We talk endlessly about making workplaces safer. We write policies, run campaigns, hold toolbox talks. But here's something actually working: serious injuries among New Zealand's oldest workers have been cut in half since 2020.

Workers aged 90 and over recorded 12,225 serious injuries in 2024, down from 28,491 just four years earlier. (Source: Stats NZ / ACC, fatal-serious-injuries) That's a 57% drop in a demographic that, by all logic, should be getting more vulnerable, not less.

This isn't a slow decline. The numbers held steady for years: 28,491 in 2020, 29,499 in 2021. Then something shifted. By 2022, injuries had dropped to 12,597. They've stayed low since: 12,672 in 2023, 12,225 in 2024.

The trend breaks every assumption about ageing and workplace risk. You'd expect older workers to be more fragile, more prone to serious harm. Instead, they're now recording injury numbers lower than any year since 2005, when the workforce looked completely different.

So what happened between 2021 and 2022? COVID changed how and where people worked. Maybe fewer nonagenarians were commuting. Maybe the pandemic forced employers to finally take workplace adaptations seriously. Maybe flexible work arrangements removed the riskiest parts of the job without anyone noticing.

Or maybe something else: the data captures fatal and serious injuries, the kind that end up in ACC claims. Perhaps older workers simply left the workforce entirely during the pandemic, and never came back. The drop in injuries might not mean workplaces got safer. It might mean the oldest workers aren't there anymore.

Either way, the numbers pose an uncomfortable question. If we managed to halve serious injuries for one group in four years, why can't we replicate it? What did we do differently, even accidentally, that made such a dramatic difference?

The answer matters because workplace injuries aren't abstractions. Each number is someone's parent, someone's colleague, someone who got hurt doing their job. If the pandemic taught us something about protecting the most vulnerable workers, we should figure out what it was before we forget.

Right now, we're sitting on evidence that something worked. We just don't know what. And nobody seems to be asking.