Fatigue Risk Management Preferences for Consumer Sleep Technologies and Data Sharing in Aviation

METHODS

Subjects

Professional opinions from pilots and crewmembers who routinely travel for work were elicited from across the global aviation community through social media, email, and word of mouth. Any crewmember who works in an aircraft while it is in transit (e.g., pilots or cabin crew) was considered eligible for inclusion; aviation professionals who do not work in an aircraft while it is in transit were not eligible for this survey (e.g., air traffic controllers or ground crew). Efforts were taken to recruit respondents from across the world and across aviation industries by directly contacting regional, national, and international airlines, pilot and crewmember associations, aviation societies, advocacy groups, and unions, as well as the use of social media tags. Fig. 1 shows a flow chart of subject inclusion based on response criteria. Respondents needed to indicate that they held a position in aviation and needed to respond to at least one survey question related to sleep or CST use to be included in the final analysis. Respondents did not need to complete the entire survey to be included in subsequent analysis but did need to provide at least one viable response to be included. A substantial proportion of respondents indicated a role in aeromedical transport via write-in response. Aeromedical transport, sometimes called air ambulance, medical evacuation (medevac), or helicopter emergency medical services (HEMS), is an emergency medical service that involves the use of an aircraft to move patients from one location to another. An aeromedical transport category was added and examined in subsequent analyses. This study was approved with exempt status by the Salus Institutional Review Board (Protocol Number IBR2023P) and these analyses were conducted in accordance with the Declaration of Helsinki.

View Figure

• Download as Powerpoint
• Download Figure

RESULTS

As shown in Fig. 1, the final study population consisted of N = 149 respondents. Distribution of respondents by job position, operation length, and organization type is summarized in Table I. All respondents completed between 38–100% of the survey, with N = 72 respondents completing 100% of the survey. Responses were unequally distributed across job positions (χ² = 99.86, P < 0.001), industry (χ² = 158.36, P < 0.001), and normal operation length, defined as short haul (SH), medium haul (MH), long haul (LH), ultra-long range (ULR), or on-call operations (χ² = 202.23, P < 0.001). As shown in Table I, the greatest number of respondents indicated “pilot” as their job position. Commercial was the most common industry and SH operations were the most common normal operation length.

Table I. Respondents by Job Position and Operation Length.

View Fullscreen

Rank order responses to Q4–Q5 are depicted in Fig. 2. Items are displayed in descending rank order and are numbered by rank. Fig. 2A shows respondents’ ranking of important sleep issues that impact their ability to perform work activities. “Nighttime operations” and “inconsistent sleep routine” ranked as the most important. “Jet lag”, “snoring/sleep apnea”, and “inability to sleep/insomnia” were ranked as the least important sleep factors affecting ability to perform work activities. Fig. 2B depicts perceived importance about information about sleep. Information about “sleep quality history” followed by “projected alertness levels” were ranked as most important for personal sleep management. “Tips for improving sleep hygiene” or “advice on when to exercise or consume caffeine” were ranked as the least important information. Respondents did have the option of entering additional write-in responses for each question but did not name any other sleep issues or information about sleep.

View Figure

• Download as Powerpoint
• Download Figure

Responses about current CST use, reasons for using or not using a CST, and brand preferences are depicted in Fig. 3. As shown in Fig. 3A, 43% of respondents indicated that they currently use a CST. Fig. 3B depicts these respondents’ main reasons for using a CST. Sleep and fatigue tracking was the most reported reason for wearing a CST in this group. Fig. 3C shows the main reasons for not using a CST by the 46% of respondents who responded negatively to Q6. No perceived need for a device and simply not owning a device were the most reported reasons in this group. Some respondents provided multiple reasons for not using a CST; each reason was independently coded into an appropriate category.

View Figure

• Download as Powerpoint
• Download Figure

CST users were additionally asked if they had a preferred brand. Largely, respondents did not have a brand preference (30%, N = 19). The most frequently named device brand was Garmin (21%; N = 13), followed by Apple (17%, N = 11), then Fitbit (10%, N = 6), Oura ring (6%, N = 4), Sleep Cycle (5%, N = 3), Samsung (5%, N = 3), Whoop (5%, N = 3), Google Fit (1%, N = 1), and CrewAlert (1%, N = 1).

For Q7, N = 65 respondents (43%) indicated a preference for a wrist-worn smart watch CST device, followed by a wrist-worn device with no smartwatch capabilities (16%, N = 24) or a device worn on the finger (14%, N = 20). Less than 10% of respondents preferred a sleep-tracking mobile app like Sleep Cycle or CrewAlert (6%, N = 9), provided a write-in response for continuous positive airway pressure (CPAP) monitoring (1%, N = 1), or preferred no device at all (5%, N = 8). No respondents indicated a preference for a digital sleep diary or head-worn device. Of the respondents, 14% (N = 21) did not provide a response.

Respondents were evenly split between those who reported using a CST (N = 64) and those who reported not using a CST (N = 68), with no answer at all from N = 17 respondents. There were no significant differences between users and nonusers concerning sleep issues that affect performance (Q4; z = 0.00,P = 1.00), information about sleep (Q5; z = 0.49, P = 0.62), preferences in CST type (Q7; z = 0.00, P = 1.00), or data sharing (Q8; z = 0.00, P = 1.00). There were no differences in ranking on any question between either CST users or nonusers compared to the overall results (all P > 0.52).

Respondents’ likelihood to share data with their organizations under different scenarios is summarized in Fig. 4. The scenario under which respondents were most likely to provide data was if they were given a sleep tracker to keep indefinitely by their organization. Respondents were least likely to provide data if they were required to complete an electronic sleep diary for up to 6 wk.

View Figure

• Download as Powerpoint
• Download Figure

Four subgroups met the population threshold (≥N = 15) for subgroup analysis. The pilot subgroup consisted of N = 108 respondents while the flight medic subgroup consisted ofN = 34 individuals. A total of N = 87 respondents flew SH operations and N = 38 flew MH operations. There was considerable overlap between job positions and normal operation length, such that 52% of pilots (N = 56) and 78% (N = 27) of flight medics flew SH; 31% of pilots (N = 33) and 6% of flight medics (N = 2) flew MH. Regarding CST use, 44% (N = 47) of pilots and 38% of flight medics (N = 13) used a CST; 36% of respondents who flew SH (N = 31) and 58% of respondents who flew MH (N = 22) used a CST. There were no significant differences in subgroup responses for Q4, Q5, Q7, and Q8 for MH, SH, pilots, or medics compared to overall results between SH and MH, or between pilots and medics (all P > 0.31). Due to the overlap in job positions, industries, and operation lengths, no further statistical tests between subgroups were appropriate.

DISCUSSION

Fatigue risk management in aviation depends upon objectively collected sleep and fatigue data that can inform decisions about flight-duty limitations. Research-grade actigraphy has historically been used to collect sleep data for FRMS, but recent advancements in commercial sleep-tracking technology could open up more opportunities for aviation organizations to collect sleep data from crewmembers in a low burden manner. Developing a strategy to update data collection methods for FRMS begins with understanding crewmembers’ preferences about CSTs and data sharing. The goal of this survey was to establish which sleep data collection methods would appeal to the greatest number of aviation professionals based on job roles, operation lengths, or specific industry. Importantly, this survey assessed demand for CSTs in aviation professionals in the context of work and FRMS data collection. Thus, findings may not generalize to aviation professionals’ preferences about CSTs for personal use.

Overwhelmingly, respondents agreed that the most important sleep issues that impacted their work performance were nighttime operations and inconsistent sleep routine (see Fig. 1A). Circadian misalignment and sleep disruption are considered two major biological factors that contribute to fatigue in aviation.^4,6,7 Aviation FRMS commonly use biomathematical models of fatigue to evaluate the fatigue risk associated with a planned schedule or operation based on sleep loss and circadian misalignment. The aviation industry also takes a proactive approach in educating crewmembers about the causes of fatigue. The findings depicted in Fig. 2A suggest that crewmembers are in agreement with the scientific and risk management communities regarding causes of fatigue that impact performance.

Sleep quality history and projected alertness levels ranked as the most important information for managing sleep, shown in Fig. 2B. Sleep quality may have higher importance than sleep duration to aircrew because their time has schedule constraints. Maximizing sleep quality may be the only option to improve performance during a limited time window that does not allow for longer sleep duration. Inconsistent sleep routines and nighttime operations can both lead to diminished sleep quality that can accrue over time. Perhaps information about how well a crewmember slept combined with a prediction of their alertness for the upcoming day is more salient than information about the overall duration of sleep or advice that may not be applicable given their work constraints.

Sleep quality history outranked similar metrics like sleep depth/sleep scoring or sleep quality score. It therefore would seem that sleep quality is viewed separately from sleep depth or sleep scores by the aviation community. Sleep efficiency, or the ratio between the time a person spends asleep over the total time dedicated to sleep, is a common measure of objective sleep quality in the scientific community¹⁶ and could represent what respondents want as feedback about objective sleep quality. Respondents could have been interested in a self-report measure of their own subjective sleep quality. An important follow-up survey would be to investigate the most salient measure of sleep quality over time to provide feedback to aviation professionals.

Another interesting point is that some information that is commonly provided by apps, such as sleep duration or recommendations to improve sleep hygiene, was ranked as having low importance in the current study while projected alertness, which is not a standard metric provided by apps, was ranked second overall. There seems to be a market gap between information that CST developers feel is important to report and information that aviation professionals feel is important to their ability to perform. There are a few mobile apps such as SleepTank,¹⁷ 2BAlert,¹⁸ or CrewAlert¹⁹ that aim to predict alertness in relation to sleep history. These apps have been developed with military or aviation populations in mind, but are not available yet for all platforms or devices.

Interestingly, recommendations about napping, exercise, optimal caffeine consumption, or sleep hygiene tips ranked as the least important perceived information about sleep, as shown in Fig. 2B. This finding seems counterintuitive given the potential benefit that tactical napping, caffeine optimization, exercise, or sleep hyigene techniques may have for an occupational population in a safety-sensitive industry like aviation. It may be that the respondents in this study are unfamiliar with how these techniques may benefit them, have already adopted these techniques, feel that they cannot adequately apply them given the operational constraints of their jobs, or are disinterested in making additional behavioral changes to accommodate their work performance. An interesting follow-up study will be to investigate why aviation professionals viewed actionable advice from a wearable as low value.

Wrist-worn devices were the preferred type of sleep tracker for work purposes. Smart watch sleep trackers and wrist-worn fitness bands with no watch face were selected by a combined 60% of respondents, with devices worn on the finger as the third most popular choice. Aviation professionals frequently sleep in different locations for work, such as an onboard rest facility or a hotel, so a device that can continuously be worn on the body makes practical sense. A previous survey asked real-world sleep researchers their preferences about CSTs and also showed a majority preference for wrist-worn devices, despite the expert respondents reporting “brain activity in combination with motor activity and biometrics” as the most accurate method of measuring sleep.¹³ Respondents in the current survey may not be familiar with portable technologies that can measure brain activity, but were provided with the option of a “headband” device on Q7. Respondents in the previous study were members of the sleep research community with experience collecting in the field¹³ and, so, were familiar with fieldable CSTs beyond wrist-worn options and were also familiar with compliance problems when working outside the lab.

Wrist-worn devices may be preferrable to real-world sleep researchers because subjects are more likely to wear them consistently. The main goal of a sleep data collection for FRMS is to assess whether the working conditions allow sufficient opportunities for sleep rather than to investigate the biological underpinnings of sleep physiology. Device removal can result in periods of no data that can skew or render the dataset useless. Subject compliance becomes more important than the ability of the device to accurately measure sleep on an epoch-by-epoch basis. The best possible technology may not always be the most accurate technology, but perhaps may be the most convenient technology in some cases.

Device brand loyalty was not strong among CST users, a finding which bodes well for organizations looking to purchase devices for FRMS. This also indicates that there is a market opportunity for CST manufacturers to create a device that can accurately measure sleep in aviation’s unique sleeping environment. Only current CST users were asked if they have a preferred device, of whom 30% indicated that they did not have a preference. The most frequently reported devices were Garmin, which were preferred by 21% of CST users (N = 13). Five of the reported preferred brands can be categorized as wrist-worn smart watch trackers (Garmin, Apple, Fitbit, Samsung, WHOOP) and one was a ring-based device (Oura). This supports the finding that most aircrew prefer a wrist-worn smart watch CST. Reifman et al. recently showed in a meta-analysis of device validation studies that currently available Fitbit and Oura devices, but not current versions of Garmin or WHOOP, produced sleep measurements that are operationally acceptable for fatigue management.¹¹ Apple and Samsung devices were not included in that meta-analysis. Three of the preferred trackers were mobile apps (Sleep Cycle, CrewAlert, Google Health) that either require completing a sleep diary or linking the app to a CST. Nonusers’ reported “no perceived need”, “no device ownership”, and “cost” as common reasons for not using a CST, as shown in Fig. 3B. While some nonusers reported concerns about privacy, comfort, the accuracy of the measurements, or unfamiliarity with the technology, these results suggest that nonusers could be convinced to wear a device if they were given one at no personal cost.

The majority of respondents reported high likelihood of providing data through the use of a sleep tracker that has been given to them by the organization. Interestingly, even respondents who were current CST users still preferred to wear a device purchased by the organization over linking their personal device to a third-party data extraction app. Respondents were least likely to provide data through a sleep diary, indicating that if an organization wants to use an app to collect sleep data for FRMS, the app should be able to link to a wearable device. Respondents were not asked if they had ever provided data for FRMS before. This limitation means that we cannot evaluate whether respondents’ likelihood ratings were tempered by past experiences. Common methods for data collection in FRMS have been actigraphy or sleep diary for a limited time period (up to 6 wk). It may be that respondents’ previous experience with data collection influenced their responses. For example, completing a sleep diary requires more effort on the part of the subject than does wearing an actigraph and can be frustrating. Someone with prior experience as a study subject may have stronger opinions about data collection methods than a respondent who has not done a data collection before.

This report is not without limitations. Responses are not evenly distributed across all sectors of aviation, restricting our ability to compare results with statistical robustness across groups. This is an ongoing study, so efforts will be taken to increase recruitment of survey respondents in meagerly represented categories, such as flight attendants/cabin crew (N = 7), LH (N = 14), or ULR (N = 5) crewmembers. Rank order tests are also qualitative in nature and are most likely not the best method to determine differences between subgroups. These reasons may be why there were no statistical differences in rank order between the overall sample population and subgroups. The results of this survey should be interpreted as preliminary and descriptive. Since the survey is ongoing, follow-up analyses can assess whether there are differences between groups based on position or operation length once a larger number of subjects have responded for each category of interest. Reasons for fatigue are known to differ by operation length,^5,20 so it is important to evaluate fatigue independently across different operational parameters.

Secondly, respondents were not asked to provide information about their region of operation, gender, ethnicity, or level of experience—information that could intersect with issues of fatigue and performance. The recruitment material was also only provided in English and posted on North American-based social media platforms, such as LinkedIn, Instagram, and Twitter/X or live presentation in North America and Europe, which may skew recruitment toward aviation professionals from those regions. Efforts were taken to increase exposure of the recruitment materials to crewmembers globally as well as crewmembers who may identify as an ethnic or gender minority, but these demographic data were not measured by the survey. Therefore, it is impossible to determine between-group differences or gauge the possibility of biased results.

Thirdly, this survey assesses demand in a population that may not fully appreciate the current shortcomings of consumer sleep trackers. Previous surveys in this series have assessed CST demand within a sleep researcher population,¹³ as well as demand for scientific accuracy in a CST in a general consumer population,¹² but respondents in this study were not asked about scientific evaluation of CSTs. This was done because any device used for FRMS would need to meet regulatory requirements for validity, so the onus of determining the appropriateness of a device falls on the regulator or operator rather than the individual crewmember. Finally, this survey is tightly focused on sleep-tracking and data sharing, thus it does not encompass the breadth of fatigue issues in aviation. Sleep is a major component of readiness however, so it is our hope that maximizing compliance to data collection within the sleep domain will lead to a better understanding of the impact of fatigue on performance in general.

In conclusion, pilots, flight medics, and cabin crew from commercial aviation, aeromedical transport, cargo aviation, business aviation, and other aviation industries would be more likely to provide sleep data to their organization through a third-party mobile app if they are given a wrist-worn CST to wear by the organization than if they need to link their personal device or complete a sleep diary. Respondents did not indicate strong preferences overall regarding brand loyalty or adopting the use of a CST for FRMS purposes. Crewmembers are most concerned about the impact of nighttime operations and inconsistent sleep routines on their ability to perform and would most appreciate feedback about sleep quality history or predictions about next day alertness. Providing crewmembers with CSTs that simultaneously collect data and provide feedback to improve sleep may doubly benefit FRMS efforts by increasing the flow of information between the organization and the individual worker.