Ebola epidemiology roundup #8

This post covers new materials made available between roughly 18 and 5 December. Travel commitments have stretched me a little thin, hence the belated posting and possible failure to catch all the new scientific articles this week.

A. Charting the Epidemic

The WHO has made some changes to how they present their data and situation reports.  Data are now updated daily on their Ebola data page, but the main SitRep is still put out every Wednesday.  I, and others, have been impressed by the addition of a new “map journal” style that links images and text together very neatly (built through ArcGIS’ ESRI Story map app, apparently).

One important note on data: as of mid-November the WHO is reporting two concurrent data series for each country.  One is the SitRep they have been providing all along; the other output from the “patient database” which should be closer to the source – i.e. what is maintained in hospitals, provincial ministries, etc. My understanding is that the “patient database” is built from a CDC-based contact tracing database, while the SitRep data is from the national MoH, but I’m not certain about this (anyone able to confirm/deny?).  Even without certainty though, more data sources at least allows for some triangulation, hopefully.

A final word on data quality.  Many people have been concerned for some time about underreporting.  Efforts to measure the undercount to date have been limited to a rough estimate made by Meltzer and colleagues based on bed capacity.  So a blog post by Les Roberts, who was working in Sierra Leone with the WHO in October/November, provides important data on this topic. He and colleagues conducted a survey of randomly selected villages, and compared reported case numbers in the national database to local assessments.  The result, a classic good news/bad news situation: bad news, it seems that only one-third of cases are making it to the government records; good news, this is roughly in line with what researchers had thought previously, so we may not need to radically re-assess the level of interventions needed to stamp this epidemic out.  As ever, though, it’s good to have some data on this.

And then one word on data access.  There has been plenty of concern about the lack of publicly-available individual-level data in this epidemic, from the WHO, CDC, national governments and others.  And quite justifiably so.  But Roberts, in a separate post, provides the best explanation of non-data-provision I have seen to date: the overburdening of on-the-ground responders means that data quality is imperfect, and the provision of data to the press/public/researchers without detailed explanation would lead to misleading and potentially harmful news stories, which would lead to increased burdens on said responders to manage the ensuing hue and cry.  So they don’t hand out the data.  Sad, but understandable.

A1. Epidemic trajectory

After reporting a jump of 1000 (largely historical) deaths in Liberia in the last week of November, the WHO has reported that this was due to a clerical error in Liberia, and removed the events. As a result, the epidemic continues at a steady but non-negligible rate in Guinea and Liberia, and continues to burn fiercely in western Sierra Leone.

There were a couple of important papers put out recently by Gerardo Chowell and colleagues highlighting the dynamics of the epidemic.  First, Chowell highlighted the dependance of predictions relating to final epidemic size on the modelling assumptions we use. (Chowell PLoS Currents Outbreaks paper). Specifically, the authors contrast exponential growth models to logarithmic growth models, showing that the latter are more useful for curve-fitting (or as Chowell says, “phenomenological”) models once some kind of control has been achieved in a given setting.  Exponential models are useful for (mass-action or random-mixing) models in situations where growth is unchecked, but not once growth rates change.  As the figure below highlights for Sierra Leone (green line, exponential; black line, logistic):


Second, Chowell highlights that the national cumulative incidence curves we have been seeing for Guinea, Liberia and SL can be broken down into county-level curves, each of which looks far closer to a logistic function than an exponential one.  The combination of multiple curves, each taking off and flattening off on different timelines has allowed the production of a exponential growth curve at the national level. (Chowell arXiv paper). Of course, we can take this several steps further – these county data are the combination of asynchronous village-level curves, and then household-level curves.  All this work highlights, for me, the importance of considering how structured (i.e. non-homogenously-mixed) contacts for Ebola are, and thus how important information on contacts is.

A2. Epidemic parameters

I haven’t seen any new individual-level data recently, but there are a few new pieces of data that might be of interest.

Given the news coverage of the man in India who arrived having recovered from Ebola, but whose semen tested positive for the virus, I will highlight this brief review on evidence for Ebola in semen by Ian Mackay and Katherine Arden.  To my knowledge, no-one has yet shown conclusively that transmission occurred through this channel, but we may not need an “abundance of caution” to believe that prevention efforts are worth expending on this topic.

I wanted to flag a couple of papers this week – one old, one new – considering how animal and human populations interact in the context of Ebola.  In the new article, Wondwossen Gebreyes and colleagues note the Global One Health paradigm, in which human and non-human health are intertwined via zoonotic infections, and the benefits that can arise from communication between public health and veterinary fields.  (Gebreyes paper). In the old article, Jean Paul Gonzalez and colleagues highlight the the possibility of non-pathogenic strains as an explanation for the 15 year gap in human Ebola epidemics in the Congolese basin up to 2000.  Of course this would also fit well with the “jump” of Ebola from central to west Africa in the current epidemic.  (Gonzalez paper). Asymptomatic/low symptom infection remains a topic of limited investigation, despite some evidence of its presence in various settings (see also this Les Roberts post, which notes anecdotal evidence for sub-clinical infection and the absence of whole-village outbreaks – although the latter may relate to the close-contact nature of infection). As Roberts says, maybe this is something we can study after the epidemic is under control.

And on the modelling front, I somehow missed this previously, but Gerardo Chowell and Hiroshi Nishiura published in early October a review of Ebola parameter estimates from previous epidemics and the early days of this one.  A tour-de-force that should be on the desk of all those building models and interventions in this epidemic.  For a briefer, and more biological, take on the same topic, please try this review by Marco Goeijenbier and colleagues.

B. Stopping the Epidemic

One overarching letter of note, following the recent claim by Declan Butler that “models overestimate the epidemic“, is the rebuttal by Caitlin Rivers and many others, that models are useful for far more than predicting case numbers – in particular, for predicting the impact of interventions, and that Butler’s “assertion that models of the Ebola epidemic have failed to
project its course misrepresents their aims”.  I couldn’t agree more.

B1. Containment.

Ebola can typically be passed on in three contexts: in the household, in healthcare settings and at funerals.  Given strong efforts to reduce risk in Ebola treatment centres/units and at funerals – both highly identifiable settings – the essence of prevention for Ebola at present lies in a single act: find infectious individuals living in the community.1 But this act is not simple. Finding infectious individuals requires communication channels for directing individuals to care, staff to bring potential cases into the healthcare system and rapid, accurate tests to determine who is in fact infected. On the first front, I note with concern the apparent lack of connection between UN-proposed social media SOPs and on-the-ground follow-up. On the second, this report of a day-in-the-life of a health surveillance team member in SL is very sobering. On the third, it seems that we are no closer to a test that will conclusively show positive within the first 2-3 days of infection, but there is good news in the form of a 15 minute rapid Ebola test for those who are beyond this “window period”, currently in trials in Guinea. Such a test should significantly reduce the need to quarantine suspected cases, and thus the fear of testing, and so potentially increase the ease of finding infectious persons.

A slightly different approach to avoiding onward transmission with communities is to have infectious individuals identify themselves and self-isolate.  This is an argument discussed in this Nature commentary by Christopher Whitty and colleagues, describing the building of community isolation centres for self-isolation by those who believe that they may be protected.  These would have minimal treatment capacity, but could help protect neighbours and family.  In the absence of sufficient hospital beds, or long distances to these hospitals, or community fears about hospitals as sources of infection, such community centres may have a vital role to play. (Whitty article)

B2. Treatment.

A couple of treatment topics this time around.  First, hydration.  It seems that much consensus exists that those patients who become massively dehydrated are at great risk of death.  While there are differing views on how aggressively to use IV fluids (see recent call for IV, nasogastic feeding tubes, but also concern for healthcare worker safety when using sharp objects around confused patients).  But the basic message is – hydrate early, hydrate often.

Second, as mentioned previously, there are several new treatments in the pipeline.  And there is a nice overview table in a recent roundup article on new treatment options in Science:

However, the article notes that scale-up of such treatment options is likely to be slow, and thus it is not clear how large role they will play in this epidemic.  When it comes to testing these drugs out – as with vaccines, see below – the big question at the moment appears to be an ethical one: can we randomize provision?  On this topic, the NYT has an in-depth debate on the ethical issues that acts as a solid primer.

Oh, and on a more nerdy epidemiologist note, Les Roberts highlighted an important point that made a lot of sense once I had read it: hospitals have been reporting very low fatality rates – sometimes 40% or less – and some have claimed this speaks to the importance of intensive care.  However, as Roberts notes, another good explanation for low CFR is survivor bias: when tests take days to come back, and entry into the hospital requires a positive test, and those who die from Ebola tend to do so relatively fast, those who are admitted are a highly non-random portion of all those infected.  Not happy news, but important things to bear in mind before trusting that current Standard of Care is likely to make a large dent in the epidemic directly (as opposed to indirectly by isolating infectious persons).

B3. Vaccines.

If the big issue for treatment testing is speed of scale-up, it looks like the question for vaccines is cold-chain.  While it is not clear yet quite how cold any vaccine will need to be, it is clear that maintaining low temperatures and transporting vaccine across West Africa is going to be a serious undertaking.

The good news this week was the announcement by Julie Ledgerwood and colleagues that phase I human safety trials for the most-advanced Ebola vaccine (ChAd3; NIAID/GSK) had shown that large doses of the vaccine were able to stimulate an immune response comparable to that seen in vaccinated non-human primates who had successfully fought off the virus. (Ledgerwood paper). The news was entirely positive, however, as noted by Helen Branswell since there were some moderate side-effects (largely fever) and the dose required was significant.  From my point of view, the fever side-effect seems particularly worrying, since vaccinated individuals would then have fever and antibodies – making a non-PCR test for the virus virtually useless.  The second-placed vaccine in the development footrace (rVSV-EBOV; Canadian government/NewLink) got a boost too, with the investment of $30-50 million by Merck.

Despite all these efforts, it is not clear that vaccines are going to play a big role in controlling this epidemic – as opposed to potentially avoiding the next one.  A modelling paper by David Fisman and Ashleigh Tuite this week notes that unless either vaccines reach millions by March 2015, or a vaccine is able to reduce the effective reproductive number well below one, then it won’t add much to the downward trend in the epidemic curve that they expect to see before June of next year. (Fisman paper).  This doesn’t make vaccine development pointless, but does highlight that it may be an investment for tomorrow, not for today.

B4. Social factors/impact.

The wide-ranging impact of Ebola is raised often in discussion, but I really liked this infographic from the MPH@GW program which highlighted quite how far the effects spread:

To which I add only a little more context for some of the figures given:

  1. Economy.  There was a new World Bank report out this week highlights both the direct costs of non-employment, and the indirect costs of lower investment both within the country and from abroad; as well as knock-on effects in neighbouring countries.  As the infographic notes, the high estimate for the overall two-year regional effect is $32.6 billion; the lower bound is a still-staggering $3.6 billion.
  2. Education. Given that no schools have been open in Guinea, SL or Liberia since the summer, there has been discussion about how best to provide education at home, potentially through technology.  While I wonder about its scalability, one programme of note is this crowdsourcing attempt to send loaded tablets to West Africa.  It’s funding until the end of December.  Do others know of other, more low-tech options?

While the impact of Ebola is clearly being felt in several social spheres, there are also efforts to understand why Ebola has been so able to spread in this outbreak.  Clearly mobility has played a large role here, and mobility has many potential causes that might be targeted to reduce infection. The Internal Displacement Monitoring Centre outlines five key displacement causes in the current outbreak: (i) fleeing the virus; (ii) fleeing quarantine; (iii) seeking health care; (iv) forced evictions/stigma; (v) violence/rights violations. I would argue, to grossly generalize, that we can see two broad categories of social determinants (which often overlap) in Distrust and Poverty.

  1. Distrust. As Helen Epstein highlights, this has been an epidemic of rumours (they so often are).  The lack of trust in authorities is deeply rooted, arising from the long history of neglect and abuse by the powerful: stretching from colonial slavery through Americo-Liberian class hegemony and the looming power of Firestone to more recent dictatorship and civil war.  Fear that the government is trying to infect, or use infection to control, you appears to run deep in these countries.
  2. Poverty. As Umaru Fofana highlights, while everyone can be infected, it is the poor who most often are, who are least able to manage their illness, and who are most affected by food prices, quarantines, etc.

The upshot of all this is a population that is hard to reach, hard to persuade to co-operate with public health officials or to access care, and thus hard to protect.  Hopefully the launch of the Ebola Response Anthropology Platform – a joint effort of LSHTM, the Institue for Development Studies and the University of Exeter – will go some way to increasing the quantity of local understanding that goes into programme design and execution. Hopefully.

C. Miscellanea

C1. Risk communication

A couple of quick notes here.  First, evidence appears to be coming in that the recent US quarantine debacle has led to a decline in volunteers to work in West Africa possibly due to the 21-day self-monitoring (or potentially quarantining) that is required on return to the US.  And second, this week I saw a wonderful take-down of the term “abundance of caution”.  Read the whole (short) article, but this was a great sentence from within:

[When] I hear “abundance of caution” being used in a sentence about Ebola, I translate it to “what I am suggesting makes no actual sense but demonstrates my extreme seriousness about fighting the very idea of Ebola”. 

C2. Journalism

I’ve linked to him several times above, but the series of blog posts by Les Roberts of Columbia University are worth reading from end to end.  Phenomenal insight into what’s going on on the ground; especially for those who geek out over epidemiology or who care about humanistic reads on the epidemic.  Hopefully the overlap of these two groups is significant. For a quicker read, this interview with a CDC epidemiologist who had worked in SL was also illuminating.

And while I’ve linked to it above too, Helen Epstein’s article in the New York Review of Books is an important read – as are so many of her pieces, whether you agree with their angle or not.  Her thesis here: historical politics is a key driver of the Liberian epidemic.

In addition, there have been a number of articles I have read as background to the epidemic and how it is changing.  In case you are in need of some bedtime reading.  I’ll note that they skew heavily towards Liberia; I’m not too sure why this is, but thought it worth highlighting.

D. Getting involved

This week, or rather next month, an opportunity to learn from several very big names in the field of Ebola science: LSHTM is running an online course for two weeks starting January 19th.  And it’s free.

As ever, if you have seen something I’ve missed (or links are broken), you can reach me @harlingg.  And as ever, thank you to all those on whose intellectual shoulders (and tweets) I am standing.

There are seven previous posts in this series and a summary of data/research sources.

1 This is not to say that there are not problems with numbers of treatment beds – there clearly are, especially in Sierra Leone – but a clear plan for expansion exists in this area and is being implemented; I’m not so convinced that a similar plan for case finding does. Of course, this is only my opinion.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s