London

81 Chancery Lane
London
WC2A 1DD
0207 112 8493
View map

Edinburgh

1 Lochrin Square
92 Fountainbridge
Edinburgh
EH3 9QA
0207 112 8493
View map

Dublin

24A Baggot Street Upper
Ballsbridge
Dublin
D04 V970
0207 112 8493
View map

Registered Office

7 Bell Yard,
London,
WC2A 2JR
0207 112 8493
View map

Send us a message

CLOSE X
Contact Us
13.05.2024

The past, present and future of wearable technology in insurance

Garmin Epix fitness watch and Whoop 4.0 band – two bits of wearable tech that the author has personally tested

Health technology has made dramatic advances in the past two decades. Mattresses can detect overnight breathing rate and conduct advanced sleep analysis. Rings can provide detailed heart rate data and insights into blood oxygen concentration. Wristbands are no longer confined to counting steps, but rather act as 24/7 trackers of physical and mental stress. Modern fitness and smart watches provide all these insights—whilst also tracking indoor and outdoor workouts with a high degree of accuracy.

In addition to their tracking capabilities, many modern wearables last days or even weeks without needing to charge and are durable enough to withstand the day-to-day of even the most adventurous. They are discreet enough that they can be worn to formal meetings and events.

This wealth of data is not only relevant to those interested in their health. The data that wearables collect, if taken in context, could also be used by insurers to gauge risk and calculate individual premiums for policyholders. Insurers could in turn provide useful insights and feedback, strengthening customer relationships whilst encouraging healthy habits. This would not only decrease premium rates—it would also contribute to healthier overall lifestyles.

This article examines the possibilities that wearables present insurers by considering the history of wearable technology, its current and future applications for gathering health data, and the major concerns preventing its widespread adoption by insurers.

Short history of wearables

Humans have been interested in quantifying their movements for hundreds of years, with designs for pedometers—step counters–tracing as far back as the Renaissance.[1] Despite renewed attention during the Enlightenment, it wasn’t until the late 20th century that wearables became more widely adopted.

The first wearables boom was largely driven by the Yamasa Corporation’s pedometer, which touted a 10,000 daily step target to counter obesity in Japan. This figure was subsequently examined in Dr. Yoshiro Hatano’s seminal 1965 study that associated 10,000 steps with improved health outcomes.[2],[3] Before you could say “pedometer”, the step-counting craze had spread across the globe. As pedometer technology advanced in subsequent decades, a sister market emerged for people interested in tracking their heart rate and distance traveled whilst exercising.

From this point, the two markets —pedometers and exercise trackers—grew in tandem and their technology frequently overlapped. Fitbit emerged in the late 2000s with a sleek and compact pedometer, something unobtrusive enough to wear all day long, and remains a leader in the wearables market today. Within a few years, research and development in the fitness watch industry saw the introduction of 24/7 stress and heart rate tracking, including sleep tracking. Other companies soon emerged offering more subtle alternatives to the clunky sports watch or flashy wristband, each with the ability to monitor steps, calories, heart rate, and stress data.

The increased popularity of smartwatches in the last five years has meant that even people who aren’t necessarily interested in health tracking are wearing devices that gather their data and synchronize seamlessly with their digital ecosystem. In 2024, 35% of the UK population owns and regularly uses a wearable health tracker, an 80% increase on 2019 usage.[4]

Uses in insurance

The use of telematics—accelerometers or “black boxes” that track driving habits—to inform motor insurance premium rates has long been viewed as an overall positive, benefiting both policyholder and insurer. Given the proliferation of wearable technology, could something similar be on the horizon for health insurance data? Many health insurance companies offer reduced gym membership costs for policyholders, or even reductions on policy premiums for frequent gym-goers. This is a blanket solution, however, and wearables offer the potential for much further customisation.

Almost all wearables track daily steps, a metric that since Dr. Hatano’s studies in the 1960s has been shown to be inversely correlated with all-cause mortality. A more recent meta-analysis covering 200,000 participants has shown that this association holds even after controlling for confounding factors such as age, gender and a variety of health indicators.

From Banach, et al. European Journal of Preventive Cardiology, Volume 30, Issue 18, December 2023, Pages 1975–1985, https://doi.org/10.1093/eurjpc/zwad229[5]

Despite the known association between daily activity and improved mortality, it wasn’t until 2015 that an insurer actively sought to implement wearable data in its policies. Discovery Ltd, a South Africa-based financial services group, implemented an incentive program whereby the company gifted its US-based policyholders a then first-of-its-kind Apple Watch as long as they met certain health goals, recorded on the watch, throughout the year. So successful was this program that it is still offered today.[6]

Other insurers developed similar initiatives. In 2017, UnitedHealthcare —no stranger to the fitness space, having sponsored a Pro Tour cycling team for almost a decade—partnered with Fitbit to offer customers credits back against the cost of their health insurance plan based on their activity data. This was the first time that a large insurer used activity metrics from wearables to directly reduce plan expenses.[7]

Both of these schemes rely on relatively simple data, like daily steps, to drive the benefit program. Most wearables, however, can measure much more than simple activity levels, and this data has the potential to be hugely informative for both the insurer and the customer.

Wearables can measure changes in resting heart rate, a metric that is strongly associated with increased mortality rates.[8] Some can detect heart rate variability (the variation in time between individual heart beats) and have been used to accurately predict the onset of COVID-19 in users.[9] Others can continuously monitor blood glucose levels, which would allow for more fair pricing strategies for diabetic policyholders who, through diet and exercise, are able to keep blood sugar levels within a stable range.

Data from wearables that offer sleep tracking and sleep stage analysis could be used to predict health outcomes and mortality. Poor sleep quality is associated with a myriad of health issues, including heart problems, immune disorders, obesity, and even all-cause mortality.[10],[11] Sleep is a relevant factor that insurers may wish to consider, and one that has not historically been addressed by traditional medical underwriting. Insurers could use sleep data from wearables to operate an incentive program encouraging policyholders to consistently get more sleep. Aetna Insurance already operates a bonus scheme for their employees who regularly report over seven hours of sleep, albeit relying on self-reporting. Could the use of wearables to track sleep be the next step?

An accurate snapshot of an individual’s baseline activity levels could provide useful information to supplement a medical examination, which would likely decrease underwriting costs in the long run.[12] More broadly, wearables offer information that a medical examination cannot. For example, two individuals might register similar medical metrics on a health examination (BMI, blood pressure, heart rate) and yet one person might be trending positively through regular exercise. Those healthy habits would ideally be considered when pricing premiums, as that individual would expect lower risk of death and fewer adverse health events.

If insurers can responsibly and efficiently leverage the wealth of data that modern wearables have on offer, they could develop schemes to attract new customers and strengthen existing customer relationships. Equally important is the potential risk reduction that this data could offer the insurer. Daily steps, exercise, and stress tracking all paint a simple picture of the general health of an individual. This picture would provide insurers more accurate information about their customer base and allow them to levy a more appropriate premium.

The upsides of using wearables–shared by customers and insurers—include the potential for more equitable pricing, risk reduction, increased underwriting efficiency, and a strengthened policyholder-insurer relationship. The process of realizing these goals is more complicated.

Data Security and Transparency

One of the primary roadblocks for wearable schemes and the wholesale adoption of insurers using wearable data is concerns around data privacy and security. A 2015 study on smartphone health data found that individuals who avoided downloading health apps did so because of data privacy concerns.[13] While individuals who already use wearables might be less apprehensive about data security issues, it could prove tricky for insurers to use wearable schemes to attract a wider berth of new customers if there is widespread apprehension about how the data is used and secured.

Insurers are no strangers to handling sensitive health data. Indeed, medical examinations have been a key part of underwriting for over 100 years. Wearables, however, present a different challenge than medical exams in that the process of data collection is not controlled by the insurer directly, meaning that it is difficult to fully mitigate data risk. One solution to this issue is to partner with reliable device manufacturers, which is what insurers have so far done.

In addition to data security, data transparency is a major concern for many individuals. While the use of wearables in insurance is simple today, advances in reliability and the continued adoption of wearable tech in the population could mean that insurers are able to develop more sophisticated pricing and risk models leveraging customer health data. While sound in principle, customers may shy away from “black box” methods in which the factors contributing to their premiums or recommendations are difficult to pinpoint.

Anti-discrimination

Wearable schemes that use steps as a target to reduce premium rates select against individuals who do not (or cannot) walk for extended periods of time (or at all), even if they are otherwise healthy. For example, not all wearables track wheelchair movement, which would class wheelchair users as sedentary even if they might be active. Other individuals may stay healthy in other ways, such as cycling or weightlifting, which is not accounted for in step data.

The possibility of using sleep data in insurance also raises a problem. In studies, women have demonstrated a greater prevalence of poor sleep quality than men.[14] Additionally, women are over 60% more likely than men to suffer from a sleep illness, for example, insomnia.[15] Given that the Financial Services Authority’s Gender Directive prohibits the use of gender as a factor in pricing and benefits, the use of sleep data should be critically considered.[16]

Wearables also present challenges specific to Black and Minority Ethnic (BAME) groups. Most glaring among these problems is that many wearables struggle to accurately track heart rate on darker-skinned individuals.[17] This problem must be fully addressed before heart rate data is used in the insurance context.

In a broader sense, using the additional health data that wearables collect could serve to further discriminate against already marginalised groups. A brief investigation into the state of healthcare in the UK highlights that BAME groups generally exhibit significantly worse health than the overall population.[18] Such is the state of this disparity that the NHS operates a Race and Ethnicity Observatory[19] to tackle inequalities experienced in health and healthcare by BAME patients, communities and the workforce. Differences in health outcomes have real consequences when it comes to premium ratings and insurability. Whether wearable data would contribute to these discrepancies should be carefully investigated before it is used in an insurance context.

The quandary of adverse selection

At the heart of the movement to incorporate wearable tech in insurance is the desire to reduce the information gap between policyholder and insurer. This would reduce adverse selection, wherein higher-risk customers receive coverage while paying rates similar to those of healthier customers.  Increasing the granularity of premium ratings to reduce adverse selection is fair in many ways. Policyholders end up paying premiums which more closely mirror their expected health outcomes. However, differentiating premiums has the knock-on effect of making insurance prohibitively expensive or impossible for higher-risk groups. On a society-wide level, this means that those who most need insurance might not qualify. Setting aside the moral implications of this, the elimination of adverse selection could lead to larger societal costs. This concept is brilliantly illustrated by Tadapar and Thomas in The Actuary. The following example and diagrams are inspired by their article.[20]

Assume the population can be cleanly divided into low-risk and high-risk customers, with low-risk customers outnumbering high-risk customers by four-to-one. An insurer charges each group an appropriate risk-differentiated premium (£0.01 and £0.04 respectively) that represents the exact amount of risk associated with each individual. Assuming that half of each group is willing and able to pay at their differentiated premium levels, the resulting loss coverage is

(4 * 0.01 + 1 * 0.01) / (8 * 0.01 + 2 * 0.02) = 50%

That is, 50% of the total possible loss value in the population is being covered by insurance.

If instead we do away with risk-differentiated premiums and charge a single pooled premium (£0.03), then it is likely that some lower-risk individuals are no longer willing or able to pay (as this represents a significantly higher premium rate). A greater proportion of higher-risk individuals, however, are able to afford insurance. The resulting loss coverage is now

(1 * 0.01 + 2 * 0.04) / ( 8*0.01 + 2 * 0.04) = 56.25%

A greater proportion of total societal loss is protected in this scenario than in the scenario of no adverse selection.

Wearables offer a means to further differentiate policyholders, which could mean that higher-risk policyholders are more easily targeted with higher premiums. Although the use of risk-differentiated premiums to avoid adverse selection is already a well-established practice in most areas of insurance, the promise of wearables which allow for even further differentiation should be considered carefully if maximal societal benefit is desired.

A model is only as good as the data that feeds it. While many wearables on the market today are consistent when it comes to tracking steps and distance, some report differences as great as 1000 steps.[21],[22] In a similar way to addressing concerns about data security, insurers must have a process for vetting what devices customers can use if the data from those devices are to be used in benefit schemes. Discovery and UnitedHealthcare, for example, address this issue by partnering directly with specific devices (Apple, Fitbit) that they have vetted.

Relying on more complex health data, while likely informative, presents additional difficulties. Many advanced health metrics, including sleep quality, resting heart rate, and heart rate variability, rely heavily on accurate heart rate data. Gold standard devices for measuring heart rate, such as an electrocardiogram (ECG) device, use electrodes placed on the chest, which can accurately detect electrical impulses from the heart. Most wearables, on the other hand, calculate heart rate using an optical light sensor that detects subtle changes in red blood cell density. Optical heart rate data is much less reliable than ECG data because it can be adversely affected by movement, external light, the presence of water, hair density, and skin color.[23] Even the best commercially available devices, if worn incorrectly, will provide inaccurate data. Insurers should be wary of using data from devices that rely on optical heart rate until the technology measurably improves.

When it comes to other predictive health indicators, such as sleep quality, few wearables reliably agree with gold-standard data. A 2022 study tracked six popular wearables over several nights and compiled heart rate and sleep data. The study evaluated the accuracy of the device sleep data compared to polysomnography (PSG) data, in which electrodes placed on the head pick up on sleeping and waking brain waves.[24]

In each case, the wearable data differs from PSG data by as much as an hour or more. Data comparing the amount of time spent in different sleep stages (light, deep, REM) was even less precise. While some devices are probably accurate enough to track sleep trends over time, none could reasonably be considered valid for serious medical insights.

Where to go from here?

It’s not all doom and gloom in the forecast for wearables. Insurers have proven that when appropriately built into incentive structures, wearable data can prove both beneficial to the policyholder and profitable for the insurer.

The possibilities for future development are promising. Modern wearables gather far more data than what insurers currently use, which could in the future provide useful insights into customer health and wellbeing. However, concerns over data quality, data privacy, and data transparency must all be addressed before the implementation of more sophisticated wearable schemes.

If insurers can verify data integrity and clearly communicate with customers how wearable data is used, there is the promise of a future of reduced policyholder premiums, lower underwriting costs, and crucially, better health outcomes for customers.

Sources and further reading

[1] Leonardo da Vinci (1938). Edward MacCurdy (ed.). The Notebooks of Leonardo Da Vinci. New York: Reynal & Hitchcock. p. 166. ISBN 978-0-9737837-3-5

[2] https://www.researchgate.net/publication/5293674_

Revisiting_How_Many_Steps_Are_Enough

[3] Hatano Y. “Use of the pedometer for promoting daily walking exercise”. ICHPER-SD J. 1993;29:4–8

[4] https://yougov.co.uk/topics/technology/trackers/brits-use-of-wearable-devices-eg-a-smartwatch-or-wearable-fitness-band

[5] https://academic.oup.com/eurjpc/article/30/18/1975/7226309

[6]https://www.discovery.co.za/discovery_coza/web/linked_content

/pdfs/vitality/apple_watch/vitality_active_rewards_apple_watch_benefit_guide.pdf

[7] https://www.cnbc.com/2017/01/05/unitedhealthcare-and-fitbit-to-pay-users-up-to-1500-to-use-devices.html

[8] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4754196/

[9] https://www.whoop.com/gb/en/thelocker/whoop-data-coronavirus/

[10] https://www.sleepfoundation.org/physical-health#:~:text=A%20lack%20of%20high%2Dquality,against

%20diseases%20and%20medical%20conditions

[11]https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2864873/#:~:text=

Those%20who%20reported%20at%20both,2.30%20%5B0.94%20to%205.60%5D

[12] https://ideas.repec.org/a/bla/rmgtin/v21y2018i3p389-411.html

[13] https://mhealth.jmir.org/2015/4/e101/

[14] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5302457/

[15] https://www.frontiersin.org/articles/10.3389/fpsyt.2020.577429/full

[16] https://www.fca.org.uk/publication/archive/fsa-gender-directive.pdf

[17] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9662769/

[18] https://www.parliament.uk/globalassets/documents/post/postpn276.pdf

[19] https://www.nhsrho.org/about-us/

[20] https://www.theactuary.com/features/2017/05/2017/05/09/appetite-selection

[21] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7241630/

[22] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5360892/

[23] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9662769/

[24] https://www.mdpi.com/1424-8220/22/16/6317/htm

 

In addition to the sources mentioned here, https://bmcmedinformdecismak.biomedcentral.com/articles/10.1186/s12911-022-01851-4,  uses a ranking-type Delphi structure to examine the challenges that wearables programs may present insurers and proved useful when researching this topic.

 

Matthias Wuest

May 2024