Midterm Study Guide Flashcards
what is evidence based dentistry?
3-part harmony, all optimize treatment outcomes
what are the three parts to evidence based dentistry?
best evidence
clinical judgement
patient values
best evidence
evaluation of the best available scientific evidence
clinical judgement
recognition of your own clinical expertise
patient values
understanding patients needs and preferences
why would you not use EBD?
Not simply just reporting the findings of the study or considering just one
study. It seeks to evaluate all scientific evidence on a specific topic. This allows us to broaden our treatment options by using more current evidence.
The steps in developing an evidence based approach to practice is:
- Determining your clinical question (What do you want to know?)
the hierarchy of evidence is based on the ability to
control for bias and demonstrate cause and effect in humans
hierarchy of evidence from top to bottom (7)
meta analysis systematic review randomized controlled trial cohort studies case control studies case series/case report animal studies
types of questions patients ask (6)
prevalence etiology/risk diagnosis therapy prognosis phenomena
prevalence
what is the frequency?
etiology/risk
what causes the problem?
cohort study>case control>case series
diagnosis
does the person have it?
prospective, blind comparison to a gold standard (RCT) or cross-sectional
therapy
what is the best treatment?
randomized control study> cohort study
prognosis
who will get the problem? (group of linked individuals)
cohort study>case control>case series
phenomena
what are the concerns?
PICO
population/patient/problem
intervention
comparison (dont always have this)
outcome
PICO
population/patient/problem
intervention
comparison (dont always have this)
outcome
most common questions of PICO
diagnosis
treatment
etiology/harm
prognosis
doctor uses given info to search for best evidence to answer the question, critically appraising the evidence and applying results in
clinical practice
what key words do you use when searching in pubmed?
PICO keywords
what order do you write your questions?
PICO
in patients with…will…as compared to…result in…
when searching for evidence, what order do you use?
I, C, P, O
what are common barriers to rational decision making?
cognitive biases
research question
presents the idea that is to be examined in a study and is the foundation of the research study
hypothesis
testable prediction; attempts to answer the research question
PICO
clinical question that cannot be tested
searching pubmed
use PICO for ket words during search; indexed with MeSH terms and newer items not indexed for searching
clinical queries
filters out all research not related to clinically related questions
Cochrane library
systematic reviews and Cochrane reviews; they have their own libraries they review
ADA’s EBD site
keeps up with hot topics, great for once you are out of school
TRIP
pulls up reviews easily
CEBD
won’t need as much and is harder to use
limited of databases (5)
language age time period/date gender/sex publication types
publication types (5)
systematic reviews meta RCT practice guidelines reviews
MeSH terms
medical subject headings
puimde is indexed with MeSJ terms and newer items not
indexed for searching
midline is indexed with only
mesh terms for search
MeSH is a
national library of medicines controlled vocabulary thesaurus
terms naming descriptions in hierarchical structure that permits searching at various level
specificity
boolean operators
a connection word or symbol that allows a computer user to include or exclude items in a test search
OR
broadens search
AND
narrows search
NOT
excludes
boolean operators must be — in pubmed
capitalized
internal validity (3) questions
are the results valid for the patients in the study?
was the study performed correctly?
was any difference between groups seen?
threats to internal validity (3)
selection bias
maturation
instrumentation
— — to groups addresses many threats to internal validity, but not all
Random assignment
external validity (2) questions
are the results valid for patients not involved in the study?
does the study population represent the larger group?
external ability to generalize findings beyond (2)
beyond subjects in the study
beyond the environmental constrains of the current study and to other temporal periods
as controls — (increasing internal validity), the generalizability of finding may — (decreasing external validity)
increase
suffer
threats to external valid (2)
publications
financial
quantitative
inquiry rooted in empiricism
only those phenomena which can be measured are “real”
measures are often numeric scales
qualitative
inquiry based in hermeneutics
the interpretation of contextual meaning
measures are subject and dependent upon perceptual bias (subjective data)
meta analysis
subset of systematic reviews; method for combining
qualitative and quantitative data from several selected studies to develop
a single conclusion that has greater statistical power
systematic reviews
provides a comprehensive review of all relevant
studies on a particular clinical topic
randomized control trials
participants are randomly assigned to an
experimental group or a control group. The only expected difference between the control and the experimental group is the outcome variable being studiedexperimental
cohort study
one or more samples (cohorts) are followed prospectively and subsequent status evaluations are conducted to determine which initial participants risk factors are associated with a disease or outcome. Outcome from participants in each cohort is measured and relationships with specific characteristics are determinedobservational
case study
compares patients who have a disease or outcome of interest (cases) with patients who do not (controls), and look back retrospectively
to determine the relationship between risk factor and disease observational
case study limits
RR calculation because cases are selected on basis of disease rather than exposure
cross sectional study
analysis of data collected from a population or representative subset at one specific point in time. Used to describe some feature of a population (i.e., prevalence of an illness)descriptive
case reports
an article that describes and interprets an individual case, often written in the form of a details story. Describes unique cases that cannot be explained, variations of a disease, unexpected events, etc. Considered the lowest level of evidencedescriptive
which studies are observational studies (2)
cohort studies
case control studies
observational studies
Researcher does not test an intervention such as a drug, but instead looks for relationships between exposure and outcomes (exposure = outcomes), such as the relationship between smoking and lung cancer.
strongest evidence for demonstrating cause and effect
randomized control trial
RCT reduces the effect of
bias due to intervening variables
assumes that confounding conditions will be equally distributed among groups
RCT includes (4)
at least one “varied condition” (treatment vs no treatment)
concurrent enrollment
random assignment to groups
follow up
RCT using blinding attempt to
reduce bias due to the expectations or preconceptions of patients or investigators (ex. double blind)
experimental research (3)
Quantitative
Investigative cause
Researcher controls or manipulates variables under investigation
observational research (4)
- May be quantitative or qualitative
- Without experimental controls (may include comparisons to natural
groups) - Sometimes called “quasi” experiments
- Designs provide for the investigation of relationships, but not cause
examples of variable (4)
unknown or known factors relevant to a study
age
ethnicity
socioeconomic status
disease history
independent factor, aka
causative factor
independent variable
a factor or condition that changes naturally or is intentionally manipulated by the investigator to observe an effect
independent variable is known and
controlled by the experimenter
dependent variable aka
response or outcome
dependent variable
observed variable in an experiment in which changes are determined by the presence or degree of one or more independent variables
dependent variable is a factor directly
affected by another
confounding variables aka
error or confounding influences
confounding variables
An extraneous variable that correlates significantly with both the
independent and dependent variable
A factor not considered or recognized by the experimenter that has
significant impact on dependent variable or outcome of interest
nominal/categorical data
Label or category without rank or order (mutually exclusive)
examples of nominal data (2)
male/female, dead/alive, pass/fail
ordinal data
Label or category with meaningful order or sequence
how to measure ordinal data
Not measured - without definite boundaries/levels
example of ordinal data
severity of disease
likert scale
strongly agree, agree, disagree, strongly disagree
ordinal data use
mann/whitney
ordinal data tests differences in
rank order
interval
continuous; scaled measure with arbitrary zero (temperature)
interval difference between — is meaningful
levels
tests for interval data
T-test or ANOVA
interval tests differences in
means
ratio
scaled measure with an absolute/true zero (test score)
odds ratio (OR)
Comparing the odds of an event in one group to the odds of an event in a
comparison group
what does odds ratio compare
subgroup and remained of population
odds ratio is an estimate of
association
odds of female disease with exposure
offs of male disease with exposure
relative risk/risk ration (RR)
Measure of risk based on a comparison of disease incidence in two
distinct groups
relative risk/risk ratio (RR) compares subgroup to
entire population
RR is the ratio of the
probability (percentage) of event occurring (or not occurring) in exposed vs non exposed group
when exposure is negative
incidence rate of people exposed to risk factor
incidence rate of people not exposed to risk factor
when exposure is positive
incidence rate of person not exposed to risk factor
incidence rate of people exposed to risk factor
Both odds ratio and relative risk compare the
likelihood of an event occurring between 2 distinct groups. RR is easier to interpret and consistent with the general intuition (comparison between subgroup and entire population rather than subgroup and remainder of population)
Experimental event rate (EER)
event rate in the treated/affected group
Control event rate (CER)
event rate in the control/unaffected group
Absolute Risk Reduction (ARR)
compares treatment effectiveness (CER-EER)
- How much does the treatment reduce the risk
Attributable Risk
opposite of ARR (EER-CER)
Relative Risk Reduction (RRR)
percentage of original risk removed (ARR/CER)
- How much is risk reduced in comparison to the baseline
- Not a good number/test (can inflate numbers or findings)
- “more or less likely to happen”
Number Needed to Treat (NNT)
number of patients needed to prevent one additional bad outcome
- 1/ARR – inverse of absolute risk reduction
Sensitivity
# of people who have the disease and test positive /# of people who have the disease (SNOUT)
o Highly sensitive tests catch the disease every time
- Sometimes they are wrong (false +)
- Good at detecting/screening
Specificity
of people who do NOT have the disease and test negative / # of people who do NOT have the disease (SPIN)
o Highly specific tests are rarely wrong
- Sometimes they miss the diagnosis (false – )
- Good at being right/confirming the diagnosis
Positive Predictive Value
chance that when the test is positive you actually have the disease
Negative Predictive Value
chance that when the test is negative you actually do NOT have the disease
Hypothesis
assumed proposition
Null Hypothesis
prediction that the observed difference is due to chance alone and not due to a systematic cause (“no difference”)
- Statistics provide the evidence to reject or fail to reject the Null
Type I error
false positive (alpha)
o No difference between groups when the study shows a difference
Type II error
false negative (beta)
o Difference between groups when the study shows no difference
confidence intervals
Most likely range within which the true size of effect lies
- Confidence that if anyone reproduced this study they would have the
same results due to independent variable(s) - Three things impact the width of a confidence interval
Confidence level
typically 95%
Variability
standard deviation
Sample size
smaller sample sizes generate wider intervals
P-value (“alpha level”)
- Tests the likelihood of differences occurring by chance alone
- Predetermined probability that the researcher is willing to make a type I
error
p < 0.05 =
5% probability that observed difference was due to chance
Chi Square
Nominal data
Mann-Whitney
Ordinal data
T-test
Interval/ratio with 2 independent groups
ANOVA
Interval/ratio with 3+ independent groups
Statistical
used in hypothesis testing
Clinical
practical importance of a treatment effect – whether it has a real,
palpable, noticeable effect on daily life