New research on Ebola: December 2015

The holidays seem to have cast their pall over my productivity, so this post is emerging three weeks late.  Hopefully January’s will be a little prompter.  As ever, do let me know if I’ve missed/misinterpreted anything important.

Epidemic dynamics

First up, Jean-Paul Chretien and colleagues take on the monumental task of reviewing all 66 Ebola modelling studies from the past 18 months. Chapeau! They highlight variability in methods and approaches and call for best practice guidelines for future outbreaks.

At the national level, Jason T. Ladner and colleagues use genomic analysis of 140 Liberian genomes to show that almost all cases of Ebola in Liberia most-likely all came from a single introduction – probably from Sierra Leone. Given the importance of intense personal contact, models reflecting network structure are often informative.  Anca Radulescu and Joanna Herron investigate the implications of community structures (internal and external, static and dynamic) for key quarantine choices (e.g. focus on breaking local or global ties), and in turn of these choices on epidemic spread. Also at the community level, Mosoka Fallah and colleagues use a stochastic model framework populated with individual-level data on cases in Montserrado county, Liberia – including contact tracing information on a subset – to suggest that the poorest communities were not only the most affected areas, but also most likely to spread infection elsewhere. Moving down to the household level, Ben Adams builds a household-structured model of a population, and shows the importance of larger household sizes in increasing initial growth rates, the basic reproduction number and the household reproduction number (how many within-household infections the average infectious person causes).  If, as seems likely, poorer households are larger households, then the Adams and Fallah papers may be approaching the same issue from different angles.

Patient-level epidemiology

Several papers in December reported on the clinical profile of the epidemic, and how this affected patient outcomes. Oumar Faye and colleagues reviewed viremia data at hospital entry for 699 patients around Conakry up to February 2015, showing that a one-log higher baseline viremia was associated with a 14% reduction in survival probability.  Samuel Crowe and colleagues showed that amongst patients in Bo District, Sierra Leone, time from symptoms to hospital admission was not associated with mortality risk, but viral load at first testing was.  JY Wong and several colleagues reviewed line-list data on all confirmed, probable and suspected Ebola cases in Sierra Leone up to the end of January 2015.  In addition to the typical inverted-u mortality curve associated with age, the authors found no increased mortality risk for women, or for healthcare workers.  Finally, Stefano Petti and colleagues noted, based on a systematic review, that the West African Ebola outbreak showed very different haemorrhagic symptoms to earlier outbreaks – notably a two-thirds drop in bleeding from gums and a tenfold drop in bleeding from the eyes and nose. It is unclear if these changes reflect host, agent or environment (e.g. healthcare) differences.

On the paediatric front, and linked to an earlier suggestion by Benjamin Black and colleagues to focus on maternal health for pregnant women with Ebola ), JM Nelson and colleagues review all published data on live births to Ebola-infected mothers since 1976, showing that all 15 known neonates died with 19 days of birth (although I believe that there is now one longer infant survivor – the last Guinean survivor in the initial outbreak). On a similar topic, Séverine Caluwaerts and colleagues report two cases of pregnant women who recovered from Ebola, but delivered stillborn babies approximately one month post-recovery with EVD in the amniotic fluid. As well as having obstetric implications, these cases suggest yet another reservoir for Ebola post-recovery.

On an operational note, F Vogt and colleagues review MSF’s triage system for admitting suspected Ebola cases in Kailahun to suspect or highly suspect wards in advance of confirmatory tests, based on positive contact history and one other relevant sign/symptom.  They find PPV, NPV, sensitivity and specificity for confirmed cases were all below 76%. Given the high risk of nosocomial infection, the authors recommend single compartments where possible, and the swift implementation of any point-of-care rapid test available. Similarly, Cristina Carias and colleagues evaluated the cost-effectiveness of providing malaria prophylaxis to Ebola case contacts, to avoid malaria and thus false-positive admissions of these contacts to ETUs. Their analysis showed cost savings based just in terms of the cost of admission/bed-stay at the ETUs, although there is also potential benefit of avoiding infection with Ebola, and of sending those with malaria (especially children) to ETUs unable to manage malaria treatment (as highlighted by an article by Gillian McKay on the ethical dilemmas of field triage for malaria/Ebola).

Vaccine and treatment trials

A common message as the West Africa epidemic wanes is that we do not know all that much more about what works in terms of products than we did two years ago. Jon Cohen and Martin Enserink provide two succinct summaries [online article, Magazine version] of the 13 clinical treatment and vaccine trials run to date, noting that only the Guinea Ebola, ca suffit! Ring vaccination trial has shown a clear benefit and had been published by the end of 2015.  Anton Camacho and colleagues provide a model that shows one reason for this dearth of evidence, showing that trials begun in the context of a waning epidemic – in this case Forécariah prefecture in Guinea, beginning in mid-2015 and enrolling 100,000 – are often doomed to failure. One reason for the delay in rolling out trials was uncertainty about the correct way to balance various ethical criteria. Francis Kombe and colleagues discuss the ethical considerations and deliberations that arose in planning a convalescent plasma trial, highlighting the need to provide access even to those typically considered vulnerable and excluded from trials (children; pregnant women), and to provide supportive services to both donors and recipients.


In contrast to treatments, there appears to have been some progress in developing rapid, point-of-care Ebola tests. Pierre Nouvellet and colleagues review rapid tests for Ebola already available and under development, and use mathematical models to suggest that the earlier isolation they might have allowed could have reduced case numbers by a substantial amount. Meanwhile, Petrus Jansen van Vuren and colleagues, and Benjamin Pinsky and colleagues provide lab evidence of Cepheid’s GeneXpert Ebola PCR test working within 90 minutes. At a conference in late October 2015, Amanda Semper and colleagues showed 100% sensitivity/specificity for the same test in field laboratory setting in West Africa.


Less this month on behavioural interventions. Umberto Pellecchia and colleagues used qualitative interviews and discussions to flag the importance of local engagement in epidemic management.  Their work highlighted tensions within communities in Liberia as they negotiated the Ebola outbreak, notably the economic strains of forced quarantines and (bribable) cremation teams, and the effectiveness of local ownership over behavioural interventions and enforcement. On a different tack, Jillian Sacks and colleagues described the process of developing, rolling out and troubleshooting an mHealth solution for electronic data collection by contact tracers in Guinea.


As the epidemic splutters out, increasing focus is turning to the health sequelae of infection. In an important piece for planning for possible future outbreaks, Rosalind Eggo and colleagues combined temporal EVD survivor data with evidence that virus can remain in semen for up to nine months for some men, to estimate how the number of potentially-infectious men might evolve over the next few months.  The authors show that the numbers are low and likely to have fallen to a handful by the end of 2015. Malcolm Hugo and colleagues highlighted the need for ongoing psychological assessment and support for Ebola survivors.  Amongst 74 discharged individuals, experiences of death, family member loss and arousal reactions were common; one-third faced stigma in their communities and one-fifth pre-PTSD-type reactions one month post-discharge. John Mattia and colleagues reviewed early data (March/April 2015) from the Port Loko Ebola survivors clinic, finding joint pain (76%) and novel eye symptoms (60%) to be very common; the latter were highly associated with acute Ebola viral load.


New research on Ebola: November 2015

A summary of research on Ebola newly published in November 2015. I’ve tried to make this round-up flow a little better.  Hopefully the dots are a bit more joined-up.

Patient-level Epidemiology

The association of Ebola infection and mortality with age, viral load and other risk factors.

Several researchers focus on the age structure of this Ebola outbreak. Jin Li and colleagues describe the clinical outcomes of 288 confirmed Ebola patients at Jui hospital from October 2014 to March 2015. The authors again highlight the tight association between viral load and mortality, as well as increasing mortality for those aged over 18 and again over 40.  This pattern was also reported by Marc-Antoine de La Vega and colleaguesAlicia Rosello et al. review all seven past outbreaks of Ebola in the Democratic Republic of the Congo using line-list data. The authors show incident cases are overrepresented amongst those aged 25-65 – perhaps reflecting nosocomial and burial-based transmission routes – with higher mortality amongst the under 5s and over 15s. More severe epidemics appear to have been controlled faster. In a letter, Leslie Libow highlights the relatively low incidence rate of Ebola amongst under 18s in both the 2014 West African and 1976 Zaire outbreaks; Libow focuses on age-specific biological risk factors, however for me this highlights once again the importance of involvement in caregiving as a risk factor for Ebola infection.

Two papers in the same journal delve into the association between EVD viral load and patient outcomes. Marc-Antoine de La Vega and colleagues show that amongst the 632 fully-documented cases of Ebola seen at the MSF hospital in Kailahun between July and November 2014, mean initial viremia of survivors was over 100 times lower than that of non-survivors, and mean viral load fell by a factor of 10 from August onwards, at the same time as Ebola-specific antibodies became more common in the population. Simone Lamini and colleagues provide additional evidence from the Emergency ETC in Freetown, moving beyond initial viral load to show levels decline rapidly 4-5 days after symptom onset in survivors, but remain substantially higher amongst those who subsequently die. Finally, Julii Brainard and colleagues conduct a systematic review of filovirus risk factors, and highlight that that only one-third of those who had direct physical contact with an infected household-member became infected; they show low risks for several other behaviours, reminding us that these diseases are thankfully relatively difficult to transmit in many circumstances.

And in a case study, Angela Dunn and colleagues describe how the admission of two individuals infected with Ebola into general medicine wards led to seven secondary cases due to limited use of PPE – highlighting the importance of careful screening and precautionary use of PPE during Ebola outbreaks.


Epidemic dynamics

How disease spreads through populations

There are two new, national-level descriptive studies of the evolution of the epidemic. Adriana Rico and colleagues provided a detailed description of the evolution of the Guinea epidemic in and around Conakry up to March 2015, exploring possible mobility and healthcare-related explanations for the continuation of transmission in this area even after infections had ended in much of the rest of the country. Tolbert Nyenswah et al. provide an overview of the Liberian epidemic, its control and its implications, highlighting the importance of a centralized management system at the national level.

Researchers are increasingly engaging with the networked nature of Ebola spread, both theoretically and empirically. Mark Burch and colleagues built a Bayesian model for the co-evolution of outbreaks and contact networks, and applied it to the 2014 DRC Ebola outbreak. Within Sierra Leone, Wan Yang and colleagues build an adjusted “gravity” model – which assumes closer, denser districts had more movement between one-another – to infer how infections passed between the 14 districts of the country. It will be interesting to see how these results compare to phylogenetic connections once all the samples are in. Marco Ajelli et al. reconstruct the transmission chain for 49 Ebola cases in one Sierra Leone district – Pujehun – by merging field and hospital notes with HCW and community interviews. The authors generate a wealth of empirical epidemiological data and highlight the role of high detection, isolation and rapid burial in controlling the local outbreak.

The effectiveness of interventions

Linked to the work on viremia (above), two more papers address the importance of detecting and isolating cases early – preferably pre-symptoms – to control Ebola epidemics. Diego Chowell and colleagues model the benefits of detecting pre-symptomatic individuals (e.g. systematic PCR testing of case-contacts), while GF Webb and CJ Browne provide similar evidence focused on very early symptomatic cases.

Contact tracers are central to early case identification, and Ashley Greiner and colleagues interviewed Ebola contact tracers in six affected West African nations in late 2014 to understand how they succeeded in following transmission chains. The article highlights many barriers (notably fear, stigma and community mis-perceptions) and some useful strategies for combating them.

Philippe Calain and Marc Poncin consider the ethical dimensions of interventions, exploring the moral basis for quarantine and isolation in the context of Ebola. The authors highlight that, even given evidence of effectiveness, such measures may be morally questionable and potentially counterproductive, if individuals and communities are coerced into compliance.  Umberto Pellicchia and colleagues at MSF provide empirical evidence for exactly such counterproductive effects: showing how top-down quarantine procedures and enforced cremations in Liberia generated stigmatization of – and resistance by – poor Ebola-affected communities, exacerbating existing social inequalities.

Health communication – including messages to induce cooperation – was central to combatting the epidemic. Mauricio Duque-Arrubla outlines in his Masters thesis how messaging in Sierra Leone shifted with phases of the epidemic, moving from top-down to bottom-up as the need shifted from nationwide action to local implementation. The author highlights the need for constant re-evaluation and engagement with community, community leaders and government via a mix of strategies to maximize effectiveness.  Joachim Allgaier and Anna Lydia Svalastog frame the spread of health messages as being in competition with the spread of disease, and of unreliable/harmful information. The authors note that combination prevention efforts include the management of all of these spreading processes.

Within Ebola treatment centres, Adam Potter and colleagues pinpoint how personal protective equipment (PPE) led to heat strain, and provide practical guidance on work/rest timings given specific types of PPE and temperature/humidity.

And finally, Adam Kucharski and colleagues link together networks, interventions and vaccination programmes – in simulating an Ebola outbreak over a network-structured population using observed contact data. The authors show that while ring vaccination can help control an epidemic in concert with other interventions (i.e. behaviour change, active case finding, isolation), such a vaccination method relies on strong knowledge of existing transmission chains. Ebola vaccination strategies therefore need to take account of the epidemic and response context in determining the most efficacious and efficient approach.



No less important than the papers covered above, but my powers of synthesis are insufficient to fit these into another catgegory.

  • Tara Smith outlines the strengths of using the West Africa Ebola outbreak to teach a cohesive and comprehensive course on global health.
  • Marc-Antoine de La Vega et al. review the evolution of the Ebola virus over the past 40 years, noting a relatively stable evolution and a lack rapid change over time.
  • P Loubet and colleagues show how the number of patients attending two HIV clinics in Monrovia dropped – and the level of follow-up delays rose – as the epidemic raged in 2014, highlighting the impact of Ebola on yet-another aspect of the healthcare system.
  • Kai-Lit Phua considers how risk factors acting at many different levels (host, agent, physical , health policy/funding and social/cultural environments) combined to increase the difficulty of turning the epidemic tide, and how a combined approach to addressing such factors might improve the chances of doing so – both now and in future epidemics.
  • A need for WHO and the world health system to reform has been highlighted by many in light of the Ebola epidemic. The Harvard-LSHTM Independent Panel on the Global Response to Ebola reported this month, and provided wide-ranging recommendations on what is needed to ensure a timely, joined-up response to future health crises; the hard part is bringing together those with power to make these changes, and persuading them to do so.   On the research policy front, the WHO and many major journal groups put out a statement on standards for sharing data in health crises – a common concern during the epidemic has been unshared data at the epidemiological and molecular levels.


Accelerated biological aging in HIV-infected individuals in South Africa

This caught my eye because it brings together things I’ve seen in two very different spheres of research – stress in social determinants research and HIV in Cape Town (I also know some of the authors – it’s always social networks in the end…).

No in-depth analysis from me here, but it’s really interesting how widely teleomere research is going and how it is being used to quantify the impact of a wide range of exposures on the broadest of health outcome measures (i.e. ‘biological aging’).  I don’t know enough details about teleomeres to know how we meaningfully convert length to life expectancy (or quality of life), but it both bothers me a little that we may be focusing on something because it is measurable, and intrigues me that we may be able show previously only-hypothesised stress effects of HIV, and in particular antiretroviral treatment, on the body.

Maybe the most interesting line for me, showing my personal interests I guess, was:

Socio-economic factors were not associated with biological aging in HIV-infected participants

My prior would have been that this is in contrast to the general population, however a recent review suggests the evidence for a link between SES and teleomere length is mixed. And in this study, SES was unrelated for both those with and without HIV.

Still lots of interesting work out there to do on this…

HIV and Teleomeres

Why does this blog exist? (in a shallow sense)

One of the main impetuses (aside: why can’t it be impeti?) for my efforts to set up this blog, was that when I googled “social epidemiology” or “social epidemiology blog“, I was severely underwhelmed by what I found. And don’t even get me started on the wikipedia page.  Even looking for “‘social determinants of health’ blog” brings up individual entries, not blogs devoted to the subject.

This was a shame, if not entirely a surprise.  A couple of years back, a group of students in the Department of Society, Human Development and Health at HSPH thought it would be a great idea to set up a group blog (à la scatterplot, among others).  This effort – under the moniker societyandhealth – was sadly rather shortlived.  I think this was partly due to it being started shortly before summer break and partly due to each of us having a slightly different idea what the site might do.

In large part, I think that the lack of material or coherence on the web reflects the breadth of the field and perhaps uncertainty regarding its epistemology.  I have heard it argued that public health in general, and social epidemiology must be a normative science and an activist discipline: if we find things that are causing ill-health, a failure to act on these through communication and policy change is little short of criminal.  Such arguments resonate with the efforts of Mayor Bloomberg.

On the other hand, there is still only a limited amount of actual evidence for health being associated with social conditions – certainly compared to more traditional risk factors such as behaviours and environmental exposures.  Thus there is a more positive science angle that says we need to run more, and better, studies to figure out which exposures cause which outcomes and through which mediating pathways.

It is also notable that social epidemiology aims to shift the discussion regarding causation in public health by changing what is a valid cause of health. (On which topic, if you haven’t heard of this book, you should get thee to a bookshop asap).  It is therefore an aggressive force for epistemological change.  And this is something I love about the field.  It does, however, often make it hard to nail down what is covered within its remit, since that keeps changing too.

All of which makes for a very interesting field, but not an easy one to follow online. Which could probably also be said of this blog post.  My point, however, is that in the absence of a blog devoted to social epidemiology and the social determinants of health, I thought that I might as well cast off from the shore and see where the currents take me.  I think we’re still within sight of the point of embarkation, but hopefully soon there’ll be new lands to discover, and maybe even pirates to fight.   But enough with the extended meataphor.  For now.

P.S.  If I’ve abused the term epistemology, my apologies.  In philosophy as is so much else, a little vocabulary is a very dangerous thing.

Inequality more relevant than Poverty for HIV in Africa

Now here is an article was sufficiently interesting to raise me from my publishing lethargy and get me to WordPress.  The research focuses on socioeconomic correlates of HIV in Africa.  It follows from past research suggesting that wealth/poverty is not driving the epidemic, as rates are (at least initially) higher amongst well-off persons in this continent.

For her dissertation, on which this paper is based, Dr Fox (no, not that one) dug into the DHS datasets from across sub-Saharan Africa and built a mammoth of a multilevel model that allowed for individual, regional and national level income wealth and wealth inequality.

There are many interesting findings in the paper, but I wanted to highlight two.  First, the headline result is that regional inequality predicts HIV infection net of personal wealth, although it is not strongly mediated by circumcision or sexual behaviour.  This is of great interest to me as someone with a working interest in the impact and mechanisms of inequality on sexually transmitted diseases, although it does beg the question of what the causal mechanisms (or confounders) might be.

Perhaps more interesting, however, is found in the nuance seen when stratifying by regional wealth:

The findings also reveal a paradox that supports a dynamic interpretation of epidemic trends: in wealthier regions/countries, individuals with less wealth were more likely to be infected with HIV, whereas in poorer regions/countries, individuals with more wealth were more likely to be infected with HIV.

I read this as supporting the idea of connectivity being key to getting one into the sexual networks that put one at risk, and the idea that having resources (e.g. wealth, knowledge) once in this network is key to reducing the risk (either by leaving the network or protecting yourself within it).  Note that ‘wealthier regions’ tends to mean urban, often capital.

There is a lot more work to be done on the dynamics of the relationship between socioeconomic factors and HIV infection, but this is an important step along the road.

Citation: Fox, AM. The HIV-poverty thesis re-examined: Poverty, wealth or inequality as a social determinant of HIV infection in sub-Saharan Africa? Journal of Biosocial Sciences, epub ahead of print. doi:10.1017/S0021932011000745 . Link.

Potentially interesting papers: Week 51 2011

I had expected this week to be quiet, what with the upcoming festive season and all, but I was pleasantly surprised to find a few titbits to look over and decide ‘yes, I should definitely read this some time soon’.   Therefore, without further ado,  I present:

1.  A simple, yet elegant, study of the interplay of individual and national income in determining who volunteers to be tested for HIV (this paper has been around for a year or so it seems).  It looks like the positive income gradient seen within countries (more income = more testing) is less pronounced in richer countries.  The analysis uses individual income in within-country quintiles, which makes interpretation difficult.  As the authors suggest, one way to look at things is to see this as evidence for the need to focus on poorer people in poor countries.  Another angle would be to see this as evidence that there may be a positive but decreasing slope to the overall relationship internationally – i.e. there is a positive relationship up to some income threshold internationally, and then the relationship levels out.  Which would suggest that focus needs to be on poor people everywhere.  Either way, food for thought.

2. A recent paper on drinking and STI acquisition in the US has me thinking about the causal relationships around sex and drugs (and rock and roll) again.  The paper finds various alcohol-related activities in adolescence linked to various risky behaviours (non-condom use) and self-reported STIs in early adulthood.   The temporality of this relationship is clear, but the almost syndemic nature of the package of behaviours that goes with alcohol and sex – can I call it a lifestyle without implying too much or too little agency? – makes me loath to attribute any direct causal effect from the alcohol to the sex/STI.  Not that the authors of this piece are claiming this, but I remain very unclear on how we might make an impact on STIs through interventions on adolescent behavioural patterns.  Of course, this may just be me banging my ‘social epi’ drum, so you may wish to let me bang it in peace.

3. Here’s something from another field that interests me greatly, but I have never had the time to get into deeply enough.  This is a nice conceptual effort to link human and pathogen behaviour together in an Ordinary Differential Equations setup to consider the mitigating/conflagratory effects of the former on the later in studies of dynamic disease spread.  It doesn’t look too technical for the non-mathematicians amongst us, so might be worth a skim for thought-provocation, if nothing else.