700
edits
Changes
Jump to navigation
Jump to search
Measures of strength of association (view source)
Revision as of 16:41, 4 May 2011
, 16:41, 4 May 2011no edit summary
Although correlation coefficients are commonly used in statistical studies, epidemiological investigations often deal with binary exposures and outcomes (such as presence or absence of a proposed risk factor for disease, and presence or absence of disease itself). Therefore, '''ratio measures''' such as the '''prevalence ratio''', the '''risk ratio''', the '''rate ratio''' and the '''odds ratio''' are commonly used as measures of strength of association in epidemiological studies.<br>
Although correlation coefficients are commonly used in statistical studies, epidemiological investigations often deal with binary exposures and outcomes (such as presence or absence of a proposed risk factor for disease, and presence or absence of disease itself). Therefore, '''ratio measures''' such as the '''prevalence ratio''', the '''risk ratio''', the '''rate ratio''' and the '''odds ratio''' are commonly used as measures of strength of association in epidemiological studies.<br>
Understanding how these measures are calculated is best approached using a contingency table (also known as a cross tabulation), as shown here. In the columns, the individuals are divided into exposed and unexposed, whilst in the rows, individuals are divided into those who are diseased and those who are not diseased.
Understanding how these measures are calculated is best approached using a contingency table (also known as a cross tabulation), as shown below. In the columns, the individuals are divided into exposed and unexposed, whilst in the rows, individuals are divided into those who are diseased and those who are not diseased. Therefore, cell 'm<sub>1</sub>' represents all diseased individuals, cell 'n<sub>1</sub>' represents all exposed individuals, and cell 'a<sub>1</sub>' represents exposed individuals who are also diseased.
{| class="wikitable"
{| class="wikitable"
|-
|-
| Disease status || Exposed || Unexposed || Total
| Disease status || Exposed || Unexposed || Total
|-
|-
| Diseased || a<sub>1</sub> || a<sub>0</sub> || Example
| Diseased || a<sub>1</sub> || a<sub>0</sub> || m<sub>1</sub>
|-
|-
| Non-diseased || b<sub>1</sub> || b<sub>0</sub> || Example
| Non-diseased || b<sub>1</sub> || b<sub>0</sub> || m<sub>0</sub>
|-
|-
| Total || n<sub>1</sub> || n<sub>0</sub> || n
| Total || n<sub>1</sub> || n<sub>0</sub> || n
|}
|}
The measures of disease frequency which could be extracted from this table will depend on the [[Study design|study design]] used, which will be [[Analytic studies#Study design|analytic]] in nature, as data regarding exposure has been collected.<br>
In the case of a [[Cross sectional studies#Study design|cross sectional study]], the '''[[Prevalence#Measures of disease frequency|prevalence]]''' can be estimated amongst exposed individuals as (a<sub>1</sub>/n<sub>1</sub>), and amongst unexposed individuals as (a<sub>0</sub>/n<sub>0</sub>).<br>
In the case of a [[Cohort studies#Study design|cohort study]], the '''[[Incidence risk#Measures of disease frequency|incidence risk]]''' can be estimated amongst exposed individuals as (a<sub>1</sub>/n<sub>1</sub>), and amongst unexposed individuals as (a<sub>0</sub>/n<sub>0</sub>). Alternatively, the [[Incidence rate#Measures of disease frequency|incidence rate]] can be estimated as (a<sub>1</sub>/[total number of animal-time units in exposed group]) amongst exposed animals and (a<sub>1</sub>/[total number of animal-time units in unexposed group]) amongst unexposed animals.