Thursday, December 3, 2015
Zika virus: trying to make sense of what we have
A friend writes the following: "the public health info released by the Brazilian government on zika and microcephaly is actually pretty good. [Still t]here is a fundamental methodological problem with all of this… zika may not be the source of the issue. The rise in microcephaly is correlated with a spike in zika, but it may not be the factor causing the condition [or it may be a] catalyst of some underlying aetiology."
He is certainly right on the mechanism: nobody knows yet. Maybe something else is causing the surge in microcephalia cases, which is the only really big news here: from 147 to 1248, for Brazil, and from 12 to 646, for Pernambuco, as of November 28 and 30). [References here and here].
For the rest, what we have is less compelling: in an interview, the head of the Dengue and Arbovirus society spoke of experimental results showing Zika to damage the brain of mice, but I have not seen—and I have looked for it—the research he was referring to. For its part, the government’s acknowledgement of a link is based on the presence of the Zica virus in the blood and tissue of a single new born baby who died in Ceará. The virus was also found in the amniotic liquid of pregnant women in Paraiba (Pernambuco's northern neighbour state).
The broader nervousness is based on a significant number of interviews of mothers of newborns with microcephalia, a large proportion of whom reported suffering from Zika-like symptoms (which could have been dengue) during the first months of pregnancy, and on a time correlation: the surge in mosquito infections, as proxied by diagnosed dengue cases, began 10 months ago and went on for 20 weeks (week 5 to 25 of 2015, see the graph below), which corresponds to the beginning of the surge in affected newborns’ beginning on week 40-41 (beginning of October) and quickly increasing after that, which is more or less what appears to be happening. Nobody knows precisely if, for how long, and where Zika had been transmitted to people by aedes aegypti, but if the hypothesis is right, the next 15 weeks will tell. Moreover, many of the zika cases identified over the last year may have been misdiagnosed as dengue. Carlos Brito, a member of the Health Ministry's Arbovirus technical committee considers for instance that "of the dengue cases identified in the State (of Pernambuco), 80% were, in fact, zika cases."
Now, 80% may look like a lot, so let's do the math. As of November 16, according to the Health Ministry, there had been 83,601 "probable cases of dengue" in Pernambuco.
Most of the cases of dengue being asymptomatic (in a proportion of between 1.8:1 13:1), this must be considered an absolute floor. But to play "safe," let's stay with the cases identified. Zika cases misdiagnosed as dengue would thus number around 67,000 for the state or about 0.8% of the its population. Assuming that Pernambuco's share of birth is proportional to its share of Brazil's population (4%), it would have had about 120,000 births in 2014 (Brazil's total was about 3 million, or 1.5% of its population). Using this as a proxy for the number of pregnancies, and factoring the fact that the peak infection period is about four months-long, this would imply a floor of about 350 pregnant mothers infected by the Zika virus in Pernambuco over the course of 2015. As noted above, there are already more than 600 cases of microcephalia this year in the state…
In other words: if there is indeed a causal link between zika and microcephalia, the number of zika infections that should be used as a basis of impact assessment has to be much larger than even 80% of the suspected number of dengue cases. In addition, the calculations made above and leading to the projected 350 cases assumes that 100% of pregnant mothers transmit the virus, and that 100% of the foetuses are affected by it. Anything lower than that would massively increase the hypothetical number of infections.