The provisional Workplace fatal injuries in Great Britain, 2020 data were released by the HSE last week showing a decline in workplace fatalities from 149 in 2018/19 to 111 in 2019/20.
While every one of the 111 fatalities is a tragic loss, the decrease in fatalities from 149, and indeed from the previous 5-year average (an annual average of 143 fatalities between 2014-19), is significant and worthy of highlight. Have we finally broken through the statistical barrier or are we seeing a reduction because of reduced workplace activity due to Covid-19?
The words ‘provisional’ and ‘significant’ in the previous paragraphs are important. We’ll come back to those in a moment.
Firstly, let’s look at the general trend. Was the decrease down to a single sector? Actually, no, as the following HSE chart shows:
Indeed, if we look at sector fatalities for 2019/20 versus 2018/2019 (rather than the 5 year average above) then every sector sector bar construction showed a drop in fatalities year on year. Some sectors had major drops, e.g. manufacturing from 26 to 15 and wholesale/retail from 19 to 6.
Leaving aside construction (which had a c30% increase from 31 to 40 fatalities), the decrease in every other sector is extremely encouraging, But, what role might Covid-19 have played?
At what point did the economy slow? Lockdown started on March 23rd but we know that GDP in March was down 5.8%. More than a week of lost production but not a month’s worth? Plus, essential construction continued. Was there a slowdown in February? Not really, GDP grew by 1% in February.
It’s difficult to be precise but 111 fatalities is a decrease of 22% over the 5-year average of 143. That’s a lot more than the proportion of production lost due to lockdown (to the end of March) suggesting cause for optimism that we are breaking through the statistical barrier that is Great Britain worker fatality rates.
Which brings us to the two words, ‘provisional’ and ‘significant’.
The figures are provisional and it may be July 2021 before we see the confirmed fatality numbers. There is a possibility that lockdown somehow impacted reporting? Let’s hope not.
Are the results statistically significant? Without going back to stats class too much, here’s the HSE’s assessment:
What do you see? I see ‘almost’ or ‘just about’. I actually see a glass half-full ‘yes’.
The chart is assessing whether, all things being equal, the 2019/20 figures fall outside the statistical bounds of what’s gone before (with a 95% confidence level). If they do (a ‘Yes’) then we can elicit statistical significance to the change. If they don’t (a ‘No’) then the decrease is not statistically significant and we shouldn’t read too much into it.
The chart also neatly splits the year into totally pre-Covid (April through January) and then includes the months where its impact is debatable. The chart further contrasts significance with last year (2018/19) and the previous 5-year average. Up until the end of January there was no statistical significance despite the large drop in fatalities. For the whole year, there is statistical significance both against 2018/19 and the previous 5-year average.
Has the lost production tipped the scale in favour of significance? We just don’t know. If the provisional figures are revised at some point, how will that play out? For now, it’s close and worthy of much needed optimism for the future.If You Like This Post, Please Share It!