Wearables in Risk Management: Interview with AIG's Ron Bellows

Written BY

Emily Friedman

July 11, 2019

I got to sit down and talk with Ron Bellows, Risk Strategist at AIG. What resulted is a fascinating- but long (it’s worth it) – read and a wealth of information. Ron will be speaking at EWTS 2019.

E: To begin, could you provide our readers with a little background on yourself and what you did at AIG? Also, when did you first encounter wearable technology?

R: I’ve been a risk management consultant with multiple insurance companies since 1980. I started playing with wearables probably as early as 1985/86. You may remember Cybermatics: Surface EMG measuring technology was connected to cyber gym equipment for rehab and prevention, so that when someone was working out – in the sports or medical world – you could actually see what the muscles were doing with the surface electromyography. It’s kind of like an EKG. Your heart is a muscle; surface EMG looks at other muscles.

Around ’86, I was working with a physical therapist doing studies on sports medicine and rehabilitation that might have application in the industrial environment. Many workers’ compensation injuries are expensive strain and sprain injuries impacting the musculoskeletal system. Biosensors, from a rehab standpoint, help us manage treatment for someone who has had a musculoskeletal injury. It began in the sports world and medicine, and around 2000 it started to become more pervasive in the industrial environment.

If you think about an athlete who over-trains, the same thing can happen in the industrial world.  Biosensors can measure posture, force, repetition, etc.; and be used to look at someone in the workplace from a preventative standpoint as well as on a pre-hiring/screening basis (i.e., can you handle the physical requirements of the job?) If you took a traditional physical, you might pass, but could you go in and work on a factory floor or warehouse for 8-10 hours a day, 6 days a week? Biosensors to better evaluate somebody before they go on the job to help assess their ability. Second value would be to evaluate somebody in the job to document the exposure they face due to fatigue, endurance, force, repetition and posture—the things that generally lead to ergonomic/ bio mechanic injuries. If you can detail that exposure prior to injury you can do a better job with prevention, training and hiring. However, if somebody does get hurt you can use those same biosensors to help assess exactly where and how badly they are injured, the best treatment options, and if they truly are ok to go back to work again. Those are the three main areas where wearables fit into the industrial arena and workers’ compensation (WC).

E: What exactly do you do at AIG?

R: I’ve consulted with large multinational customers to help them find solutions to their risk management issues. Often, they were most interested in workers’ comp risk because it tends to drive loss frequency and severity, impacts the workforce and absenteeism, and reduces efficiency and profitability. Workers tend to be 30-50% of a company’s operating expense, so if you can reduce injuries you can increase efficiency, profitability, etc. Today with the shortage of workers that we see, a lot of companies are working at a 20% absenteeism rate. Imagine what happens when you can’t find enough people to man the tasks in a factory. If you also have extensive injuries that put people out of work or on restrictive duty, it’s even more difficult to run the business. Making sure people can work safely and come back to the job every day is very important to risk managers. I also help risk managers with issues like fleet, liability, supply chain, business continuity, and disaster recovery—anything that keeps them up at night.

E: You just mentioned a bunch of pain points like the shortage of workers. What are the challenges and pain points for AIG’s clients that are driving interest in wearable technologies?

R: There are really two things: One is traditional safety, making sure we document exposure properly so that we can prevent injuries and do better training. It’s not just job hazard analysis but also the workers’ comp system itself, which is very difficult to manage as the venues are all different and every state has different rules. If we can document exposure, we can better manage an individual pre and post-loss. Many times, we see that older, high tenure workers are driving losses. We’re seeing the average age of workers going up, especially in manufacturing, warehousing, trucking, etc. where you have extensive injuries to the shoulder and back. Those injuries are the most difficult to diagnose, treat, and return to work. If you’re older and you get hurt, it may take you weeks to get back to where you were pre-loss. Our average workforce is in the 40- to 50-year range, so when they have an injury it’s impacted by comorbidity – hypertension, diabetes, obesity – making it more difficult for them to get back to pre-injury status.

Second, many companies today are looking at exoskeletons or other interventions to reduce exposure. When you put an intervention in place you don’t know for sure how much of an impact it’s having on the individual, because everyone is different. With biosensors, we can measure the impact of different interventions and see which ones are having the greatest impact on the worker based on their exposure. For example, I would use different exoskeletons for the upper extremities versus the back, versus the legs. So, it depends on what kind of difficulties workers are having in the workplace. For example, if I have to do static standing all day on a conveyor line, exoskeletons may not be valuable, but the biosensors can tell me what’s going on with the static stress on the lower extremities, which impacts the entire body. I can then look for alternatives like automatic massage therapy, continuous stretching, compression sleeves, to improve endurance and reduce fatigue where the exoskeletons don’t work.

E: What kinds of wearable technologies are you looking at for risk mitigation? Have any solutions made it past the pilot phase to rollout?

R: There are a lot. The biosensor marketplace has exploded in the last several years. We can use biosensors like we’ve talked about from a musculoskeletal standpoint and that’s where most of the impact is seen. But you can also use biosensors to look at an environment: A construction worker going into a pit that may be lacking oxygen can use a biosensor attached to an iPhone that sends a safety signal. You can use a posture monitor for the back like I did with the Visiting Nurse Association. Nurses visiting patients by themselves can be attacked, chased by dogs, fall down stairs, etc. Having an inclinometer or GPS monitor can send an automatic ‘man down’ signal if they’re horizontal. If they can’t push a panic button, their supervisor and local authorities can be alerted to the fact that something is wrong. That’s just one example. Biosensors in chemical plants can look at oxygen-deficient environments and exposure to chemicals and send an alert right away to the individual or supervisor. So, if you’re working remotely in a plant and there’s an ammonia tank with a small leak, the biosensor can alert you to very low levels before you’re overcome. There are so many different ways to use biosensors to alert you to exposure before it creates injury.

E: In most cases are you looking for over-the-counter, out-of-the-box solutions or bespoke devices? Where is the software being made?

R: I review what’s in the market and what’s in development. I try to stay abreast of what’s available so that I can help clients make the best and most informed decisions about how to reduce exposure. There are always several intervention options that could have an impact, so I usually demo and then pilot test the options that fit the particular exposures but also their organization structure and culture. So, I’m always looking to kick the tires on everything around the market.

E: I imagine biosensors come in every form factor at this point. Is it one sensor per device or are you testing multiple metrics?

R: Let’s take posture monitoring as an example, which is huge in workers’ comp because 30-50% of a company’s losses are from strains. Everyone wants to work on musculoskeletal disorders, which also happen to be the most expensive loss type. Inclinometers which measure posture are great because force, repetition and posture are the lead drivers of strain and sprain injuries. You can do heavier work in the power zone between your shoulders and hips, but outside of neutral posture a task becomes more egregious to your body.

Many companies are doing posture monitoring; some are focusing on the upper extremities, some on the low back. Several biosensor companies have produced very good software programs to go along with the inclinometers, showing not only when someone is out of neutral posture but also how many times a day that person is out of neutral posture, for how long, at which tasks, or what times of day, etc. Some biosensors give an automatic response to the employee (like a buzz). That can be good or bad. If I can’t change my posture because the task is set up so that I have to bend a certain way, the buzzer is going to be continuous and become really annoying. That’s where I would take the data to management and operations and say: Here’s Joe and Mike doing the same job but Mike can’t handle the postures. Why? Because he’s a little older and can’t bend as well at the knees. So, posture monitoring without the dashboard is not as effective. The better the dashboard, the better data we have and the more opportunity we have to provide valuable interventions to the physical task.

E: Can the intervention involve changing the way the task is done?

R: Yes. In fact, we can even link biosensors through a software program to a camera, so that as a person moves, we can see both the physical video and an overlay of what’s going on with his or her posture and force over time. While seeing the person do their task in space and time, we’re capturing their force and posture. That becomes really powerful. We can do that over time, creating a dashboard for different tasks, and then give an employer a prioritized list of the most egregious tasks, where high force and high repetition are most likely to generate a musculoskeletal disorder. So, biosensors with a dashboard and video overlay are very powerful in exposure documentation.

E: Can you talk about some of your recent biometrics and exoskeleton projects?

R: Well, anybody familiar with meat processing knows that it’s a very high endurance, high repetition task impacting the upper extremities and back. It’s static stress on the legs, leaning, twisting and bending at the waist, and moving your arms to process meat. Every part of the body is impacted; the repetition is so intense that you’re moving a carcass every 2 seconds. You’re constantly moving, standing in one place doing the same motion over and over, and you’re usually working a 10-hour shift, 6 days a week. Operations, safety and HR know it’s a difficult task but to change the process is very expensive. You might have to move the conveyor circling the entire plant or slow it down, which operations won’t like. Or, you’re going to have to build adjustable stanchions for people to stand up on. Oftentimes in fixed manufacturing plants, it’s very difficult to change the physical process, so we look at other interventions. The biosensors give us data on where the most difficult task/positions are and where management can spend their nickels to make the best impact. You can give them engineering solutions but if they don’t have the money for re-engineering there are alternative solutions like endurance and fatigue management or job rotation, or even just ongoing stretching throughout the day. You mitigate the exposure if you can’t eliminate it. We look for engineering solutions first, but older plants especially have a hard time putting those automation or engineering changes in place.

E: How are you measuring the ROI of the different solutions you’re implementing? What are the KPIs you’re looking for?

R: Primarily, I look at two things when it’s workers’ comp-related: Loss frequency rate: The number of injury accidents per hundred employees (for example, how many strains and sprains we have per task before and after a solution is implemented) and average cost of claim: How does that cost change after the solution is implemented? We try to reduce both frequency and severity of loss.

Here’s a good example: One 24-hour plant of 400 employees had 50 visits to the nurse everyday looking for splints, gauze wraps, and other assistance. You know that the more times people are going to the nurse, the greater the likelihood you’ll have a claim eventually. We implemented endurance / fatigue solutions and then looked at the number of individuals visiting the nurse and in some tasks the number dropped by 80%. That’s telling because it takes a while for the claims numbers to mature enough to tell you statistically significant results. If I have a short change over time, is it just that introducing the solution made everyone more aware? 18 months is about where you have to be to really see a material change in losses. So, we look at other metrics like online perception and symptom surveys. I’ve used therapy to reduce endurance and fatigue injuries and after each session, we give a quick survey asking how the person felt before and after the fatigue management program. We can then see if we’re going down the right road and match up the results to the loss analysis in the future.

E: RFID devices, body-worn (biometric tracking) wearables, and exoskeletons—which category is most mature and deployable today?

R: Posture monitors. The inclinometers and GPS are the most robust and have some of the best software. RFID is good but you have to understand what the exposure is and what end result you’re trying to get to. RFID chips are really good in environments like construction, where it’s noisy, dark and dusty and vision and hearing are impaired. RFID chips give me another sense to avoid exposure. It can also be used for equipment or where people are working very remotely, to see where somebody is working in a plant and where they’ve been. But posture monitors are probably the most robust in terms of software because, again, everyone’s trying to mitigate the strain and sprain injuries. Industrial hygiene (IH) exposure doesn’t have the same frequency of loss as strains and sprains and has been controlled very well over the last 20 years; it’s gotten a lot of attention and there are so many good programs in place.

E: Is ergonomics slightly newer?

R: Ergonomics has been developing since the mid-80s, but it’s interesting that we haven’t found a silver bullet solution, so we’ve done a lot of training. Office ergonomics got a lot of attention. ‘Ergonomic’ became a buzz word and a marketing ploy, and now a lot of equipment is considered ‘ergonomic.’ For example, you can buy a snow shovel that’s “ergonomic”, but the actual exposure to the individual hasn’t really changed. Carpal tunnel syndrome was huge in the late 90s and early 2000s, then the Mayo Clinic and other studies said that the aging workforce is driving CTS more than typing. Today in the workers’ comp arena, an individual’s physical condition can be as much a factor in injury development as the workplace exposure. The comorbidity or illness can make a simple injury so much more difficult to diagnose and treat and this is why wellness and ergonomics need to be considered together. Wearables are helping us communicate exposure to the operations managers who usually hold the intervention purse strings. Ergonomists haven’t done a great job of this in the past, but the biosensors give us data on an individual or task basis that is very telling for operations, human resources and safety teams.

E: How have employees received wearables? What has been the feedback? Are there privacy concerns and how are you dealing with that?

R: A lot of the biosensors are applied directly to the skin and managers are very skeptical or skittish about that. So, in looking at which wearable is going to work for a company you have to consider the unions, female workers, people that don’t speak English etc. You have to think about having an interpreter, if someone will have an allergy to the spray or adhesive used to attach the biosensor… What if you have a lot of hair on your back? Part of my focus is always communicating these considerations to the risk manager: Given the exposure model they face and the loss profile they have, which tasks are driving the losses, what’s the exposure model for the people doing those tasks, and what are the right biosensors to use to fit their organization’s culture.

E: Are they more receptive if the sensor is in a bracelet?

R: You get better, deeper data especially from a force standpoint if you can attach something to the skin. If you can’t you have to use a halter monitor around the chest or a belt-worn device, something on the biceps if upper extremities are the issue, a bracelet on the arm etc. That’s why it’s important to know the loss profile and exposure model for the risk before adopting a wearable product–what tasks are driving loss and what options the company is willing to consider for solutions.

E: What is your hope for the future as for how wearables are developing? What’s a big development you’d like to see?

R: Right now, biosensors are really good at looking at exposure, giving us a dashboard and helping us come up with solution options. Of course, you need to know what’s available and understand the organization culture; but we’re not using biosensors to their full effectiveness in the hiring, and screening process or the post loss injury management.  In WC, early objective medical diagnosis is critical to managing loss costs especially with strain and sprain injuries, and biosensors can be a substantial benefit in that area – including developing telemedicine programs.  We’re also not always closing the loop between risk management, safety, HR and operations in terms of exposure understanding and the implementation of interventions. Consider how many workplace tasks are developed with the idea that there will be one, two or three injuries per year in that task? The answer is none, but we accept those types of metrics as part of the cost of production. We’re collecting good loss and exposure data but not integrating safety intervention into process the way we do with quality. Biosensors give me the detailed exposure information I need to express business and human cost and help qualify the rationale for the interventions needed to reduce exposure. If I can provide detailed documentation of exposure, I can communicate better to Risk Management so they can do a better job of funding exposure reduction solutions and provide the insight for stronger diagnosis, treatment and return to work practices if a loss occurs. You’d be amazed how many loss profiles show repeat injuries, which get exponentially more expensive.  Biosensors can therefore have a significant impact in all three areas of the WC exposure control model:  Hiring, Screening and Deployment; Prevention and Training and Post Loss Injury Management…

Photo credit: Lara de Oliveira Corselet dB via photopin (license)

Ron will present a case study at the upcoming 6th Enterprise Wearable Technology Summit on September 19.

Further Reading