We can't really measure vaccine efficacy outside of trials the same way it is measured in trials. In trials, you give some people the vaccine and others not, and then compare the two groups. When you start vaccinating the public, you don't have a randomized control group anymore, you have differences in the groups including A) Higher vaccination rates among vulnerable groups, and B) self-selection for vaccination once the vaccine is broadly available.
It's also difficult to measure because maybe some of the people who would not get vaccinated are the same people who would not get tested to confirm that their symptoms are indeed being caused by COVID-19, or perhaps the same people who would engage in riskier behaviors like traveling without masks and meeting in large groups while transmission rates are high.
However, the Associated Press looked specifically at 18,000 people who died in the US in the month of May 2021. They found that only 150, less than 1%, of those who died were fully vaccinated, despite about half of people in the US being vaccinated. It's not possible to directly convert this into an efficacy for preventing death due to the caveats I mentioned above, but it is at least highly suggestive evidence that vaccination is highly effective.