UPDATE: The latest numbers do not indicate a weakening of the outbreak. (See list of new cases below. Several graphs have been updated as well)
UPDATE: More detailed discussion of transmission of Ebola
UPDATE: I note with sadness the death of my neighbor (though I did not know him) of Patrick Sawyer, of the Liberian Ministry of Finance, who died in Nigeria of Ebola contracted in Liberia. He was on his way home to Minnesota at the time.
There is an Ebola Outbreak currently underway in several West African countries, mainly Sierra Leone, Liberia, and Guinea. This is the most extreme known Ebola outbreak to date. The first known outbreak of this virus was in 1976, and there have been several instances since then ranging from single cases (which by definition are not outbreaks) to 425 confirmed cases (with 224 deaths in that instance, in Uganda, 2000-2001). The current outbreak is significantly larger with about double that number or more.
There is some confusion in the press (most notably in CNN) about the nature of Ebola and perhaps about some of the details of this outbreak. Here, I want to provide some basic data to help clear some of this up. CNN reported at one point that you can get Ebola only after a person is symptomatic, and (in the same story) at any time a person is infected even if they are not symptomatic. It is probably the case that as long as Ebola is in a person’s system, they can spread it. It is only spread through contact with bodily fluids, but that is not such a hard thing to do; mucus membranes can absorb the virus, as well as cuts or other injuries. It is probably sexually transmittable. It does not appear to be airborne, but bodily fluids that are in or on needles, hospital equipment, etc. can carry the disease to another person.
Another issue with reporting is the difference between suspected cases, likely cases, and confirmed cases. Even within the health community these numbers are all over the place because they are always changing as cases go from suspected to either eliminated or confirmed. Wikipedia and CNN both recently stated that there have been 1,093 human cases with 660 deaths so far. However, this includes both confirmed and suspected cases. There is a good chance that the total number of cases is in fact close to this, but the data are of lower than ideal quality. If we want to look at mortality rates and changes over time in this outbreak, it is better to look at a smaller subset of the better confirmed data. That’s what I’ll do here. But, when looking at the numbers, keep in mind that although most of the data I show in graphics below show several hundred fewer cases than being widely reported, the actual number of people affected by the disease over the last four months or so is probably not only higher than the cleaned up data set but also, likely higher than the reported 1,093. Furthermore, the data I’m using here only go up to July 24th.
One of the most egregious errors at CNN is the frequent statement that Ebola has a 90% death rate, but that the current outbreak has a much lower death rate. This is rubbish. Ebola simply does not have a 90% mortality rate, and stating that the current outbreak is much lower in mortality gives the impression that this particular form of Ebola, or this particular outbreak of the disease, is somehow not as bad as usual. In fact, this outbreak is worse than any previous outbreak for several reasons. For one, it is larger. Also, it seems to be not burning itself out like most previous outbreaks did. Ebola outbreaks in the past have tended to happen in relatively isolated areas, because the population that includes victims is in close proximity to the presumed reservoir of the disease (probably fruit bats) and interacts directly with the intermediate hosts (eg. primates or other mammals that picked up the disease from fruit bats*). But there is plenty of reservoir and intermediate reservoir in some areas near major population areas. Apparently, Ebola broke into the human population in one or more areas of high population density, and this density together with relatively high mobility is allowing the disease to persist.
The following graphs are based on data I collected from the WHO reports. For March, I use only very likely cases, for April through July, I use only confirmed cases (not available for March). And, July does not include the last week for that month (a few more days have been added to this information bringing us to July 23rd, added on July 30th).
The following charts show the total number of cumulative cases conservatively estimated, and total number of cumulative deaths. When the outbreak starts to weaken, we would see a leveling off, but that is not indicated here (UPDATED).
The last several reports from WHO (including confirmed, probable, and suspect cases) are as follows:
July 21st through July 23rd: 108 NEW
July 18th through July 20th: 45
July 15th through July 17th: 67
July 13th through July 14th: 18
July 08th through July 12th: 85
July 06th through July 08th: 44
July 03rd through July 06th: 50
The exact time spans for each of these reports may not be the same, but I believe the number of cases do not overlap; each listing is a separate set of new cases. Clearly, for the last several days of available information, there is variation in, but no let up in, the number of new cases.
Looking at the number of new cases reported (and for the most part confirmed) and the number of deaths (the same data as used to make the cumulative graphic above, but by month) we have this (Updated):
Keep in mind that the data for July are short by several days.
Another area where MSM, and for that matter, Wikipedia, could do a better job is in reporting the mortality rate for the disease. Wikipedia states that “The disease has a high death rate: often between 50% and 90%.” This is misleading because the outbreaks with 90% mortality rates are not typical, and the statement seems to be based on a set of data that includes a lot of data points one would do better to ignore. I assume CNN is taking this information (from Wikipedia or elsewhere, which perhaps repeats the Wikipedia claim) and exaggerating slightly when they say that Ebola normally has a 90% mortality rate.
The Ebola affecting people right now in Africa is one of a handful of similar viruses known over a larger geographical range. Some of the deaths found in the larger data set of all known outbreaks are from individuals who showed up in a hospital nowhere near where they got the disease, or laboratory workers. The best way to estimate mortality rates related to the present outbreak in West Africa is to take only field cases — actual outbreaks in normal populations — in Africa only, and to not count “outbreaks” that are not outbreaks because only one person is in the sample.
The following chart compares mortality rates for all of the “outbreaks” listed in Wikipedia page regardless of size of sample, geography, or circumstances, with only those that are African Ebola in the field. The latter set also excludes the present outbreak.
Notice that the clean data are bimodal; some outbreaks have mortality rates between 0 and 90%, others between 40 and 60%, and not much in between. Also, there are several in the all-data set that have a mortality rate of zero. This bimodality is not necessarily a persistent statistical characteristic of the sample; I could make it go away by changing the histogram intervals. But it is a convenient place to break the sample into “more severe” and “less severe” outbreaks.
The zero cases in the full data set are all odd cases. Seven are not in Africa and include in some cases lab workers or animal handlers, and most are not African (Zaire type) Ebola. One is a scientist who caught the disease from doing a necropsy on a chimp in the Ivory Coast, examining an outbreak among the non-human primates there. There is one case where the fatality rate is 100%, but this was only one person, and the case was discovered post hoc. We don’t know if anyone else there had the disease. A 90% mortality rate occurred in a remote part of the Congo, with 143 people affected including health care workers. It appears that several individuals contracted the disease butchering non-human primates. This occurred during suboptimal conditions during the Second Congo War. One case of 88% mortality occurred early on in the history of the disease (the second known outbreak) also under very poor conditions. Although the data are too sparse to draw firm conclusions, it seems that the more severe outbreaks in terms of mortality tend to have occurred under more difficult conditions.
Ebola probably has a very high mortality rate when an infected person gets no medical treatment, and a mortality rate closer to 50% when a person quickly gets medical attention. There is no cure, but when a patient is given IV solutions in a hospital setting the chance of survival goes way up. This might suggest that smaller outbreaks that run their course before intervention would have a higher mortality rate, or that the mortality rate would be higher near the beginning of the event. Similarly, one might expect mortality rates to be higher in the early years of Ebola than later, as treatment methods developed.
There is some, but not much, evidence for these effects.
The following chart shows mortality over size of the outbreak, using only the cleaned up data set:
There is not a relationship between size of outbreak and mortality rate.
This chart shows the mortality rate over time, for the cleaned up data:
This seems to show that lower mortality has been achieved in recent outbreaks, though the statistical significance of this is non existent. But, the data set is small. The above chart also indicates the average morality rate across all of these events, which is 64% across 18 outbreaks. Not “usually 90%” as CNN states.
The following chart shows the approximate mortality rate for the current outbreak by month.
This is calculated from confirmed or highly likely cases. This is not a true mortality rate because people who got the disease in one month may have died the next month. But it does give an approximate indication of change over time in rates. The rate at the beginning of the outbreak could be high, or this large percentage could be a function of how cases were counted. In any event, this is an indication of higher mortality rates calculated at the beginning of an outbreak, and there are likely two reasons for that high rate, either or both applying in a particular case.
The current outbreak is settling in at about 60% mortality rate. There is no indication from WHO that the epidemic is slowing down.
UPDATE: Is Ebola Only Transmitted By Symptomatic Individuals?
According to the usual sources (WHO and CDC for example) the following is probably true. When someone gets Ebola, typically, after a while they get sick. This means they show symptoms. If they did not show symptoms they would not be “sick” even if the virus was in them and even if the virus is multiplying in them. Presumably people are infected with a sufficient number of viroids that they become a host for the disease, the virus starts to multiply above some level that makes the person sick, and we can say at that point that they “have Ebola.” This is when the infected person is able to transmit the disease to others through bodily fluids that might come into contact with wounds or mucous surfaces in the downstream patient.
This is what the WHO and CDC literature on Ebola says, and this has lead bloggers and news outlets to state incorrectly that Ebola is only transmitted to others when the person shows symptoms. Unfortunately this is not true in one or possibly two ways.
It appears that people who have had Ebola, live, and get “better” (i.e., their symptoms go away) can still carry Ebola for a period of time, and in this state, they can still transmit it. What has probably happened is their immune system has started to fight the virus enough that it is attenuated in its effects, but it isn’t’ entirely gone yet. Medical personnel like to send someone home only after the virus has cleared. Even so, men who are supposedly virus free by that standard, when sent home after surviving Ebola, are told to avoid sex for several weeks because there is still the possibility of sexual transmission of the virus. Meaning, of course, that the virus is still knocking around in some individuals at this point, and still transmittable. It is not clear how likely that is to happen.
This is very important. Most people would interpret “only transmitted by people showing symptoms” (or words to that effect) when they read it in a news outlet as meaning – well, as meaning exactly what it says. But post-symptomatic patients may still transmit the disease.
Is it possible that pre-symptomatic people can transmit the disease too? Personally I think it is possible even if it is generally unlikely. In a disease that kills over half of those who get it, “unlikely” is not comforting. A small percentage of people who never seemed to have had Ebola, or to have been exposed to it, seem to have antibodies that would probably only develop if exposed to Ebola. Some studies have shown immune reactions to Ebola in those known to have been exposed but also known to not have gotten sick. This is important but not shocking. There are a number of different situations where a normally icky disease that makes you really sick seems to have infected a certain percentage of people asymptomatically. Are these people carriers at some point, i.e., people who have the virus in them, can transmit it to others, but don’t get sick themselves? There is no evidence to suggest that this is the case with Ebola, but the total number of known human cases of Ebola is very small and the conditions for study of the disease in the field very poor, so the safest thing to conclude is that we simply don’t know, but it is also reasonable to say that asymptomatic carriers don’t seem to be a problem, or this would likely be noticed.
The important point here is that there is not a perfect correspondence to being infected and having symptoms, and transmission post-treatment and survival is possible and of sufficient concern that WHO and CDC assume it, so it would be unwise to make too many assumptions about pre-symptomatic transmission.
Imagine you are a health care person addressing an Ebola epidemic. An jet liner flies over a very long flight, say 10 hours long, on Monday. On Friday five people who were on the plane come down with Ebola and you have reason to believe that they were all infected before the flight. Would you determine that it was impossible for the nearly 300 people stuck on a tube with five pre-symptomatic Ebola carriers to become infected? No. You would watch those people and test them.
An additional point to underscore; it has been touched on but not emphasized. The symptoms of Ebola include vomiting and bleeding from places one normally does not bleed. Put another way, the symptoms of Ebola include spreading around bodily fluids. This is often how diseases spread. The disease results in a bodily reaction that spreads the disease (look up “virulence”). So, no matter what, the most likely transmission by far is during the period of symptomatic reaction to the disease, or for some time after death while the virus is still viable. That does not mean that there is no transmission before or after, but it does mean that the most obvious transmission will be from symptomatic patients or recently diseased symptomatic patients.
* Fruit bats will drop fragments, or stones, of fruit they feed on, sometimes in discrete piles. It is almost impossible to imagine a ground dwelling frugivore, such as a chimp or a duiker, not stopping to munch on this detritus. Since Ebola is spread through bodily fluid contact and can be spread via mucous membranes, and fruit bat spit counts as a bodily fluid, I’m personally of the opinion that this is how Ebola may often transfer from its natural reservoir, where it seems to exist without harm, to other animals. Of course, I figured this out after having discovered and handled several such piles of fruit bad wadge.