How do we know if a factor is associated with a disease?

3 posts / 0 new
Log in or register to post comments
Last post
Sun, 10/09/2011 - 11:33
#1
Praveen Nirmalan
How do we know if a factor is associated with a disease?
  1. How do we know or determine if there a particular factor or exposure is associated with a disease (example: is sex associated with blindness)?
  2. How doe we convey the strength of an association ( example- smoking is strongly associated with a type of cancer)
  3. How do we convey the direction of association (example- drinking tea is protective for cardiac events)
  4. Is the relative risk the same as odds ratio? When do we use a relative risk and when do we use an odds ratio?
  5. How do we interpret the relative risk and odds ratio?

 

Top
Mon, 10/10/2011 - 19:54
Praveen Nirmalan
Is a risk factor associated with a disease?

There are several considerations to determine if a risk factor is associated with a disease.

  • Association does not imply that the factor causes the disease. It implies that the factor is present along with the disease.
  • What do we need to show to determine if a risk factor is associated with a disease?
  • We need to show that the risk factor is either more common among those with the disease or increases the risk of the disease ( a harmful association) or those with the disease are more likely to have the exposure to the risk factor OR
  • We need to show the risk factor decreases the risk for the disease or the disease is less among those have the risk factor (a beneficial association) or the risk factor is less common among those with the disease 
  • Thus, we say, person with risk factor- more risk for disease or less risk for disease
  • How do we show if there is a difference in the risk factor?
  • We can look for the difference in incidence or an excess risk or we look for the ratio of incidence rates (see example below)

 

  Population A Population B
Incidence of disease in exposed (%) 30 80
Incidence of disease in non exposed (%) 10 60
Difference in incidence rates (%) (30-10)=20 (80-60)=20
Ratio of incidence rates (30/10)=3 (80/60)=1.3

Let us use Diabetes Mellitus as an example.

Based on the above table,

In population A,

The incidence of  Diabetes Mellitus among those who have a BMI > 30 (exposure) is 30%

The incience of Diabetes Mellitus among those who have a BMI < 30 (not exposed) is 10%

The excess risk is 30%-10%=20%

In population B

The incidence of  Diabetes Mellitus among those who have a BMI > 30 (exposure) is 80%

The incience of Diabetes Mellitus among those who have a BMI < 30 (not exposed) is 60%

The excess risk is 80%-60%=20%

 

In both populations the excess risk is the same (20%)

 

In population A,

The incidence of  Diabetes Mellitus among those who have a BMI > 30 (exposure) is 30%

The incience of Diabetes Mellitus among those who have a BMI < 30 (not exposed) is 10%

The incidence rate ratio is 30/10 =3

In population B

The incidence of  Diabetes Mellitus among those who have a BMI > 30 (exposure) is 80%

The incience of Diabetes Mellitus among those who have a BMI < 30 (not exposed) is 60%

The incidence rate ratio is 80/60=1.3

 

The Incidence rate ratio differs between the two populations (3 and 1.3). It is higher in population A although the excess risk of both populations were the same.

Do both measures tell us the same thing? Are both measures useful?

Top
Wed, 10/12/2011 - 20:50
Praveen Nirmalan
Relative Risk

What is the ratio of disease in persons exposed to a risk factor

to

The risk of disease in persons not exposed to the risk factor

 

This ratio is the relative risk. 

 

  Disease + Disease - Total
Exposure + a b a+b
Exposure - c d c+d

Risk in exposed= a/(a+b)

Risk in non exposed= c/ (c+d)

Relative risk= Risk in exposed / Risk in non exposed

 

Interpreting Relative Risk
RR=1, OR 95% CI limits on either side of 1 Risk in exposed equal to risk in non exposed (no association)
RR>1 AND both 95%CI limits >1 Risk in exposed greater than risk in non exposed (positive association, maybe causal)
RR<1 AND both 95%CI limits <1 Risk in exposed lower than risk in non exposed (negative association, probably protective)

 

 

Top
Log in or register to post comments