module 1

Question 1

numerator and denominator

Answer 1

numerator = cases of disease
denominator = population
numerator over denominator = freq. of dis-ease
always start with population (D)

Question 2

what do epidemiologists do and why

Answer 2

measure freq. of health and dis-ease to identify causes of poor health, how to prevent, how to improve, find out whether a drug worked or not.

Question 3

dis-ease occurrence

Answer 3

transition between non dis-ease and dis-ease

• event = when transition is observable
• state = not easily observable

Question 4

gate frame components

Answer 4

PECOT:
p = participant popl/study popl, split into sub-denominators
e, c = exposure & comparison groups EG & CG, can be divided into 2< categories (usually only 1 CG)
o = outcomes; numerators. split into a (EG dis-ease), b (CG dis-ease), c (EG not dis-ease), d (CG not dis-ease)
t = time. same point in time arrow horizontal; over period of time arrow vertical (down)

Question 5

EGO & CGO (numerical and categorical data)

Answer 5

EGO (exposure group occurrence) = a/EG
CGO (comparison group occurrence) = b/CG
numerical data:
some = categorically classified but outcomes still numerical - avg. for each group
calc. mean outcome - EGO = Σa/EG; CGO = Σb/CG

Question 6

incidence

Answer 6

counting number of onsets of disease (events) during a period of time - can be easily observed, represented as proportion D (%)
I = no. of persons in group who have dis-ease outcomes (N) / total no. of persons in group (D) during study time T
incidence in EG = [a/EG] OR [a/a+c] during time T
incidence in CG = [b/CG] OR [a/b+d] during time T

Question 7

prevalence

Answer 7

measure of disease occurrence at one point in time - when transition from non dis-ease to dis-ease if not easily observable
static measure of dis-ease - when counting prevalence “pool” need to take into acc “water lost” (death) and “evaporated water” (cured) –> not useful for causes of disease
useful to state when p was measures - p doesn’t have unit of time
P in EG = EGO vice versa

Question 8

cross sectional vs cohort study

Answer 8

cross sectional = the exposure and outcome is measured at the SAME point in time (horizontal arrow) –> not possible to measure incidence in c-s study
cohort = EG & CG followed for a period of time, count dis-ease events over a period of time (downwards arrow). P can be measured in cohort studies - at a point in time during the study

Question 9

incidence and prevalence strengths and weaknesses

Answer 9

INCIDENCE
determined only by dis-ease risk in a population (“clean” measure of dis-ease occurrence) ✅
includes events (N), population (D) and time (T) ✅
difficult to measure as have to observe over time ❌
PREVALENCE
(relatively) easy to “stop time” (snapshot) and count dis-ease occurrence ✅
only include N and D –> less info than incidence ❌
determined by incidence, cure rate, death rate (“dirty” measure of dis-ease occurrence) ❌

Question 10

2 ways to compare disease occurrences

Answer 10

ratio of occurrences (risk ratio/relative ratio RR) EGO/CGO but can be CGO/EGO so clarify risk in one group more than other for eg.
difference in occurrences (risk difference RD)
EGO-CGO vice versa
same units used in calc of EGO & CGO eg. events per 10,000 ppl

Question 11

relative risk reduction RRR, relative risk increase RRI

Answer 11

RR = any value more than 0; no difference in effect of E or C in study outcome - “no effect value”
if RR >1 –> RRR = (1 - RR) x 100 (%)
if RR 1< –> RRI = (RR - 1) x 100 (%)

Question 12

randomised control trials

Answer 12

ppts randomly allocated to EG/CG; only advantage is that all ppts more likely to be similar at the beginning of a study (therefore differences due to the factor that experiment is controlling - LESS CONFOUNDING)
randomisation only possible when ethical - exposure is safe as comparison
sometimes impractical - lack of long-term maintenance

Question 13

risk difference vs risk ratio

Answer 13

decisions should be made based off of rD NOT rR
rD more important - more info w units (eg. per 1000 ppl)
most ppl “understand” rR - but its RELATIVE so can be deceptive and provide less info

Question 14

two types of error in epidemiological studies

Answer 14

RANDOM errors - by chance

NON-RANDOM errors - faulty study design/conduction

Question 15

sources of non random error in epidemiological studies (names not details)

Answer 15

RECRUITMENT
ALLOCATION
MAINTENANCE
BLIND & OBJECTIVE MEASUREMENT
ANALYSIS

Question 16

recruitment (non-random error in studies)

Answer 16

external validity error - study findings not applicable to wider pool
random sampling = best
important when study objective = measure characteristics of popl
some studies = unnecessary to have representative sample
selection bias - when many/all ppts in EG recruited from very diff source from CG
triangle in gate frame - open top = setting in which eligible popl. recruited; middle = eligible popl; tip = ppl who AGREE to study. if small portion agree –> important to check if representative

Question 17

allocation (non-random error)

Answer 17

incorrectly allocated to EG/CG
bias from researcher in allocation –> confounding
RCT good
measurement errors (self report –> eg/cg low validity) –> allocation error so use biological tests, valid questionnaires etc.
tampering w/random allocation (bias from researcher) –> confounding in RCT. therefore use concealment of allocation - someone independent to experiment allocates. (sometimes not needed- eg placebo & pill look the same)

Question 18

analysis (non-random error)

Answer 18

were differences in EG, CG adjusted?
stratify study onto sub studies –> result of analysis in strata can be combined if similar; if diff –> report differently - stratified results
age standardisation

Question 19

maintenance (non-random error)

Answer 19

not a problem in cross-sectional studies
ideally - maintain e/c status throughout study; not exposed to other factors; not drop out of study
if ppl move groups (eg–>cg vice versa) –> groups not as diff as they would be if no one moved –> underestimate true effect of E on O
prevention - ppts & researchers blind to EG, CG. easier w drugs

Question 20

blind and objective measurement of disease outcomes (non-random error)

Answer 20

similar to allocation measurement errors
blind - ppl who measure disease outcome blind to ppts e/c status. knowledge of e status –> influence ppt/pract. perception/interpretation of signs/symptoms
objective - validated and standardised questionnaires, physical tests more objective

Question 21

ecological studies

Answer 21

popl = groups of people (eg. whole countries), allocated to EG/CG –> more prone to confounding