A fundamental task in epidemiology is to quantify how commonly or frequently disease occurs in a population (measures of disease frequency). This is important since it allows animal health decision-makers to:
Measuring disease frequency can be done by simply counting the number of affected individuals in a population. However, in order to compare levels of disease across different groups of individuals, time frames and locations, we also need to consider counts of cases in context of the size of the population from which those cases arose. For example, 10 cases of disease in a population of 1,000 is a very different story than 10 cases in a population of 1,000,000.
It is also important to note that the level and distribution of disease in a population depends on the interplay of individual, spatial and temporal factors.
The figure on the right illustrate changes in the frequency and distribution of bovine tuberculosis (bTB) cases in the United Kingdom from 1986 to 2010. The growing clusters in south east of England and Wales are related to a high density of badger populations which act as wildlife reservoirs that continuously spread the disease to cattle populations. Scotland has remained virtually bTB free since 2013 thanks to requirements for cattle imported from high-risk regions to be tested for bTB before and after moving across the border.
Before discussing the methods for quantifying disease frequency any further, we first need to define a few key terms:
Prevalence measures existing cases of disease in a population and is calculated as the proportion of individuals in a defined population that has disease at a specified point in time (point prevalence) or during a specified period of time (period prevalence).
When we measure prevalence, we are essentially ‘stopping the clock’ and determining how many individuals have disease at that point in time. However, the point in time is not necessarily a single point in calendar time. To illustrate this, consider a study research in which we want to estimate the prevalence of dental decay in a sample of 400 cats.
It is not feasible to examine every single cat on the same day or even in the same week and therefore, the visits for the study will have distributed over several weeks throughout the year. We are still examining each cat only once, but the time point for measurement becomes the office visit during the year rather than a specific calendar date.
In the example to the right, we have a population of 18 cats with 5 found to have dental decay. The prevalence of dental decay in this population is therefore 5 / 18 or 27.7%
Incidence risk measures new cases of disease occurring in a population and is calculated as the proportion of initially disease-free (susceptible) individuals in a population who become new cases during a defined follow-up period.
The key here is that you are observing individuals at two or more time points – at the first time point you are checking to make sure the individual does not have disease and then you follow the individual over a period of time to see how many of them develop disease.
In the example to the right, let’s pretend we want to investigate the risk that healthy cats will get a cat bite abscess over a 10-week period. We take a group of 5 cats who do not currently have a cat bite abscess and follow them for a period of 10-weeks. Cat 1 develops a cat bite abscess in Week 5 and Cat 3 develops a cat bite abscess in Week 8, while the rest of the animals in our study remain puncture free. The incidence of cat bites is 2 / 5 cats or 40% over a 10-week period.
It is always important to specify the time period when you are reporting incidence because 2 / 5 cats in a 10-week period present a very different level of concern compared with 2 / 5 cats in a 10-year period.
Incidence rate is the number of new cases of disease that occur per unit of individual time at risk during a defined follow-up period. Incidence rate is expressed in terms of events per unit of time-at-risk and is calculated as follows:
Incidence rate is used when the population is dynamic, that is individuals are entering and leaving the population over the course of the study and this allows for inclusion of individuals with varying duration of follow up. The key to the calculation of incidence rate is the concept ‘time-at-risk’. Individuals contribute time-at-risk for as long as they are in the study or until they experience the event of interest.
Consider the example below where we have some cats that were recruited into the study late (Cat 2 and Cat 3) and some cats that left early either because they developed disease (Cat 1 and Cat 3) or dropped out from our study for another reason (Cat 2). Even though we intended to observe all cats for a 10-week period, Cats 1, 2, and 3 were only observed for 5 weeks each. In this case, we would count up the total number of cats that developed disease (2 cats) and divide that by the total weeks that cats were observed (35 weeks) to get our final incidence rate of 2 cases per 35 cat-weeks or 0.06 cases per cat-week.
You can get some very strange units here like dog-days, cow-months or horse years. Both incidence risk and incidence rate are useful for counselling animal owners in preventative care discussions about the likelihood of their currently healthy animal developing disease in the future.
Another way of thinking about prevalence and incidence is imagining a bucket filling and emptying with water. The bucket represents the total number of individuals in your population and the level of water in your bucket represents prevalence of disease. The water flowing from the tap into the bucket represents incidence or new cases that will increase the water level. The hole in the bottom of the bucket represents recovery (or possibly death!) from disease and will act to lower the water level in the bucket. The balance between rate in and rate out determines your disease prevalence.
An interesting example of this is when they introduced anti-viral agents for the treatment of HIV in people. Even though the incidence of new HIV cases remained roughly the same, the prevalence of HIV actually increased because people were living longer with the disease (i.e. the hole at the bottom of the bucket was plugged so the water levels rise).
The odds of disease is a slightly strange measure of the likelihood that an event will take place and is calculated as the number of diseased individuals (events) divided by the number of non-diseased individuals (events). You have probably heard the term most commonly used in the context of placing bets for things like horse races. In the example below, we have 3 diseased cats and 2 non-diseased cats which would make our prevalence 3 / 5 or 60% and our odds 3:2 (which reduces to 1.5:1).
The following table summarises the key take home messages about the measures of disease frequency.
| Item | Prevalence | Incidence risk | Incidence rate | Odds |
|---|---|---|---|---|
| Numerator | All cases counted on a single occasion | New cases occurring during a specified follow-up period | New cases occurring during a specified follow-up period | All cases counted on a single occasion |
| Denominator | All individuals examined (cases and non-cases) | All susceptible individuals present at the start of the study | Sum of time periods during which individuals could have developed disease | All non-cases |
| Time | Single point or period | Defined period | Measured for each individual from study start until disease event | Single point or period |
| Interpretation | Probability of having disease at a given point in time | Probability of developing disease over a specified period | How quickly new cases develop over a specified period | The odds of having disease at a given point in time |
Advancing Animal Care through Shared Learning