Brustsitus 4

number of people with new disease during a time period / total person-time when people were at risk of getting the disease

• taking person-time is actually the same as average people in a time period (see Box 2.5)
• merke: incidence rate gibt Anzahl cases pro person-time als outcome!!!
• attention: person-time wird streng genommen!! Sobald eine Person die Disease hat, werden diese nicht mehr dazugezählt!!

• to compare incidence rates between different countries, standard population should be used to age-standardize (as Brazil and Germany have different Alterspyramiden)

  direct standardization (age for example)

• given:
  • age-dependent mortality rate from the study population
  • standard population (with age distribution)
• now the mortality rates can be calculated for the standard population (Annex 1 in the book)

•  

  indirect standardization

• age-dependent mortality rate of study population unknown or unreliable (small sample size)
  • then we can calculate the SMR (Standardised mortality ratio)
    • given:
      • age-dependent mortality rate from standard population
      • age distribution of study population
    • now the calculated cases (from mortality rates of standard population) and the observed cases in the study population will be put in a ratio

• is used, when only the proportion of deaths are known but not the population at risk

PMR=(Total number of deaths from all causesNumber of deaths due to a specific cause)×100

• Z.B. Tod durch Unfall bei Bergsteigern ist 2%, in der Gesamtbevölkerung 0.5%. Gibt eine PMR von 2/0.5 = 4

• is used in a short period of time (z.B. 28 days), otherwise survival rates are used

CFR=(Number of diagnosed cases of the diseaseNumber of deaths from the disease)×100

It is usually expressed as a percentage.

estimates loss of life

Wird pro Erkrankung berechnet.

Anzahl Todesfälle * verlorene Lebensjahre. Verlorene Lebensjahre = Alter bei Tod – durchschnittliche Lebenserwartung

Gelebte Jahre werden der Lebensqualität angepasst. Ein Jahr gelebt mit 50% Lebensqualität entspricht 0.5 Jahre. 

Rechte Lunge: Abdruck von Herz, V. cava, V. brachiocephalica & Oesophagus

Linke Lunge: Herz, Aorta, Oesophagus

Beidseitig: der A. subclavia

estimates loss of life

1 DALY = 1 lost year of healthy life

Berpcksichtigt auch verstorbene Jahre, nicht kur mit eingeschrönkter Lebensqualität

DALY=YLL+YLD

Good:

- case-control studies

- ecological sutdies

Bad:

- cross-sectional studies

- RCT

- cohort

Good:

- RCT

- cohort (if a sample selected which is exposed)

Bad:

- ecological

- cross-sectional

- case-control

Good:

- cohort

- RCT

bad:

- cross-sectional

- case-control

same group is exposed and unexposed (with timely gap)

-> refers to interventional study

Group 1: Takes Drug A for 2 weeks → Washout → Takes Drug B for 2 weeks

Group 2: Takes Drug B first → Washout → Then Drug A

cluster (groups) will be randomized (not individuals)

if you have very expensive tests and you don’t want to do them to all the cohort, you can do it just for a small group. In the end, it will be done as well for all the ones which develop the outcome. You can then compare the result of the test of the small group and the cases.

it’s basically a case-control study but not chosen from the general population but from a cohort. For every case will be chosen an matched pair (age and sex).

each case is at the same time case and control; works for transient exposure (see example with sex and MI)

for each unit increase in BMI the odds of CHD is increased by a factor of 1.05 (95%-CI 1.03, 1.07), or equivalently, increases by 5% (95%-CI: 3%, 7%)

Alcohol consumption is associated with an 80% increase in risk of developing cancer

non-differential --> towards the null (underestimates)

differential --> under- or overestimates

Can overestimate, underestimate of completely hide the real association. Can even reverse the direction of an effect (Simpson's paradox). 

Confounding

Selection Bias

and these to remember: 

• Temporality: exposure must precede the disease. Best answered by cohort studies.
• Consistency: an effect which is found across a range of studies gives reassurance that it is not an artefact.
• Strength of association (doesn't need to be strong), rather consistent throughout the studies
• Dose-response relationship: can add weight to an evaluation of causation, but its absence need to
• Biological plausibility
• Specificity: there must be some degree of specificity to evaluate causation. E.g. bicycle helmets would be expected to reduce the risk of head injury but not of other types of injury.

The disease:

- severe, common and a public health program

- early detection should give a better outcome

- long pre-clinical phase

Screening test:

- accurate (high sens & spez)

- safe and acceptable

- cheap and simple

• volunteer bias --> people who participate might have healthier lifestyle
  • can be avoided if the volunteers get randomized
• lead time bias --> if earliert diagnosed, time to death may seem longer
• length bias --> only slowly growing diseases may be detected and rapid growing, more aggressive ones will be missed