Public health Flashcards
cross sectional study
- frequency of disease and frequency of risk-related factors are assessed in the present
- What is happening?
- measures prevalence and risk factor association
case-controlled study
- compares a group of people with disease to a group without
- looks to see if odds of exposure or risk factor differs
cross sectional study in my words
- takes a slice
- takes place at the present
- measures prevalence
case-controlled study in my words
- always in the past
- split by groups that do and don’t have disease
- looks for if something was a risk factor
- “if they have the disease, did something put them at risk”
cohort study in my words
- “if they have a risk, did they develop the disease”
- measure relative risk
- can be in the past or future
triple blind
blind to analysis of data
case-fatality rate
number of fatal cases / total number of cases
95% is how many SDs
1.98 (99% = 2.58)
standard error
- standard deviation of a number of sample means
- SD/ square root of n
confidence interval
- accounts for variability of sample means
- = mean +- z score + standard error
standard error takes into account what two factors
standard deviation and sample size (n)
hawthorne effect
- observer effect
- subjects change their behavior because they know they are in a study
Berkson’s bias
- specific selection bias created by choosing hospitalized patients as control group
lead-time bias
- using a screening method that detects a disease earlier, leading to an apparent prolongation of survival
- example: screening for PSA diagnoses prostate cancer earlier leading to longer survival times
Pygmalion effect
researchers belief in efficacy of treatment can change its actual effect
- example: if you think people are stupid, your intervention wont work as well
recall bias
- inaccurate recall of past exposures
- usually seen when asking questions
cumulative incidence
number of new cases of a disease / number of people at risk
- remember to subtract people that already have the disease because they are not at risk
power
the ability to detect between groups when one actually exsists
power equation
1-beta (type II error)
beta
study does not reject the null hypothesis when it should
ecological study
like a cross sectional study but it uses population data instead of individual data
effect modification
when the effect of an exposure on an outcome is modified by another variable
how effect modification differs from confounding
stratification is used to distinguish the difference. When the confounder is removed, there is no longer a difference between the two
attrition bias
when loss to follow up is greater for one group than another
selection bias
systemic difference between groups
misclassification bias
exposure and outcome are not measured properly, it affects both group equally
- example = nonfunctioning blood pressure cuff
purpose of matching
prevent confounding variables, or differences between groups
selection bias
nonrandom sampling so a group may not be representative of the population as a whole
chi squared test
used for the association of two categorical variables
definition of odds
probability of an event happening / probability of the event not happening
case control vs odds ratio mnemonic
CCOR and CRR - cant have double letter together
- case-control uses odds ratio and cohort uses relative risk
relative risk
(people who have the disease / total exposed) / (people with disease / total not exposed)
- gives an answer in relationship to 1 - the higher the numerator, the more likely the disease was related to the risk
attributable risk
(people who have the disease / total exposed) - (people with disease / total not exposed)
or, the percent of cases that are attributable to the risk
relative risk reduction
1 - relative risk
or, the proportion of risk reduction attributable to intervention
absolute risk reduction
difference in risk attributable to the intervention as compared to the control
number needed to treat
1/ARR
number needed to harm
1/AR
which two variables are swaps of eachother
attributable risk and absolute risk reduction (both used to calculate NNT and NNH)
selection bias
nonrandom sampling of subjects so that it can’t be extrapolated to greater population
- usually and sampling bias
Berkson bias
study population from hospital is less healthy than general population
healthy worker effect
subjects are healthier than general population
non-response bias
participating subjects differ from non-respondents in meaningful ways
ways to decrease selection bias
randomization and ensuring the right comparison group
recall bias
awareness of disorder alters recall by subjects, common in retrospective trials
measurement bias
information is gather in an unorganized manner
Hawthorne effect
participants change their behavior because they know they are being studied
procedure bias
subjects in different groups are not treated the same
observer-expectancy bias
belief in efficacy changes the results
confounding bias
a factor is related to both exposure and outcome, but not on causal pathway
how to decrease confounders
- repeat studies
- crossover studies
- matching
- restriction and randomization
lead-time bias
early detection leads to increase survival
type 1 error (alpha)
when the null hypothesis is rejected, but it should not be
use: gives the p value in studies
trick for remembering type 1 and 2 errors
type 1 (1 word) = DO reject null type 2 (2 words) = DO NOT reject null
type 2 error (beta)
when you do not reject the null hypothesis, but you should
use: for the power of a study (1-beta)
what increases power
- increased sample size
- increased effect size
- increased precision of measurement
t-test
checks difference between means
ANOVA
checks difference between means of 3+ groups
Chi squared
checks difference between 2 categorical variables (Chi-tegorical)
capacity vs competency
capacity = from a physician competency = from a judge
HMO
- patients are restricted to a network
- payment is denied if it does not fit evidence based guidelines
- requires referral to see specialist
point of service insurance
- patients can see providers out of network, but have higher costs
- requires referral
preferred provider organization insurance
- patients can see anyone and have higher copays
- does no require referral
exclusive provider organization insurance
- limited to certain doctors
- does not need a referral
capitation
- set amount of money per head over a period of time
discounted fee for service
patient pays for each service at a discounted rate
global payment
patient pays for everything associated with a single incident of care
