Glossary
Internet Healthcare Class Glossary
Absolute Risk Reduction (ARR) is the difference in the event rate between control group (CER) and treated group (EER): ARR = CER - EER.
Abstract A very brief summary or digest of the study and its results. It should tell you what the study tried to show, how the researchers went about it, and what they found. The abstract can be very misleading though. This is often the only part of the content of an article that will show up on a database.
Adverse reaction Experience of a bad, unpleasant, or dangerous effect in response to a treatment.
Algorithm A formula. Each search engine uses a different algorithm to create its index. These formulas give different weight to title words, meta-tags, text, and frequency of words.
Alpha and Beta errors (See also Type I/Type 11 errors) Alpha (Type I) errors are 'false positives': that is, the results suggest that a treatment works, when in fact it does not work. Beta (Type 11) errors are 'false negatives': that is, the results suggest a treatment does not work, when in fact it actually does. Large enough numbers can help to try and avoid these kinds of errors.
Arm A controlled trial has at least two arms: that is, at least two groups of people experiencing different treatment. One of those is the control arm or group (that is, the people not receiving the experimental treatment). The experimental arms are the groups of people getting the treatment (or treatments) being tested. There might be one or more experimental arms, but there is always only one control group.
Ascertainment Identification or assessment of outcomes. Ascertainment bias is bias occurs when there is a difference in assessing outcomes of people in different study groups.
Association A known link, or statistical dependence, between two or more conditions or variables: eg, statistics demonstrate that there is an association between smoking and lung cancer. A 'positive' association is one where the incidence of one condition increases if the other condition or variable increases (as with smoking and lung cancer). There is a 'negative' association when an increase in one thing is apparently associated with a decrease in something else.
Automatic Term Mapping In PubMed, unqualified terms that are entered in the query box are matched (in this order) against a MeSH heading, translation table, a journal translation table, a phrase list and an author index. (Truncation turns off automatic term mapping and also automatic explosion of a MeSH term for more specific terms indented under it in the MeSH hierarchy. For example, heart attack* will not map to the MeSH term, Myocardial Infarction.)
Bias Something that introduces a difference or trend that distorts (or could distort) results of a study. Bias introduces systematic error into a study, because what is being observed may not be the effect of the treatment being studied, but rather it may be the effect of bias.
Blind/ing Blinding (also called masking), in research, refers to the process of preventing participants and/or experimenters from knowing whether a particular person is receiving the experimental or control treatment. A single blind trial: only one party (eg participants or practitioners administering treatment) has been masked. Double blind: both the experimenter and participant are unaware of the individual assignment. Triple blind: experimenters, participants, and those assessing outcomes are unaware of what treatment the participant received.
Boolean Logic Use of the terms AND, OR, NOT in formulating online search commands. Many search engines use the plus character (+) instead of AND, and the minus character (-) instead of NOT.
Case Sensitivity Treating capital letters differently from lower-case letters. Most search engines apply case sensitivity if a word is typed with a capital letter, but otherwise they are case-insensitive.
Case study or series A case study is a report of a single example (generally this is an anecdote about one 'interesting' or 'unusual' person or situation). A case series is a description of a number of such 'cases'.
Case-series is a report on a series of patients with an outcome of interest. No control group is involved.
Case-control study Compares people with a disease or condition ('cases') to another group of people from the same population who don't have that disease or condition ('controls'). A case-control study can identify risks and trends, and suggest some possible causes for disease, or for particular outcomes.
Case-control Study involves identifying patients who have the outcome of interest (cases) and control patients without the same outcome, and looking back to see if they had the exposure of interest.
CER Control Event Rate: see Event Rate.
CINAHL Cumulative index to nursing and allied health literature.
Clinical Guidelines recommend a particular protocol for prevention or treatment. Their authors synthesize large amounts of literature to discover the evidence for best practices. Many of them carefully rank the evidence based on the study design used to achieve the results. In general, randomized clinical trials are considered the best or the "gold standard," rated above other study designs and expert opinions. However, it is not often feasible or ethical to conduct randomized clinical trials.
Clinical trial A trial involves administering a treatment to test it: it is an experiment. Clinical trial is an umbrella term for a variety of health care trials, whether controlled or uncontrolled. Types include uncontrolled trials, controlled clinical trials (CCT), community trials, and randomised controlled trials (RCT). (A clinical trial is also sometimes called a 'therapeutic trial'.)
Cochrane Collaboration The Cochrane Collaboration is an international effort by researchers, practitioners, and consumers to sift through research on the effects of health care. The Collaboration prepares, maintains, and disseminates systematic reviews of the effects of health care. The reviews are published in the Cochrane Database of Systematic Reviews - one of the components of a regular, electronic publication called The Cochrane Library.
Cohort (study) A 'cohort' is a group of people clearly identified: a cohort study follows that group over time, and reports on what happens to them. A cohort study is an observational study, and it can be prospective or retrospective.
Cohort Study involves identification of two groups (cohorts) of patients, one which did receive the exposure of interest, and one which did not, and following these cohorts forward for the outcome of interest.
Co-morbidity In a study looking at treatment for one disease or condition, some of the individuals with that disease will also have other diseases or conditions that could be affecting their outcomes. Any other such condition is called a 'co-morbidity'. (A co-morbidity is a confounding variable. )
Comparison group The control group. (See controls)
Concept Search A search for documents related conceptually to a word, rather than for the word itself.
Confidence interval (CI) What happened in a study is a fact
for the people that were in it, but for everyone else, the
result shown by a study is always just an estimate of what might
happen to them. There will always be a level of uncertainty
around predicting the result for anyone else: a confidence
interval (CI) is a way of quantifying the range of uncertainty
around a result. The CI stretches between two limits: one is the
lowest possible result that others could face, and the other is
the highest - the mean (average) result of the study lies
somewhere in between. Eg, say a friend is coming to stay at your
place, and they ask how long it takes to drive into the city. If
you answer "Oh, half an hour - give or take 10 minutes", you are
giving them an estimate (30 mins), with a confidence interval
around it (20-40 mins).
The CI also shows the precision of the estimate for anyone who
wants to apply the results to others, because the power of a
study to detect a real difference is reflected in the narrowness
of the range: the shorter the length of the CI, the more precise
the estimate is able to be. (See also confidence limits,
confidence levels, probability, and p value)
Confidence levels Statistical results and estimates are usually calculated at the 195% confidence level'. This means that if someone were to keep repeating the study in other populations, 95% of the time a similar result would occur: ie, a result that falls somewhere between the upper and lower limit of the confidence interval. Sometimes, researchers who want to be more confident about a result will do calculations at a 99% confidence level. Sometimes, results will be calculated at a 90% level of confidence. However, the 95% confidence level is the most usual one. (See also confidence intervals, confidence limits, probability, and p value)
Confidence limits See confidence interval: the upper and lower boundaries of a confidence interval are called its upper and lower 'limits'.
Confounder or confounding variable (See also variable) Another factor or effect that confuses the picture. A confounder distorts the ability to attribute the cause of something to the treatment, because something else could be influencing the result. Eg, if people are receiving a mixture of therapies, it would be possible to confuse the effect of one with the other. Or if a group of people who did poorly also all lived in the same street, while everyone who did well lived on the other side of town, their worse outcome might be the result of socioeconomic or environmental factors, not the treatment.
Contamination The treatment assigned to people in one group is also used by some or all members of the other group.
Continuous/dichotomous Different categories of measuring results: a dichotomous result is a clear 'either/or' one, eg yes/no, alive/dead. A continuous one is something like time (1 hour, 2 hours 20 mins, 3 hours, etc) or weight (75kg, 82kg, 110g, etc).
Controls The people in the 'control' group or arm in a controlled trial or a case-control study (also called the comparison group). In a trial, people who are the 'controls' represent the status quo, against which the effectiveness of a treatment is tested. These could receive no treatment, a placebo treatment, or the standard or conventional treatment. The people in the other arm of a trial are the 'experimental' group or arm. In a case-control study, the controls are the people who don't have the condition being studied: the 'cases' are the people who have the condition.
Controlled (clinical) trial (CCT) (See clinical trial)
Convenience sample A population being studied because they are conveniently accessible in some way. This could make them particularly unrepresentative, as they are not a random sample of the whole population. A convenience sample, for example, might be all the people at a certain hospital, or attending a particular support group. They could differ in important ways from the people who haven't been brought together in that way: they could be more or less sick, for example.
Conventional treatment Whatever the standard or usual treatment is for a particular condition at that time.
Cost-Benefit Analysis converts effects into the same monetary terms as the costs and compares them.
Cost-Effectiveness Analysis converts effects into health terms and describes the costs for some additional health gain (e.g. cost per additional MI prevented).
Cost-Utility Analysis converts effects into personal preferences (or utilities) and describes how much it costs for some additional quality gain (e.g. cost per additional quality-adjusted life-year, or QUALY).
CRD The Centre for Reviews and Dissemination in York: a kind of sister centre to the UK Cochrane Centre. The CRD produces DARE (see below).
Critical appraisal The process of assessing and interpreting evidence, by systematically considering its validity, results, and relevance.
Cross-over trial A trial where each of the groups will receive each of the treatments, but in a randomised order: that is, they will start off in one arm of the trial, but will deliberately 'cross over' to the other arm(s) in turn.
Crossover Study Design: the administration of two or more experimental therapies one after the other in a specified or random order to the same group of patients.
Cross sectional study Also called or prevalence study: an observational study. It is like taking a snapshot of a group of people at one point in time and seeing the prevalence of diseases, etc, in that population.
Cross-Sectional Study the observation of a defined population at a single point in time or time interval. Exposure and outcome are determined simultaneously. Decision Analysis is the application of explicit, quantitative methods to analyse decisions under conditions of uncertainty.
DARE The Database of Abstracts of Reviews of Effectiveness, prepared by the Centre for Reviews and Dissemination (CRD) in York, and included in The Cochrane Library. (See also Cochrane Collaboration)
Data dredging Digging through the data of a trial trying to find a significant result.
Data monitoring committee An expert committee set up to monitor the results of a continuing trial periodically, and assess whether or not the trial should continue or stop on ethical grounds. That is, if a treatment appears to be dramatically effective or harmful, and providing it or denying it to half the participants has become unethical. (See also equipoise)
Descriptive Literally, a study that simply describes the current situation.
Dichotomous (See continuous/dichotomous)
Dose dependent A response to a drug may be related to the amount received (ie, the dose). A drug may not work with a smaller dose than is in ordinary use, but on the other hand a smaller dose may work just as well. Similarly, an adverse reaction may be dependent on the size of the dose as well. Sometimes trials are done to test the effect of different dosages of the same drug.
Double blind (See blind/ing)
Dynamic Categorization Refining a search by the automatic creation of a thesaurus based on document themes. See AltaVista's "Refine" function for an example. Similar but not identical to the See Related Articles function in PubMed.
Ecological Survery based on aggregated data for some population as it exists at some point or points in time; to investigate the relationship of an exposure to a known or presumed risk factor for a specified outcome.
EER Experimental Event Rate: see Event Rate.
Effectiveness The extent to which an intervention does people more good than harm. An effective treatment or intervention is effective in real life circumstances, not just an ideal situation.
Efficacy The extent to which a intervention improves the outcome for people under ideal circumstances. Testing efficacy means finding out whether something is capable of causing an effect at all.
EMBASE (Excerpta Medica database.) A major European database of medical and health research. (See MEDLINE for a more detailed explanation.)
Empirical Based on observation, not on reasoning or belief alone.
Endpoint An outcome. Sometimes, a longterm outcome (such as mortality or disability) cannot be measured within the time range of a study. An 'intermediate endpoint' is sometimes substituted. It is some point along a path that commonly leads to the 'endpoint'. Eg, serious seizures in newborn infants are often the first sign of neurological damage: damage may not be able to be diagnosed for years, but seizures can be seen and recorded early. (See also surrogate measure)
Epidemiology The study of the health of populations and communities, not just particular individuals.
Equipoise A theoretical state of belief, where someone is equally balanced between being 'for' or 'against' a treatment: a state of uncertainty about whether or not a treatment works. It is seen as a necessity so that randomly assigning people to one or the other becomes ethical (so that treatments known to be effective, for example, are not deliberately withheld for no good reason).
Event Rate is the proportion of patients in a group in whom an the event is observed. Thus, if out of 100 patients, the event is observed in 27, the event rate is 0.27. Control Event Rate (CER) and Experiemental Event Rate (EER) are used to refer to this in control and experimental groups of patients respectively.
Evidence-based healthcare explains how the evidence can be applied to health policy and management decisions for populations, rather than for individual patients. Evidence-Based Healthcare extends the application of the principles of Evidence-Based Medicine to all professions associated with health care, including purchasing and management.
Evidence-Based Medicine is 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.
Experimental An experimental study (a 'trial') is one in which the investigators are testing something, and they are determining the conditions of the experiment. In a controlled trial, the people receiving the treatment being tested are said to be in the experimental group or arm of the trial. (See also controls and arm)
Explosion Expanding the subcategories beneath a term. PubMed automatically explodes MeSH terms and searches these as well as the main term.
External/internal validity Internal validity is the extent to which a study properly measures what it is meant to. External validity is the extent to which the results of a study can apply to people other than the ones that were in the study (how generalisable the results are to others, or how applicable they are in the real world).
Factorial design A randomised controlled trial usually compares a number of treatments with each other. In a trial with a factorial design, people get randomised twice: they get randomised to a group for one treatment, and then the people within that group get randomised again to another two options. This is a way of assessing the results of two or more treatments without needing extremely large numbers, and enables researchers to test combinations of treatments as well. At the end of a factorial design trial of two treatments, instead of having just two groups (one of which got the treatment, and the other of which was the control), you have four groups of people: one that got two treatments, one that got control treatments twice, one that got the first treatment and control for the second, and one that got control treatment and then the second experimental treatment.
False positive/negative Two types of mistaken results. In people, these mistaken results could refer, for example, to a diagnostic test. A 'false positive' diagnoses someone as having the condition, when in fact they don't. A 'false negative' gives them the all clear' when in fact the person actually has the condition. A study can have a false positive when it suggests a treatment is effective, but the treatment in fact is not effective. A false negative is when results of a study suggest a significant result does not exist, when in fact it actually does. (See Alpha and Beta errors)
Field A categor in a database record, such as author or title.
Fuzzy Search A search that will find matches even when words are only partially spelled or misspelled.
Generalisability Whether or not the results of a study are applicable or relevant to another group of people or population. (See also external validity)
Grey literature Grey literature is the kind of material that is not published in regular accessible journals, and does not show up on databases. It includes things like conference proceedings that include the abstracts of the research presented at the conference, unpublished theses, and so on.
Handsearching Looking for articles, not by searching electronic databases, but by systematically going through copies of journals yourself.
Heterogeneous(geneity) The opposite of homogeneous. If a set of studies on the same subject have varied or conflicting results, the results of the group of studies are heterogeneous. Examining and explaining this heterogeneity is an important part of reviewing the research on a particular subject.
Hierarchy Ordering things or concepts by levels. A common system in medicine is that used to describe the physical body: organism, organ, tissue, cell. MeSH uses a hierarchy of concepts to organize medical concepts.
Historical control trial Comparing a group of people receiving a particular treatment now, with a similar group from the past who did not receive the treatment.
Homogeneity See heterogeneous: when results are similar and consistent from one study to another, then the results are said to be homogeneous.
Hypothesis A theory or suggestion that is being tested with a piece of research. For an experimental study to properly test a hypothesis, it needs to be prespecified and clearly articulated so that the design, conduct, and interpretation of the study can properly test it. A prespecified hypothesis is also called a 'prior hypothesis'. Studies are often framed to test what is called the 'null hypothesis': that is, that the treatment in question has no effect.
Incidence The number of occurrences of something in a population over a particular period of time: eg, the number of cases of a disease in a country over one year.
Intention to treat analysis Analysing the results according to the intended treatment to which someone was allocated in a randomised controlled trial (as opposed to the treatment they actually received in the end).
Internal/external validity Internal validity is the extent to which a study properly measures what it is meant to: whether the conclusions are 'true' for the people in the study. External validity is the extent to which the results of a study can apply to people other than the ones that were in the study (how generalisable the results are to others, or how applicable they are in the real world - whether the results will be true for people outside the study).
Intervention Anything meant to change the course of events for someone: surgery, a drug, a test, a treatment, counselling, giving someone a pamphlet - all of these are interventions.
Key words A string of words attached to an article to be used to index or code the article in a database. If someone is interested in a particular subject, they search the database for entries indexed according to a word they are interested in (eg 'consumer', or 'stroke'), and the database will produce a list of all the items it has that have been coded with that key word. (See also MeSH and MEDLINE)
Knowledge Mapping To group search results by meaning, synonyms, or related terms.
Likelihood Ratio is the likelihood that a given test result would be expected in a patient with the target disorder compared to the likelihood that the same result would be expected in a patient without that disorder.
Limit To restrict the number of results of a search by applying parameters, typically done with boolean connectors.
Logistic regression (See regression analysis)
Marker An indicator of something else. Sometimes there is an association with a factor that people might think is the cause. However, rather than being the cause, it could just be a marker for the real cause: it just indicates or points to people who are experiencing the real cause as well as the 'marker'. Eg, if something happens more frequently among smokers, on occasion this may have nothing to do with the effects of tobacco: smoking could be a marker for something else, such as lower socioeconomic status (because people on lower incomes are more likely to also smoke).
Matching In a case-control study, if the controls are chosen because of particular similarities to the people who are in the 'case' group, the cases and controls are said to be 'matched'. Researchers match people according to particular variables that are thought to be important: eg choosing people in the same age group and gender.
Mean The average. (Add up the results of something for each participant, and divide by the number of people to find the average - or mean - outcome for people in the study.)
Median The number exactly in the middle of the range of outcomes that people had. It is not the same as the average, which is called the mean.
MEDLINE MEDLINE (MEDLARS online) is published by the National Library of Medicine (NLM) in the USA. It is an electronic database listing thousands of pieces of biomedical research published in a selected list of journals. It includes abstracts of the studies. MEDLINE is available through medical libraries, and can be accessed by CD-ROM and through a phone line using a computer with a modem. Another such major database is the European EMBASE.
MeSH 'Medical Subject Headings': terms, or key words, used by the National Library of Medicine to index the entries on the MEDLINE database. The MeSH headings get used by many others for indexing and classification as well.
Meta-analysis
A statistical technique which summarises the results of several studies into a single estimate of their combined result. It is a key element of many systematic reviews or overviews. It is also what is meant when people talk about 'pooling data'. Meta-analysis is the technique which produces a graph depicting the confidence intervals of all the studies.
In a systematic review, the data from individual studies may be pooled and re-analysed. This technique is called meta-analysis. There are two types of meta-analysis, depending on the source of the data analysed:
MAL meta-analysis in which the data are abstracted from published papers in the literature, i.e. data on groups of patients.
MAP meta-analysis in which the data have been obtained by asking the authors of published papers for the original single patient data and by asking drup companies for unpublished data.
MAP meta-analyses are more accurate that MAL meta-analyses.
Morbidity Illness or harm. (See also co-morbidity)
Mortality Death.
Multicentre Research is mostly conducted within a prescribed area, or within a single institution. A trial is sometimes conducted between a number of collaborating institutions - particularly when very large numbers of people are needed to complete a trial. Those trials are called 'multicentre' trials, or trials conducted at several or multiple 'sites'.
Multivariate analysis Measuring the impact of more than one variable at a time while analysing a set of data, eg looking at the impact of age, gender, and occupation on a particular outcome.
N-of-1 Trials The patient undergoes pairs of treatment periods organised so that one period involves the use of the epxerimental treatment and one period involves the use of an alternate or placebo therapy. The patients and physician are blinded, if possible, and outcomes are monitored. Treatment periods are replicated until the clinician and patient are convinced htat the treatments are definitely different or definitely not different.
Natural Language Processing The ability of a computer program to accept statements or questions phrased in common syntax, rather than a computer language, for example "What is the cure for the common cold?"
Negative Predictive Value (-PV) is the proportion of people with a negative test who are free of disease.
Negative trial A trial is often called 'negative' when it does not find a treatment effective. Negative trials are less likely to end up published. (See publication bias)
Nesting Using parentheses to create a commmplex search phrase. For example: (burn AND infection) NOT staphylococcus ... or (management OR therapy).
Null hypothesis (See hypothesis)
Number needed to treat (NNT) A number which gives you an estimate of how many people need to receive a treatment before one person would experience the beneficial outcome. Eg, if you need to give a stroke prevention drug to 20 people before one stroke is prevented, then the NNT for that stroke prevention drug is 20.
Number Needed to Treat (NNT) is the number of patients who need to be treated to prevent one bad outcome. It is the inverse of the ARR: NNT = 1/ARR.
Observational study A survey or non-experimental study. The researchers are examining and reporting on what is happening, without deliberately intervening in the course of events.
Odds The Odds or probability of something happening. It is used the same way as in gambling: odds can be, say, 2 to 1, or 150 to 1.
Odds are a ratio of nonevents to events. If the event rate for a disease is 0.1 (10 per cent), its nonevent rate is 0.9 and therefore its odds are 9:1. Note that this is not the same expression as the inverse of event rate.
Odds ratio (OR) This is a common way of estimating the effect of a treatment. An OR greater than one (> 1) means the treatment is estimated to increase the odds of something: < 1, and it decreases the odds. If the OR is exactly 1, then the treatment appears to have no effect on that outcome. Eg, a treatment with this estimate: an OR 2.0 (95% CI: 1.0-3.0), apparently increases the odds of a person experiencing this. effect, while an OR of 0.5 is an estimated decrease in the chances of experiencing that result. This is similar to a risk of experiencing of something - but not exactly the same thing. (See also confidence interval, confidence limits, relative risk)
Odds Ratio describes the odds of an experimental patient suffering an adverse event relative to a control patient.
Open label trial This is a non-blind/non-masked trial: one where both the doctor and patient know what drug someone is taking, and at what dose.
Overview is a systematic review and summary of the medical literature. (see Systematic Review)
P value The findings of a study may be just an unusual fluke. Calculating the p value can determine whether or not the results of the study are likely to be a fluke or not. The p (probability) value shows whether or not the result could have been caused by chance. If the p value is less than 0.05, then the result is not due to chance. A result with a p value of less than 0.05 is statistically significant. The 0.05 level is equal to odds of 19 to 1 (or a 1 in 20 chance). (See also confidence level, power, and probability)
Peer review A refereeing process, used to check the quality and importance of reports of research. An article submitted for publication in a peer reviewed journal is reviewed by other experts in the area. It aims to provide a wider check on the quality and interpretation of a report.
Phase 1, 11, 111, IV trials A series of levels of trials required of drugs: Phase I trials assess toxic effects on humans (not many people in them, and usually without controls); Phase 11 trials assess therapeutic benefit (usually a few hundred people, usually with controls, but not always); Phase III trials compare the new treatment against standard (or placebo) treatment (usually a full randomised controlled trial). At this point, a drug can be approved for community use. Phase IV monitors new treatment in the community (usually a full randomised controlled trial evaluating longterm safety and effectiveness).
Phrase Search The ability of a search engine to look for a phrase, rather than individual words. PubMed phrase searching is done by placing quotation marks around the phrase; for example, "myocardial infarction".
Placebo A fake (sham, dummy, inert) treatment, given to people in a control group so they can't know whether or not they are in an experimental or control group. Placebos are meant to be both useless and harmless, so that they don't have any effect. However, there is also such a thing as a 'placebo effect': that is, if someone believes they are getting a real or active treatment, they can experience effects (positive or negative), even if there is nothing in the actual 'treatment' that could be the cause.
Point estimate Odds ratios and relative risks are called 'point estimates'.
Population In research, this term is applied to the group of people being studied, which may or may not be the population of a particular geographical area. The population in question in a research study, for example, may be 'all those people with cancer'. The study of the health of populations, as opposed to health of individuals, is epidemiology.
Positive Predictive Value (+PV) is the proportion of people with a positive test who have disease.
Power Statistical power of a study: a study needs to have a specific level of 'power' in order to be able to reliably detect a difference that a treatment might cause. To be powerful enough, the study needs to have enough participants, who experience enough of the outcomes in question, to be able to come up with statistically significant results.
Pragmatic trial Aims to test a treatment policy in a 'real life' situation, when many people may not fully start the treatment, and may use other treatments as well. This is as opposed to a 'causal' or explanatory trial', which is done under ideal conditions and is trying to determine whether a therapy has the ability to make a difference at all (ie, testing its efficacy).
Prevalence The proportion of a population having a particular condition or characteristic: eg, the percentage of people in a city with a particular disease, or who smoke.
Primary study A study of other studies is called a review, or secondary study. A primary study is one of the individual studies within that review.
Probability Probability is the 'chance' or 'risk' of something happening. (It is the word from which springs the 'p' in the notion 'p value'.)
Prospective/retrospective In a retrospective study, the outcomes of a group of people are examined in hindsight, using existing records. In a prospective study, the study is designed ahead of time, and people are then recruited and studied according to the study's criteria. A randomised controlled trial, for example, is always prospective, while a case-control study is retrospective ('after the event'). Retrospective studies are generally more limited in the data available for analysis, as the data have rarely been collected with the needs of that particular study in mind. This kind of limitation means that a retrospective study is usually less reliable than a prospective study.
Protocol Both randomised controlled trials and systematic reviews should be undertaken according to a clearly defined protocol, which prospectively sets out what is being tested, why, and how it will be done. The trial or review should then adhere strictly to the pre-set actions in the protocol to maintain uniformity and minimise bias.
Proximity Search The ability of a search engine to look for a word if it is near another, for example within five or ten words, or in the same paragraph. Also called adjacency search.
Publication bias This results from the fact that studies with more dramatic results and studies that show a treatment works well are more likely to end up published than those with results that are either uncertain, or show that a treatment has no useful effect.
Quasi-random Methods of allocating people to a trial which are not strictly random. Quasi-random methods include: allocation by the person's date of birth, by the day of the week or month of the year, by a person's medical record number, or just allocating every alternate person. Quasi-random allocation may look random, but it is not, because the group to which a person will be allocated is predictable, and people can start to manipulate who goes in. Eg, if someone wants to be in the experimental group, but not the control group, they can be placed in the experimental group if their number has come up, and simply excluded from the trial if it doesn't. One, other, or both arms of the trial can then be biased. (See also randomised controlled trial)
Query A structured question that is posed using a search engine or database, typically by searching on one or more fields.
Random sample When a group of people is being selected for study, one of the ways to try and ensure that the group studied is representative, is to try and recruit people who have been selected randomly from the population. This means that everyone in the population has an equal chance of being approached to participate in the survey, and the process is meant to ensure that a sample is as representative of the population as possible. It is a method that has less bias than the other option, which is to use a convenience sample: that is, a group that the researchers have more convenient access to.
Randomisation The process of randomly assigning participants into one of the arms of a randomised controlled trial (see next entry).
Randomised controlled trial (RCT) In a randomised controlled
trial, participants are assigned by chance to receive either an
experimental or control treatment. The idea behind the
randomised controlled trial is that when it is done properly,
the effect of a treatment(s) can be studied in groups of people
who are:
(1) the same at the outset, and
(2) treated the same way,
except for the intervention(s) being studied. Any differences
then seen in the groups at the end can be attributed to the
difference in treatment alone, and not to bias or chance.
(See randomisation above, and quasi-random)
Although RCTs are the "gold standard" for evaluating the effects of a therapy, they are more difficult to organize with increasing complexity of intervention, either technically, because the number of services randomly allocated may be too few to ensure the trial has adequate power, or politically, because the politicians may be reluctant to admit that a new policy should be subject to a trial.
Consequently, studies are sometimes investigated using research methods other than the RCT, notably:
the cohort study
the case-control study
Ranking The ordering of search results. In PubMed, the results of a basic search are ranked in reverse chronological order (most recent first) according to the date they were entered into Medline (Entrez date).
Recall bias Arises from mistakes in recollecting events, both because of failures of memory, and looking at things 'with hindsight' and possibly changed views.- People's reports of what is happening to them currently, therefore, can be more accurate than their recall of what happened two years ago and how they felt about it at the time.
Regression analysis A statistical modelling technique. Regression analysis is used to estimate or predict the relative influence of more than one variable on something: eg, the effect of age, gender, and educational level on the prevalence of a disease. There are different types of these models, including 'linear' and 'logistic' regression.
Relative risk (RR) Also called the 'risk ratio'. It is a common way of estimating the risk of experiencing a particular effect or result. A RR > 1 means a person is estimated to be at an increased risk, while a RR < I means a person is apparently at decreased risk. A RR of 1.0 means there is no apparent effect on risk at all. Eg, if the RR = 4.0, the result is about 4 times as likely to happen, and 0.4 means it is 4 times less likely to happen. The RR is expressed with confidence intervals: eg, RR 3.0 (95% CI: 2.5 - 3.8). This means the result is 3 times as likely to happen - anything from 2.5 times as likely, to 3.8 times as likely. It is statistically significant. On the other hand, RR 3.0 (95% CI: 0.5 - 8.9), means it is also estimated to be 3 times as likely, but it is not statistically significant. The chances go from half as likely to happen (0.5 a decreased chance), to nearly 9 times as likely to happen (8.9 an increased chance). (See also confidence interval, confidence limit, odds ratio)
Relative Risk Reduction (RRR) is the percent reduction in events in the treated group event rate (EER) compared to the control group event rate (CER):
RRR = (CER
- EER) / CER * 100
Risk Ratio is the ratio of risk in the treated group (EER) to
the risk in the control group (CER): RR = EER/CER. RR is used in
randomised trials and cohort studies.
Relevancy Ranking Ordering results by how relevant they are to a word, concept, or set of pages. Relevancy is determined by an algorithm which is different for each search program. Results from the See Related Articles search in PubMed are ranked by keyword relevancy.
Replicate/reproduce Do the same thing to other people in order to achieve the same benefits that occurred in a study. Also, repeating the circumstances of a study to test whether the results and outcomes are similar in another sample or population.
Reproducible Able to be done the same way elsewhere. (See replicate/ reproduce)
Retrospective/prospective (See prospective/retrospective)
Review, See Systematic Reviews
Risk difference Also called absolute risk reduction. It is literally the difference in size of risk between two groups. Eg, if one group has a 15% incidence of a disease, and the other has a 10% incidence of the disease, the risk difference is 5%.
Robot A small program used by automated search engines to look for and index new Web pages. Rough synonyms: spider, bot, worm.
Search engine A program which indexes text documents or Web pages.
Secondary study A study of studies: a review of individual studies (each of which is called a 'primary study').
Sensitivity The degree to which a search or test discovers correct results (true positives). If a search has 100% sensitivity, it finds all items which fit the question. Note that it still may have false negatives, however. Sensitivity is the proportion of people with disease who have a positive test.
SnNout when a sign/test has a high sensitivity, a negative result rules out the diagnosis; e.g. the sensitivity of a history of ankle swelling for diagnosing ascites is 92 per cent, therefore is a person does not have a history of ankle swelling, it is highly unlikely that the person has ascites.
Specificity The degree to which a search or test avoids false negatives. If a search has 100% specificity, it does not turn up any false results. Note that it still may have false positives, however. Specificity is the porportion of people free of a disease who have a negative test.
SpPin when a sign/test has a high specificity, a Positive result rules in the diagnosis; e.g. the specificity of fluid wave for diagnosing ascites is 92 per cent. Therefore, if a person has a fluid wave, it is highly likely t hat the person has ascites.
Also ~ Sensitivity and Specificity
In searching, these two terms have an analogous meaning to the way they are employed in diagnosis. Specificity refers to the proportion of documents which you retrieve which are relevant, while sensitivity refers to the proportion of all the documents which are relevant that your search manages to retrieve. (These terms are also sometimes called precision and recall respectively).
Put another way:Sensitivity is the likelihood of retrieving relevant items;
Specificity is the likelihood of excluding irrelevant items.
If you're doing a search and it yields an unmanageably large number of hits, you probably need to increase the specificity of your search; conversely, if you get too small a number of hits, you probably need to increase sensitivity.Increase specificity by:
narrowing your question
Increase sensitivity by:
broadening your question
Standard deviation A set measure of how far things vary from the average result (the mean). The mean shows where the value for most people was centred. The standard deviation is a way of describing how far away from this centre, or average, the values spread. Eg, a mean waiting time in a hospital emergency room might be two hours, but to cover most people's waiting time, you might have to give or take an hour: the waiting time is therefore 2 hours ± 1 hour. That extra one hour is the standard deviation. A person who waited 4 hours to be seen would therefore be 2 standard deviations from the mean.
Standard treatment (See conventional treatment)
Statistical significance The findings of a study may be just an unusual fluke. A statistical test can determine whether or not the results of the study are likely to be a fluke or not. That test calculates the probability of the result being caused by chance: it provides a p value (probability). If the p value is less than 0.05, then the result is not due to chance. A result with a p value of less than 0.05 is statistically significant. The 0.05 level is equal to odds of 19 to 1 (or a 1 in 20 chance). (See also p value, confidence level, power, and probability)
Statistical power (See power)
Stemming The ability for a search to include the stem of words, through the use of a thesaurus. For example, searching on nursing and also retrieving pages with the word nurse.
Stopwords Common words which are automatically excluded from a search, such as all, also, each, many. PubMed has more thatn 350 stopwords.
Stratification In research, 'strata' are sub-groups of the population that share a characteristic: eg, an age group, gender, or socioeconomic status. It is possible to compare these different layers separately, to try and see if the groups respond differently to a treatment, or if the effects of a treatment differ between the groups. Stratification is the process by which groups are separated into these distinct layers.
Surrogate measure A measure substituted for one that cannot be measured in the course of a study, eg measuring blood pressure instead of cardiac events. (See also endpoint)
Systematic error Systematic errors are caused by bias: something that is introducing (or could introduce) a system-wide distortion of results.
Systematic review A review of studies in which evidence has been systematically searched for, studied, assessed, and summarised according to predetermined criteria. Also called an overview, systematic reviews often use meta-analysis to summarise results of comparable studies.
Systematic Reviews are written for decision makers, who are users of research, and also researchers because every project for which new data is to be collected, i.e. "primary research", should be preceded by a systematic review of what is known and be concluded by incorporating that new data into a revised systematic review. (This procedure is not always followed, also, reviews may be biased and misleading).
Primary Research usually published in scientific journals of variable quality and written primarily by researchesrs for other researchers. As the primary research base is enormous and is increasing rapidly, the primary literature should be accessed only after attempts have been made to find the evidence needed from systematic reviews.
Summary A systematic review is an overview of primary studies that used explicit and reproducible methods. A meta-analysis is a mathematical synthesis of the results of two or more primary studies that addressed the same hypothesis in the same way. Although meta-analysis can increase the precision of a result, it is important to ensure that the methods used for the review were valid and reliable.
Therapeutic trial (See clinical trial)
Thesaurus A list of words and synonyms. Also called controlled vocabulary. MeSH is the thesaurus for Medline.
Toxicity The degree to which a medicine is poisonous. How much of a medicine can be taken before it has a toxic effect.
Truncation To truncate is to shorten. Using a database, truncation signifies the ability to search for a shortened form of a word, typically done with the asterisk (*). For example, searching for card* will find cardiac, cardiopulmonary, and other words.
Type I and 11 errors (See Alpha and Beta errors)
Validity The degree to which a result is likely to be 'true' and free of bias. (See also internal/external validity)
Variable A variable is a factor which differs among and between groups of people. Variables include things like age and gender, as wen as things like smoking or employment. There can also be treatment or condition variables, eg in a childbirth study, the length of time someone was in labour. AU these factors can potentially have an impact on outcomes. (See confounding variables)
Washout period/phase See cross-over trial: the washout period is the stage in a crossover trial after the first treatment is withdrawn, but before the second treatment is started. The washout period allows time for any active effects of the first treatment to wear off before the new one gets started.