Measuring health and illness

The basic measurements in epidemiology include frequencies, proportions, ratios, and rates. Proportion, ratio, and rate measures involve comparison of numerator (cases) and denominator (population) and relates to a population rather than an individual person.  Denominator is what differs depending on whether using proportion, ratio, or rate.

Frequency. This is the simplest form of quantitative measure is a count. This is the number of individuals who meet the case definition.   

Proportion (or percentage): The numerator is some portion of the denominator.  The denominator is the entire group, and individuals in numerator are always included in denominator.  Always expressed as percentage (from 0% to 100%) or proportion (from 0.0 to 1.0).

Example: 20 Asians / 80 total population

Interpretation: Asians represent ¼ of entire population

Or convert denominator to 100 and interpret as %.

Example: multiply numerator & denominator by 1.25.  This converts 20/80 into equivalent value of 25/100.

Result:  25/100 – or 25% of population is Asian.

Ratio: This is the comparison of 2 quantities. The numerator is divided by denominator. BUT … individuals in numerator are not included in denominator. 

Example: ratio of women to men in population

40 females / 20 males

Equivalent interpretations:

There are 40 females for every 20 males

Women outnumber men 2 to 1.

Rate: Major measure used in epidemiology. Frequency with which disease, illness, or other event occurs in defined population during specified period of time. Numerator is the number of cases or events. For ex: number of deaths; number of cases of disease; number of injuries; etc.  Denominator is the population, group of individuals, etc

It is IMPORTANT to note that individuals in denominator MUST have potential to be in numerator.  That is, people in population (denominator) must have chance of having disease or illness (numerator).  Rates (instead of raw numbers) are used because rates adjust for differences in size of populations.

It is also important to note that a large denominator is needed to calculate a reliable rate.  For example suppose infant mortality is 20%. This seems like an alarmingly high number, but if the rate was based on 2 observed infant deaths for 10 births the death of 1 infant has a huge impact on the rate and therefore it is unreliable.

Example: Mortality rate during 2008  =  # deaths/ total population

If result is 3.8, this suggests that mortality rate is 3.8 per 100,000 people (or some other measure of population).

Rates are generally expressed per 1,000, per 100,000, or per 1 million population. And you can multiply the rate by an amount to yield desired denominator

Example: If 3 deaths in 10,000 population and want to express rate as deaths per 100,000; multiply numerator & denominator by 10

Result: 3 deaths per 10,000 population same as 30 deaths per 100,000 population (also same as 300 deaths per million population).

Need to specify (a) time period and (b) population measure to make rate meaningful. For example, the value of rate per 1,000 is very different from rate per million. Or the rate may be different if for 2000 vs. 2004 vs. 2008 vs. any other year.