EBPS Flashcards
p-value
determines the strength of evidence against a null hypothesis
null hypothesis (H0)
assumption that there is no significant difference, effect, or relationship between two or more groups or variables being studied
Alternative Hypothesis (Ha)
It usually suggests the presence of a significant effect, difference, or relationship.
hyperlipidemia
elevated levels of lipids in the bloodstream, including cholesterol and triglycerides, which can increase the risk of cardiovascular diseases
Normal blood pressure for adults is
120/80 or lower
Retrospective Study
looks at past data or events to examine the relationships between variables to draw conclusions about potential associations or outcomes.
Prospective Study
gathers data from participants moving forward in time, starting from the present and following them into the future to observe and measure outcomes as they occur, often through the design of cohort studies or clinical trials.
Cohort studies
follow a group of individuals, known as a cohort, and track their experiences and health outcomes over an extended period
confidence interval
a range of values that is calculated from sample data and is used to estimate the range within which a population parameter, such as a mean or proportion, is likely to fall with a certain level of confidence.
Epidemiology
- the study of the distribution and determinants of disease frequency in human populations
- the application of this study to control health problems and improve public health
- understand and to control its causes
Biostatistics
concerns with analysis and summarization of raw data in interpretable messages related to human health
Evidence-based medicine (EBM)
using the current best evidence in decision making in medicine in conjunction (together) with expertise of the decision-makers and
expectations and values of the patients/people
clinical research
- studying groups of people who are ill
- studies humans in clinical facilities such as outpatient clinics or inpatient facilities
- the interventions are often about therapy in sick people
- experimental design
- small to moderate size
epidemiological studies
- study people in communities
- preventive interventions
- observational studies
- large sample size
the “Big 6”
- description
- causation
- attribution
- mediation
- interaction
- prediction
Description
addresses how frequent or common are various risk factors, exposure, conditions, or diseases
Causation
addresses establishing causal relationships among biological, behavioral, environmental and other factors within humans.
Attribution
addresses what fraction or how many cases of disease Y can be eliminated if a causal exposure X is eliminated or reduced?
Mediation
- addresses the mechanisms of causal relationships
- Given that X does cause Y, how does X cause Y? What is the mechanism?
Interaction
- addresses when and for whom does X cause/predict Y?
- closely related to causation
Prediction
addresses as to whether some feature A or a combination of features A, B, and C predict the concurrent presence or future occurrence of Y?
How could we determine causes of diseases?
Conduct population studies using epidemiological methods
Why should pharmacists care about Epidemiology?
- practice evidenced based medicine (EBM)
randomized controlled trials (RCTs)
- scientific experiments in which participants are randomly assigned to receive different interventions or treatments
- assess the efficacy and safety of these interventions while minimizing bias
Case-control studies
observational research designs that compare individuals with a specific outcome or condition (cases) to those without it (controls) in order to identify factors associated with the development of that outcome or condition
Steps in Practicing EBM
- identify a good question
- find relevant literature
- critically evaluate data
- synthesize and apply to patients
- recognize gaps and design solutions
Two types of study designs
- experimental
- observational
Experimental study can be categorized as
two
- randomized control trial (RCT)
- non-randomized control trial
observational studies can be grouped as
two
- analytical
- descriptive
analytical studies can be
two
- case-control
- cohort
descriptive studies can be
cross-sectional
internal validity
How well do the study estimates represent what was intended in the study plan?
external validity
- How relevant are the study estimates to the research question?
- AKA: Generalizability
- Are study results applicable to the patient/population/problem in front of me?
three threats to validity
- chance (random error)
- bias (systematic error)
- confouding
chance (random error)
- errors that occur by chance
- improved by increasing sample size
- measured by CI
- can affect precision
- Lots of random error/chance = poor precision
- There can be random error in both sampling and measurement
bias (systematic error)
- can include selection bias, volunteer bias, measurement bias
- can affect accuracy
- errors caused by choices, compromises and mistakes we make in how we conduct our study and not by random processes
- Lots of systematic error/bias = poor accuracy
- There can be systematic error in both sampling and measurement
confounding
third variable associated with both exposure and outcome
prevalence
currently have a disease
incidence rate
definition
new cases per unit person-time
point prevalence
proportion with disease at a particular point in time
period prevalence
proportion with disease at any point in time during the period (short lived ex: COVID-19, migraine)
point prevalence formula
number of people with a disease or trait / total # of people in a study population
period prevalence formula
number of people with a disease or trait in specific time period / total # of people in study population
cumulative incidence formula
number of new cases of the disease during a specific time period / number of people in study population
incidence rate formula
number of new cases of a disease / number people in population * time period
RCT scheme
Participants are randomly assigned to different groups: one group receives the intervention being tested (the treatment group), and another group (the control group) receives either a placebo or a standard treatment, depending on
the study design.
RCT strengths
- Provides strongest causal evidence
- Randomization minimizes confounding and blinding minimizes measurement bias
RCT weaknesses
- $$$ and time-consuming
- Low external validity
- Ethics: some exposures impossible to study
RCT association measures
relative risk
cohort scheme
- Follow 2+ groups with different exposures over an extended period of time and compare outcomes
- Ideal for common diseases and rare exposures
cohort strengths
- Can measure incidence of multiple outcomes
- Can study effects of multiple risk factors
- Provides strong causal evidence due to time sequence
cohort weaknesses
- $$$ and time-consuming
- Vulnerable to confounding because we know the exposure beforehand
- Difficult to assess rare outcomes
- Loss to follow up
association measures
relative risk
case-control scheme
- Identify 2 groups with and w/o an outcome interest, collect data and compare odds of exposure
- Ideal for rare outcomes and outcomes with a long latency
case-control strengths
- $ and not time-consuming, quick
- Can assess multiple outcomes and study rare outcome
case-control weaknesses
- Recall and selection biases → low internal validity
- Vulnerable to confounding - Can only study one outcome
association measures
odds ratio
cross-sectional scheme
- Sample at one point in time to describe the distribution of exposures and outcomes
- Ideal for hypothesis generation
cross-sectional strengths
- $ and not time-consuming, quick
- Can measure prevalence
- Can assess several exposures & outcomes
cross-sectional weaknesses
- No temporal ordering → weak causal evidence
- Vulnerable to confounding
- Cannot measure incidence
association measures
odds ratio
PICO
Population
Intervention
Comparison
Outcome
Relative-risk (RR)
Odd ratio (OR)
RR >1
exposed are X times more likely to have a disease compared to unexposed
OR > 1
cases have X times higher odds of exposure compared to controls
RR = 1
exposed and the unexposed are equally likely to have a disease
OR = 1
odds of exposure in cases and controls are the same
RR < 1
exposed are (1-X)% less likely to have a disease compared to unexposed
OR < 1
odds of exposure in cases is (1-X)% lower than in controls
cumulative incidence
definition
new cases during time period
The frequency of of new COPD diagnosis in smokers is 33 per 1,000 person-years
incidence or prevalence
Incidence rate
In the same study, 5 had active wheezing at baseline exam
incidence or prevalence
Period prevalence
In the same study, 11 reported at baseline that they had taken opioid medications for pain at some point in the last year
prevalence or incidence
Point prevalence
In a study of 1,000 young adults, 24 developed diabetes over 10 years
prevalence or incidence
Cumulative incidence
measurement
making observations about the individuals who are sampled for the study
measurement can be
numeric (quantitative)
thematic (qualitative)
intervention
intentionally expose people to something
stages of conducting a research study
1) Specifying a research question
2) Making a study plan
3) Implementing that plan
point estimate
best guess about what the truth is
95% confidence interval
The interval within which the TRUE parameter will be found 95% of the time (this helps us understand the precision of an estimate)
bias is a problem with
accuracy
Lots of random error =
chance is playing a large role = poor precision
Lots of systematic error =
bias is playing a large role = poor accuracy
Sampling
Choosing particular individuals from a population
Census
we study every individual in a population
Experimental study
You manipulate an exposure by doing an
“intervention” (usually with one or more control groups), and then see what happens
Observational study
You just observe without any intervention
Three types of Observational study
Cross-sectional
Cohort
Case-control
Target population
The population for whom the research question is relevant
Accessible population
The population the researchers have access to and plan to
study
Study sample
the actual study subjects who were included in the study and whose data were analyzed and included in the study estimates
Target phenomenon
The thing you want to learn about
Intended variables
The things you think you can realistically measure in a research study
Actual measurements
The measurements that are actually made (with error) for a study
Inference
A conclusion reached on the basis of evidence and reasoning
Estimate
Numerical best guess informed by data
prevalence
current disease
incidence
new disease
average incidence rate
number of events / person-time
Kaplan-meier
- the graph can be for the rate of mortality or rate of survival
time-to-event analysis
cumulative incidence over time
rate
- must have time in denominator
- ex: incidence rate = events/person-year
proportion
- numerator is subset of denominator (between 0-1)
- ex: cumulative incidence = # developing disease / # total
