Professional Documents
Culture Documents
I.THE "VARIABLE"
-when reduced to lowest terms, all medical research is simply the study of relationships
among variables
-most often, medical investigators are interested in studying either associations or
differences among variables
-a variable is any:
-quality
-characteristic
-constituent
of a person or thing
that can be measured
-by definition a variable is subkect to change
1
b.Examples
-patients status or condition may be classified as: unimproved, stable,
or improved
-note that, although it is known that a patient classified as
unimproved is more ill than a patient classified stable, it is not
known how much more ill the first patient is
-also, the difference in illness status from unimproved to stable is
not necessarily the same as the difference in status from stable to
improved
-cancer staging typically uses an ordinal scale to classify disease
according to the degree and nature of involvement of body tissues
-for example, the staging of Hodgkin's disease is based on the
degree of lymph node involvement, with stage I limited to a single
lymph node region or single extralymphatic site and stage IV
characterized by diffuse or disseminated involvement in
extralymphatic tissues.
3.Interval Scale
a.Definition an interval scale assigns each measurement to one of an unlimited
number of categories that are equally spaced
-the scale has NO true zero point (i.e., the zero point on the scale does not
represent the true or theoretical absence of the variable being measured)
-with an interval scale, it is possible to determine exactly how much more
(or how much less) of the variable being measured is represented by each
category
b.Examples
-temperature expressed in degrees Fahrenheit or Celsius is a variable
measured on an interval scale
-note that 00C is the point at which water freezes it does not
representabsence of temperature
4.Ratio Scale
a.Definition on a ratio scale, measurement begins at a true zero point and the
scale has equal intervals
-so ratio scale is equivalent to interval scale except the zero points
including the unlimited number of categories (both have unlimited number
of categories
b.Examples
-variables measured on a ratio scale are: length, time, mass, volume and
temperature in degrees Kelvin
B.Types of variables
-variables can be broadly classified as:
-qualitative
-quantitative
1.Qualitative variables
a.Definition are variables that are measured at a nominal level
b.Examples a diagnostic test for pregnancy gives a result of either "positive" or
2
"negative"
-the diagnostic test variable is a qualitative variable
2.Quantitative variables
a.Definition are variables that are measured on an ordinal or interval/ratio
scale
b.Examples
-a measurement of serum sodium concentration (e.g., 140 mEq/l)
expresses the exact amount of sodium in the serum
-serum cholesterol level, systolic blood pressure, and blood ureea nitrogen
(BUN) level are other quantitative clinical variables
3
3.Measurement variation
a.Definition measurement variation (sometimes called measurement error) is
variation among clinical observations that is attributed to the measurement
process
-measurement error may have both a:
-random component
-systematic component
-random measurement error
-is governed by the laws of chance and
-results in a measurement that is either above or below the
true value with equal probability
-a series of measurements affected only by random variation will
center on the true value of variable being measured
-systematic measurement error
-occurs when, as a result of a flaw in the measurement process, the
measurements no longer center around the true value but around a
value that is systematically higher or lower than the true value
b.Examples
-for a given sample of urine, a series of measurements of pH made with
the same meter by the same analyst under theoretically identical
conditions will not be exactly equal
-random variations in these measurements is the sum of many
components, including instrument precision
systematic variation results if the instrument is out of calibration
D.Withinpatient variation
1.Variation in the value of a clinical variable within a given patient may result from:
-biologic variations that result from:
-true biologic (random) variation
-systematic variation ("bias")
-varying conditions under which the measurement is made (systematic
variation)
-measurement error
-a pathologic change in the biologic state of the patient
2.The physician is concerned with identifying the sources of variation within a given
patient, because intervention should occur only if there is a true pathologic change
4
2.The physician is concerned with variation among patients, because most answers to
clinical questions for an individual patient are derived from information obtained from groups of
patients with similar conditions for example, a physician would study variation among patients
to answer questions such as:
-"how unusual is Mr.Smith's creatinine kinase (CK) value relative to a group of
persons known to be free of myocardial infarction?"
-"do groups of patients who receive antihypertensive agent A have lower diastolic
blood pressure measurements, on the average, than those who receive agent B?"
Types of variations:
-Biologic variation can be:
-"true" due to random changes
-"bias" due to sampling error (a systematic change or variation)
-Temporary systematic variations (changes) this can be ONLY systematic, unlike the
other 2 variations
-due to observations made under different conditions
-Measurement variations = error
-can be:
-random (true)
-systematic = bias
A.Bias
-here the discussion of "bias" is limited to cases in which multiple measurements are
obtained from one patient
-other types of bias (e.g., selection bias, confounding bias) arise when comparisons are
made among groups of patients (e.g., to determine the most effective treatment of a
disease)