JAMAevidence Glossary

Terms are derived from Users' Guides to the Medical Literature: A Manual for Evidence-Based Practice, 2nd Edition and The Rational Clinical Examination: Evidence-Based Clinical Diagnosis. Updated December 2009.
Download a PDF of the glossary (179 KB). (Adobe Reader is required to open or print PDF files.)





Random
Governed by a formal chance process in which the occurrence of previous events is of no value in predicting future events. For example, the probability of assigning a participant to one of two specified groups is 50%. See also Randomization; Random error.

Random error
We can never know with certainty the true value of an intervention effect because of random error. It is inherent in all measurement. The observations that are made in a study are only a sample of all possible observations that could be made from the population of relevant patients. Thus, the average value of any sample of observations is subject to some variation from the true value for that entire population. When the level of random error associated with a measurement is high, the measurement is less precise, and we are less certain about the value of that measurement. See also Random sample.

Random sample
A sample derived by selecting sampling units (eg, individual patients) such that each unit has an independent and fixed (generally equal) chance of selection. Whether a given unit is selected is determined by chance; for example, by a table of randomly ordered numbers. See also Random error.

Random-effects model
A model used to give a summary estimate of the magnitude of effect in a meta-analysis that assumes that the studies included are a random sample of a population of studies addressing the question posed in the meta-analysis. Each study estimates a different underlying true effect, and the distribution of these effects is assumed to be normal around a mean value. Because a random-effects model takes into account both within-study and between-study variability, the confidence interval around the point estimate is, when there is appreciable variability in results across studies, wider than it could be if a fixed-effects model were used.

Randomization
Allocation of individuals to groups by chance, usually done with the aid of a table of random numbers. Not to be confused with systematic allocation or quasi-randomization (eg, on even and odd days of the month) or allocation at the convenience or discretion of the investigator. See also Random sample; Random error.

Randomized controlled trial
Experiment in which individuals are randomly allocated to receive or not receive an experimental diagnostic, preventive, therapeutic, or palliative procedure and then followed to determine the effect of the intervention. See also Nonrandomized controlled trial.

Recall bias
Occurs when patients who experience an adverse outcome have a different likelihood of recalling an exposure than patients who do not experience the adverse outcome, independent of the true extent of exposure. See also Bias.

Recessive
Describes any trait that is expressed in a homozygote but not a heterozygote, ie, 2 copies of that allele are necessary to manifest its effect.

Recursive partitioning analysis
A technique for determining the optimal way of using a set of predictor variables to estimate the likelihood of an individual experiencing a particular outcome. The technique repeatedly divides the population (eg, old vs young; among young and old, the men and the women; and so on) according to their status on variables that discriminate between those who will have the outcome of interest and those who will not.

Referral bias
Occurs when characteristics of patients differ between one setting (eg, primary care) and another setting that includes only referred patients (eg, secondary or tertiary care). See also Bias.

Referred care
Medical care provided to a patient when referred by one health professional to another with more specialized qualifications or interests. There are two levels of referred care: secondary and tertiary. Secondary care is usually provided by a broadly skilled specialist such as a general surgeon, general internist, or obstetrician. See also Primary care.

Reflexivity
In qualitative research using field observation, whichever of the three approaches used, the observer will always have some effect on what is being observed, small or large. This interaction of the observer with what is observed is called reflexivity. Whether it plays a positive or negative role in accessing social truths, the researcher must acknowledge and investigate reflexivity and account for it in data interpretation.

Regression
A technique that uses predictor or independent variables to build a statistical model that predicts an individual patient’s status with respect to a dependent or target variable.

Rehabilitation
A set of actions designed to restore, following disease or injury, the ability to function in a normal or near-normal manner.

Relative benefit increase
The proportional increase in rates of good outcomes between experimental and control participants. It is calculated by dividing the rate of good outcome in the experimental group (experimental event rate, or EER) minus the rate of good outcome in the control group by the rate of good outcome in the control group.

Relative diagnostic odds ratio
The diagnostic odds ratio is a single value that provides one way of representing the power of the diagnostic test. It is applicable when we have a single cut point for a test and classify tests results as positive and negative. The diagnostic odds ratio is calculated as the product of the true positive and true negative divided by the product of the false positives and false negatives. The relative diagnostic odds ratio is the ratio of one diagnostic odds ratio to another.

Relative difference
The absolute difference (risk difference) in rates of harmful outcomes between experimental groups (experimental event rate, or EER) and control groups (control event rate, or CER), calculated as the rate of harmful outcome in the control group minus the rate of harmful outcome in the experimental group (CER – EER). Typically used to describe a beneficial exposure or intervention (eg, if 20% of patients in the control group have an adverse event, as do 10% among treated patients, the ARR or risk difference would be 10% expressed as a percentage or 0.10 expressed as a proportion).

Relative risk
Ratio of the risk of an event among an exposed population to the risk among the unexposed. See also Relative risk reduction.

Relative risk increase
The proportional increase in rates of harmful outcomes between experimental and control participants. It is calculated by dividing the rate of harmful outcome in the experimental group (experimental event rate, or EER) minus the rate of harmful outcome in the control group (control event rate, or CER) by the rate of harmful outcome in the control group ([EER – CER]/CER). Typically used with a harmful exposure.

Relative risk reduction
The proportional reduction in rates of harmful outcomes between experimental and control participants. It is calculated by dividing the rate of harmful outcome in the control group (control event rate, or CER) minus the rate of harmful outcome in the experimental group (experimental event rate, or EER) by the rate of harmful outcome in the control group ([CER – EER]/CER). Used with a beneficial exposure or intervention. See also Relative risk; Risk; Treatment effect.

Reliability
Reliability is used as a technical statistical term that refers to a measurement instrument's ability to differentiate between subjects, patients, or participants in some underlying trait. Reliability increases as the variability between subjects increases and decreases as the variability within subjects (over time, or over raters) increases. Reliability is typically expressed as an intraclass correlation coefficient with between-subject variability in the numerator and total variability (between-subject and within-subject) in the denominator.

Reminder systems
A strategy for changing clinician behavior. Manual or computerized reminders to prompt behavior change. See also Alerting systems.

Reporting bias
The inclination of authors to differentially report research results according to the magnitude, direction, or statistical significance of the results. See also Bias.

Residual confounding
Unknown, unmeasured, or suboptimally measured prognostic factors that remain unbalanced between groups after full covariable adjustment by statistical techniques. The remaining imbalance will lead to a biased assessment of the effect of any putatively causal exposure.

Review
A general term for all attempts to obtain and synthesize the results and conclusions of two or more publications on a given topic.

Ribosome
The protein synthesis machinery of a cell where messenger RNA translation occurs.

Risk
A measure of the association between exposure and outcome (including incidence, adverse effects, or toxicity). See also Absolute risk reduction; Relative risk reduction.

Risk aversion
People are said to be risk averse if they would accept a fixed outcome with certainty rather than a lottery with a higher expected value. For example, they would choose $10 for sure rather than a 50/50 chance of $0 or $30.

Risk factors
Risk factors are patient characteristics associated with the development of a disease in the first place. Prognostic factors are patient characteristics that confer increased or decreased risk of a positive or adverse outcome from a given disease.

ROC curve
A figure depicting the power of a diagnostic test. The ROC curve presents the test’s true-positive rate (i.e., sensitivity) on the horizontal axis and the false-positive rate (i.e., 1 – specificity) on the vertical axis for different cut-points dividing a positive from a negative test. An ROC curve for a perfect test has an area under the curve = 1.0, while a test that performs no better than chance has an area under the curve of only 0.5.

Rome criteria
A series of proposed combinations of findings used to diagnose irritable bowel syndrome. The Rome III criteria is the most recently proposed set of findings. See Table 55-2 in The Rational Clinical Examination.

Rovsing sign
A sign related to the rebound tenderness test for appendicitis. Press deeply and evenly in the left lower quadrant and then release pressure suddenly. The presence of tenderness in the right lower quadrant during palpation or referred rebound tenderness in the right lower quadrant during release is considered a positive Rovsing sign.