In this session, we will discuss some practical uses of epidemic curves. In the first session of this course by Professor Peiris, we learned that an epidemic is the occurrence of diseases in excess of what would normally be expected in a community or geographical area. For example, smallpox has been globally eradicated since 1980, so the expected number of cases is 0 worldwide. This means that a single case of smallpox today anywhere will constitute an epidemic. For some diseases, cases are expected to occur throughout the year at some background rate. For these diseases, an epidemic corresponds to an abrupt and significant increase of cases above this expected level. For example, this figure shows the weekly number of hepatitis A cases in Pennsylvania between year 2000 and 2011. In this figure, we can clearly see that there was a hepatitis A epidemic in 2003. When we describe an epidemic of a disease that spreads from person to person in a population, the primary cases refer to the first infected cases in the population. The secondary cases refer to the cases infected by the primary cases. The tertiary cases refer to those infected by the secondary cases, and so on. For many infections, such as influenza and polio, a substantial proportion of the infected people have no clinical symptoms but are still able to infect other people. These sub-clinical or asymptomatic cases are typically not registered by public health surveillance. When we investigate an epidemic, the first step is usually to create the so-called line list which records the time of illness onset, location and other relevant information for each identified case. The next step is usually to look at the number of new cases over time. An epidemic curve, or epi curve, is a graph that shows the number of cases by the date of illness onset. Let us now look at a simple example on how to draw an epidemic curve. In this example, we assume that people show symptoms as soon as they are infected. On day 1, one person is infected by an external source. For example, an animal reservoir. On day 2, this person infects two people. Each of these two persons then infects another two people on day 3, therefore, a total of four new cases on day 3. To draw the epi curve for this epidemic, we label time along the horizontal axis and the number of new cases on each day along the vertical axis. We then count the number of new cases on each day and plot these counts in the graph. If the number of cases is not too large, we can also show the case ID in the epi curve for easy cross-reference with the line list. The epi curve can give us some very useful information about the epidemic. First, we can use the epi curve to estimate the size of the epidemic, which means how many people have been infected since the start of the epidemic. Second, we can stratify the epi curve by age, gender, locations, symptoms, and other characteristics to look at the distribution of cases among these subpopulations. Let us now look at a real example. This is the epi curve of the SARS epidemic in Hong Kong in 2003. The epi curve shows the epidemic peaked 6 weeks after it started on February 15, with a total of 1,755 cases over a period of 15 weeks. By stratifying the epi curve by the residence location of these cases, epidemiologists found that 19% of the cases came from a single residential building, while the other cases were much more geographically scattered in the community. This observation suggested that the mechanism of disease spread for this cluster was very different compared to that for the other cases in the community. Besides epidemic size and case distribution, the shape of the epi curve may help us identify the cause of the epidemic and estimate the incubation period, which is defined as the time from infection to illness onset. If the epidemic is caused by point-source exposure with no or very little person-to-person spread, for example, a food-contamination outbreak, the epi curve typically has one peak with all cases occurring within one incubation period after exposure. For example, this is the hepatitis A epidemic in Pennsylvania that we saw earlier in this session. The epidemic was caused by consumption of contaminated food in a restaurant. Outbreak investigation of this epidemic found that 79% of the cases had eaten in this restaurant between October 3 and 6. These people began to show hepatitis A symptoms 2-8 weeks after eating at this restaurant, which was consistent with the known incubation period of hepatitis A. In contrast with point-source epidemics, an epidemic that is propagated by person-to-person spread may comprise multiple waves with progressively taller peaks, where each peak roughly corresponds to the infections caused by the wave preceding it. The time interval between the peaks of successive waves is known as the serial interval which we will learn more about later this week. For example, this figure shows a small measles epidemic in South Dakota in 1971. The epi curve apparently had three waves that were apart by roughly 10-12 days, which is the average serial interval for measles. Finally, by overlaying the epi curve with events that affect the spread of disease, such as vaccinations and school closure, we can assess the impact of these events on disease transmission. For example, this is the measles epi curve for England between 1950 and 2004. This epi curve shows that the annual number of new cases of measles had decreased by more than 90% since childhood vaccination against measles began in the 1970s. To summarize, in this session, we have discussed some uses of epidemic curves.
