Injury rates tell an incomplete story. David Provan and Drew Rae explore how adding an injury severity metric transforms safety performance conversations. They examine the 2026 Journal of Safety Research paper "Injury Rates Tell an Incomplete Story, but a More Complete Measurement Narrative May Be Possible" by Kevin Geddert, Sidney Dekker, and Drew Rae. Across three large organizational case studies, the researchers presented senior executives with their own injury data calculated three ways: total recordable injury frequency (TRIF), average injury severity, and a weighted combination called the Severity Adjusted Injury Frequency (SAIF). The results reshaped which periods, business units, and countries looked safest.
Drew and David discuss how falling injury rates can hide rising severity, why one country's data showed serious injuries but almost no minor ones, and how different reporting practices can make aggregate numbers meaningless. Most importantly, they describe what happened in the boardroom: with a single number, leaders speculated about causes, but with a second metric, conversations slowed down and became curious and data-driven. As Drew Rae puts it, "I can't give you a better metric. I can give you a metric that will give you a different type of conversation." The episode closes with four practical takeaways, including the case for continuous severity measures over categories.
Discussion Points:
Quotes:
Drew Rae: "If you've got one number, that number can only do two things. It can go up, it can go down."
David Provan: "Once you've got one metric with nothing else to have the conversation around, you can tell any story you want about that metric."
David Provan: "Multiple metrics will result in better conversations than any one metric."
Drew Rae: "I can't give you a better metric. I can give you a metric that will give you a different type of conversation."
Drew Rae: "When I'm lying in hospital, the last conversation I want two senior managers in my organization to be having is should they classify me as a moderate injury or as a high potential event."
Resources:
Injury rates tell an incomplete story, but a more complete measurement narrative may be possible - Kevin Geddert, Sidney Dekker, Drew Rae. Journal of Safety Research, Vol. 98, 2026 (open access)
Paper: Signs of safety: An investigation of how OHS professionals interpret injury metrics. James Pomeroy and Colin Pilbeam, Journal of Safety Research, 2025.
Organization:
Construction Safety Research Alliance.
Previous episodes referenced:
Ep. 136: What is the symbolic purpose of injury rates?
Ep. 133: How do policies and metrics shape the outcome of investigations?
Ep. 109: Do safety performance indicators mean the same thing to different stakeholders?
Ep. 104: How can we get better at using measurement?
Ep. 97: Should we link safety performance to bonus pay?
Ep. 85: Why does safety get harder as systems get safer?
Ep. 74: Is a capacity index a good replacement for incident-count safety metrics?
The Safety of Work on LinkedIn
David Provan - cohost [0:00 - 2:05]
You're listening to the Safety of work podcast, episode 138. Today we're asking the question, how can we improve the conversations we have around injury reporting? Let's get started. Hey, everybody. My name's David Provan and I'm here with Drew Rae and we're from Griffith University in Australia. Welcome to the Safety of Work podcast. In each episode, we ask an important question in relation to the safety of work or the work of safety, and we examine the evidence surrounding it. Today we're going back to one of the most common questions and controversies in safety. I guess at least as long as I've been practicing it. And Drew, I'm sure as long as you've been researching it as well. How do we measure and report safety in the first place? And we've done a number of previous episodes on metrics. Two episodes ago, in episode 136, we talked about the symbolic purpose of injury rates. Episode 133, how target shape investigations 113 on 0 as a target. 109, how different stakeholders interpret performance measures differently. 104 questionable measurement practices. 97 linking safety performance to bonus pay. 85, the paradox of being very safe. And back in 74, which is a couple of years ago now, the capacity index or a capacity index instead of recordable injury rates. So, you know, and we also talk through, in, in a way about measurement behind almost every evaluation of any safety practice that we do in an organization to understand what. What it's doing, the impact it's having or not having. So measurement is something that we are never going to get away from in safety. And I guess today's conversation is about can we have better conversations which might lead to greater insight and better decision making within our organizations. So, Drew.
Drew Rae - cohost [2:05 - 3:16]
Yeah, David, I've got to say, measurement has got to be my least favorite topic, I think, to talk about in safety, just because we've had so many conversations about it and the conversations typically go absolutely nowhere. We have people who want to defend the absolute indefensible when it comes to recordable entry frequency as a measurement. We have people who want to invent brand new measures just like a rabbit out of a hat with all of the problems of trifle, but with the novelty of it's not trif. We have people who want to solve the world's problems by doing an entire PhD inventing new safety measurements. And yeah, a big part of this is we just need to, like, get away from thinking that this is something with easy solutions. And I'm really quite happy with the PhD project that this paper that we're looking today has come out of and a brand new PhD project that I was talking to someone just kicking off this morning, which gets to less about how do we fix measurement and just more about what is measurement doing to the way we think and behave and talk inside our organizations.
David Provan - cohost [3:16 - 3:44]
Yeah, Andrew, I think this is one where can we be better, not necessarily even good or best at what we're doing, but you know, the current organizational practices today can clearly be better around the measurement space. So it's kind of exciting. And sure, this PhD is good because I know the supervisors are definitely good at this. But Drew, do you want to introduce the paper because you're a co author of it? Sure.
Drew Rae - cohost [3:44 - 4:20]
So the paper is called Injury Rates Tell an Incomplete Story, but a More Complete Measurement Narrative May Be Possible. Interesting title. Not sort of like my usual style, but I quite like it as a clear statement about what the paper says and does. The authors are Kevin Gethert, Sydney Decker and Drew Ray. Fairly obviously there. Kevin is current PhD student in our group. The two supervisors are Sydney Decker and myself. This is the second publication to come out of this PhD and so jury
David Provan - cohost [4:20 - 4:40]
is very recent published 2026 Journal of Safety Research in the US or largely out of the US. So do you want to talk a little bit about the background to this, this research and then we can talk a little bit about how you went about the research, the findings and what it means in practice. I guess for our listeners, at least
Drew Rae - cohost [4:40 - 6:08]
as described in the paper, this paper is almost a direct follow up to another paper that we reviewed, David, a couple of episodes ago by James Pomeroy and Colin Pilbeam over in the UK, published in 2025. In that paper, they were looking at the way metrics were used in organizations and they, you may remember, they highlighted 17 different ways injury rates are interpreted, only a couple of which actually have to do with measurements, as measurements. So in that paper they sort of like talked about injury rates almost as signs or symbols rather than as numbers or statistics. Which is kind of interesting because if the number is not a number, if the number is just some sort of other symbol, then it doesn't make a lot of sense to argue about whether it's statistically valid or not. You know, if people are using it in a statistical way, then by all means criticise. It's called validity. But if it's just kind of a narrative generating engine, then statistical validity isn't really the point. And so that's really what this paper was Looking at is the narratives that go around injury rates and the narratives that come from injury rates. And can we change those narratives without needing to compare completely change the way we do measurement? So this sort of like, speaks directly what you were saying, David. It's not about being perfect or even great. It's about where are we now? How can we change this? We don't need to find the perfect metric to change the conversations we have around metrics.
David Provan - cohost [6:08 - 6:45]
And it might be stating the obvious, Drew, though here that organizations use injury rates as their ultimate trying to avoid using the word measure, but the ultimate narrative of their safety performance, of how safe they are. So I think when injury rates become the thing that people use to determine if something is safe or not, then it is something that's important that we actually explore how we can have better conversations around them, because that's where I guess the statistical validity is called into question, that they aren't a reflection of how safe an organization is.
Drew Rae - cohost [6:46 - 7:24]
And one of the initial observations of this paper, which we'll come back to again in the findings, is if you have any single number that claims to represent safety, and it really doesn't matter what that single number is, if it's counting the number of coffee cups in your break room, if it's measuring the injury rate, if it's some fantastic new view index of safety, if you've got one number, that number can only do two things. It can go up, it can go down. And your conversation with your senior executives or in the boardroom starts with safety has gone up or safety has gone down. What are we going to do about this?
David Provan - cohost [7:24 - 8:25]
And one of the ways I explain that a little bit, when organizations just have the conversation around one number. So say, well, when you think about the financial performance of the company, you will have one headline number and it might be earnings before incomes, income and tax, right? Ebitda. But if you look at a company's annual report and financial statement, there will be 60, 70, 100 plus pages of graphs and measures which include, you know, underlying profit, total shareholder return, return on capital employed. There'll be, you know, like a whole book written about all of the different ways you might want to think about financial performance. Because, you know, organizations know that if this number was this last year and this number was this this year, it doesn't necessarily mean that the organization is better or worse from a financial performance point of view. Yet we've never, I guess, in safety gone. We want to tell a much bigger story here about the way that to enable a much better Conversation about what's actually happening in our business.
Drew Rae - cohost [8:25 - 8:52]
Yeah. And unlike finance, very often those other numbers are not there. People are just drilling into the details of that same number. So this leads to some very uninformed conversations. Safety's got worse and immediately the conversation just goes off onto folk theories about why it's got worse or it's got better. Folk theories about why it's got better, which don't lead to, like, great executive decision making or support or even better investigation on how we can improve things.
David Provan - cohost [8:52 - 9:28]
Yeah. And I also had this conversation with the team yesterday through about. You can almost tell any story you want around a single number because you might have entry rates and go, our entry rates are going up. Oh, we are having more injuries, we're getting worse. Or our injury rates are going up. Oh, we have improved our reporting culture and the injuries were always there, but now we're just reporting them. So we're getting better. We want our injury rates to go up for a little while while we strengthen our culture. So once you've got one metric with nothing else to have the conversation around, you can tell any story you want about that metric. Yeah.
Drew Rae - cohost [9:28 - 9:48]
I have to admit, David, I have cynically said multiple times before that injury rates are the perfect metric because you can't be wrong. If it goes up, then you've got improved reporting. Great. You're doing well. If it goes down, you've got improved safety, you're doing well. You can't look bad with a single number that you can explain it going up or down? Both. As you're doing a better job.
David Provan - cohost [9:48 - 10:00]
Yeah. So, Drew, how did you go about this? So a little bit more. Anything else you want to say about the background? But, you know, maybe what the research question kind of was and how you went about the research or how Kevin went about the research.
Drew Rae - cohost [10:00 - 13:02]
Yeah. So the design of this is, I guess you'd call the broad design case study research, something that we've talked about before. These are very applied case studies. So what we're doing in each case is basically taking the injury data from each of three organisations, each one being a different case study. Where then? So the injury rates are originally just expressed as total recordable injury frequency. Taking the raw data and using some different metrics to create different answers to that question about safety, reporting those different metrics back to the organization and seeing what happens. How do people react when you give them their own data back? Again, but calculated in different ways, there are three different things that we're giving back to them. All of them are from the same reportable injury. So we're not really changing the underlying data that's collected. The first one is the one that is most common for organizations to use, which is the total recordable injury frequency. This is what we sometimes call injury rates, or trifa. The way it's usually calculated is you calculate the total number of injuries, you divide that by a number of person hours that are worked to turn it from a raw number into a rate. In this particular case, for each organisation, it's on a 12 month rolling basis. So over the past 12 months, how many injuries per hours worked. Second, statistics that's used is the average severity of each injury. So we won't go through the full calculation here, but the key thing here is that this isn't just dividing them into low severity, medium severity, high severity. It's actual. A number for severity normalized around a standard injury being two weeks. If it's less than two weeks, it's lower than one, if it's more than two weeks, it's greater than one. And there's a couple of complications built in, like if there's no lost time, but there's medical treatment that's given a number. If there's an actual fatality that's given a different number because obviously a fatality doesn't have lost time, it's permanent lost time. But the key thing here is that you've got a number from 0 to basically infinity for each injury based on the severity. And then the third number is a weighted combination of the two. Kevin calls it the severity adjusted injury frequency, which you pronounce as the safe rate, which is. David, I'll leave you to guess just how much I like the acronym spelling out safe. Spoiler alert. Not at all. But yeah, this is a kind of tradition in safety measurement to have clever sounding acronyms, but the idea here is that this is kind of like the total volume of safety. So the three things we've got are average number of things that happen, average severity of things that happen, and combined average of number and severity. And you kind of think that those would tell you very similar things. But yeah, the interesting thing is they give you quite entirely different graphs when you plot those three things for the same organization.
David Provan - cohost [13:02 - 13:51]
And I think, Drew, you know, our listeners may have some experience with severity based measures. I definitely have used them at different points throughout my career to try to tell a more nuanced or broader story about incident performance. And many of our listeners would have also seen those graphs, you know, and the, the things that hurt people aren't the same as the things that kill people. If we want to quote some of the Construction Safety Research alliance work where you see this this year by year, reducing recordable injury rate and this very flat fatality rate in a. In a business. So I think it's really interesting, though, to then combine these and go, well, but within that recordable rate, there's a volatility of severity. And, you know, is that, I guess, as we get toward the end of this episode. Well, what does that do to change the conversation that you're having when you start presenting that data?
Drew Rae - cohost [13:51 - 14:26]
Yeah, and one of the interesting things that we keep encountering in this particular project that we might come back to later, David, is organizations say, oh, we're already including severity because we break it up into low severity and high severity, or injuries and fatalities. But that's not the same as doing a weighted metric. When you divide things into binary classifications or three classifications of severity, you're still really just dealing with a single number for each type rather than different metrics, and it doesn't really fix the problem. So, David, can we jump into talking about each of these case studies?
David Provan - cohost [14:26 - 14:37]
Yeah, and I think, as we'll talk about at the end, I'm just thinking when you said it doesn't really fix the problem, it doesn't fix the problem mathematically, but also definitely doesn't fix the problem socially inside an organization. Yeah.
Drew Rae - cohost [14:37 - 14:43]
And that's the key thing here, is this is not about perfect numbers. This is about incentives and conversations.
David Provan - cohost [14:43 - 14:48]
Yes, Drew, so talk about what we learned in. Or what you learned in each of the three case studies.
Drew Rae - cohost [14:48 - 15:47]
Okay. So the first organization, and this is podcasts, are not a visual medium, so we can't put up a graph in front of you, but this is a graph that will look very familiar to most of our readers. This is an organization where over the past 13 years, you've generally got a downward trend in the injury frequency, which is there are a couple of bumps in the road periods where it temporarily went back up a little bit, but each time it went back up, there was just this steady decline. There's one key particular point in that time when they got their lowest injury rate across that entire period. This is like the result of a very clear turning point where they've managed to drive the injuries down over a sustained period of a couple of them, 2021-2020. But more recently, that injury rate has stagnated. They drove it down very, very low. Last couple of years, it stayed basically the same, trending slightly upwards. David, is that A story you've seen before, A graph you've seen before?
David Provan - cohost [15:47 - 16:07]
Yeah, I think we've all, if you take any industry in any organization over the last, over a 13 year period, I think you would see that long steady decline as I'm not going to say why I think that's happened, but I think our listeners will all have their own hypotheses about why that long steady decline has happened.
Drew Rae - cohost [16:07 - 16:51]
Yeah, so next thing we look at is what about the safe rate? So this is the combined injury and severity. Very first thing we notice is that that early period which in the injury rate was there historically, we were bad in the past, we're good now. That early period was not when the company was performing the worst overall. The worst was actually in the middle of the period in 2020. In fact, they actually started fairly low and improved, followed by a sort of like sustained increase up to a bump that's gone down again more recently. So just a totally different shape of what periods of the company's history was safety worse in. David, your immediate reaction?
David Provan - cohost [16:51 - 17:31]
Yeah, look, I think if you look at this period and again our listeners can't see the graph, but the numbers, like in one 12 month period, your recordable injury rate goes from like 16 to 7. So you're jumping up and down going, you know, we've got a 56% reduction in our recordable injury rate. And the same period you get this 250% increase in severity from like 0.23 to 0.55. So you're kind of going, are we actually two times better or are we two times worse? Like what is, what is the result here? So it's, yeah, I think it's a, it's a real interesting. I would be fascinated to hear the conversation change and the debate that might happen as a result.
Drew Rae - cohost [17:31 - 20:09]
Yeah, so, so just having those two things doesn't really tell you what's going on. The next thing you need to do is add in, okay, what about just like the raw average severity of injuries and this is what explains the difference between the injury rate and the combined rate. And in each of those periods where there's a difference, it's because the injuries are getting more severe or less severe in ways that seem to be actually quite contradicting the total rate of injuries. So in the periods where they were getting progressively better in terms of raw numbers of injuries, their injury rates going down, there were quite worrying things going on with severe injuries. Severe injuries were on an uptick not just as a percentage of all injuries, but actually like the total number of severe injuries was going up. It was just being hidden by driving down the number of minor injuries. And the opposite was true as well. In periods where the company thought that they were stagnating, actually there were real safety improvements going on, but those real safety improvements were hidden by the noise coming from the low severity events. So that stagnation was actually quite a steady improvement in the most worrying types of injuries. This gets even more interesting when you look at where in the company this is happening. So the next thing that Kevin did is he broke it down by the four internal businesses and looked at their relative performance. So no names here, so we're just calling them Business one, Business two, Business three. Business. If you look at their injury rates, then Business three is by far the best performer. And then there's quite a clear progression. Three is the best, then four, then one and Business two is by far the worst business. But in terms of safe rate, it's a completely different order. The safest one is the same. Business three is still the safest and that's very easily explained just in terms of risk profile. What Business 3 is doing is fundamentally safer. So they're always going to have a better safety performance. But Business two, which was by far the worst in terms of injury rates, is actually the next best, once you take the severity into account, Followed by Business1and4 Being about the same. So you've kind of swapped which businesses you're worried about based on including this extra information. And then you start to wonder, okay, so how come we've got this business which is not our worst performer, but they're generating this very large number of low severity events? Are they having lots of low severity events? Are they just much more honest reporters than the other businesses? Are they got some reason why they're doing a better job of collecting data about these low severity events. Anything else you want to say about that, participant?
David Provan - cohost [20:09 - 20:14]
No, not at all. I was going to sort of. We could see if you wanted to move on and tell the story of the other two.
Drew Rae - cohost [20:14 - 21:30]
Yeah. So we won't sort of like give you things that are the same across these participants. Just there's a bit of extra information we learn with each case study. The interesting thing we got from Participant 2 is even though they had a very similar kind of history and that contradiction between injury rates and severity, the extra finding is that the combined rate was a lot more stable than the injury frequency, so it was less likely to have large swings from year to year. And the reason for that is that every Time the TRIF went up, the severity rate went down and vice versa. And so the combination of the two stayed roughly the same. There are two possible inferences you could draw from that. The first one is that you are actually driving down the number of low severity events just because they're easier to manage if you place lots of attention on them. The second, possibly more likely, is that when you try to drive down the number, you actually drive down reporting and you can hide low severity events a lot easier than you can hide high severity events. But in either case, when you focus on trif, you're not actually focusing on the injuries that are making most contribution. So the underlying rate of safety is actually really just the same. It's just those injury rates are going up and down. David, thoughts about that?
David Provan - cohost [21:31 - 22:15]
Yeah, look, I think that's nice to see inferred from the research because I think our listeners, or at least in my experience, I think because we've had recordable injury rates as a target in many organizations and industries for a decade or more, and at least LTI for lost time injuries for maybe two decades. I think there is always that question about have we actually got safer or just have we either, you know, out classified certain things from reports. When I say out classified, I don't even know that's a term, but just not reported something. And we've obviously invested kind of heavily in injury management as well. So, you know, certain things maybe be seen to be less severe as well, perhaps.
Drew Rae - cohost [22:16 - 23:57]
Yeah. And you could argue to the cows come home why your trip has gone up or down and you simply don't have the explanation for it. You don't have data to draw conclusions about why it's gone up or down. Whereas building the severity into the calculation and having both of them sort of takes away from that conversation, because it doesn't really matter if it's gone up or down, you can see your underlying amount of safety has not gone up or down. The third participant is a little bit different from the other two. This is an international organisation and unlike the others, which had discrete business units doing different things, this organisation basically does the same thing in each country. So you would expect each country to have roughly the same risk profile, the same types of injuries, roughly the same sort of data. But what they found was that even though they're doing similar work, they think the company has standardised their definitions for safety and their reporting practices. Actually, each country has very, very different profiles of events. And you can see this very clearly once you have the three measures. The TRIFA and the safe and the average severity, you can see that like the data, it really doesn't make sense from some of the companies. So, for example, there's one country that had similar serious events to all of the other countries in terms of the type of events and the number of events, but they had hardly any low to medium severity events. And that literally just doesn't make sense, I think. David, can you have a business that is generating lots of, sorry, not lots of, but a measurable number of serious events and just no minor events going on?
David Provan - cohost [23:57 - 24:50]
Look, I think for our listeners who work in international organizations, I can hypothesize which countries will report in different ways. And you know, I think there are something. And it's not, it's not a reflection on bad intent, it's just in some different countries around the world, cut finger in a workplace in one country is something that is noteworthy and cut finger in a workplace is in maybe another country through local supervision and local workforce, just not something that means anything in terms of reporting. I know that might sound a little bit strange to some people, but I've seen similar injury profiles with organizations saying, hey, we've had fatalities in this part of our business around the world, but it's our best performing part of the business in terms of recordable injury rates, you know, yet we have our most fatalities there.
Drew Rae - cohost [24:50 - 25:56]
Yeah. And there are other things that can be going on, like, you know, in countries like the US and Australia, if you go to hospital, it's got to be paid for somehow. It's going to be paid for by the insurer, it's going to be paid for by work cover, it's going to be paid under Medicare and that's going to leave a trace of medical treatment and lost time. Whereas companies with different medical systems, different medical practices, even though you've formally got the same definition, different things are going to get end up recorded in the system in different ways in different amounts. And that's really what the company realised is they couldn't draw conclusions about safety here. They just realised that they had such different reporting practices that they shouldn't be drawing the conclusions that they were drawing. And it was kind of like encouraging because the aggregate data showed that safety had been gradually getting a bit worse year for year. And once you break it down and realise that the contributions are coming from completely different data sets that mean completely different things, you realise that that total number is also meaningless. So maybe they didn't actually have a problem they needed to worry about. Maybe they did, but certainly their data that seemed to show that they had a problem wasn't reliable data.
David Provan - cohost [25:56 - 26:48]
And I think in, in all of this conversation about data, Drew is, you know, that's part of the point. You know, we try to aggregate this data up to a picture at even a business unit or a country or an organizational level, which is really meaningless because what, what are you going to do with that? Right. Like it, you can't really use it to make any decision around safety priorities or safety improvement. But then if you take it down to a local unit where you really want to understand risk at a, you know, in the physical world, at a particular work site or on a particular project, that just number is so widely variable that 11 months of the year it's zero, and one month of the year something happens. So you can't use it to do anything meaningful at a work site level either. So as much as we report it, I think in some ways this data, cutting the data in different ways almost just shows that we still shouldn't probably really be talking about it anyway.
Drew Rae - cohost [26:48 - 27:23]
Possibly. But remember, the overall purpose of this research isn't to actually like, evaluate the metrics. It's not just, this isn't research about whether TRIF is good or bad. And it's not even like the new measure that we're using here, the weighted average safe. This paper can't possibly prove whether SAFE is a good or bad measure. That's not the question. What we were really observing was what were the conversations that were happening when this data was reported back to the companies? What questions did they ask? What speculations did they make? What were their reactions to having a couple of extra numbers?
David Provan - cohost [27:23 - 28:02]
I'd love to explore that a little bit because if when we're talking about, maybe I'll sort of play back what hypothesis and you can tell me if what you learned in relation to that, if you present one number, just total recordable injury rate, you can have a conversation about good, bad, better, worse. What are we going to do about it? Right. Or everything's fine. Once you introduce other numbers alongside those numbers, do people start having more conversations about the numbers themselves, you know, and the quality of the numbers and the reporting, or are they still having conversations about which is better and which is worse? And what do we, what do we do?
Drew Rae - cohost [28:02 - 28:45]
Yeah. So your hypothesis there, David, is pretty much spot on to what we noticed. So when you just have one number, no matter what that number is, people kind of assume that that number tells them about reality. So either safety is getting better or safety is getting worse. And the conversation immediately moves on away from the number towards what are we going to do about it? Or what's the explanation for this? But that conversation isn't a data driven conversation. That conversation is a highly uninformed and highly unedifying conversation that happens because people are just speculating, they're giving their own opinions about why something is happening with no data fed into those which for
David Provan - cohost [28:45 - 28:51]
me cynically might result in someone sending an email or, or a safety stand down happening across a part of a business.
Drew Rae - cohost [28:51 - 30:19]
Yeah. And you know, you don't want a conversation to be 100% driven by numbers, particularly when those numbers are unreliable. But you also just don't want senior people doing speculation about something that they're not informed about and have no data to talk about. What happens when you immediately just introduce one second metric, one point of comparison is the conversation slows down and it starts being about why are these numbers different? And about the reliability of the numbers and sources of inaccuracy and different things that might be going on and asking questions back to the person, giving you numbers, asking for more numbers, asking to see that next page which drills down into the different groups and do that same comparison at the different groups. So it's a much more data driven conversation and a lot more curiosity and asking questions rather than making speculations. So yeah, that's really the effect of simply adding, in this case we were giving three numbers. So we're giving the injury rate, the average severity and the combination. But really you can sort of think of that as giving two numbers and then also showing the combination of those two numbers. And so simply that adding of one extra piece of information, that severity as well as the rate makes a big, big difference in the quality and type of conversation that is happening. So it doesn't require you to collect different data about the injuries. It doesn't require to get rid of the data. All it does is requires putting up that one extra piece alongside that first data.
David Provan - cohost [30:20 - 30:27]
So, Drew, anything else you want to sort of talk about in terms of the research or the findings and then we, we can probably go into some takeaways.
Drew Rae - cohost [30:27 - 30:33]
No, I'd be happy to go straight into takeaways and what we can like reasonably conclude or not conclude from this sort of work, David?
David Provan - cohost [30:33 - 31:12]
Sure. Well, I'll start. I was sort of thinking of these for myself a little bit while you were still doing the prep of the, the episode. But I think we know, and I don't think we need to restate it, but we know that recordable injury rates is a sort of A quite meaningless and blunt information. But if you do want to have a discussion about lagging injury measures in your organization, pet, you know, like, it's more meaningful to talk about those lagging indicators in kind of like a risk adjusted or severity adjusted way, adding some nuance to the discussion, rather than just a raw count where someone who loses a leg is exactly the same as someone who, I don't know, has cuts their fingers.
Drew Rae - cohost [31:12 - 31:19]
Yeah. Just to be clear, this particular study doesn't tell you that. That's almost like a starting point of your study.
David Provan - cohost [31:20 - 31:26]
All right, well, assuming that the person who lost their leg had more days off work than the other person.
Drew Rae - cohost [31:26 - 31:45]
Yeah, yeah, yeah. The problems with injury rates have been done to death. And this is not actually more evidence against injury rates. What it is is evidence that we can improve it without needing to throw them out. And, yeah, just throwing in that extra information about severity does seem to significantly help the conversations get better.
David Provan - cohost [31:45 - 31:46]
Do you want to keep going truth?
Drew Rae - cohost [31:46 - 32:27]
Yeah. So the second takeaway is one that I put in, which is that just even changing the problem we're trying to solve, I think helps. People always ask, okay, if we're not going to use truth, what will we replace it with? What's a better measure? What's a valid measure? None of these new measures are any more valid than the old measure. If you get away from that question of what metric should we use and get onto the actual problem here, which is what types of conversations do we want to be having? At what level in our organization do we want to be having those conversations? What type of data sparks the conversations that we want to have? I think that's a much better question to ask, and it's a much easier question to answer. I can't give you a better metric. I can give you a metric that will give you a different type of conversation.
David Provan - cohost [32:27 - 33:29]
Yeah. And I think along. I think one of the problems that we've had with de emphasizing the discussion about TRIFA in our organizations is the search for a single, better replacement for trifa. And I think the third takeaway here, Drew, is that multiple metrics will result in better conversations than any one metric. When we can start to compare different metrics and data that are all talking about how safe an organization is, then we can have a better conversation. Individual metrics wildly fluctuate, whereas I think this research shows that when you pair metrics or combine three metrics together, you throw a bit of confusion into the discussion, which requires, like you said, Drew, requires a bit of a slowdown. More questions get Asked there's more things to unpack. Hopefully going to create a little bit of time to, you know, have a slightly deeper discussion than this busy unit. This business unit is going badly. Go and do something unhelpful to try to fix it.
Drew Rae - cohost [33:29 - 34:18]
Yeah, I think the new information we have from this research, we already knew that multiple metrics was better, but there's still this constant pressure in organizations to get things down to a single number with a single answer. And the new information we have is you don't need lots of extra metrics. You don't need a big suite of metric to replace that single number. Even just adding in one more metric, if you like, show it on the same graph with two lines, one for the first metric, one for the second metric. Just that little bit of extra information can be enough to start shifting the conversation and having a better conversation. You just one extra column on that graph on the table that shows each of your business units. So you don't need to fight for that big suite of metrics or get people to be engaged in lots and lots of data. Just pairing the metric with something to give it balance can be enough.
David Provan - cohost [34:18 - 34:28]
Andrew, the fourth one, I think I'm most. I think that's been probably the contribution that I've taken away. So do you want to sort of talk a little bit about a continuous metric?
Drew Rae - cohost [34:28 - 35:40]
Sure. So this is a little bit more complicated to explain, but it's fairly important. What people tend to do with severity is they tend to break it into categories. So they might even like have one graph for all injuries, another graph for high potentials or for severes, or separate into like injuries and major injuries and fatalities that, you know, binary or tertiary classification of severity. The trouble with doing that is it makes a massive difference on how each individual event is reported and places a lot of pressure on someone to record something as minor rather than as major and severe. If you take away that classification and instead you create a continuous metric, like the number of days that is lost, then you take away that pressure. So instead of. Yeah. Instead of forcing these classifications early on in the process, you just turn severity into a continuous range. Even if you need to be a little bit artificial in how you define that, it still removes a lot of the internal under classification and under reporting pressure, which in turn means you're going to have better conversations about those results.
David Provan - cohost [35:40 - 36:38]
Yeah, I think for me, Drew, with that, is that you know all of this. Well, I guess all of the severity categories that I can think of, is it a recordable, Is it not a recordable Is it, Is it this or that? Is it a high potential or not a high potential? Even if someone's scoring it against their five level consequence risk matrix, it's a big step to go from a 2, which might be moderate, to a 3, which might be serious. We're talking about Even like that one step in the matrix from a 2 to a 3 is moderate to serious. And so you know, what that does is generate a whole lot of anxiety, politics, time, effort, where it's like it was seven days or it was six days. It's like no one's really going to debate or discuss or burn time on, on that. So I actually think thinking about something as a continuous variable, that's what I took out of this paper. I thought that was a really nice way. Now I can argue about is day's loss the best thing? And I'm sure our listeners are going to have, you know, other ways. But that principle I think is really powerful.
Drew Rae - cohost [36:38 - 37:06]
Yeah. I have to tell you, David, when I'm lying in hospital, the last conversation I want two senior managers in my organization to be having is should they classify me as a moderate injury or as a high potential event rather than what happened, how do they take care of me and how do they fix it? Yeah, that's the trouble with these binary classifications is that's what the conversations are, is how do we classify this? How do we justify reclassifying it? Not how do we stop it happening?
David Provan - cohost [37:06 - 37:08]
Absolutely. So, Drew, anything else?
Drew Rae - cohost [37:08 - 37:14]
No, slightly shorter episode, David, but I think an interesting paper and a useful conversation to have.
David Provan - cohost [37:14 - 37:20]
Yeah, wonderful. So the question that we asked this week was how can we improve the conversations that we have around injury reporting?
Drew Rae - cohost [37:21 - 37:35]
Yeah, the answer for this is add one more metric. If you want it to be safe, we've got the details of it in the paper, publicly available, but there's nothing special about safe. It's just about adding that second continuous metric to go along with the first one.
David Provan - cohost [37:35 - 37:50]
Thanks, Drew. That's it for this week. We hope you found this episode thought provoking and ultimately useful in shaping the safety of work in your own organization. Join us in a discussion on LinkedIn or send any comments, questions or ideas for for future episodes directly to us @feedbackafetyofwork.com.