HS 326 key terms

A structured way to turn a clinical uncertainty into an answerable question by naming the Population, the Intervention or exposure, a Comparator, and the Outcomes of interest.

A question narrow enough to guide a literature search and broad enough to matter to patients, usually specifying who, what, compared with what, and which result counts as success.

The documented combination of databases, keywords, controlled vocabulary, and limits used to find studies; it must be reproducible so another person could repeat the search and find the same records.

Reports not produced by commercial publishers, such as conference abstracts, theses, government reports, and trial registries; including it can reduce the effect of publication bias on a review's conclusions.

Allocating participants to groups by chance so that known and unknown prognostic factors are, on average, balanced; it supports causal inference and reduces selection bias in comparative studies.

Hiding the upcoming assignment from the person enrolling participants, so that the decision to enter someone cannot be influenced by knowing which treatment that person would receive.

Withholding information about group assignment from participants, clinicians, assessors, or analysts to limit performance and detection bias; the term specifies who was masked, not merely that masking occurred.

A third factor associated with both the exposure and the outcome that distorts their apparent relationship; randomization addresses it at design, while statistical adjustment can only address measured confounders.

Distortion arising when the people studied, or the data available for them, differ systematically from the population the conclusion is meant to apply to, often through nonrandom entry or attrition.

Participants who leave a study before outcome ascertainment; if departure is related to both treatment and prognosis, results can be biased, so reasons and numbers in each arm must be reported.

Distortion caused by systematic differences between participants who remain in a study and those who drop out, particularly when dropout rates or reasons differ between comparison groups.

Analyzing participants in the group to which they were assigned regardless of what treatment they actually received; it preserves randomization but can dilute the estimated effect of actually taking treatment.

Analyzing only participants who adhered to the assigned protocol; it can estimate the effect of treatment received but is vulnerable to bias because adherence may be linked to prognosis.

The likelihood that a study's design, conduct, or reporting produced a result departing from the truth; appraised domain by domain rather than reduced to a single quality score.

A hierarchy ranking study designs by their capacity to support causal or therapeutic claims, typically placing systematic reviews of randomized trials above single trials, cohort studies, case series, and expert opinion.

A framework for rating certainty of evidence for each outcome, starting high for randomized trials and low for observational studies, then downgrading or upgrading for specified concerns such as inconsistency or imprecision.

A review that poses a specific question, searches comprehensively using a predefined strategy, appraises included studies for bias, and synthesizes results, often with meta-analysis and an assessment of certainty.

Statistical pooling of results from multiple studies to produce a summary estimate; appropriate only when studies are similar enough in populations, interventions, and outcomes for a combined number to be meaningful.

Real differences among study results beyond chance, arising from varied populations, doses, settings, or definitions; quantified statistics such as I-squared describe its magnitude but do not explain its cause.

A graph displaying each study's effect estimate with its confidence interval alongside a pooled estimate, letting readers see the spread of results and how much each study contributes to the summary.

A range of values consistent with the observed data at a stated probability, usually 95 percent; it conveys both the size of an estimate and the precision with which it was measured.

The probability of observing data at least as extreme as those obtained if the null hypothesis and its assumptions were true; it is not the probability that the null hypothesis is true.

A result unlikely under the null hypothesis by a preset threshold, usually 0.05; it says nothing about the size or clinical importance of an effect and is sensitive to sample size.

Uncertainty in an estimate due to few events or participants, reflected in wide confidence intervals that include both trivial and important effects; a common reason to downgrade certainty of evidence.

The arithmetic difference in event risk between comparison groups, expressed as a proportion; unlike a relative measure, it reflects how many people are actually affected by the intervention.

The ratio of event risk in the exposed or treated group to that in the comparison group; it can look large even when absolute differences are small, so it should be read alongside baseline risk.

The ratio of the odds of an outcome in one group to the odds in another; it approximates relative risk only when the outcome is rare, and it overstates risk when events are common.

The number of patients who must receive an intervention for one additional person to benefit, calculated as the reciprocal of the absolute risk reduction; a clinically intuitive measure of effect size.

The smallest change in an outcome that patients would regard as worthwhile; it gives meaning to statistical results by defining the threshold an effect must cross to matter in practice.

The proportion of people who truly have a condition and are correctly identified by a test; a highly sensitive test is useful for ruling out disease when the result is negative.

The proportion of people without the condition who are correctly identified as negative by a test; a highly specific test is useful for ruling in disease when the result is positive.

How much a test result multiplies the pretest odds of disease; it combines sensitivity and specificity and, unlike predictive values, remains stable across settings with different disease prevalence.

The estimated chance a patient has the target condition before a test is ordered, based on history, examination, and local prevalence; it anchors how a result should change management.

Distortion of a test's measured accuracy when the studied patients differ in severity or comorbidity from those in whom the test will be used, typically exaggerating performance in milder cases.

The number of new cases arising in a population over a defined period, divided by the population at risk; it measures how quickly disease develops and is the basis of risk estimates.

The proportion of a population that has a condition at a given time; it reflects both how often disease occurs and how long it lasts, and it drives the predictive value of tests.

The total time each person was observed and remained at risk, summed across the group; dividing events by person-time yields an incidence rate that accommodates unequal follow-up.

A design that samples people with an outcome and comparable people without it, then looks back at prior exposures; efficient for rare diseases but vulnerable to recall and selection bias.

A design that follows groups defined by exposure status forward in time and compares outcome occurrence; it can estimate incidence and establish temporality, though unmeasured confounding remains a threat.

The probability of receiving an exposure given measured baseline characteristics, used to balance treated and untreated groups in observational data; it cannot adjust for factors that were never measured.

A conversation in which clinician and patient exchange evidence about options and the patient's values, then agree on a plan; it requires communicating absolute risks and uncertainties in plain language.

A tool that presents options, outcomes, and probabilities in a balanced, understandable format to support patient choice; good ones report absolute risks and disclose their sources and conflicts.

A participant's or patient's voluntary agreement after receiving understandable information about purpose, procedures, risks, benefits, alternatives, and the right to withdraw without penalty.

An independent committee that reviews research involving humans before it begins, judging risks, benefits, equitable selection, consent procedures, and protections for vulnerable participants.

A research participant's mistaken belief that study procedures are designed solely for personal benefit rather than to produce generalizable knowledge, which undermines the validity of consent.

The three ethical foundations of human research: respect for persons, beneficence with nonmaleficence, and justice; they underpin modern consent requirements and fair subject selection.

A financial or personal interest that could compromise objectivity or appear to do so; disclosure and management protect credibility, and undisclosed industry funding is a recognized source of bias.

The tendency for studies with positive or striking results to be published, submitted, or cited more readily than null ones, which inflates pooled effects in reviews that rely on published literature.

A scatterplot of study effect against a measure of its precision, used to look for asymmetry suggesting small negative studies are missing; asymmetry can also arise from genuine heterogeneity.

A person's capacity to obtain, process, and understand basic health information needed to make decisions; low health literacy calls for plain language, teach-back, and numerical formats people can grasp.

Asking a patient to restate information in their own words so the clinician can confirm understanding and correct gaps; it checks communication rather than the patient's intelligence.

The process of moving research findings into routine practice through tailored tools, reminders, audit and feedback, and local adaptation, recognizing that dissemination alone rarely changes behavior.

Systematically developed statements that translate evidence into recommendations for specific clinical circumstances; trustworthy ones grade the evidence and disclose conflicts and the basis for each recommendation.

The study of methods that promote the systematic uptake of evidence into practice and policy, examining barriers, facilitators, and strategies rather than assuming good evidence speaks for itself.

The extent to which findings from a studied sample apply to other populations and settings; it depends on who was enrolled and how closely the care delivered resembles usual practice.

Efficacy is benefit under ideal, tightly controlled conditions, while effectiveness is benefit under ordinary practice with real patients and imperfect adherence; the two often differ substantially.