| About PMID Numbers: We frequently utilize a PMID number in place of a citation. Where PMID numbers are available, enter that number into the PubMed search box to retrieve that citation and listing. A B C D E F G H I J K L M N O P Q R S T U V W X Y Z Absolute
risk reduction (ARR) – A measure of outcomes which reflects
the actual percent difference in study outcomes between groups.
See a visual description here. Accuracy – The ability to correctly identify that which one intends to identify. Active agent — Active drug (i.e., not the placebo). Adjustment - Statistical means for minimizing differences in study and comparison groups (e.g., standardization and regression analysis).Adverse event (ADE) — see Safety Aim, study — see Objectives Allocation concealment — see Concealment of allocation Alpha error — see Error A priori – In advance. For example, issues such as study questions, outcome measures, subgroups for analysis and p-values should be determined in advance of the actual study (i.e., determined a priori). Ascertainment bias - See bias Assessment bias - See bias Association – Statistical relationship between two or more events, characteristics or other variables. Attrition bias - See bias Baseline characteristics — Key characteristics of interest at the start of a study which are potential prognostic factors. These usually include demographic information such as age, gender, race and other characteristics related to the area under study such as disease severity, baseline measurements for clinical items of interest, etc. Collecting and reporting on baseline characteristics helps to establish how similar the study groups are, describes the actual population studied and helps to measure improvements during the course of the study. Benchmark – A standard or point of reference used in measuring and/or judging something such as quality or value. Beta error - see Error Bias – Study processes which may result in, or lead to, conclusions differing from truth in a systematic way (meaning not due to chance). Bias may occur at various study stages such as assigning subjects to study or comparison groups, intervention or exposure, performance, provision of services or conduct of processes affecting subjects, data collection, subject follow-up, measurement, analysis, interpretation and/or publication of data. Bias frequently occurs as a result of some inequality between the study and the comparison group. The inequality may be important because it may affect the study outcome.
Blinding
– An important study procedure to keep secret certain study
procedures such as which is an active drug and which is a placebo.
Bias can result when study subjects and those involved in study
procedures know treatment assignment of individual subjects. Blinding
is a method to help avoid the introduction of this kind of bias.
Double-blinding refers to when neither patient nor persons performing
the intervention or exposure — or working with the subjects'
data — know if the patient is in the study group or the comparison
group. Case — Research subject who is a member of the group possessing the characteristic under study such as an exposure, risk factor, outcome or receiving the intervention in an experimental study. Case-control study – An observational study in which subjects with outcomes of interest (cases) are compared to those without the outcome (controls). Histories are examined to attempt to determine exposures. Case series – A group of patients receives an intervention and outcomes are assessed. There is no comparison group. Case series is almost never “evidence.” All-or-none results are required, which rarely happens. Case study — An observational of one person. Also referred to as a case report. A case study is not evidence. Chance
— Research results happening by "accident," meaning
that they did not happen because of truth or because of bias. Likelihood
is assessed by using appropriate statistical tests. To critically
appraise, examine p-values and confidence intervals with associated
point-estimates (none of which can address bias, which is a systematic
error - meaning an error not due to chance). Class effect - Class effect refers to a concept that certain drugs share sufficient similarities that they can be thought of as a class. It is a determination that a set of agents with similar chemical structures, mechanisms of action and pharmacological effects has similar therapeutic and adverse effects. There are no universally accepted criteria for defining class effect. (For all intents and purposes, class effect treats the agents as if they are the same clinically, but some other factors, such as cost, may vary.) Clinical practice guidelines – Clinical practice guidelines are systematically developed statements to assist practitioners and patients in choosing appropriate healthcare for specific conditions. They should be based on valid and useful evidence, and they should be critically appraised before adoption. Clinical significance — Research should benefit patients in ways that are clinically significant (i.e., that matter to them). Clinically significant areas are morbidity, mortality, symptom relief, functioning and health-related quality of life. Results from valid research in these areas should be large enough and useful enough to provide meaningful clinical benefit. Clinical trial — A research methodology which evaluates a drug or other intervention to assess safety and efficacy by evaluating its effect on a group of subjects. The subjects are divided into two or more groups with the study group receiving the drug or intervention and the comparison group or groups receiving a comparison drug, placebo or intervention. Cohort study – An observational study in which exposure occurs in the study group and not in the comparison group. Exposure is not introduced by the investigator, but happens through other means such as natural occurrence or individual choice. Groups are followed over time and outcomes are compared. Cohort studies can be done prospectively where a group of interest and a comparison group are followed forward in time. Cohort studies can be done retrospectively when the group of interest and comparison group are identified from the past, based on an exposure or characteristic, and then information about them is reviewed from that time forward to assess development of outcomes. Comparison groups — Subjects in groups being compared to the group of interest (e.g., intervention group would be “group of interest” and placebo group would be the “comparison group”). Composite endpoints — Refers to an endpoint in which single endpoints are grouped together to form one endpoint such as “cardiovascular events.” Concealment
of allocation – The process used in a randomized controlled
trial to hide the assignment to a study group. The purpose is to
ensure that no one can influence or control which study subject
gets assigned to which group (e.g., assignment made through a call-center,
etc.) Confidence
interval (CI) – For a valid study, the range in which the
likelihood of a true value is expected to be, within a given degree
of certainty, usually 95 percent. Confounding – A special type of bias in which another factor associated with the study variable of interest may have "traveled" with that variable and is the true reason for the study conclusion instead of the variable under study. There are known confounders and unknown confounders. Randomization is a method which attempts to minimize confounding by randomly allocating subjects to their groups in hopes that any potential confounders are equally distributed between the groups. Contamination – Any member of a comparison group receiving the intervention under study or being exposed to the variable of interest. Control - Research subject who is a member of the comparison group. Controlled
clinical trial – An experiment in which investigators include
a control group which is used as a comparison group. The comparison
group may receive another treatment, a placebo or “usual care”.
Subjects are assigned to the study or to the control group. Intervention
or exposure is then performed. Patients are followed to determine
outcomes such as improvements or harms. Critical appraisal – A scientific evaluation of evidence (e.g., research data) to appraise validity (closeness to truth) and usefulness (e.g., generalizability to one's own patients or circumstances, meaningful benefit, etc). Cross-over design - A study design in which patients start out having one intervention and then are “crossed over” to a different intervention. Crossover
trial – A comparison of two or more interventions in which
the participants receive one treatment and then are “crossed
over” to another intervention. Care must be taken with this
study design because effects of the first treatment could contaminate
the effects of the subsequent intervention. Cross-sectional study – An observational study in which a sample from a larger population is examined, at one single point in time – like a snapshot - for prevalence of an exposure or characteristic along with an outcome of interest. Cross-sectional studies are also frequently used to evaluate diagnostic testing. Database research — Statistical relationships between two or more variables are assessed from databases. Frequently database research is used to report outcome measures. Database research, however, is a type of observation and, like all observational research, is highly prone to selection bias, observation bias and confounding — and the likelihood of finding "statistically significant" relationships merely due to chance is high to certain, depending upon the way the analysis is conducted. Definite cause and effect conclusions should not be drawn from database research. Data gathering validity – Data gathering validity refers to the methods used to obtain numerators and denominators during the data gathering phase of performance measurement. Delta — See Equivalence trials and Non-inferiority trials Denominator
– The lower portion of a fraction used to calculate a rate
or ratio. In a rate, the denominator is usually the population (or
population experience, as in person-years, etc.) at risk.
Dependent
variable – In a statistical analysis, the outcome variables
under study. Detection bias - See bias Diagnostic testing — see Measures of test function Disease spectrum — A range of symptoms, signs, lab results and results of other diagnostic tests, and rate of disease progression, response to therapy and disease severity. Double-blind – see Blinding Double-dummy treatment design — A treatment design involving two placebos. Control subjects receive one or the other of two placebos, but not both. Study subjects receive the intervention drug. For example, if you were studying an antibiotic being given at two different doses, you might need two placebos because of the weight and size difference of the doses. Drug utilization review (DUR): A system for evaluating appropriateness of drug therapy. Prescribing patterns are evaluated to determine if drugs are being misprescribed, possibly resulting in problems with safety or effectiveness. Effectiveness
– The extent to which a given intervention is likely to produce
beneficial results for which it is intended in ordinary circumstances. Efficacy
– The extent to which a given intervention is likely to produce
beneficial effects in the context of the research study. Empirical – Empirical results are those which are based on experience or observation. Encapsulation
— A drug blinding method whereby drugs (active or inert) are
crushed and put into gelatin capsules to disguise the active agent
as compared to the placebo. Endpoint — see Outcome measure Epidemiology – A study of the distribution and determinants of health-related states or events in specified populations, and the application of this study to the control of health problems. Equipoise -In healthcare, a state in which it is unknown which interventions will result in best outcomes. Equipotency - See Class effect Equivalence trial - Equivalence trials are usually used to demonstrate that the effects of two treatments do not vary more than a prespecified clinically acceptable amount and can therefore be considered clinically equivalent. Delta is the name given to the range of results within which results are judged to be equivalent. Error —
Estimate of effect – see Measures of outcomes Evidence-based medicine (EBM) – Delfini definition: "The use of the scientific method and application of valid and useful science to inform health care provision, practice, evaluation and decisions." Experimental
study – A study in which the investigator controls an exposure
or intervention, then follows subjects to compare outcomes between
research subjects who are exposed or who receive the intervention
and those who are not exposed or who do not receive the intervention.
Tip: If an intervention is "assigned" through the research,
it is an experiment. If it is chosen, then the study type is observational. Exposed group – A group whose members have been exposed to a supposed cause of disease or health state of interest, or possess a characteristic that is a determinant of the health outcome of interest. External
validity – Whether a study's results are generalizable either
to a patient population (see Population bias) or to "real world"
circumstances (see Intensity bias). Also referred to as First-line therapy: The drugs to be utilized first in treating a patient. Follow-up bias - See bias Foraging
tools - Tools that assist with keeping up-to-date with new information.
Formulary: List of therapeutic agents AQ29available in a particular practice for caring for patients. The term “preferred drug list” is also used.
Formulary system: A system that provides for the processes for establishing and managing the formulary. Generalizability – see External validity Generic substitution: Replacement of one agent with a different agent having the same chemical structure. This may be done when the patent on a brand-name drug expires. Bioequivalence is frequently assumed (i.e., it is assumed that the generic agent is equivalent to the brand-name drug). In some cases the effects of other components of the generic preparation (e.g., the vehicle in a dermatological preparation) may vary and result in outcomes that differ from those reported for the brand-name agent. Gold standard – The intervention generally believed to be the best available against which new interventions should be compared. Halo effect - How an observer's perception, knowledge or recollection, may bias results. Also used as a synonym for Placebo effect when medical attention and services can bias results. Hawthorne effect - How the knowledge of being in a study can influence a subject's behavior (usually favorably) which can bias results. Hazard ratio - The slope of the survival curve to provide estimates of how rapidly subjects are dying in the treated versus the control group. The hazard rate is the probability that if the event in question has not already occurred, it will occur in the next time interval The time interval is short, so the hazard rate represents an “instantaneous” rate. Hunting
tools - Tools which aid the process for searching to answer a clinical
question. Hypothesis
— Tentative explanation that forms the basis of a research
study. Imputation
— As in "imputation of missing variables." Principles
of Intention-to-Treat (ITT) analysis require analyzing all patients
in the groups to which they were assigned. Investigators are to
"impute" or assign outcomes for missing data points. Example:
worst case scenario in which study subjects with missing outcomes
are assigned as "treatment failures," and comparison subjects,
assigned as "successes." Frequently "last-observation-carried-forward"
(LOCF) is used, however, this has been shown to be a method prone
to bias. Imputing outcomes for missing data points is not a method
for "determining the truth of what may have happened if we
had no missing data," but rather a method to test the strength
of the outcomes (i.e., statistically similar results to those reported)
considering all the data points in such a way that does not favor
the intervention. LOCF would be especially biased in patients with
a progressive illness, for example, because patient outcomes would
appear better than would be expected in a progressive illness. Incidence
– The proportion of new cases of the target disorder in the
population at risk during a specified time interval. Incidence rate – A measure of the frequency with which an event, such as a new case of illness, occurs in a population over a period of time. The denominator is the population at risk; the numerator is the number of new cases occurring during a given time period. Independent
variable – An exposure, risk factor, or other characteristic
being observed or measured that is hypothesized to influence an
event or outcome (i.e., the dependent variable). Inference, statistical – In statistics, the development of generalizations from sample data, usually with calculated degrees of uncertainty. Intensity
bias - An external validity bias in which the circumstances of the
study differ meaningfully from the circumstances to which study
methods might otherwise be applied (e.g., "real world"
settings as contrasted with the highly controlled environment often
found in controlled trials). Intention-to-treat
– Analyzing results for all patients in the groups to which
they were assigned whether or not they received or completed the
intervention or experienced the exposure. The number randomized
to each group should equal the number analyzed in each group —
and they should be the same people. Intermediate
Outcome Markers – Outcome measures, such as a biologic factor
(biomarker) or lab/imaging test, that are “assumed”
to represent clinical outcomes (e.g., blood pressure used as a surrogate
end point in studies of stroke). Internal
Validity - Closeness to truth within the context of the study (i.e.,
truth of the study not taking into account external validity). Assessing
internal validity entails "ruling out" bias, confounding
and chance as possible explanations for an observed association
between an element of interest in a study and resulting outcomes. Interventions — Includes screening, prevention and treatments. Only valid randomized controlled trials with useful results should be used to inform decisions about interventions in these areas. Intervention bias - See bias Lead
time bias - A bias resulting from a disease found through screening
as compared to when it might otherwise have been detected. This
kind of bias can result in a treatment seeming to be very effective
if the lead time is long (e.g., “increased” survival
time). Length
time bias - A bias that can occur when certain characteristics or
conditions under study differ in the speed of progression. This
kind of bias can result in findings favoring screening. An example
is tumors. Faster-growing tumors causing symptoms will be more likely
to be found outside of screening. Slower-growing, asymptomatic tumors
will have a longer duration and be less likely to be found outside
of screening. Thus, screening will identify more slower-growing
asymptomatic tumors which could then result in a conclusion that
screening helps prevent mortality. This kind of bias is most likely
to occur in screening studies and case-control studies where prevalence
cases are included, rather than incidence cases because incidence
cases are assumed to be new starts. Likelihood
ratios (LR) – Likelihood ratios can be helpful for comparing
one test to another, and results can help rule in or rule out a
condition. Line
of no difference – The point at which there is no greater
benefit or risk one way or another (meaning the meeting point for
"favors intervention" versus "favors placebo.")
Matching – A mechanism to reduce selection bias by attempting to make a study group and a comparison group as comparable as possible by matching similar variables such as age and sex, etc. Mean, arithmetic – The average. Measure of association – see Measures of outcomes Measurement bias - See bias Measurement of risk – see Measures of outcomes Measures
of outcomes — Statistics that show the size of differences
between the results from the study groups.
Measures
of test function — Measures to help determine the accuracy
and usefulness of diagnostic tests.
Median – The data midpoint. Meta-analysis
– A quantitative technique for summarizing results of more
than one study using predetermined criteria. The goal is to provide
a summary estimate of effect based on the scientific weight of the
studies. Meta-analysis can be achieved through use of the results
of individual studies or actually pooling data from those studies. Mode – The most frequently occurring value in a dataset. Monograph - A written review and analysis, often of a single agent, containing a list of usage recommendations. Multivariable analysis - A statistical method for determining the specific contributions of various factors to a single outcome. This allows investigation of a single variable while controlling for the effect of other variables. Multivariable analysis determines the independent contribution of each independent variable to the dependent variable (i.e., development of coronary heart disease).
Multivariate analysis - A statistical method for determining the specific contributions of various factors to a more than one outcome. Narrative review: An article in the medical literature summarizing other studies for a given topic, characterized by a lack of a transparent, scientific, and systematic approach; thus, the summary is highly likely to be misleading. Instead, systematic reviews should be sought (and appraised). Negative
likelihood ratio – The number of times the percent of false
negatives occurs over percent of true negatives.
Negative
predictive value (NPV) – Of all testing negative, percent
who do not have a condition, based on that population's prevalence. Non-inferiority trial - Non-inferiority trials aim to show that an intervention is not inferior to a comparison intervention by more than a prespecified clinically acceptable amount (Delta). Judgment is required to establish what is meant by a clinically acceptable amount. Non-parametric - Non-parametric refers to instance in which the distribution is unknown. Parametric refers to instances in which parameters of the distribution of the variable of interest is known (e.g., knowledge that a distribution of a variable is expected to follow a bell curve. Null
hypothesis – The first step in testing for statistical significance
in which it is assumed that the exposure is not related to a disease.
This is done for mathematical purposes to level the playing field
to begin with the assumption that study outcomes are the same as
what would have occurred by chance. Number-needed
to diagnose (NND) — From Bandolier: "For any chosen clinical
endpoint the NNT is the reciprocal of the fractional improvement
in a treated group minus the fractional improvement in an untreated
group NNT = 1/(fraction improved with active - fraction improved
with control)" or NND = 1/[Sensitivity - (1 - Specificity)].
Primarily useful for comparing the NND values between different
tests. Best outcome is as close to 1.0 as possible. (http://www.jr2.ox.ac.uk/bandolier/band27/b27-2.html) Number-needed-to-treat
(NNT) – The number of patients who need to be treated in order
for one patient to benefit within the study time period. NNT is
the reciprocal of the ARR. Numerator
– The upper portion of a fraction.
Objective — Goal or aim of the study. Observation bias - See bias Observational
study – Epidemiological study in which observations are made,
but investigators do not control the exposure or intervention and
other factors. Changes or differences in one characteristic are
studied in relation to changes or differences in others, without
the intervention of the investigator. Observational studies are
highly prone to selection bias, observation bias and confounding.
Tip: If an intervention is "assigned" through the research,
it is an experiment. If it is chosen, then the study type is observational. Odds - The likelihood of an event occurring compared to not occurring--e.g., odds of two to one mean that likelihood of an event occurring is twice that of not occurring. Odds
ratio (OR) – A point estimate used for case-control studies
which attempts to quantify a mathematical relationship between an
exposure and a health outcome. Odds are used in case-control studies
because the investigator arbitrarily controls the population; therefore,
probability cannot be determined because the disease rates in the
study population cannot be known. The odds that a case is exposed
to a certain variable are divided by the odds that a control is
exposed to that same variable. Odds are often used in other types
of studies as well, such as meta-analysis, because of various properties
of odds which make them easy to use mathematically. Open
label study - Not blinded as to the intervention. Outcome — see Outcome measure Outcome
measure – What we are interested in studying (e.g., mortality,
use of rescue medications in asthma patients, incidence of no flares
in atopic dermatitis). Overdiagnosis - A finding of a disease at an asymptomatic stage in a patient who would not have become symptomatic or harmed by the disease. Override - Process of setting aside a prescriber’s choice of a medication and usually substituting another medication. Parametric - Parametric refers to instances in which parameters of the distribution of the variable of interest is known (e.g., knowledge that a distribution of a variable is expected to follow a bell curve. Non-parametric refers to instance in which the distribution is unknown. Pathophysiology - Disruption of normal biochemical or physical function by a disease or disorder. Pearson
correlation co-efficient - An analysis method to measure the extent
of the linear relationship of two variables to determine how independent
and dependent variables change together (e.g., salt intake and blood
pressure). Performance bias - See bias Performance measure – A quantitative assessment of a process of care, outcome or service usually used in quality improvement work. It consists of a denominator (e.g., population of interest), a numerator (i.e., a count of events of interest occurring within the denominator) and a frequency (i.e., the specified interval for measurement). May target various levels or units such as a system, specialty group or individual. Usually expressed as a rate, ratio or percentage. Sometimes used interchangeably with “quality indicator.”
Pharmacy & Therapeutics Committee: Committee charged with making formulary management decisions, along with performing other formulary management functions. Pharmacy benefit manager (PBM): A company that manages pharmacy benefits and formulary management for health care systems and/or insurance companies. Phase (clinical) studies –
Placebo
– An inactive substance or procedure administered to a patient
for comparison to the intervention. Used in clinical trials to blind
people to their treatment allocation. Pivotal trial - A controlled trial to evaluate the safety and efficacy of a drug in patients and usually the basis for the New Drug Application (NDA) filing with the FDA. Point estimate – see Measures of outcomes Population – For research studies, the term "population" may refer to a group of people living in a specific geographic area or it may mean a unique set of individuals who share some characteristic such as exposure to a disease or a group of patients of a single clinician, as examples. In statistics, this can mean a population of data. Population
bias - An external validity bias in which the population under study
differs meaningfully from the population to which the results might
be applied. Positive
likelihood ratio (LR+) – The number of times the percent of
true positives occurs over percent of false positives. This represents
the change from pre-test odds to post-test odds.
Positive
predictive value (PPV) – Of all testing positive, percent
who have the disease, based on that population's prevalence. Post-test
odds for positive test — The odds of a person having a condition
if the test is positive. Post-test
odds for negative test — The odds of a person not having a
condition if the test is negative. Post-test
probability for a positive test — After learning test result,
the probability that a person testing positive has the condition. Post-test
probability for a negative test — After learning test result,
the probability that a person testing negative has the condition. Power - See Statistical power. Precision – The ability to provide sufficient detail, such as small incremental units, to be useful. Predictive
value – A diagnostic measure of function which is used to
address the probability of having or not having a condition. Predictive
value is affected by prevalence. Pre-test
likelihood — Pre-test probability of having a condition based
on the prevalence of the condition in the population studied. Pre-test
odds — Pre-test odds of having a condition based on the prevalence
of the condition in the population studied. Prevalence
– The number or proportion of cases or events or conditions
in a given population. Prevalence rate – The proportion of persons in a population who have a particular disease or attribute at a specified point in time or over a specified period of time. Prior authorization: Requirement that a clinician obtain approval before a drug can be dispensed and/or covered. Probability - The likelihood of an event occurring expressed as a number between 0 and 1. It is measured by the ratio of the event to the total number of possible events. Example: the probability of flipping a coin and coming up with heads is .5. Proportion
– A type of ratio in which the numerator is included in the
denominator. The ratio of a part to the whole, expressed as a “decimal
fraction” (e.g., 0.2), as a fraction (1/5), or, loosely, as
a percentage (20%). Proxy outcome markers – see Intermediate outcome markers. Publication bias – The bias toward the publishing of studies showing statistically significant “positive” results. P-value
– The probability that the results observed in a valid study
could have arisen by chance. The p-value of <.05, often set as
the point of statistical significance, is conventional and simply
means there is less than a 1 in 20 chance that the findings observed
are due to chance. Qualitative research - Any research that utilizes non-quantitative methodology, e.g., case studies or narrative description or interviews. Quality indicator – Oftentimes used interchangeably with “performance measure.” Quality indicators are specific and measurable elements of health care that can be used to assess the quality of care. Randomization – Mechanism to assign, by chance, “intervention / no intervention” assignments to study participants in randomized controlled trials.
Randomized
controlled trial (RCT) – An experiment in which investigators
control an exposure or intervention, and subjects are randomly assigned
to the study group or to the control group or groups. Intervention
or exposure then is performed. Patients are followed to determine
outcomes such as improvements or harms. Random sample – A sample derived by selecting individuals such that each individual has the same probability of selection. Rate
– Derived by dividing a numerator by a denominator. The numerator
is a subset of the denominator. Example: The percentage of diabetic
eye exams for Type I diabetics was 80% for our clinic. Ratio
– A numerator and denominator (the numerator is not required
to be a subset of the denominator). Example –The ratio of
women to men in these studies was 1 to 5 (which can also be expressed
as 1:5). Or, the rate and ratio for diabetic eye exams in our clinic
would be 80/100. Rebate: A cost offset that pharmacy systems receive when purchasing large quantities of drugs. AQ30Selection of drugs for large bulk purchase is based on the drugs’ volume of use. Recall
bias – Inaccuracies resulting from data collected from study
participants who are asked to retrospectively self-report on study
items of interest. This kind of bias is most likely to occur when
certain behaviors are hard to track (e.g., diet) or are often hidden
for some reason such as being socially sensitive (e.g., sexual behaviors,
smoking behavior, etc.). Regression model – A method for estimating the relationship of a dependent variable to a variety of independent variables. Regression-toward-the-mean – Returning to the “average” state. Meaning that extreme test values are statistically likely to move to an average over time. When patients present with extreme values and then seem to have improvement, it may be falsely attributed to an intervention when it is truly due to regression-to-the-mean. A comparison group with no intervention can help expose this effect. Relative
risk (RR) - A point estimate in which the risk of some health-related
event, such as disease or death, is compared between a study group
and a comparison group. Relative risk is expressed as the number
of times one group may be at risk over another. A relative risk
of less than 1 is a lower risk. Relative
risk reduction (RRR) – A point estimate in which the percent
reduction in events in the intervention or exposed group is compared
to the comparison group — relative risk reduction is the proportional
difference in size between outcomes. For example, if we had 15%
mortality in a comparison group as compared to 10% mortality in
the intervention group, since 10 is one-third smaller than 15, the
RRR would be 33%. Relative risk reduction begs the question "relative
to what?" Because RRR overestimates the actual difference,
it should always be used in conjunction with another point estimate
such as absolute risk reduction or number-needed-to-treat when communicating
study results. Representative sample – A sample whose characteristics correspond to those of the original population or population of interest. Retrospective study – A study method to look at cause and effect after outcomes have already occurred. Generally this refers to a case-control study; however, cohort studies can use a retrospective method. Risk – The probability that an event will occur. Risk adjustment – The process of adjusting performance rates or other outcomes of care to level the playing field due to differences in health status between populations. Risk factor – An aspect of personal behavior or lifestyle, an environmental exposure, or an inborn or inherited characteristic that is associated with an increased occurrence of disease or other health-related event or condition. Risk ratio – see Relative risk Risk stratification – The process of, or result of, separating a sample into subsamples based on health status or risk factors such as age, comorbidities, etc. Safety
— Safety in research studies pertains to harms of various
interventions. Sample
– A selected subset of a population. A sample may be random
or non-random and it may be representative or non-representative. Sampling error - See chance Second-line therapy: Therapy for patients failing on drugs established as “first-line” therapy. Selection bias – See bias Sensitivity
(SN) – Correct identification by a screening test or case
definition as having disease – of all those with a disease,
the percent testing positive (true positives). Sensitivity is derived
from calculations based on people who are known to have the condition.
Sensitivity is especially useful when it is important not to miss
a disease. High sensitivity is considered to "rule in"
a condition. Specificity
(SP) – Correct identification by a screening test or case
definition as not having disease – of all those without a
disease, the percent testing negative (true negatives). Specificity
is derived from calculations based on people who are known not to
have the condition. Specificity is especially useful when it is
important to avoid false positives. High specificity is considered
to "rule out" a condition. Standard deviation (SD) – A widely used measure of dispersion of a frequency distribution, equal to the positive square root of the variance. Standard deviation can tell you how widely or how tightly variables are distributed around the average. One standard deviation accounts for approximately 68 percent of the distribution. Two standard deviations account for approximately 95 percent. Statistical
power - The ability to reliably determine a statistically significant
relationship between interventions if one exists. Power comes into
play in planning and interpreting studies. Investigators want to
know the risk of drawing erroneous conclusion about efficacy and
so they do power calculations to estimate the number of subjects
needed in a study to show a statistically significant result if
one exists. It is conventional to accept a 5% risk of concluding
that an outcome is truly different in the study groups when they
are not (Type I or Alpha error). It is also conventional to accept
a 20% risk of concluding that an outcome is not statistically significant
when in truth there may be a true difference (Type II or Beta error).
Power can be determined by considering Alpha, Beta, the event rate
in the comparison group, and what is judged to be a clinically significant
difference between the groups. As a practical matter, many users
focus on the confidence intervals. For valid studies, consider what
you judge to be a reasonable range for clinical significance –
this need not be hard and fast. For statistically significant findings,
is the confidence interval wholly within bounds for clinical significance?
For non-significant findings, is the confidence interval wholly
beneath your limit for clinical significance? A yes to these two
questions means likely conclusive findings for valid studies. No,
means findings are inconclusive. Statistical
significance – The extent to which study results are unlikely
to be due to chance. Expressed as a p-value and it is typically
set at <.05 due to convention. Statistical significance may be
set lower, but must be determined in advance of the study. Subgroup analysis — Analysis of research results on subsets of patients in a study, such as a subgroup of Caucasian women in a study of asthmatics of all races, ages and genders. Subgroups for study must be determined in advance of a study — otherwise the results are highly likely to be due to chance. Superiority trial - Clinical trials which are conducted to determine if one intervention is superior to another. Surrogate end points – See Intermediate outcome markers. Systematic
review – Studies that examine more than one study on a given
topic in a systematic way using predetermined criteria. The goal
is to provide a summary estimate of effect based on the scientific
weight of the studies. Therapeutic substitution: Substitution of a drug with an agent having a different chemical structure but similar clinical benefits. Trend - A term used to describe observed results in several studies which consistently go in the same direction, but are not statistically significant. Triangulation issues: A catchall term for all the various considerations that need to be made in making a decision. Type 1 error - see Error Type 2 error - see Error Two-by-Two Table — Table to report results and from which statistics are calculated. See examples below: Intervention Example
Diagnostic Testing Example
Variable – Any characteristic or attribute that can be measured.
Related: Dependent variable, independent variable Variance – A measure of the spread of a variable about its mean value. In a data set, a single point lies above or below the mean for the entire dataset. A deviation score is the measure of how much each point lies above or below the mean for the entire dataset. Variance is the mean of all the deviation scores. It is a way of mathematically stating how far deviations are from expected results. Washout period – Used in a cross-over trial to discontinue patients on a treatment before the second treatment is given to reduce the likelihood that the first treatment will affect the outcome even after it is stopped. |
©
Delfini Group, LLC, 2002-2008. All Rights Reserved Worldwide.
Use of this website implies your agreement to our Notices »
Counter courtesy of www.amazingcounters.com and sponsors below: