PubMed Health. A service of the National Library of Medicine, National Institutes of Health.

Irwig L, Irwig J, Trevena L, et al. Smart Health Choices: Making Sense of Health Advice. London: Hammersmith Press; 2008.


Absolute risk reduction (ARR) or risk difference

the difference in the incidence of poor outcomes between the intervention group of a study and the control group. For example, if 20 per cent of people die in the intervention group and 30 per cent in the control group, the ARR is 10 per cent (30–20 per cent).

Abstract (of a study)

a summary of the main features of a study. Major journals now use subheadings (similar to those in the main paper) to make it a structured abstract. These subheadings are, for example: introduction, methods, results and conclusions.

Adverse reaction

a side effect or unintended bad outcome of a treatment.

Anecdotal evidence

evidence that comes from an individual experience. This may be the experience of a person with an illness or the experience of a practitioner based on one or more patients outside a formal research study.

Benefit (of a health intervention)

the extent to which one’s lifespan is increased and/or quality of life improved


something that distorts the real effect in a study, so that the researchers get the wrong answer. The term does not suggest that the researchers are biased, but rather that sources of error can easily occur in studies.

Blinding and double-blinding

methods of preventing individuals, healthcare providers and those assessing outcome in a study from knowing whether the participant is in the experimental or control group.

Case–control study

involves selecting people who have the outcome of interest (cases), and a control group without the outcome of interest, and looking backwards in time to see if the groups were exposed to a supposed cause. These studies are not considered high-quality evidence unless the association is very strong; for example, case–control studies on lung cancer and smoking. Case–control studies are either population based or hospital based. The latter is the less reliable of the two.

Clinical practice guideline

a systematically developed statement designed to assist practitioner and patient to decide on appropriate care for specific clinical circumstances. Not all practice guidelines are based on the best available evidence.

Clinical trial

a study in which an intervention is being tested (see Trial).

Cochrane Collaboration

an international organisation that prepares, maintains and disseminates systematic reviews of the effects of healthcare. Publication is electronic. Abstracts are available free of charge on the internet.

Cohort study

an observational study that involves classifying people by their exposure to a study factor (for example, environment or lifestyle) of interest and following them over a period of time to see whether those exposed are more or less likely to develop disease than those not exposed.

Complementary medicine

any alternative or ‘eastern’ doctrine of healthcare. For example, naturopathy, homeopathy, reflexology, acupuncture, traditional Chinese medicine.

Confidence interval

even if studies are perfectly designed and carried out, the results may show variability because of the play of chance. A confidence interval covers the likely range of the true effect. For example, the result of a study may be that 40 per cent (95 per cent confidence interval 30–50 per cent) of people are helped by a treatment. That means that we can be 95 per cent certain that the true effect is between 30 and 50 per cent.


something that could explain an association between a study factor and outcome. For example, workers in a factory may get more lung cancer than those working elsewhere not because of their work but because they happen to be exposed to another known cause of lung cancer, cigarette smoking. In this example, smoking is the confounder and factory work may no longer be associated with lung cancer once the confounder has been taken into account.

Controlled trial

an intervention study in which a group given some intervention is compared with a control group.

Empirical evidence

evidence provided by experiments or observational studies rather than theory, assumptions or recall of single experiences.

Evidence-based medicine (healthcare)

an approach to making health decisions that uses the best available evidence from good studies in combination with information from the patient.

Experimental study

see Intervention study.

Harm (of an intervention)

the extent to which one’s lifespan is shortened or one’s quality of life deteriorates.


the number of new cases of a disease in a given period of time as a proportion of the population.


any therapy, surgical procedure, diagnostic or screening test or change in lifestyle or behaviour intended to have an effect on health.

Intervention study

an experimental study in which people are given an intervention to assess its effects. Examples are clinical trials, controlled trials and randomised controlled trials.

Lead time bias

a bias that occurs in the assessment of screening. As people who have been screened have their disease detected earlier, they live longer from the time of diagnosis, even if screening has no beneficial effects. This means that they have been given more years of disease rather than more years of life.


an electronic database of the titles and abstracts of many medical journals. A version called PubMed is available free of charge on the internet.


see Systematic review.

Number needed to treat (NNT)

the number of people who must be treated to result in benefit in one person. It is the inverse of absolute risk reduction.

Observational study

a non-experimental study that examines the association between a study factor (for example, exposure or lifestyle) and outcome. Examples are cohort and case–control studies.

Odds ratio

a way of measuring relative risk.

Orthodox or mainstream medicine

conventional or ‘western’ doctrine of healthcare.


any identified change in health status after a disease, an exposure to something or a preventive or therapeutic intervention. By comparing the outcomes of two experimental groups, one that receives the intervention and one that does not, the effect of an intervention can be assessed. Most often, the frequency of bad outcomes (that is, poor health or death) is measured.


an inert treatment (or procedure); one that is not expected to have any pharmacological effect.

Placebo effect

a change caused by an expectation that the treatment or procedure will have an effect rather than directly by the treatment or procedure itself. The placebo effect is usually but not necessarily beneficial.

Post-test probability

the probability of having a disease after having a test. A positive test result will increase the probability of disease above the pre-test probability. A negative test result will decrease the probability of disease below the pre-test probability.


anyone who offers any form of healthcare in a professional capacity. For example, doctors, nurses, homeopaths, naturopaths, dentists, nutritionists, pharmacists, chiropractors, physiotherapists, occupational therapists, surgeons, veterinarians.

Pre-test probability

the probability of having a disease before having a test (see Post-test probability).


the total number of people who have the disease or condition at a particular time expressed as a proportion of the population.


allocation based on chance to either an intervention group or a control group of a study. This ensures that the outcomes in the groups are expected to be the same if the intervention has no effect.

Randomised controlled trial (RCT)

a trial in which people are allocated randomly to either an intervention group or a control group.

Randomised trial

same as randomised controlled trial.

Recall bias

differences in the completeness or accuracy of recall of prior events, for example, mothers whose children died of leukaemia may be more likely to remember exposure of the unborn infant to X-rays than a control group of mothers.

Regression to the mean

many illnesses get better on their own and many abnormal findings become more normal on re-measurement in the absence of any intervention.

Relative risk (RR)

the rate (risk) of poor outcomes in the intervention group divided by the rate of poor outcomes in the control group. For example, if the rate of poor outcomes is 20 per cent in the intervention group and 30 per cent in the control group, the relative risk is 0.67 (20 per cent divided by 30 per cent). The relative risk is 1 when the intervention has no effect, below 1 when it does good and above 1 when it does harm (see Absolute risk reduction).

Relative risk reduction (RRR)

the extent to which the risk of a poor outcome is reduced by an intervention. In the example given in Relative risk (above), the relative risk reduction, expressed as a percentage, is 33 per cent (1.0 – 0.67 = 0.33) (see Absolute risk reduction).


the probability that an event will occur, for example, that an individual will die or become ill within a stated period of time.

Selection bias

error caused by systematic differences in characteristics between those who are selected and followed up in the intervention and control groups or the groups being compared in an observational study.

Sensitivity of a test

the probability of a positive test in people who have the disease of interest.

Specificity of a test

the probability of a negative test in people who don’t have the disease of interest.

Systematic review

a review in which all relevant studies are identified and those of adequate quality selected. Results from adequate studies are usually pooled (meta-analysed) to give the best single estimate of effect.


an experiment or intervention study where a specific treatment or other intervention is ‘tried’ on a group of people; it is a synonym for clinical trial. Unless otherwise stated, a trial might not include a control group (see Clinical trial, Controlled trial and Randomised controlled trial).

Copyright © 2008, Professor Les Irwig, Judy Irwig, Dr Lyndal Trevena, Melissa Sweet.

All rights reserved. No part of this publication may be reproduced, stored in any retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the publishers and copyright holder or in the case of reprographic reproduction in accordance with the terms of licences issued by the appropriate Reprographic Rights Organisation.

Cover of Smart Health Choices
Smart Health Choices: Making Sense of Health Advice.
Irwig L, Irwig J, Trevena L, et al.
London: Hammersmith Press; 2008.

PubMed Health Blog...

read all...

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...