While a dog attack in Timaru makes headlines today, here's the workplace safety story nobody's explaining: New Zealand recorded 312 fatal workplace injuries in 2024, virtually unchanged from 315 in 2023. Three years of consistency at just over 300 deaths per year.

But look at what happened two years earlier. In 2021, this same dataset recorded 61,194 serious injuries. By 2022, that number became 357. Not a 99% reduction in workplace injuries. A complete restructure of how ACC counts what matters.

The contrast tells you everything about how we measure workplace safety in New Zealand. Fatal injuries: steady, countable, undeniable. Serious injuries: definitional, changeable, suddenly incomparable to last decade's data.

Between 2020 and 2021, serious workplace injuries climbed from 59,148 to 61,194. Then the methodology changed. What we called "serious" in 2021 became something else entirely in 2022. The new counting system gave us 357 serious injuries alongside those 357 fatal ones, a ratio that makes no statistical sense unless the definition of "serious" became drastically narrower.

This isn't about data quality. It's about what happens when measurement systems change mid-crisis. For two decades, New Zealand tracked both fatal and serious workplace injuries on comparable terms. You could see trends, spot problems, allocate resources. Then in 2022, the serious injury data became something else, leaving us with a 24-year dataset where the most recent three years can't be compared to the previous 21.

Meanwhile, 312 families lost someone to a workplace death last year. That number we can trust. It hasn't been redefined. A fatal injury in 2024 means the same thing it meant in 2000.

The stability of the fatal injury count since 2022 is itself noteworthy: 357, 315, 312. After decades of fluctuation, we've landed at a plateau just above 300 deaths per year. That's not improvement. That's stasis. Three hundred workers dying annually in a country of five million is a rate we've apparently decided to live with.

What we've lost is the ability to see whether the injuries that don't kill people are getting better or worse. The data that would tell us whether workplace safety is improving for the 99% of injured workers who survive simply isn't comparable anymore. We can count the dead with precision. Everything else is a methodological question mark.

The dog attack that made news today will be investigated, documented, potentially prosecuted. The 312 workplace deaths last year each generated their own investigations, their own ACC claims, their own coronial inquiries. But the data that would tell us whether New Zealand workplaces are becoming safer or more dangerous for everyone else? That story ended in 2021.