Saturday, 23 April 2016

On Zika and microcephaly: causality, consensus and checklists....

Over the past month the World Health Organization (WHO) and then the United States Centers for Disease Control and Prevention (CDC) have associated infection of by Zika virus (ZIKV) with cases of microcephaly and central nervous system disease (M&CD).[1]
The first WHO statement cited a scientific consensus identified at a meeting called to examine the evidence linking ZIKV infection with foetal malformations and neurological disorders.[1] Later in March the WHO pointed to case studies as the origin for that scientific consensus.[2] I've talked about my views on gaps in some of these studies before - for example here and here.

The CDC based their much stronger comments on a review of the literature and its application to address historically robust checklists - Shepard's criteria and the Bradford Hill criteria - that have heretofore proven themselves useful to identify teratogens (causes of embryonic malformation).[3] The CDC authors of this paper note that there is no "smoking gun" at this point and Dr Tom Frieden, CDC Director said it could take years before answers to other questions are found.[12] But not about ZIKV causing microcephaly. That question is answered according to two of the world's leading public health agencies. 

Checking the checklists.

The checklists CDC used were applied to resolve the questions around the cause of an apparent surge of M&CD diagnoses - the so-called congenital Zika syndrome.[4,5] To date this has been almost exclusively occurring in north east Brazil. Some instances have been reported in other countries with current - or past - ZIKV epidemics, but the numbers are small enough for questions to linger about whether these represent part of "normal" M&CD figures, perhaps brought to attention because of the enhanced focus on congenital deformities in 2015 and 2016 or ZIKV-caused disease.

Shepard noted in 1994 that..

"the rare malformation/rare exposure 
or case report method is far easier, 
less expensive, and more common than 
full epidemiologic studies."

He described "Examples of this rare defect/rare exposure "proof" (or better stated strong association)"[sic] including the virus driven congenital rubella. I'm not sure about rare in terms of ZIKV infection tough. While data are near on-existent from Brazil, we've been told that over a million people have been infected with ZIKV.[19] That seems to be a much less stringent use of Shepard's criteria than that of the CDC's "no longer any doubt".[12] Shepard's criteria have gone on to be used in legal definitions [7] which also take a broader view on causality in the study of teratology..

"causation is demonstrated between an 
exposure and an outcome if the outcome 
would not have occurred but for the exposure. 
The but-for test is typically modified by a 
substantial factor test, that is, the 
exposure was a substantial factor in 
bringing about the outcome, or by 
consideration of the exposure as a 
contributory cause"

The other checklist was the tabulated criteria of Bradford Hill, described in 1965 in his occupational medicine-focussed paper, The Environment and Disease: Association or Causation? which sought to relate sickness, injury and conditions of work.[18] I find Bradford Hill's comments in the Experiment section of his criteria interesting as they discuss whether a preventative action in fact reduces the frequency of the event...

"Here the strongest support for the
causation hypothesis may be revealed."

The CDC interpret this wholly in relation to an animal model.[3] Perhaps this meaning has evolved in subsequent analyses that I admittedly haven't read, but I see this criterion differently. To me it is addressing the need to wait and watch for any impact on reduced mosquito breeding and presence either because of seasonal variation or human interventions; a long wait.

In a statement similar to Koch's about his postulates, Bradford Hill stresses that none of these criteria can be used as if they were set in stone and none bring indisputable evidence for or against a cause-and-effect hypothesis. They are intended to..

"help us to make up or mins on the fundamental 
question - is there any other way of 
explaining the set of facts before us, is 
there any other answer equally, or more, 
likely than cause and effect?"

This seems - to me at least - at odds with the CDC's strident use of these criteria to define causality here.

Neither checklist necessarily hits the mark perfectly for what we're seeing in Brazil but, as with Koch's original postulates, these have been "matured" and tuned over time to fit the need of the moment-presence of a common virus infection causing a rare syndrome. 

Is the strong language essential to a response?

But let's back quietly out of the courtroom and return to the world of science, research and causality. We don't have a smoking gun but that has not prevented some important triggers being pulled. Principal among these was that the WHO called a Public Health Emergency of International Concern (PHEIC) on the 1st of February 2016.[8] Well before the need to use stronger language in March 2016, the PHEIC generated recommendations [9] for...
  • To interrupt ZIKV transmission using enhanced surveillance, diagnostics, risk communication, vector control measures, counselling and more
  • research and develop vaccines, therapeutics and diagnostics and increases  relevant services in affected areas
  • provide uptodate advice on travel to affected areas, disinfection of aircraft but do not restrict travel or trade
  • ensure rapid and timely reporting and sharing of information of public health importance relevant to the PHEIC
Obviously those things are expensive. We know from recent experience that the WHO struggled to get the pledged funds they'd requested to mount for an effective Ebola virus disease reposes fast enough and to match the requested spend. Perhaps stronger language is intended to free up the purse strings.[10,11] The main sigh of relief outcome from making such a strong statement by the CDC was...

"Now that we've determined the causal the 
relationship, we can use this information to 
redouble our efforts to prevent Zika, more 
narrowly focus our research and communicate
 even more directly about the risks of Zika."

And herein lies one of my concerns. Narrowed research, by definition, could miss things that have contributed to congenital Zika syndrome. Things that might include...
  • other viruses - rubella and cytomegalovirus are teratogenic viruses that are sometimes sought and not often found but that search can use a hodgepodge of methods. But what about new viruses and new variants of existing viruses? 
  • the impact of chemicals or toxins - the pesticide issue has not gone away [13]
  • the very complex immune responses that to date have mostly been a topic for discussion as a problem for antibody detection in the lab, but may be a part of the process [14,16] although did not seem to play a role in Guillain-Barre syndrome [15]
But all those things may still be included in a narrowed research focus. Those things aside we do know that ZIKV loves to grow in epithelial cell-derived neural stem cells; there is a lot of IgM antibody to ZIKV in babies born with microcephaly [20] and ZIKV has been found in the brain tissues of foetuses with disease.[21]

Devil's advocate - what other things might we consider?

We have not yet addressed whether ZIKV is just as harmless as we used to think it was and whether it is found in these tissues as a passenger and not a pathogen. 

Might it also be in the brain tissues of ZIKV-infected foetuses who do not develop any congenital anomalies? There has been little or no exploration of controls in most papers to date. Mostly - this would be unethical, but there might be other reasons for related tissues to be sampled which could then be leveraged fro important testing. Its important yet missing information.

Excerpted from WHO Zika virus microcephaly
and Guillain-Barre syndrome situation report. [17]
There is also the Colombian elephant in the room. 

This week saw the number of M&CD diagnoses in Colombia double...okay, from 2 to 4... having decreased the week before. This might be normal and part of the 140 annual cases reported in Colombia annually-that's an average of 2.6 per week. 

More time has now passed in Colombia than had in Brazil before Brazil reported its first concerns - and 141 M&CD diagnoses - over ZIKV and M&CD, after it identified local ZIKV transmission. As I've discussed previously, there could be many reasons for that difference - and a rise in cases in Colombia this week may herald that the starting line has been crossed indicating the beginning of some important supporting evidence for ZIKV causing M&CD. Or we might just be seeing normal levels of M&CD that also happen to be ZIKV infected-during an epidemic of ZIKV.

To my mind, the studies used to check off the lists leave some important things unanswered. Patience might have been advisable since the previous announcement of a PHEIC mean that there was no obvious need for the issue to be forced. Or perhaps this is more about the politics of finding a better way to secure the funding needed to address congenital Zika syndrome, and avoid the pitfalls of funding gaps dug during the fight to contain Ebola virus syndrome.

Only time will tell whether such strong WHO and CDC language was needed, helped or even hindered the response and understanding of congenital Zika syndrome. 


  1. WHO Director-General briefs the media on the Zika situation
  2. Zika situation report 31-March 2016
  3. Zika Virus and Birth Defects — Reviewing the Evidence for Causality
  6. "Proof" of Human Teratogenicity
  7. Causation in Teratology-Related Litigation
  18. The Environment and Disease: Association or Causation?