Terms in this set (293)
What components are included in basic science
Epidemiology and Biostastics
What components are considered applications of science
preventative medicine and public health
Why is it important to study populations?
-to prevent disease
-to learn about what matters for health
-to study interventions we can apply to individuals
Who was John Snow
Anesthesiologist who worked for the queen Victoria and aided her delivery of children. He worked in London where there was a prevalent outbreak of cholera that allowed him to perform the first epidemiological study that lead to the ability to prevent a disease.
How did cholera toxin affect people in London?
There was no centralized sewage system leading to the use of the river for dumping of sewage. This river was also the source of drinking water that was supplied by public water pumps. The Broadstreet pump was a highly desired public water pump because it was cool and slightly carbonated.
Pioneer in understanding the power of statistics in medicine. Before him, there was no one collecting data and storing it in a centralized location. Farr helped Snow by collecting data from death certificates about address and cause of death.
Where did Farr keep his data collected from death certificates?
General Register Office
Theory of Miasma
idea that there was an invisible gas that would seep around that air. Once inhaled, this gas would cause disease and death
Farr's conclusions from death certificate data
Farr concluded that death rates from cholera were higher in areas over lower elevation which failed to reject the miasma theory because the gas would accumulate in lower elevation areas.
How did Snow use Farr's data
Snow went out into the community and began asking questions to the families of deceased to figure out a commonality.
Shoe Leather Epidemiology
practice of working in the field to conduct research. Snow practiced this type of research when he asked around the community about the deaths.
What was Snow's Ghost Map
A map of the town with the cholera outbreak that had the deceased houses highlighted. This helped him visualize the cluster of deaths and seek out common causes
Result of using Shoe Leather Epidemiology for Snow
was successful in determining that the water pump was a likely link between all of the victims. All victims got their water there and then would suffer from cholera. Snow advised the city to remove the handle of the pump to prevent the spread of cholera
was hired to chief engineer of the new sewage system in London to get rid of horrible smell.
Why was Snow's conclusion unique to prevention?
He didn't know the underlying mechanisms of the cholera toxin but could still prevent the disease from spreading.
Mother of Nursing who was visiting hospitals to determine what was killing soldiers during the Crimean war in Turkey.
What was the importance of Nightingale's findings?
She found that more people were dying from preventable diseases than warfare. She was the first person to look at complex data and present it in a pie chart comparing frequency and type of soldier death
Was the chief resident at an OB hospital. He discovered that the death toll was 2X higher at his facility than another clinic that used midwives. He collected data about sepsis and why it was occurring in his clinic more often
infection of the blood
Importance of Semmelweis's findings
After a colleague died from a scalpel stick, he figured out that disease was being transmitted through bad hygeine practices. He observed students working with cadavers and then delivering a baby. He began to use bleach before deliveries because he believed the smell from cadavers could be causing the disease. After implementation of the bleach hand wash, the death toll in his clinic significantly decreased
studying health in populations
study of the distribution and determinants of health-related states or events in specified populations, and the application of this study to control or human health problems
what are the two types of epidemilogy
classical - focused on populations, want to identify risks
clinical - focused on patients, seeks to guide clinical decision making
application of statistics to human data, very important for epidemiology, confined to analysis of data
Model of how disease starts
time between exposure and disease inititation
individual already has the disease but does not show clinical signs or symptoms
levels of prevention
primary - stop disease from happening in the first place (vaccination, quit smoking)
secondary - identify disease early to improve outcome (screening tests, colonoscopy)
tertiary - do nothing to alter full manifestation of disease or injury but attempt to reduce long-term effects (cardiac rehab)
quaternary - mitigate or avoid the consequences of unnecessary or excessive intervention of the health system (overmedication)
what level of prevention can occur after the patient knows he has been infected?
tertiary or quaternary
during what period in the model of disease does the secondary prevention occur
during what period in the model of disease does the tertiary prevention occur
after disease manifestation
the epidemiologic triad
originated for infectious disease but can be applied to a lot of things.
interaction between a host that occurs in an enviornment
epidemiologic triad for malaria
host - human
vector - mosquito
agent - blood smear showing agent causing malaria
environment - rice field with standing water
what is the most preventative measure for malaria
bed mosquito netting that kills mosquitos upon contact
for any given disesase, there is a much larger problem with carriers and asymptomatic individuals than we realize
definition of public health
the science and art of preventing disease, prolonging life, and promoting health through the organized efforts and informed choices of society, organizations, public and private communities and individuals
Winslow (intervene at a social level)
medical literature, analysis, and retrieval system online
national library of medicine's online data base (NLM)
web-based retrieval system to search medline
produced by elsevier
citation back to 1974
European based (not government funded)
targeted more towards consumers
can be websites and usually not journals
What is in pubmed
citations and abstracts of biomedical journal articles
Who selects journals for medline
the literature selection technical review committee
does pubmed require quality control by the NLM
what would a citation look like on pubmed if it was out of the scope for medline
The Cochrane Library
product of Cochrane Collaboration (volunteers)
controlled trials register (CENTRAL)
database of systematic review (CDSR)
searchable database the includes medline, embase, and others
almost one million trials
like to have prospective trial registration
what is the benefit of having prospective trial registration
eliminate publication bias
reviews of primary research in health care and health policy
any one can review
trained in systematic method to analyze literature
many different review groups
pros of cochrane library
saves you work (standard for what works and what doesn't work)
first search location
cons of Cochrane library
limited number of reviews
what three reasons why a certain result might arise are included in the mantra?
Medical paradigm on "the cause"
most people think science occurs through breakthroughs but in reality it occurs through slight forward advancements that are slow
what did Newton say about scientific proof?
in science you do not prove something true but you fail to prove something false
Who believed that his idea's proved Newton's false?
Black Swan phenomenon
in Europe people only saw white swans so they believed that all swans were white. One European traveled to Australia and saw a black swan. This sighting threw off what so many people believed to be true
how does science work?
many small experiments
articulate a specific and falsifiable hypothesis
what is "association"
relationship in which a change in frequency of one variable is associated with a change in frequency of another
just notice the change
ex: stork population and birthrate
look at populations and make false assumptions about individuals in that data
what is "causation"
relationship in which on condition precedes and must be present in order for another outcome to occur
ex: smoking causes cancer
association vs causation
association may be causal, determined by statistics and epidemiology
any association may be defined by chance, bias, or confounding
determining causality is a qualitative process
Sir Bradford Hill on statistics
X2 is an excellent servant and a bad master
can be mislead by statistical tests
how much disease we saw with treatment divided by disease without treatment
does not tell us about chance, bias, or confounding
what is "chance"
association was a fluke due to random variability
statistics helpful in determining the roll of chance using p-values or confidence intervals
tells statistical significance (where 95% certainty the number lies between)
confidence interval and relative risk
if no difference between treatment and no treatment, the RR would be 1
if the confidence interval does not include 1 then there is not a likelihood of chance
if wide range, there could be measurement error or small sample size
post hoc application of science
apply statistics to an event that already happened
meaningless because it did happen
going out with a big, wide net and get some meaningful data only because of the size of net you used
no meaningful predictions before experiment
ex: different disorders among male barbers by race
could be hypothesis generating
an unusual number, real or perceived, of health events grouped together in time and location
almost always due to random variability
nonrandom appearance of random variability
how do you investigate a disease cluster?
confirm that it is a cluster
study fresh population
(clusters can be used to generate hypothesis)
what is bias
a systematic error involving one group in the study
must involve one group preferentially and be systematic
causes you to see things that are not real
what is a "confounder"
third variable that is independently associated with both the exposure and the outcome
ex: increased rate of pancreatic cancer among coffee drinks. Coffee drinkers are more likely to smoke
must be a parallel pathway to the same result
must be linked to both exposure and disease and not an intermediate in the cause-effect relationship
a third variable that is an intermediate in the cause-effect chain and influences the frequency of disease
ex: increase in EBV in areas with malaria
a form of effective modification where the two variables work together to cause the disease
ex: asbestos and smoking leading to lung cancer
how can you tell a confounder from an effect modifier?
when we stratify for different levels of a confounder the risk will go to 1 (no relationship)
when we stratify for different levels of an effect modifier the association change persists
go from specific observation to generate a general principle
smoking causes lung cancer
what epidemiologists do
apply general principles to specific situation
what clinicians do
tell people to quit smoking because see a general trend of cancer (might not happen to that individual)
Rothman on causes of diease
infinite number of component causes
these can interact in complex ways
there can be an infinite number of causal relationships that act in complex ways
one cause may be in someone's chain but not in another
first person to provide causal criteria for human disease
thought that microorganisms were always the cause and they must be present in all cases, inoculations of pure cultures must produce in animals, and cultures must be obtained from inoculated animal
John Stuart Mill
not specific to human disease but related to general causal inference
had idea of cause
chief viewpoints on the cause of disease
clinicians level of proof
anything is possible
tend to focus on this
definition of health
state of complete physical, mental, and social well-being and not merely the absence of disease
encompasses all aspects of well being
what is usually the numerator in health data? denominator?
which one is harder to get/
number of deaths or cases
number of people at risk
denominator is harder because we don't really know how many people are at risk (people who could die)
what is the source of denominator data?
generally census (10 years)
some countries don't have resources
birth and death data (done continuously)
what are examples of vital statistics?
information taken continuously
birth, death, marriage
how are birth and death certificates filled out?
what is the national notifiable disease surveillance system?
system in which physicians need to inform government about disease incidents of certain disorders.
HIV, TB, cholera
what is the national center for health statistics? What data collection do they use?
through the cdc and is a variety of databases that are important for vital statistics
uses National health interview survey.
provides demographic and socioeconomic information about specific health problems through household interviews
how sexual orientation influences health care
what is the behavioral risk factor surveillance system? (BRFSS)
phone interviews assessing risk factors for disease (risk-behaviors)
looks at things like exercise, smoking, obesity, alcohol
what is the surveillance, epidemiology, and end result program? (SEER)
information system about cancer
most US states feed into national database that looks at specific cancer
avg age of onset, life expectancy gender
what is the national vital statistics survey (NVSS)?
vital statistics that are collected locally but complied nationally (births, deaths, marriage, divorce)
what is the national health care surveys?
provide surveys of providers and health care settings to answer questions about health care policy and resources
use of medical records, disparities in health care, patterns of care
how do you record mortality?
physician must pronounce the death in a standardized manner
document time of death, cause of death based on state regulations
what is the immediate cause of death
final disease, injury, or complication resulting in death
usually follows a chain of events
what is the underlying cause of death
disease or injury that initiated the chain of events leading directly to death
precedes immediate cause
what is a contributory cause of death
condition which played a role in death but NOT part of the underlying causes
not within the causal chain
what is the mechanism of death
how the death occurred (not the disease)
For public health what is considered the cause of death?
the one disease of injury that started the whole chain of events. The lowest underlying cause
how are deaths coded now?
international classification of disease (ICD-10)
owned by WHO and has all scenarios classified into numbers
what is the DSM
diagnostic manual of mental disorders from american psychiatry association.
what is the controversy of DSM-10
-lack of transparency
-not evidence based
-poor reliability because different people can reach different diagnosis
-medicalization of nonmedical problems (Dr. Martin's daughter as example)
what is incidence
the number of new cases that occurred in a population over a set time
what is prevalence
how many people have the disease in a population right now
why is it important to measure death?
mortality is an obvious outcome and is severe but it does not capture the whole outcome.
what is the crude mortality rate?
number of people who die in a given year divided by number of people in a population in the middle of that year
usually per 100,000 people
can you directly compare crude mortality rates between countries?
no because there is a difference in age structure. The magnitude of difference is underestimated because there are more older people in more advanced countries
what is an age specific mortality rate?
rate calculated for a specific age group to compare mortality across different ages at same time, or same age over time
death of people in age group divided by midyear population of age group
why is the value of under 5 mortality important?
measures the crude mortality rate of age group under 5 years old
direct measurement of health infrastructure because infant health fails the quickest if infrastructure is bad
what is cause-specific mortality rate?
mortality rate for a specific disease
deaths from disease divided by total midyear population (x100,000)
what is the case-fatality rate?
calculates the number of deaths due to a disease divided by the number of total cases of the disease (x100)
gives us an idea of the severity and risk
case-fatality rate for reproductive and perinatal case-fatality
always have to assume that the woman in the reproductive age group is pregnant because you don't want to count them out.
use live births for the denominator because it is a product of conception showing a sign of life
40-50% of pregnancies end in spontaneous abortion (sometimes not even known)
what the time span known as "fetal"
(unborn) 0-28 weeks
what is the time span known as "stillborn"
(unborn) 28-40 weeks
what is the time span known as "perinatal"
28 weeks prenatal to 1 week postnatal
what is the time span known as "neonatal"
birth - 4 weeks
what is the time span known as "infant"
birth - 1 year
what is the infant mortality rate?
number of infant deaths under a certain age divided by the total number of live births (x k (usually 1,000))
what is the maternal mortality rate
number of women who die as a result of complications of pregnancy or childbearing in given year
number of maternal deaths divided by the number of total live births (x100,000)
what is the proportionate mortality rate?
deaths from a specific cause expressed as a percent
what is an advantage of using proportionate mortality rate?
two numerators are used in this equation
can calculate without population data (do need death data)
why will the causes of mortality change over time?
when you prevent one cause of death people will begin to die from another cause
what is something you need to do when comparing mortality?
standardize to prevent confounding variables
need to pick a standard and make the other population mirror it
what is simpson's paradox?
before correcting for a confounder you will see one results and after correcting you will see a different one
what is indirect standardization?
picking the other population to be standard (other than original) and calculating a new death based off of death rate
indirect standardization and standard mortality rate?
use standardized death number for expected and use observed death number
if over 100 there is a higher mortality rate in the observed town
made randomization in clinical trials a standard practice
also discovered a method of survival analysis known as Kaplan-Meier curve
what is survival analysis
technique in medical statistics that takes a group of people and follows them over a period of time
can perform on specific population with a disease using start point
what happens during a study if there is no information on a subject?
subject is "censored" and data will stop being collected on them
Kaplan-Meier curve will preserve data from people who are lost in a study
what is a "step-down" in a kaplan-meier curve?
death in the study
what is a hashmark on a kaplan-meier curve? (pointed up)
what are two tests used to compare kaplan-meier curves?
log rank (mortality constant)
wilcoxon (mortality changes)
what is a proportional hazards model?
regression analysis performed on survival curves
also called Cox
what is point prevalence
number of cases at a specific point in time
cross-sectional of that specific time, don't know when developed or how long
don't know temporality
when is period prevalence?
number of cases that occurs in a specified time frame
sum of the prevalence point at beginning of time frame and new incidence
what is a person year?
could be a person followed for a number of years, or multiple people followed for a number of years
what is cumulative incidence
total number of new cases over a specified period of time
useful when there is a known start time of the first case (outbreak)
what is incidence density
accounts for variable follow up between subjects because each person contributes person-years for the length of time they were followed
individual person on chart and how many person years they contribute
how is prevalence a steady state measurement?
you are gaining people due to new incidents but also losing people due to death or cures
equation for prevalence
incidence x (average)Duration
this is why prevalence goes up with increased treatments, the individual still has the disease but symptoms are managed for a longer duration
what could be the cause of a large increase in cases at a certain time?
a new definition of the disorder
there would be a large backlog of patients that have the disease but are now considered affected
what are factors that can increase both prevalence and incidence?
greater number of cases
what are factors that can increase prevalence but not incidence
improved long term treatment
out migration of healthy people
in migration of affected people
how do you determine which study design to use?
level of knowledge about a subject
which studies are known as hypothesis generating?
qualitative and descriptive/ecological quantitative
case reports, correlational studies, cross-sectional
what studies can be hypothesis generating and hypothesis testing?
analytical and interventional/experimental quantitative studies
observational (case control, cohort), clinical trial
what is a frequent sequence of study in the human population and what has the most strength?
randomized trials (most strength)
characteristics of qualitative studies
usually conducted to generate a hypothesis
flexible, open ended
very little quantitative data
want to understand peoples feelings or thoughts (many possible answers)
how do you sample for qualitative studies?
purposeful (seek out subjects because of the purpose and aim)
snowball (each participant is asked to recruit additional)
what methods are used in qualitative studies?
interviews, observation, examination of existing records, not numeric data
methods can change while study is going on
validity of qualitative studies
addressed at the end of the study
make sure you get everyone you can (saturation)
look at different sources of information (triangulation)
present results to responders and get feedback
have different people look at recording and see if similar results
characteristics of quantitative studies
validity addressed at outset by design
involve numerical data
methods pre-specified and rigid
what is descriptive epidemiology
described general characteristics of disease
comparisons to exposure often implicit or indirect (cancer compared in US or Japan)
what is analytical epidemiology
comparison of disease to exposure is direct and explicit
case control, cohort, and interventional studies (compare US to Japan specifically looking at stomach cancer and diet)
what are descriptive studies?
usually unplanned or natural experiments (noticed in nature)
often provide first evidence of association
results may be explored further by analytical methods
what are three types of descriptive studies?
case reports/case series
what is the difference between case-reports and case series?
case reports are on one case of a disease and case series are more than one case
what are some advantages of case reports/series?
easy kind of study to publish
usually first info on a disease
may be a rare condition that is hard to study otherwise
not a lot of money or follow up needed
what are some disadvantages of case reports/series?
only circumstantial evidence
cannot determine causal relationship
what is a correlational study?
compare disease frequency in relation to another factor at a population level
often called ecological study (birds eye view of data)
why is it hard to think about causal relationships with correlational studies?
different causes are not mutually exclusive (genetics and environment)
how would you distinguish if something is environmental or genetic base in a correlational study between countries?
look at immigrants (because you are maintaining the genetic integrity but changing environment)
what is a residual environmental effect?
after leaving a population, an immigrant is not fully assimilated into the new culture right away and may maintain cultural habits from earlier environment
what are some advantages and disadvantages to correlational studies?
adv: easy to do and reasonably easy to get data
dis: lots of confounding data, cannot draw inferences on individuals, individual trends can be lost in population data
what is a cross-sectional study?
measure number of cases of disease among individuals at one point in time in a defined population (sometimes called prevalence study)
usually use interviews and questionnaires
what is different from a cross-sectional than a correlational study?
cross-sectional give you data on individuals
what is the NHANES
surveys to provide current statistical data on the amount, distribution, and effects of illness on disability in US
example of cross-sectional data
what is a challenge of cross-sectional data?
cohort effect: groups may have had different experiences before being studies
could lead to wrong conclusion
what is an advantage of cross-sectional study?
want to know what is occurring in population right now
what are disadvantages of cross-sectional studies
temporality is not satisfied (cannot determine causation)
what is a cohort study?
study of individuals who are outcome free and follow them in time to monitor exposures and determine disease status
can either be prospective or historical
what is a two by two table?
table used to set up a cohort study
rows = exposed/unexposed
how do you find the relative risk or risk ratio of cohort study?
incidence of disease in exposed group divided by the incidence of disease in the unexposed group
helps to determine how much the exposure is related to the disease
what is a prospective cohort study?
look at subjects that do not have the disease at the start of study. Will follow these subjects into the future to see who gets the disease
subjects in two groups: exposure/no exposure and observe outcomes
what is the Framingham study?
large sample of residents from Framingham, Mass that had a biennial cardiovascular exam and looked at hospital data and other sources.
none of the individuals in the beginning had heart disease but this study looked at what exposures give you more risk
prospective cohort study
still getting data from these
what is a historical (retrospective) cohort study?
want to look back on people that already have the disease.
when you start the study people already have disease and want information on exposure status before development of disease
have to go back in time before they had the disease and follow them to present
what is key about the sources of cohort?
need to have one exposed and one unexposed (or less exposed) and want everything else between the populations to be relatively similar
what are some advantages of a cohort study?
can control and standardize data collection
can check outcomes as they occur
estimate of risk from the study is a true risk
can study risks that were not anticipated
efficient for rare exposures
what are some disadvantages of cohort studies?
cannot add additional exposures once began
attrition (wearing down) can be common in these
results take awhile
not good for rare diseases or outcomes
what is a case-control (retrospective) study?
subjects selected based on if they have disease or not and study will look back in time to see differences in exposure
when do you start a case-control study?
present time (disease vs no disease) and look back at exposures
what is the key difference between a historical cohort study and a case-control study?
in case-control you separate based on disease and look at exposure
in historical cohort you separate based on exposure in history and follow forward to look at disease
how do you choose a case for a case-control study?
have to define the criteria of being considered a "disease" and what should be excluded
must resemble a case in everyway besides developing the disease
what are some sources of cases for case-control studies?
people treated at hospital
people from general population
how do you choose a control for a case-control study
should be representative of the population that produced the cases
important for minimizing bias
what is "matching"
selecting the controls so that they are similar to the cases in certain characteristics
allows to control for things that could confound the study
what are two ways to match?
individual match: before and after disease
match pair: pair two different individuals (match for gender)
what is "overmatch"
occurs when cases and controls are matched on a characteristic that could potentially be causative for the exposure
can end up stratifying on exposure when unentended
how is the two-by-two table different for a case-control study than a cohort study?
more focus on the columns because these are the values we assigned
cannot calculate incidence in this table only exposure
what is "odds ratio"
ratio that someone will have exposure to someone will not have exposure
compare with disease to without
ad/bc (from table)
what are some advantages of a case-control study?
good for rare disease (already have cases)
what are some disadvantages to a case-control study?
prone to bias (recall, selection)
difficult to ensure matching parameters when measure disease
cannot obtain incidence or prevalence data
what is a nested case-control study?
start with a cohort study and look for individuals that developed disease or did not develop
select those who developed disease for cases and subgroup of those who didn't for control
go back and look at different exposure levels
what is the equation for attributable risk?
AR= Risk(exposed) - Risk(unexposed)
what can attributable risk tell you?
the burden of a disease and how much exposure affects your liklihood of developing a disease
what is the equation for attributable risk as percentage?
[Risk(exposed) - Risk(unexposed)] / Risk(exposed)
tells the total risk of disease attributed to a certain exposure
what is population attributable risk?
Risk(total) - Risk(exposed)
uses general population risk to tell how common a disease is
divide by Risk(total) and multiply by 100 to get a percentage
what is a clinical trial?
a planned experiment of an intervention
looks at efficacy
how is a clinical trial different than a cohort study?
in a clinical trial you are manipulating the behavior of the subjects
assigning clinical groups?
best assigned based on randomization
what is an equivalence trial?
testing to see if a treatment is "not too different" in characteristics from what treatment is previously available.
no better or worse
what is a non-inferiority trial?
testing to see if treatment is not worse than another
make sure treatment is equal or better
what is a superiority trial?
see if treatment is better than a placebo or what is previously available
How is an intervention study set up?
need homogenous group without disease
need explicit exclusion criteria because this can help eliminate bias
what are some considerations that should be taken into account when setting up an intervention study?
justify why you chose a sample size
large numbers required
expensive and follow uo
some may be hard to blind (surgery)
why might an intervention study stop early?
the treatment has a clear benefit (or harm) to the patient
another treatment comes along that is proving to be beneficial
continues to fail over long term follow up
what are the phases of a drug trial?
Phase I-safety and pharmacology
Phase II-pilot efficacy
Phase III-extensive clinical trial
Phase IV-long term effects
what is pre-clinical testing
first phase in a clinical trial
have to file an investigational new drug application with the FDA
helps to identify the NOAEL
what is the NOAEL
safe dosing range
no observable adverse effect level
hallmark of pre-clinical testing
looks for the highest dose without adverse effects
what is Phase I of clinical trial?
safety and pharmacology
introduced to a small number of healthy human volunteers
what is the goal of phase I?
determine safety and mode of action looking at pharmokinetics and side effects
look at different dosing levels (lower than preclinical trials)
what is the end point of phase I?
what is phase II?
larger sample of randomized volunteers that lasts about 2 years
intervention study design occurs
demonstrate safety and EFFICACY
WHERE TRIAL USUALLY FAILS
what is phase III?
expand population to 1000 for a longer time period of 3 years so get a more complete assessment of safety with longer use
much more extensive than phase II
what do you have to file during Phase III?
new drug application with the FDA
usually takes 2.5 years and contains all pre-clinical, manufacturing, and clinical data
what is phase IV?
occurs after new drug registration
much larger group over a much longer time period to look at long term effects
this phase is more reflective of routine use
some drugs are actually removed at this point
what is post marketing surveillance?
surveillance based on reports from health care providers
could be used to find weird side effects
what is a parallel trial design?
what is a cross over trial design?
useful to look at two different treatments at the same time
what are some pros of a cross-over design?
compare both treatments within a group
can get rid of individual differences between group
one treatment group can serve as its own control
what are some cons of a cross-over design?
effects of first treatment can have some impact on the response of treatment B
what is a factorial design?
can look at two experimental treatments at the same time
can help to look at drug interactions
what is a pro of a factorial?
allows you to assess for interactions between two different experimental groups
Who invented blinding?
what was the first blinding study?
Frank Mesmer believed that he could control a fluid in human bodies and cure all disease. Benjamin Franklin wanted to see if this was real so he had the physician do animal magnetism on trees and blindfolded children. The children could not pick the correct trees
what are the different levels of blinding?
none: open label
single: participants blinded
double: participants and investigator blinded
triple: participants, investigators, and data analysts are blinded
what is required for a blinded study?
blinding is the best way to reduce
what is allocation concealment?
concealment of an experimental group assignment to participants and investigators delivering the intervention in an RCT
limits selection bias
usually requires centralized service
done before the individuals are grouped into their levels
what is a placebo
an intervention designed to simulate medical therapy but not believed by the clinician or investigator to be a specific therapy for the target condition
who is John Haygarth and what is the Perknis Patent Tractor?
There was a belief that these certain rods could heal anything. John used a fake/dummy rod made of wood to see if people would get better and people actually did
this proved that there is a psychological component
what is the nocebo effect?
people's pain will get worse if they believe there is a stimulus
people with headaches got told they were getting an undetectable electric shock and 70% reported headache after they believed they had gotten the shock (never got the shock)
what does pain appear to be a part of?
situational cues, anxiety, and/ow expectation of pain
what are methods to deal with non-compliance?
run-in or wash-out period
what is a run-in or wash-out period?
participants receive treatment or placebo before randomization and those who do not comply are removed from the study.
what is a problem with the run-in or wash-out period?
cannot be applied for a acute treatments
influences the generalizability of the findings
what are some ways to monitor compliance?
pick high risk patients
testing for presence of drug
what is an intention to treat analysis?
all patients in each group are followed up and analyzed according to what they were assigned even if they did not complete or comply with the assigned therapy
once in a group, always in a group
why is it important to use an intention to treat analysis?
guard against conscious or unconscious attempts to influence results of the study by excluding odd outcomes
preserves baseline comparability between treatment groups achieved by randomization
reflects the way treatment will perform in population by ignoring adherence when data are analyzed
what are the three E's of an intervention?
what is efficacy?
does is work under ideal conditions of a RCT (Phase I-III)
what is effectiveness?
does it work in the real work (Phase IV)
what is efficiency?
cost versus benefit.
what is NNT?
number of people you'd need to treat for the prescribed period of time in order to prevent one episode of disease
what is NNH?
the number needed to harm for side effects for a period of time in order to prevent one episode of disease
what is ARR?
absolute risk reduction
Risk(w/o treatment) - Risk(w treatment)
take the reciprocal
what is relative risk reduction?
risk(w/o treatment) - Risk(w treatment) divided by Risk(w/o treatment)
what is privacy?
the expectation that a participants "data" will be safe from unwanted intrusion
i can protect data from someone else accessing it (password)
what is confidentiality?
the expectation that a participant enrolled in the study cannot be identified
make sure people cannot be identified (remove identifiers)
what is accuracy?
how close the measurement is to the true value
declines as measured values approach the detection limit
what is reliability?
consistency on repeat measurements
what is precision?
relates to the limit of detection of the test
lowest value that can be recorded
what is validity?
whether the test measures what it purports to measure
doesn't measure what supposed to measure
what are some reasons for ordering a test?
what is the difference between screening and surveillance
screening looks at an individuals and is a form of secondary prevention
what is sensitivity?
the probability that someone has a positive test given that they have the disease (probability of a true positive)
ability of a test to decrease amount of false negatives
what is specificity?
probability of having a true negative result in someone that does not have the disease
ability of a test to decrease amount of false positives
what is the positive predictive value
probability of having a disease given a positive test result
value does change
what is a negative predictive value?
probability of not having disease given a negative test result
value does change
how do you calculate accuracy?
(true positive+true negative)/everyone
how do you calculate prevalence?
people with disease/everyone
what is the problem with most continuous tests?
there is a gray zone where it does not necessarily fall in spectrum
what happens if the specificity is 0%?
say that everyone has the disease and you cannot miss anyone
sensitivity will be 100%
what do the predictive value of a test depend on?
prevalence or pre-test likelihood of disease
Describe a test with high sensitivity?
high false positive rate and low specificity
will not miss people that have the disease but will include those that don't.
better for a lethal disease
Describe a test with high specificity?
high false negative value and no false positives
everyone with positive result WILL have disease
how do you decide the cut off for a diagnostic test?
depends on the disease you are screening for
if disease is irreversible and lethal, you would want to have no false negatives. Everyone that has the disease should test positive
what measurement of a diagnostic test is most important for screening someone?
what measure of a diagnostic is most important for confirming a disease?
Who was Avicenna
first person to look at causal relationship that WAS NOT SPECIFIC TO HUMAN DISEASE
What is likelihood ratio?
ratio between the probability of a defined test result given the presence of disease and the probability of the same test result in the absence of disease
test result with disease/test result without disease
What is a likelihood positive ratio?
ratio of sensitivity of a test to the false positive error probability
(a/(a+c)) / (b/(b+d))
(a/b) / [(a+c) / (b+d)]
odds of disease among those who test positive divided by the odds of disease in entire population
P(T+ I D+) / P(T+ I D-)
not influenced by the prevalence of disease
what value is the ideal LP+?
>1 because that says that you have more TP than FP
what is Likelihood ratio negative?
ratio of false negative probability to the specificity
probability of FN/TN
[c/(a+c) / d/(d+b)]
(c/d) / [(a+c)/(b+d)]
odds of disease in those that the test is negative/odds of disease in everyone that you test
what is the ideal LR(-) value?
<1 because you want to have more TN than FN so you want the denominator to be larger than the numerator
odds vs probability
Odds: ratio of event/event not happening
Probability: ratio of event/total number of events
What are pre-test odds?
subjects with disease/subjects without disease
look at subjects in the study
pre-test probability/(1- pretest probability)
what is the pre-test probability?
subjects with disease/everyone
when is LR+ maximized?
when there is a high cut off to denote a + test
more TP than FP
when is LR- the smallest (maximized)?
when there is a low cut off to denote a - test
this will create more TN than FN
What are some advantages of using a likelihood ratio?
not influenced by the prevalence of disease like PPV and NPV
can be used on continuous tests unlike sensitivity and specificity
What is an ROC?
receiver operator characteristics
method to determine best single cutoff value when the test yields continuous results
what are the axis on an ROC?
Y - true positive
X - false positive (1-specificity)
where is the ideal test located on an ROC?
farthest away from the line of zero discrimination
Top left (want high TP and low FP)
what is the line of zero discrimination?
line on an ROC curve that represents TP:FP is 1 (equal likelihood of either occurrences
What does the area under the ROC curve represent?
discrimination or accuracy of the test
what are interpretive tests?
test is not an objective measurement but an interpretation of a visual image or subjective decision
what is the kappa statistic?
measure of inter- and intra- observer agreements
calculates the agreement beyond chance alone
How do you calculate the overall percentage of agreement between observers?
numbers that are the same/total
what is the max agreement between two observers?
How do you calculate the number of cases that could be due to a chance agreement?
percentage of cases in 1 x percentage of cases in 2
How do you calculate the total agreement by chance?
add number of cases by chance and number of absences by chance
divide by total
How do you calculate kappa statistic?
[(%agreement observed) - (%agreement by chance)] / [100% - %agreement by chance]
what is the ideal kappa?
get as close to one as you can (perfect agreement)
YOU MIGHT ALSO LIKE...
Comprehensive USMLE Step 2/COMLEX Level 2 Guide
Epidemiology Exam 3
EPI test 1
OTHER SETS BY THIS CREATOR
USMLE Step 2 CK-Renal