Clinicometrics Flashcards
crude mortality
total # of deaths in a population / entire population
cause specific mortality
deaths from disease / entire population
case fatality mortality
deaths from disease / # of cases of disease
proportionate mortality for disease
deaths from disease / total # deaths in a population
prevalence
total number of cases in a population at a given time
cases at time T / population at time T
does NOT determine risk or probability of becoming a case
what type of study determines prevalence
cross sectional studies
can CS studies calculate risk/probability
no because it ignores disease duration
what factors influence prevalence
- population dynamics
- duration of disease
prevalence = incidence x disease duration
incidence
risk/probability of disease occurring in a time period
used to compare disease occurrence in one group versus another (exposed vs not exposed OR treatment vs controls)
what type of study determines incidence
clinical trials
cohort studies
what is required in order to calculate incidence
defined time interval (study length)
incidence proportion
new cases during study period / susceptible subjects
ranges from 0 to 1
assumes subjects do not have disease at the start of the time period and are not at risk of being removed from the population
limitations of calculating incidence proportion
- diseases with recurrence
- dynamic populations - requires calculation of incidence rate
incidence rate
new cases during study period / time that subjects are at risk of disease
denominator = sum of time at risk for all individuals in the population
- accounts for dynamic populations
measured in units of time (cases per X number of days)
2X2 tables - A, B, C, D
A: diseased + exposed
B: diseased + not exposed
C: healthy + exposed
D: healthy + not exposed
risk ratio (RR)
incidence proportion in exposed / incidence proportion in unexposed
relative risk, incidence proportion ratio
(# exposed new cases / total # exposed) / (# unexposed new cases / total # unexposed)
=
(A / A+C) / (B / B+D)
how do you interpret risk ratio
RR probability of developing disease in exposed vs unexposed
range of risk ratio
0 to infinity
RR > 1: exposure is a risk factor for disease
RR < 1: exposure is protective against disease
risk difference
incidence proportion in exposed - incidence proportion in unexposed
(A/ A+C) = (B/ B+D)
absolute measure of risk
determines the importance of the disease
interpretation of risk difference
for every (total # exposed), only X develop disease due to exposure
incidence rate ratio
incidence rate in exposed/ incidence rate in unexposed
(# exposed new cases / time exposed are at risk) / (# unexposed new cases / time unexposed are at risk)
interpretation of incidence rate ratio
patients will get disease IRR times faster if exposed
hazard rates
similar to incidence rate but estimated from a type of statistical model that estimates time to event of interest in different groups
cohort study uses
good for COMMON diseases to calculate RISK RATIO
start with population at risk (disease free) –> determine exposure
why can you calculate risk ratio of a cohort study
because samples are randomly selected (all subjects are disease free at enrollment)
limitations of cohort studies
rare diseases make it difficult because you would have cases «< controls
inefficient to follow all controls just to find an adequate number of cases
- can compare the cases to a smaller sample of controls and statistically still get the same results
case control study uses
similar enrollment as cohort study except controls are sampled from the cohort
- allows study of RARE diseases
cases: enrolled from the cohort if they develop the disease over a set time period
controls: enrolled from the cohort if they are at risk of getting the disease but do not have it yet
can you calculate risk ratio from a case control study
NO because the number of cases and controls were selected by the investigator (not randomly sampled like cohort)
subjects are STARTING with disease vs no disease
case control study steps
- start with a cohort
- determine the marginal totals of cases and controls (sampled from cohort)
- determine exposure status of cases and controls
what can be calculated from case control studies
- odds of exposure
- case control (exposure) odds ratio
odds of exposure
probability of being exposed / probability of not being exposed
p / 1-p where p = probability
odds of exposure among cases
proportion of cases exposed / proportion of cases not exposed
(A / A+B) / (B / A+B)
=
A / B
odds of exposure among controls
proportion of controls exposed / proportion of controls not exposed
(C / C+D) / (D / C+D)
=
C /D
case control odds ratio
odds of exposure among cases / odds of exposure among controls
exposure odds ratio
(A/B) / (C/D)
=
AD / BC
can risk ratio be calculated from a case control study
NO - but can be approximated from the case control odds ratio
risk ratio = disease odds ratio = case control odds ratio ONLY when the disease is rare
can you calculate risk from a case control study
NO and you can NOT approximate it either
can only approximate RISK RATIO (relative risk)
exposure odds ratio vs disease odds ratio
EOR: measured in case control studies
DOR: measured in cohort studies
how are EOR and DOR related
EOR is always equal to DOR
how are DOR and risk ratio related
DOR = risk ratio (relative risk) when the disease is RARE
when is a disease considered rare
when risk < 5% in a period of time
odds is bound from 0 to infinity
risk is bound from 0 to 1
when p = 0.05 –> odds curve diverges from probability/risk curve –> odds ratio no longer equals risk ratio
how are controls selected for a case control study
3 rules
1. control should only be selected if someday, if it were to become a case, then it would be eligible to be included in the study as a case
2. controls should be selected for reasons unrelated to exposure
3. select incident (new) cases NOT prevalent (old) cases
how many controls should be enrolled in a case control study
want to have a minimum of 1 control per case
more controls per case is better, but no need for >4 cases per control
what is the use of case control studies
hypothesis generating - examine a large number of exposures for one outcome/disease
