Understanding and Interpreting Relative Risk

As a public health professional, you rely on a variety of research findings to inform your practice. In this video, we will help you understand the concept of relative risk, which allows you to evaluate the effectiveness of an intervention, or the relationship between a risk factor and an outcome of interest. Relative risks not only tell you if a relationship exists, but also how strong that relationship is. For example, a relative risk can tell you if a relationship exists between smoking and lung cancer, as well as how great the risk of lung cancer is when one smokes. Put simply, relative risks, known as RRs, describe the likelihood of a certain event happening in one group versus another. An RR can only be calculated when there are only two possible outcomes for the event being measured, such as being compliant with food handling legislation or not being compliant. In addition, RRs can only be calculated for studies in which one group is exposed to either an intervention or a risk factor, and the other group is not exposed, and in which both groups are followed forward in time. This is in contrast to studies that start by selecting those with an outcome of interest and looking back in time to see who was or was not exposed to the risk factor. For example, looking at people with cancer and investigating if any were smokers. In these studies, one would calculate odds ratios rather than relative risk. To learn more about odds ratios, watch How to Calculate an Odds Ratio in this video series. Now, let us take a look at a hypothetical example while we learn how easy it is to calculate and interpret NRR. Remember, the data presented here are purely hypothetical and shouldn't be used for decision making. Suppose we have 1,000 children who receive oral hygiene education for preventing dental caries and another 1,000 children who do not. Among those exposed to the education, 25 get dental caries and 975 do not. Among children not receiving dental education, 50 get dental caries and 950 do not. The risk of dental caries in the group receiving the education is calculated by dividing the number with dental caries by the total number receiving the education. So 25 divided by 1,000 gives us a risk of 0.025. The risk of dental caries among children not receiving the education is calculated in the same way. This time, we divide the number of children with dental caries in this group by the total number of children who did not receive the education. So, this is 50 divided by 1,000, which gives us a risk of 0.05. Now that we know both the risk of dental carries among children receiving oral hygiene education and the risk among those not receiving the education, we can calculate the ratio between them. Risk ratios are calculated by dividing the risk of the event among those exposed to the intervention by the risk of the event among those not exposed. In our example, we divide the risk of dental carries among children receiving the education, 0.025, by the risk of dental caries among children not receiving the education, 0.05. This gives us a relative risk of 0.5. Now that we have calculated the relative risk, we need to interpret it. An RR of 1 would mean the risk of dental caries is no different for children who receive education than it is for children who do not. An RR below 1 would mean children who receive the education have a lower risk of dental caries than those who do not. And an RR above 1 would mean children who receive education have a greater risk of dental caries than those who do not. In our example, the RR is 0.5, which is less than 1. This tells us that children who receive the education are at a lower risk for dental caries than children who do not. However, this RR tells us not only that children receiving the education are at lower risk for dental caries, It also tells us how much lower their risk is. Let's take a look. Remember, an RR of 1 tells us there is no difference in risk of dental carries among the two groups. If the RR is 0.5, that is half of 1. So we say children who receive the education are at half the risk for dental carries as children who do not receive the education. In other words, they have a 50% lower risk of developing dental carries. This would be considered an important reduction in risk and likely would support public health programs that encourage oral hygiene education. If the RR had been 0.25, we would say that the risk for developing dental caries in children receiving the education was 75% less than that of children not receiving the education. We arrive at this number by subtracting the RR, 0.25, from 1. So, 1 minus 0.25 equals 0.75, or 75%. If the RR was 0.10, we would subtract that from 1, giving us 0.90. This would mean that children receiving the education had a 90% lower risk for dental carries than children not receiving the education. In this example, the farther away from 1 the RR is, the lower the risk of dental carries, the more effective the intervention is. As the RR moves closer to 1, there is less difference in the risk for dental carries among those who receive the education and those who do not, indicating that our intervention had less of an effect. In this example, an RR below 1 constitutes a positive outcome because we hope to reduce the risk of dental caries. However, there are many instances in public health when we want to see more of an event, in which case an RR above 1 would represent a positive outcome. Let's look at another hypothetical example. Suppose we want to know if community-based interventions are an effective way to increase the likelihood that adolescents participate in one hour of daily physical activity. And this example, an RR above 1, will represent a positive outcome because we hope to see an increase in physical activity among those exposed to the intervention compared to those not exposed. Let's say we have 100 adolescents exposed to a community-based intervention to promote physical activity and another 100 not exposed. Among those exposed, 75 participate in physical activity for one hour each day and 25 do not. Among those not exposed, 45 participate in physical activity and 55 do not. To calculate the likelihood of participating in physical activity among those exposed to the intervention, we divide those participating in an hour of physical activity by the total number exposed to the intervention. So 75 divided by 100. which gives us a likelihood of 0.75. Among those not exposed to the intervention, we divide the number who participated in physical activity by the total not exposed, so 45 divided by 100, which gives us a likelihood of 0.45. To calculate the RR, we divide the likelihood of participating in physical activity among those exposed to the intervention by the likelihood of participating in physical activity among those not exposed. So, 0.75 divided by 0.45, which gives us an RR of 1.67. We know from the previous example that an RR of 1 would mean that those exposed to the intervention would be no more or less likely to participate in physical activity than those not exposed. Furthermore, in this example an RR greater than 1 means those exposed are more likely to participate in physical activity than those not exposed, and an RR less than 1 means those exposed are less likely to participate in physical activity than those not exposed. In our example looking at ways to promote daily physical activity, an RR of less than 1 would suggest that community-based intervention was not effective. However, the RR of 1.67 tells us that those exposed to the community-based intervention are more likely to participate in physical activity than those not exposed. The RR also gives us an indication of how large this effect is. Again, if we know that an RR of 1 means the groups are the same, then an RR of 1.67 means those exposed were 1.67 times more likely to engage in physical activity than those not exposed. We can subtract 1 from 1.67 to give us 0.67 or 67%, meaning adolescents exposed to the intervention were 67% more likely to participate in physical activity than those not exposed. Generally, most would agree that an effect size this large would constitute an important finding. A relative risk is an important statistic in public health. because it helps us determine not only if a relationship exists between exposure to either an intervention or risk factor and an event, but also how strong that relationship is. An RR can only be calculated for studies that have only two possible outcomes and when the research designs follow both groups forward in time. While RRs won't be the only evidence used to guide your public health decisions, they provide an important type of evidence to consider when assessing the effectiveness of interventions as well as the association between risk factors and outcomes. With a little practice, you will be able to confidently calculate and interpret RRs to help inform your practice decision.

Put simply, relative risks, known as RRs, describe the likelihood of a certain event happening in one group versus another. An RR can only be calculated when there are only two possible outcomes for the event being measured, such as being compliant with food handling legislation or not being compliant. In addition, RRs can only be calculated for studies in which one group is exposed to either an intervention or a risk factor, and the other group is not exposed, and in which both groups are followed forward in time. This is in contrast to studies that start by selecting those with an outcome of interest and looking back in time to see who was or was not exposed to the risk factor.

For example, looking at people with cancer and investigating if any were smokers. In these studies, one would calculate odds ratios rather than relative risk. To learn more about odds ratios, watch How to Calculate an Odds Ratio in this video series. Now, let us take a look at a hypothetical example while we learn how easy it is to calculate and interpret NRR.

Remember, the data presented here are purely hypothetical and shouldn't be used for decision making. Suppose we have 1,000 children who receive oral hygiene education for preventing dental caries and another 1,000 children who do not. Among those exposed to the education, 25 get dental caries and 975 do not. Among children not receiving dental education, 50 get dental caries and 950 do not.

The risk of dental caries in the group receiving the education is calculated by dividing the number with dental caries by the total number receiving the education. So 25 divided by 1,000 gives us a risk of 0.025. The risk of dental caries among children not receiving the education is calculated in the same way.

This time, we divide the number of children with dental caries in this group by the total number of children who did not receive the education. So, this is 50 divided by 1,000, which gives us a risk of 0.05. Now that we know both the risk of dental carries among children receiving oral hygiene education and the risk among those not receiving the education, we can calculate the ratio between them. Risk ratios are calculated by dividing the risk of the event among those exposed to the intervention by the risk of the event among those not exposed.

In our example, we divide the risk of dental carries among children receiving the education, 0.025, by the risk of dental caries among children not receiving the education, 0.05. This gives us a relative risk of 0.5. Now that we have calculated the relative risk, we need to interpret it. An RR of 1 would mean the risk of dental caries is no different for children who receive education than it is for children who do not.

An RR below 1 would mean children who receive the education have a lower risk of dental caries than those who do not. And an RR above 1 would mean children who receive education have a greater risk of dental caries than those who do not. In our example, the RR is 0.5, which is less than 1. This tells us that children who receive the education are at a lower risk for dental caries than children who do not.

However, this RR tells us not only that children receiving the education are at lower risk for dental caries, It also tells us how much lower their risk is. Let's take a look. Remember, an RR of 1 tells us there is no difference in risk of dental carries among the two groups.

If the RR is 0.5, that is half of 1. So we say children who receive the education are at half the risk for dental carries as children who do not receive the education. In other words, they have a 50% lower risk of developing dental carries. This would be considered an important reduction in risk and likely would support public health programs that encourage oral hygiene education.

If the RR had been 0.25, we would say that the risk for developing dental caries in children receiving the education was 75% less than that of children not receiving the education. We arrive at this number by subtracting the RR, 0.25, from 1. So, 1 minus 0.25 equals 0.75, or 75%. If the RR was 0.10, we would subtract that from 1, giving us 0.90.

This would mean that children receiving the education had a 90% lower risk for dental carries than children not receiving the education. In this example, the farther away from 1 the RR is, the lower the risk of dental carries, the more effective the intervention is. As the RR moves closer to 1, there is less difference in the risk for dental carries among those who receive the education and those who do not, indicating that our intervention had less of an effect.

In this example, an RR below 1 constitutes a positive outcome because we hope to reduce the risk of dental caries. However, there are many instances in public health when we want to see more of an event, in which case an RR above 1 would represent a positive outcome. Let's look at another hypothetical example.

Suppose we want to know if community-based interventions are an effective way to increase the likelihood that adolescents participate in one hour of daily physical activity. And this example, an RR above 1, will represent a positive outcome because we hope to see an increase in physical activity among those exposed to the intervention compared to those not exposed. Let's say we have 100 adolescents exposed to a community-based intervention to promote physical activity and another 100 not exposed. Among those exposed, 75 participate in physical activity for one hour each day and 25 do not.

Among those not exposed, 45 participate in physical activity and 55 do not. To calculate the likelihood of participating in physical activity among those exposed to the intervention, we divide those participating in an hour of physical activity by the total number exposed to the intervention. So 75 divided by 100. which gives us a likelihood of 0.75.

Among those not exposed to the intervention, we divide the number who participated in physical activity by the total not exposed, so 45 divided by 100, which gives us a likelihood of 0.45. To calculate the RR, we divide the likelihood of participating in physical activity among those exposed to the intervention by the likelihood of participating in physical activity among those not exposed. So, 0.75 divided by 0.45, which gives us an RR of 1.67. We know from the previous example that an RR of 1 would mean that those exposed to the intervention would be no more or less likely to participate in physical activity than those not exposed. Furthermore, in this example an RR greater than 1 means those exposed are more likely to participate in physical activity than those not exposed, and an RR less than 1 means those exposed are less likely to participate in physical activity than those not exposed.

In our example looking at ways to promote daily physical activity, an RR of less than 1 would suggest that community-based intervention was not effective. However, the RR of 1.67 tells us that those exposed to the community-based intervention are more likely to participate in physical activity than those not exposed. The RR also gives us an indication of how large this effect is.

Again, if we know that an RR of 1 means the groups are the same, then an RR of 1.67 means those exposed were 1.67 times more likely to engage in physical activity than those not exposed. We can subtract 1 from 1.67 to give us 0.67 or 67%, meaning adolescents exposed to the intervention were 67% more likely to participate in physical activity than those not exposed. Generally, most would agree that an effect size this large would constitute an important finding. A relative risk is an important statistic in public health. because it helps us determine not only if a relationship exists between exposure to either an intervention or risk factor and an event, but also how strong that relationship is.

An RR can only be calculated for studies that have only two possible outcomes and when the research designs follow both groups forward in time. While RRs won't be the only evidence used to guide your public health decisions, they provide an important type of evidence to consider when assessing the effectiveness of interventions as well as the association between risk factors and outcomes. With a little practice, you will be able to confidently calculate and interpret RRs to help inform your practice decision.

Transcript for:Understanding and Interpreting Relative Risk

Transcript for:
Understanding and Interpreting Relative Risk