As I write this the day before Martin Luther King, Jr., Day I thank Dr. Norville Coots, CEO of Holy Cross Health in Silver Spring, MD, for leading me on a little discovery of Dr. King's words that were new to me. Apparently often misquoted due to lack of a prepared text or transcript, he was speaking at a press conference before and then during a speech at the convention of the Medical Committee for Human Rights in Chicago on March 25, 1966. According to scholar Dr. Charlene Galarneau, the correct quote is:
"Of all the forms of inequality, injustice in health is the most shocking and the most inhuman because it often results in physical death."
Sadly the pandemic reminds us these words are still true.
Good News for Vaccinated Adolescents
This week saw the publication of updated information that the Pfizer vaccine offers significant protection against serious COVID-19 illness in 12- to 17-year-olds. In this case-control test-negative study design (a design commonly used to estimate vaccine effectiveness [VE] for influenza, for example), VE was 98% for both ICU admission and receipt of life-support. VE for hospitalization was 94%. All 7 deaths in the study group were in unvaccinated individuals. One caveat, however, the study time period ended in late October, before omicron.
A Variant Early Warning System?
I was intrigued by a preprint (i.e. not yet peer-reviewed) article describing a structural and machine-learning model for an early warning system (EWS) to predict whether new SARS-CoV-2 variants will develop into significant problems. Let me state 2 things up front: first, I am in no way capable of understanding the mathematics/artificial intelligence principles used to develop this EWS; and second, investigators at BioNTech contributed to its development. Pfizer/BioNTech funds the COVID-19 vaccine trial that I oversee at Children's National, but I don't think this presents a significant bias for me in assessing this article.
All of this starts with the basic genome sequence of a new variant. The sequence data are used to predict binding affinity of the virus to host cells and potential for immune escape (monoclonal, vaccine, and natural infection antibody failures). Machine learning analysis tries to predict what sequences contribute to these properties. All of this is combined into immune escape and fitness (transmissibility) scores for each variant.
The EWS claim to fame is that it predicted exactly what omicron is now doing, in the same week that the omicron sequence was first reported. In fact it did really well in predicting pandemic behavior of 12 of the 13 named variants (alpha, beta, gamma, epsilon, zeta, iota, theta, eta, kappa, lambda, and mu) very early. Alert clinicians and Greek scholars will quickly see that what the EWS missed, at least early on, was the behavior of the delta variant. The study authors attribute some of this delay to the fact that delta likely originated in India where government regulations prohibited export of biologic data and sequencing capabilities in the country were limited. In other words, the early sample size was too low to predict how bad delta would be.
The EWS is intended to be run on a continuous basis, reanalyzing new data as available. No mention of whether the results will be made available to the public; I fear that the involvement of for-profit corporations will mean less access.