On the internet, I came across a story that a doctor in Liberia who was desperate for Ebola treatments, tried lamivudine, an anti-viral used in hepatitis and HIV on 15 patients infected with Ebola. This caught my attention, as there’s currently no approved treatments for Ebola.
I checked the estimated death rates on the WHO Ebola Factsheet page. They have a table of the number of people infected with Ebola, and the number of subsequent deaths. Although there are lots of references to this table with an estimate of a 50% death rate, I calculate that there were 1590 deaths in 2387 Ebola cases, giving an estimated death rate of 67%.
The statistical analysis from this point is fairly easy – it’s a simple binomial distribution. If the “true” death rate is 67%, then the probability of seeing only 2 deaths or less out of 15 is 3.2 x 10^5, which would usually be represented as p<0.001. In other words, it would be extremely likely that the lamivudine is having a beneficial effect.
Another way of looking at the analysis is using a 2 by 2 table – whether the patients survive or not, and whether they received lamivudine or not. Using the previous Ebola outbreaks data (1590/2387), and carrying out a Fisher’s Exact test to see whether the lamivudine data come from the same underlying distribution, gives p<0.001. The odds ratio for surviving ebola is 13.0. Using the numbers from this outbreak (3865/8033), gives a Fisher’s Exact result of p=0.008 (odds ration 6.0).
If these results are confirmed, then it would appear that this is very strong evidence that lamivudine is helpful in saving lives in the fight against the Ebola virus, and should be investigated in further trials as a matter of urgency.
There has been a paper published in the Lancet online that details the reasons why randomised clinical trials, the usual gold standard for collecting medical evidence on the effectiveness of drugs, is difficult and potentially unethical in the current Ebola situation. Given this report of some effectiveness of lamivudine, I expect there will be other reports of success or failure of the treatment of patients with other drugs. It seems imperative that a global resource is set up to capture this information and ensure that each patient treated with Ebola contributes information and knowledge to the treatment of subsequent infected patients. A global registry of information on treatments given and survival outcome is needed as a matter of urgency.