Hacker News Reader

Comment by newbamboo | Hacker News Reader

Wow for 18-49 the only “deaths while hospitalized” were in the vaccinated group. No deaths in the unvaccinated in that age group. Over all age groups paints a different picture entirely, with more deaths for unvaccinated. Glad the broke out the results by age group.

View on HN

I wouldn’t read too much into one figure that may represent 1-2 total observations (there were a grand total of 6 deaths in the vaccinated group, not broken out by age, and 14 in the unvaccinated). It doesn’t even explicitly state if they died due to Covid.

Might read something from the zero unvaccinated deaths, though.

Vaccination status has a ton of confounding variables, including age, immunocompromised status/other comorbidities, immunity from prior infection etc.

This was at one singular hospital with unknown population level immunity. The fact that there were zero deaths in what was likely a small subgroup is not something to base a broad likelihood estimate off of.

There is more to it than that one bar. One example, note (from Table 1)

---

Unvaccinated (n = 657)

Death while hospitalized

¶¶ Denominator excludes 129 patients who remained hospitalized as of January 27, 2022

---

Approximately 20% cannot be assigned either way for this outcome yet.

They read the figure correctly.

Unfortunately the study authors didn’t provide the underlying data that created the figure in a consumable format, so it’s hard to draw conclusions.

Table 1 tells us that there were 6 total Omicron-period deaths while hospitalized in the fully vaccinated group. It is almost certainly a single death. It's conceivable that it's two deaths, but more than that appears mathematically impossible.

Precisely, but what gp said is technically correct.

Honestly it was fairly irresponsible for the study authors not to contextualize that graphic better.

Hey, even if he's wrong, you're not going to convince anyone with personal attacks. I was surprised at the data too.

Unless I'm reading the report wrong, it's from a single hospital. We shouldn't be taking any lessons from such a limited dataset as there are far too many possible confounding factors that would be ameliorated in a larger dataset from more hospitals.

That's a fair point, but the literature here does say what the original comment stated, which makes the whole reading-comprehension thing excessive.

Reply on news.ycombinator.com