How Contagious is BA.5?
The headlines are scary. "Hopes of Covid-19 reprieve fade as BA.5 subvariant takes over" (Wall Street Journal). "The worst variant is here"(CNN). "You are going to get COVID again…and again…and again" (The Atlantic). The stories accompanying these headlines may be nuanced but they're still pessimistic. A common theme is that you're likely to be infected by BA.5, but it probably won't be serious.
In this newsletter, I'll be discussing what the stats tell us about the impact of BA.5 on infection rates. Next week I'll expand the discussion to include hospitalization and mortality stats. (In short, contagiousness this week, dangerousness next week.)
This newsletter will culminate in good news – I view the stats more optimistically than most journalists do – but I need to start with the not-so-cheerful observation that COVID-19 infections are currently on the rise. Specifically, following several months of relatively low rates, the average number of new cases per day in the U.S. has exceeded 100,000 most days since mid-May and, as of this week, is approaching 130,000.
Are reported case rates accurate?
Reported case rates are underestimates, because some people don't get tested, and some people who test positive don't report their results. How much higher are true case rates? Estimates vary from expert to expert, but here's some reassuring news: Major news organizations tend to latch onto experts who estimate the biggest discrepancies. For example, people like Katelyn Jetelina (UT Health Sciences) and Jeffrey Shaman (Columbia) have been quoted recently saying that actual case rates are now roughly 8 to 10 times higher than reported rates. Much less attention is paid to experts, such as those at Imperial College London, whose models project true case rates for the U.S. that are only about 3 times higher than reported ones. Although models like these have larger margins of error, even their worst-case projections are much more reassuring than what you typically see or hear in the news.
"Testdemic" skepticism
Even assuming the smaller estimated discrepancies between reported and actual case rates, skeptics argue that the discrepancies are still too large for us to be sure whether true case rates are increasing or not. Their argument is that it wouldn't be inherently contradictory to find that true case rates stay the same, or even decline, while reported case rates are increasing. Here's a hypothetical example: At time A, there are 100,000 reported cases per day on average but the true number is 300,000. Later, at time B, there are 120,000 reported cases per day on average, but the true number is 290,000. The increase in reported cases would lead us to infer, erroneously, that true case rates are rising. There's nothing implausible about this scenario. Perhaps between time A and time B more people got tested – or more people chose to report positive tests, because they were concerned about a new variant.
The argument I just outlined is called the "casedemic" or "testdemic" view when skeptics refer to increased testing as the cause for increased case rates. These skeptics include unreliable cranks as well as a handful of reputable folks such as John Dee, a former public health statistician at the U.K.'s National Health Service.
Why the testdemic view is wrong
I wish I could accept the testdemic argument, because it would mean that true case rates aren't really increasing. Unfortunately, the argument collapses once it's confronted by additional statistics.
For one thing, reported case rates have risen and fallen independently of testing rates throughout the pandemic (see here and here). In other words, more testing doesn't necessarily yield proportionally more cases. And, although this may sound like a paradox, we're actually more confident that true case rates are increasing than we are about the actual number of cases. The main reason for our confidence is corroborating evidence from other stats. For example, since June, reported case rates have increased in each of over 40 states, and the percentage of tests yielding positive results has increased (nationally, it's just over 16% this week).
In short, although it's hard to say how many cases there are, case rates are clearly increasing. The increase has been attributed to BA.4 and BA.5, not just because of correlations between the spread of these variants and upticks in case rates, but also because BA.4 and BA.5 are better than earlier variants at bypassing the antibodies we obtain through vaccination and/or prior infection. The jargony way to put is that we believe the new variants are causing more infections because we've identified a physical mechanism that accounts for observed correlations between variant proportions and case rates.
The bad news, in brief
On July 12, the CDC reported that during the previous week, 65% of new cases were BA.5 (up from 54% the preceding week, with BA.4 contributing another 16.3%). Given the exponential nature of BA.5's growth, a conservative estimate from existing data is that by early August, something like 90 to 95% of new cases will be BA.5. We also expect case rates to increase accordingly (again, by how much is unclear – I wouldn't trust specific estimates that some experts trot out). Because BA.5 is better than earlier variants at evading key antibodies, vaccinations and prior infections provide less protection than they did against earlier variants.
The good news
Vaccines and prior infections do provide some protection against BA.5. Perhaps a lot. The stats don't warrant very specific definitions of "some" and "a lot" just yet, but here are two trends that seem to be emerging in state-level as well as international data: (a) Having one or two boosters provides slightly more protection against BA.4 and BA.5 than the traditional full course (two Pfizer or Moderna shots, or one J & J). (b) Being fully vaccinated (or vaccinated and boosted) roughly halves the risk of contracting BA.4 or BA.5.
Here's more good news: With respect to personal risk, your behavior will still be the main determinant of whether or not you get infected (or re-infected) in the coming months. To put it crudely, if you're unvaccinated and socialize unmasked with a variety of people in densely crowded spaces, you're going to get COVID-19. If you're vaccinated and boosted, you avoid socializing unless there's good ventilation, and you wear a properly-fitting mask in public at all times, etc., you almost surely won't get infected.
Behaviorally speaking, most of us, including myself, fall somewhere between these two extremes (hopefully, closer to the latter). Now that we're in the third year of the pandemic, it's hard for most of us to maintain earlier levels of vigilance. Meanwhile, you can't know the extent of risk you face in most situations, because you don't know whether the people you encounter are infected, you don't know your own antibody levels, and so on. All we can say for sure is that the more careful you are, the safer you'll be – as has been the case throughout the pandemic.
So, how contagious is BA.5?
New data show that compared to earlier omicron variants, BA.5's reproduction number (the average number of people infected by one person in a susceptible population) is twice as high, while its resistance to antibodies from vaccines or prior infections is three to five times higher. These findings are preliminary – I hope to share more information next week – but even if we trust them, they only tell us that BA.5 is more contagious, all other things equal. This doesn't change the fact that your behavioral choices are much more influential.
So, even though BA.5 is more contagious, you're not necessarily at greater risk of infection than you were with earlier variants, because your personal behavior, along with variables beyond your control, have always had a much stronger impact on risk than the particular variant you're exposed to. Statistical generalizations don't fit individual cases equally consistently.
Next week, I'll provide some updates and discuss the relative mildness of BA.5 infections, along with some surprising twists. Stay tuned – and be safe!