Midterm

Question 1

IMRAD

Answer 1

introduction 
materials and methods 
results 
(and)
discussion

Question 2

other components of a paper

Answer 2

title
abstract
keywords 
references
supplemental

Question 3

steps of a paper critique

Answer 3

read 
analyze
establish research context 
evaluate 
establish significance of the research

Question 4

steps of experimental design

Answer 4

background research 
formulate research question
identify variables
generate hypothesis 
determine experimental design

Question 5

what is optimization of an experiment?

Answer 5

set of experiments on its own to optomize variables i.e. cell type, treatment time, concentration of drug

see what works best and then use it

Question 6

what are controls for? negative? positive?

Answer 6

scientific controls minimize the effects of variables other than the independent variable (i.e. control for confounding)

negative - no response expected
positive - effect when there should be effect

Question 7

when do you use loading controls?

Answer 7

western blot - look for a house keeping gene to make sure all lanes have been loaded equally

they are a type of positive control

Question 8

technical replicates

Answer 8

the same sample being used in 3 wells etc

• control for human error
• improve accuracy

note: don’t use technical replicates for western blot

Question 9

biological replicates

Answer 9

more than one biological sample i.e. using 2 mice or passaging cell lines and repeating experiment
-account for individuals with differences

Question 10

descriptive statistics

Answer 10

discrete, quantitative analysis of data

summary of one sample of the population

not based on probability

ie demographic data, individual GPA scores etc

summary of your data set, don’t extend to population level

Question 11

inferential statistics

Answer 11

generalized, extended analysis of data

assumes properties of a population from an observed data set

based on probability

ie efficacy/significance of treatment paradigms on general population

generalize to population level

Question 12

data set

Answer 12

recorded raw values of a variable of interest

Question 13

mean

Answer 13

average value of the data set

Question 14

SD

Answer 14

how much the raw values of the data set spread across the mean

• best used for descriptive statistics
• describes variability of raw data across mean
• dependent on n value

(for technical replicates)

Question 15

SEM

Answer 15

how much the sample mean differs from the population mean

• best used for inferential statistics
• attempts to find SD of a sampling distribution
• dependent on n value

(for biological replicates)

Question 16

why use a large sample?

Answer 16

mean of a large sample is likely to be closer to the true population mean than that of aa small sample
i.e. with large sample know the value of the mean with a lot of precision even if the data are very scattered

Question 17

what is significance?

Answer 17

p-value

probability that the changes between two sets of data are true

95% CI/p<0.05 is standard

Question 18

standard t-test

Answer 18

used to compare 2 sets of independent data

Question 19

paired/repeated measures t-test

Answer 19

used to compare 2 sets of related data

Question 20

One-way ANOVA

Answer 20

used in comparing 3+ sets of data (1 variable)

Question 21

two-way ANOVA

Answer 21

used in comparing 3+ sets of data across 2 independent variables

Question 22

What is ANOVA?

Answer 22

basically multiple t-tests performed in sequence but is better because it is more conservative i.e. less chance for type I error (false positive)

Question 23

What is a post hoc?

Answer 23

further analysis of treatment groups after running an ANOVA
reduces probability of discovering a false positive
Turkey’s and Bonferroni are popular ones

Question 24

Turkey’s post hoc

Answer 24

comparison of each mean to every other mean (similar to multiple t-tests)

Question 25

Bonferroni post hoc

Answer 25

corrects the CI (alpha) depending on number on comparisons made (alpha/n) - the more comparisons you make the lower your significance - overcorrects test each hypothesis at a lower alpha to reduce chances of making type I errors (false positive) when doing multiple comparisons

Question 26

additive

Answer 26

combined drug effects are consistent with individual drug effect ie NOT them added together

Question 27

synergistic

Answer 27

combined drug effects produce an effect which is above what is expected from the individual drug effects

Question 28

independent variable

Answer 28

what you are manipulating

Question 29

dependent variable

Answer 29

depends on the independent variable (i.e. what you are measuring)

Question 30

minimum requirements to conclude that E is a cause of O?

Answer 30

correlation/association E precedes O replication don't need: - theory - randomization

Question 31

define epidemiology

Answer 31

the study of the occurrence and distribution of health related states or events in specified populations, including the study of the determinants influencing such states, and the application of this knowledge to control the health problems

Question 32

occurrence

Answer 32

incidence and prevalence

Question 33

distribution

Answer 33

occurrence by characteristics of person, place and time

Question 34

health related states

Answer 34

things that endure over time | ie a chronic disease

Question 35

health related events

Answer 35

things discretely dated in time | i.e. heart attack

Question 36

what is a case definition? what should it be?

Answer 36

definition of health state/event that is valid and reliable

Question 37

valid

Answer 37

measures what is intended | specificity and sensitivity

Question 38

reliable

Answer 38

stable, repeated, consistent, over time or by different raters

Question 39

sensitivity

Answer 39

probability that a diseased person in the population tested will be identified as diseased by the test i.e. true positive probability

Question 40

specificity

Answer 40

probability that a person without the disease will be correctly identified as non diseased by the test i.e. true negative probability

Question 41

a, b, c and d

Answer 41

``` a = true positive b = false positive c = false negative d = true negative ```

Question 42

sensitivity formula

Answer 42

a / (a+c)

Question 43

specificity formula

Answer 43

d / (b+d)

Question 44

descriptive epidemiology

Answer 44

- count cases in the population - present by characteristics of person, place and time - devise testable hypotheses based on unexpected of unusual trends, differences etc

Question 45

analytic epidemiology

Answer 45

- test etiologic hypotheses of risk and protective factors - replicate - minimize confounding - identify protective and risk factors

Question 46

experimental epidemiology

Answer 46

randomized clinical trials of interventions i.e. can see if protective factors you identified are as good as you thought they were

Question 47

confounding

Answer 47

due to correlation of apparent risk factor and true cause, there is confounded estimate of effect of risk factor on outcome as a result of confounding can find something that is strong, statistically significant, replaceable and completely wrong

Question 48

what can confounding do?

Answer 48

- create false associations - mask true associations - change estimated association from risk factor to protective factor or vice versa

Question 49

what are 5 strategies to control confounding?

Answer 49

design: randomization, matching, restriction analysis: stratification, multivariable analysis

Question 50

experimental vs observational study design

Answer 50

experimental: - use randomization to assign exposure - randomization turns confounding into random error (more as n increases) - works for known and unknown confounders observational: - exposures selected by self, parent, provider, insurer, happenstance etc - use other means (matching, restriction, stratification, multivariable analyses) to minimize confounding

Question 51

what do both prevalence and incidence need to include?

Answer 51

the time period i.e. annual incidence or prevalence on december 5th

Question 52

prevalence

Answer 52

all population cases/all people in population x100 good for bookkeeping i.e. monitoring trends, comparisons between regions and planning health system bad for identifying causes contaminated measure because it is affected by incidence, mortality and recovery

Question 53

incidence

Answer 53

all new cases in population/all ppl at risk in population base-10 i.e. per 100 000 etc essential for studying causes harder to collect so fewer around need to determine who is at risk pure measure, reflects all etiologic forces behind a disease

Question 54

Who is considered at risk when finding incidence?

Answer 54

at risk = people who have not been diagnosed or experienced outcome, but could not at risk = ppl who have already been diagnosed or cannot biologically have it

Question 55

how could decreased prevalence be bad? increased good?

Answer 55

decreased could be due to increased fatality | increased could be due reduced mortality

Question 56

what should you critical appraise incidence and prevalence measures?

Answer 56

because they are sample estimates of population parameters so difference by person, place or time can be real or artifact of different methods ie don't over interpret small changes in the presence of substantial error

Question 57

case-control studies

Answer 57

always control on the basis of OUTCOME cases have it (i.e. diabetes), controls do not select your cases and controls and then measure variables of interest relatively cheap

Question 58

cohort studies

Answer 58

always control on the basis of EXPOSURE exposed are (i.e. obese), unexposed are not determine exposure of interest and choose ppl that have or don't, follow up to observe outcomes (i.e. no on has disease at the beginning of the study - see who becomes a case) very expensive

Question 59

what are case-control studies good for? what kind of bias are they prone to?

Answer 59

good for rare diseases (can study all) long latency periods (if a disease takes a while to develop start with it instead of waiting for it to develop in a cohort study emergencies (ie find out what is causing food poisoning) prone to recall bias ie was your pregnancy stressful

Question 60

what are cohort studies good for? what kind of bias are they prone to?

Answer 60

good for rare exposures (study all) frequent outcomes prone to attrition bias

Question 61

what are the null and range of values for correlation? when do you use correlation?

Answer 61

null = 0 goes from -1 to 1 use for 2 continuously distributed variables i.e. dietary Na and BP

Question 62

what are null and range for ration measures of association (RR and OR)?

Answer 62

null = 1.0 | range - o.ooooo1 to infinity

Question 63

risk

Answer 63

prospectively observed probability of outcome

Question 64

risk ratio

Answer 64

risk of exposed / risk of unexposed 1.0 is null ie RR = 2.0 means that outcome is twice as likely in exposed group used for cohort studies

Question 65

a, b, c and d for ratios

Answer 65

``` go case (i.e. bad outcome) in first column, exposed on top, not on bottom non-case on other side, exposed on top and not on bottom ```

Question 66

risk exposed formula? unexposed?

Answer 66

a/ (a+b) | c/(c+d)

Question 67

odds

Answer 67

odds = probability of exposure / (1 - probability of exposure)

Question 68

odds ratio

Answer 68

ratio of the odds of being exposed in the cases/controls ``` null = 1.0 OR = 2.25 would be 2.25 more times likely to have been exposed in cases than controls ``` used for case-control studies

Question 69

cross product for OR

Answer 69

(a*d) / (b*c)

Question 70

relative risk?

Answer 70

often synonymous with RR, but sometimes used for any ratio (i.e. OR)

Question 71

if you match can that variable confound?

Answer 71

no i.e. if you age match you are guaranteed that your results are not confounded by age

Question 72

crude vs adjusted RR/OR

Answer 72

crude indicates the crude association between exposure and outcome with all confounding present adjusted cannot be confirmed bc regression has been used to adjust for the effects of known, measured and analyzed confounders ->can see how much confounding is going on by how large change is between crude and adjusted

Question 73

incidence rate ratio

Answer 73

ratio of 2 incidence rates -picks up speed of occurrence ie RR could be the same, but IRR could be larger because something is happening faster generally better than RR, especially if time matters

Question 74

incidence rate

Answer 74

* always count the last day* measured in events per time unit ie 111 events per 1000 person years

Question 75

hazard ratios

Answer 75

based on time-to-event i.e. death can also be called survival analysis also measures speed of occurrence ratio of the two speeds cumulated over the entire period of observation on everyone HR = 2.0 means that if a patient in one group has not died etc at a certain time point they have twice the probability of having died etc by the next time point compared to someone in the other treatment group

Question 76

what is an etiologic study?

Answer 76

what are risk and protective factors for ____ disease

Question 77

what is a prognostic study?

Answer 77

what are the risk and protective factors for specific outcomes in people with disease ____ i.e. everyone in study has disease

Question 78

what does prospective mean?

Answer 78

measure as they do it