Incidence/Prevalance & Critical Appraisal/Bias

epidemiology

the branch of research dedicated to exploring the frequency & determinants of disease or other health outcomes in populatons

incidence

a descriptve epidemiological estimate that is concerned with how many persons have ONSET of a condition during a given span of time

cumulative incidence

the number of new cases of a disease during a specified time period divided by the total number of people at risk; the proportion of new cases of a disease in a population

prevalence

defined as a proporton of a total population of people who have a particular health-related condition

bias

Any infuence that may interfere with the valid relatonship between variables, potentially resulting in misleading interpretation of outcomes

systematic error

a form of measurement error, where error is constant across trials (e.g., mis-calibrated scale)

random error

a measurement error that occurs by chance, potentally increasing or decreasing the true score value to varying degrees (e.g., person misreading & recording an incorrect measurement)

methods to reduce measurement error

‒ Standardized assessment

‒ Methods of informing, training, & ensuring accuracy among raters

‒ Repeated measures

pretrial bias

‒ Study design

• Data collection methods

• Data collection protocols (e.g., training, blinding)

‒ Recruitment

during trial bias

Data collecton & recording

post trial bias

Data analysis

‒ Publicaton

reliability coefficient

the ratio of the variance of the true score to the total variance observed on an assessment

Interpretaton: ranges from 0 to 1.0 (1.0 = no error)

test-retest reliability

measures the ability of an assessment to remain stable

over time in what it aims to measure

‒ Estmated from a single assessment when data is gathered from the same group of subjects on two or more occasions within a short tme frame

split-half reliability

when investigators divide the items of a questonnaire into two smaller questonnaires (usually dividing it into odd an even items, or frist half-last half) & then correlates the scores obtained from the two halves of the assessment to test the reliability of the entire measure

unidimensional

constructs that are expected to have a single underlying dimension (e.g., weight)

multidimensional

constructs that consist of two or more underlying dimensions (e.g., quality of life)

parallel forms reliability

involves administraton of the alternatve forms to subjects at the same tme

internal consistency

the extent to which the items that make up an assessment covary or correlate with each other (i.e., homogeneity)

Cronbach’s

Can be used with nominal & ordinal data

• Alpha is the average of all split-half reliabilites for the items that make up the assessmen

item-to-total correlatons

Each item is correlated to the total test score

• Advantage over alpha: allow an assessment developer to identfy individual items that may be inconsistent with total score & contributng error to the assessment

interrater reliability

the ratings of more than one rater on a single assessment of a single subject are compared to estmate the ability of the assessment to be rated consistently across users

assessment validity

A measure derived from an assessment represents the underlying construct that the assessment is designed to measure

‒ Validate an interpretaton of the scores the assessment yields

face validity

an assessment has the appearance of measuring an underlying construct

‒ Weakest evidence of validity

content validity

the adequacy with which an assessment captures the domain or universe of the construct it aims to measure

‒ For example: self-care

criterion balidity

involves collectng objectve evidence about the validity of an assessment

− Refers to the ability of an assessment to produce results that concur with or predict a known criterion assessment or known variable (results closely match those of another existng assessment performed on the same criterion)

concurrent validity

an approach to establishing criterion validity that refers to evidence that the assessment under development or investgaton concurs or covaries with the result of another assessment that is known to measure the intended construct

• ‘gold standard’

predictive validity

an approach to establishing criterion validity that involves generating evidence that an assessment is a predictor of a future criterion

construct balidity

the capacity of an assessment to measure the intended underlying construct

‒ It is the ultmate objectve of all forms of empirically assessing validity

known groups method

identfying subjects who are demonstrated to difer on the characteristic the assessment aims to measure

discriminant analysis

evaluate the ability of the measure derived from the assessment to correctly classify the subjects into their known groups

convergence

two measures intended to capture the same underlying trait should be highly correlated

divergence

different traits show patyerns of association that discriminate between the traits

factor analytic method

an approach to demonstratng construct validity that examines a set of items that make up an assessment & determines whether there are one or more clusters of items

sensitivity

test’s ability to obtain a positve test when the target conditon is present (true positive rate)

• Detection of presence of conditon

specificity

test’s ability to obtain a negative test when the conditon is absent (true negative rate)

• No detection when conditon is not present

norm referencing

interpretation of a score based on its value relative to a standardized score

‒ Standardized according to statstcal norms established within groups of people (sample dependent)

criterion referencing

interpretaton of a score regarding what is considered adequate or acceptable performance based on a defned criterion

standardized scores

t-scores or z-scores

‒ The number of standard deviatons that a given value is above or below the mean of the distributon

‒ Typically 0-100 scale, mean = 50

standard error of measurement

a reliability measure of response stability, estimating the standard error in a set of repeated scores