not rejecting the null hypothesis, i.e., declaring that a difference does not exist between sets of data when in fact a difference does exist.

Question 89

Validity

Answer 89

The extent to which a measurement actually measures what it purports to measure.

Two varieties of study validity are distinguished:

1. Internal validity: The comparison groups (e.g. in an RCT) are comparable with regard to potential confounders.
2. External validity (or GENERALISABILITY): The degree to which the inference can validly be drawn from a study and extrapolated to the wider population from which the study subjects are drawn.

Question 90

Variable

Answer 90

Any quantitative measurement of an entity, attribute, or phenomenon that can vary. In epidemiological and other scientific studies, the aim is often to assess how variables relate to one another. For this purpose, a distinction is made between independent and dependent variables. An independent variable is one that is assumed to remain constant or uninfluenced by changing conditions or circumstances during a study. A dependent variable changes in way that are determined by the influence of attributes or external factors. The aim of a research study is often to identify how these attributes or factors influence the behavior of a dependent variable or variables, preferably to develop a generalizable model or mathematical rule to explain and predict the relationship.

Question 91

Washout phase

Answer 91

The phase of a clinical trial after stopping the test medication when its effects are wearing off. It is comparable to “drying out” after acute or chronic alcohol intoxication, but the washout phase varies in duration, according to the metabolic half-life of the medication.