Enhancing Sleep Efficiency on Vessels in the Tug/Towboat/Barge Industry (2016)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

13 The overall objective of this research was to develop a compendium of best practices for enhanc- ing sleep efficiency on towboats in the U.S. tug/towboat/barge industry. This chapter describes the research approach. The research team has leveraged its prior work/experience with the U.S. tug/towboat/barge industry and the data collected in what it termed Phases I through IV, as well as a new assess- ment termed Phase V, to address the tasks proposed as part of NCFRP Project 45. A summary of these assessments is provided in Table 1, including the study phase, group(s) assessed, number of individuals studied, and the type of the assessment. Data from all five phases are used in this report, but particular focus is given to Phases IV and V. 2.1 Task 1. Kickoff Meetings The first task (Task 1) of this NCFRP research was a teleconference with the NCFRP project panel. A separate stakeholder presentation was given at the AWO Combined Annual Meeting in January of 2014. 2.1.1 NCFRP Panel Meeting At the initiation of the project, the three key members of the research team participated in a kickoff conference call with the NCFRP project panel. At this meeting, the research plan was reviewed. 2.1.2 Stakeholder Meeting A stakeholder kickoff meeting was then held at the AWO Combined Annual Meeting of the Midwest, Ohio Valley & Southern Regions in New Orleans in January 2014. The NCFRP research plan was presented to the attendees and questions. 2.2 Task 2. Evaluate Current Operational Practices Task 2 was to identify and describe the metrics to be used to evaluate current operational inter- ventions (e.g., educational materials and programs; noise abatement; sleep disorders screening— especially sleep apnea, and wellness and nutritional programs) for their effectiveness in improving sleep efficiency on tug/towboat/barge vessels. In order to evaluate the current operational practices related to improving sleep efficiency, two main strategies were used: 1. Assess operational practices already in place across the industry using an online survey and structured interviews with key personnel involved in fatigue management, and Research Approach C H A P T E R 2

14 Enhancing Sleep Efficiency on Vessels in the Tug/Towboat/Barge Industry 2. Survey crew members from the previous Phase IV study to assess the impact of current opera- tional practices on their sleep efficiency. This survey also included an assessment of the edu- cational materials distributed as part of the previous Phase IV study. While the assessments focused on key areas in the evaluation of current practices related to sleep efficiency, open-ended questions were also asked to elicit discussion about related practices that may be unique to a particular individual, company, or region within the U.S. tug/towboat/ barge industry. The key areas included in the survey were: • Noise abatement (ear plugs, white noise generators, quiet times), • Sleeping quarters [e.g., light reduction, beds (size, mattress), sharing, sound proofing], • Training/education related to sleep or sleep-related health factors, • Screening for sleep disorders, • Wellness programs, • Stress management, • Diet/nutrition, • Exercise, • Medical conditions that impact sleep, • Medications that impact sleep, and • Identification of other practices. In addition to collecting information on current practices related to sleep efficiency, the research team asked the representatives of the companies whether they monitored the success of the prac- tice, and in an attempt to determine whether these practices were successful, crew members were asked whether they thought the current fatigue management practices of their company influenced their behavior. This assessment was done as a Likert Scale to rate the practice from “very effective” to “not at all effective” in influencing sleep efficiency. In addition, if a company had data they were willing to share on the effectiveness of current practices, the research team collected that information, assessed its validity, and, where appro- priate, included it in this report. Phase Who Assessed # Assessed Type of Assessment I – White Paper: Split Sleep & Napping NA NA • Review of literature II - Assessment and Intervention of Anchor- Sleep/Nap-Sleep Strategy All crew on vessel 50 • Actigraphy/sleep log • Questionnaire • Anchor-sleep/nap-sleep intervention III – Develop Education Intervention and Assessment of Sleep All crew on vessel 70 • Actigraphy/sleep log • Questionnaire • Education intervention IV – Comparison of Sleep on Vessel and at Home Captains and pilots only 161 • Sleep log and questionnaire • No intervention • Educational materials provided at conclusion of the study V – Develop Best Practices • Crew • Management • Other Industries • Government • Academics • Consultants • N=40 • N=53 • N=6 • N=3 • N=11 • N=4 • Survey/questionnaire • Interview • Review of publicly available sources Table 1. Summary of research in American waterways, 2008 to 2014.

Research Approach 15 The team also included a few key questions in the survey of crew members to determine whether they had taken advantage of the programs provided by their company, and, if so, what the per- ceived benefits or weaknesses of the program were, particularly in impacting sleep efficiency. Crew without access to these types of programs were asked to rate how likely they would be to use such a program and how beneficial they believe such a program would be if available to them. 2.2.1 Management Survey to Evaluate Current Operational Practices With the aid of the AWO, the research team reached out to approximately 200 AWO member company representatives to complete an online survey (see Appendix B for full survey). The research team then invited several of the representatives that completed the survey to participate in a phone interview. Representatives were selected for an interview based on their responses to the survey questions, including their willingness to participate in an interview, com- pany size, primary shift schedule utilized, and whether they had various fatigue management practices in place. This selection process allowed the research team to interview representatives from a range of companies and to gather further information about specific practices and how they may have been implemented. See Table 2. 2.2.2 Crew Member Survey to Evaluate Operational Practices For the second part of Task 2, 231 crew members who had previously consented to participate in the Phase IV study were approached. In the original Phase IV study sufficient data was received from 163 of the consented crew members. Initially, crew who were determined from their Phase IV data to be short, moderate, or long sleepers were approached to participate; however, due to recruitment difficulty all Phase IV crew were invited to participate (N = 231). Crew members were contacted (up to a total of seven times). The crew members who agreed to participate were given a start date that coincided with their next rotation on the vessel. Crew participation in this research was between June and October 2014. # # Crew Employed Primary Schedule Secondary Schedule 1. 1078 6:6:6:6 2. 126 6:6:6:6 3. 262 6:6:6:6 4. 225, 100, 550* 6:6:6:6 12:12 5. 40 4:8:4:8 6. 149 6:6:6:6 7. 152 6:6:6:6 12:12 8. 34 12:12 9. 160 6:6:6:6 12:12 10. 900, 1250* 6:6:6:6 11. 12 6:6:6:6 12:12 12. 100, 550* 6:6:6:6 4:4 13. 36 12:12 14. 50 6:6:6:6 12:12 * Indicates that responses from multiple representatives of the same company (different operations of the same larger company) were received that included different total numbers of crew employed. All responses are listed. Table 2. Summary of the characteristics of the companies from the management survey respondents invited to be interviewed.

16 Enhancing Sleep Efficiency on Vessels in the Tug/Towboat/Barge Industry There were two levels of participation for the crew. Level 1 was a survey only (see Table 3 in Sec- tion 2.4 for a summary of survey items and Appendix B for a full copy of the survey), which was given to all crew members who consented to be part of this research. Level 2 was a work and sleep diary that was to be completed at the end of each rest interval (see Appendix B for a copy of the diary). Those who were determined to be short, moderate, or long sleepers (based on sufficient data from Phase IV to group crew by average sleep duration; a total possible invitees was N = 136) were invited to complete this diary (see Section 2.4.1). The diary was to be completed for a total of 14 days, days 1 through 7 of rotation on the vessel and days 15 through 21 of rotation on the vessel. 2.3 Task 3. Evaluate the Use of Anchor-Sleep/ Nap-Sleep Strategies The completion of this task involved two main steps: • Determining the proportion of crew who use anchor-sleep/nap-sleep strategies: – Re-analysis of Phase II to IV trials to identify crew members who used anchor-sleep/nap-sleep strategies, – Surveying all crew that participated in the Phase IV trial (for Phase V). • Using mathematical models to predict performance based on actual sleep/wake and work schedule. 2.3.1 Determine the Proportion of Crew Who Use Anchor-Sleep/ Nap-Sleep Strategies For Phases II and III, sleep diaries and wrist actigraphy monitoring were completed by all participating crew on a total of 16 vessels for up to 12 days. For Phases IV and V, sleep diaries were completed for up to 14 days for each sleep opportunity, regardless of whether the crew member slept or not (see Appendix B for a copy of the diary). The analysis of the existing data set from the Phases II through IV trials specifically assessed the prevalence of the use of the anchor-sleep/nap-sleep strategies. In order to do this, a variable (yes or no) was created to identify whether a crew member is deemed to use an anchor-sleep/ nap-sleep strategy. If a crew member consistently slept or attempted to sleep for more than 1 hour in both of the sleep opportunities per 24-hour period on at least 90% of the recorded days, they were deemed to use an anchor-sleep/nap-sleep strategy. 2.3.2 Use Mathematical Models to Predict Performance Based on Actual Sleep-Wake and Work Schedule As part of the evaluation of nap-sleep/anchor-sleep strategies within the tug/towboat/barge industry, a currently available mathematical model was used to predict performance from sleep- wake history of actual crew members working in the industry. These models are a cost effec- tive and scientifically valid way of predicting performance using a combination of sleep-wake and work history. There are many different models available, but most share the same math- ematical “guts,” the “two-process model.” One such model, the Sleep, Activity, Fatigue, and Task Effectiveness/Fatigue Avoidance Scheduling Tool (SAFTE/FAST) model (Hursh et al. 2004), is in broad use by the commercial aviation and rail industries, as well as government organizations, including the USCG and the U.S.DOT. The SAFTE/FAST predicts performance as the interaction of the linear decline in perfor- mance resulting from increasing time awake or the homeostatic sleep drive (Process H) and the

Research Approach 17 sinusoidal variation in performance resulting from the circadian rhythm, the 24-hour rhythm in body temperature (Process C). This two-process model was originally developed to predict high- amplitude, slow electrical activity on the surface of the brain and, thus, is firmly rooted in brain physiology (Borbely 1982). It was later discovered that the same two-process model accurately predicted alertness, sleepiness, and performance. As input, the SAFTE/FAST model takes the sleep-wake history of a study participant. The sleep-wake history derived from a wrist activity monitor (actigraph) or created from a sleep-work diary is a minute-by-minute record of whether the participant is awake or asleep, with waking typically coded as a “1” and sleep as a “0.” Thus, a 10-minute sleep-wake history beginning at 08:00 with the participant awake for the first 5 minutes and asleep for the last 5 minutes would look like the following: 08:01 hours – 1 08:02 hours – 1 08:03 hours – 1 08:04 hours – 1 08:05 hours – 1 08:06 hours – 0 08:07 hours – 0 08:08 hours – 0 08:09 hours – 0 08:10 hours – 0 Taking this sleep-wake history as input, and estimating the position of the circadian rhythm relative to sunrise and sunset, the SAFTE/FAST model produces a performance prediction on a scale of 0 to 100, updated every minute. The model remembers the performance prediction from the previous minute and updates this prediction, decreasing predicted performance if, in the current minute, the person was awake and increasing predicted performance if, in the current minute, the person was asleep or awake as modulated by time of day to take into account the circadian rhythm. The model accurately predicts performance. The SAFTE/FAST model was used to translate the minute-by-minute sleep-wake history into a minute-by-minute performance prediction, thus, interpreting the meaning of the sleep-wake history in terms of its effects on operational performance. Figure 1 is an example of how a mathematical model (in this case, SAFTE/FAST) can be used to predict the performance of a single crew member. In the Phase IV study, sleep-wake histories were collected for both on-duty days and off-duty days using a sleep-work diary. Simply looking at a sleep-wake history provides limited information as one would like to know how this history translates into performance as modulated by circadian rhythmicity (Process C) and homeostatic sleep drive (Process H). Another approach is to directly measure performance using, for example, a psychomotor vigi- lance task (PVT). A PVT takes at least 5 minutes to complete and needs to be executed at regular and frequent intervals, and the tester must be focused on the test; thus, the PVT is too obtrusive and not flexible under the research conditions. An alternative (as described above) is to take the sleep-wake history and use it as input to a mathematical model using homeostatic sleep drive (Process H) and circadian rhythmicity (Process C) to predict minute-to-minute performance. Modeling in this way predicted performance using the sleep-wake histories from the sleep-work diary of a captain who participated in the Phase IV study across approximately 1 week while on board. The results of this modeling are depicted in Figure 1 and is presented here as an example of how the SAFTE/FAST model can be used to predict performance based on sleep-wake history and circadian phase.

18 Enhancing Sleep Efficiency on Vessels in the Tug/Towboat/Barge Industry In Figure 1 nighttime is indicated by gray bars, sleep by blue bars, work shift by black bars, and predicted performance by black (waking) and blue (sleeping) curved lines. The X-axis indicates times (days and hours) and the Y-axis indicates predicted performance. The background colors (green, yellow, and red) indicate increasingly impaired performance. In this model, a night of sleep is missed and the effect of this can be seen by a dip in performance. As the week progresses, however, the participant gets sufficient sleep and performance improves. There are troughs in performance in the early hours of the morning and peaks in the evening; this is due to the circadian rhythms produced by the internal circadian clock. Good performance is maintained throughout the week despite the participant splitting their sleep into two periods over each 24-hour period. 2.4 Task 4. Identify Barriers to Adopting Good Sleep Management Practice and Develop Practices to Overcome These Barriers The completion of this task involved four main steps: • Identifying factors that predict the 20% best and 20% worst sleepers. • Identifying crew members who changed sleep behaviors since the Phase IV trial. • Evaluating current best practices. • Using the identified factors to model changes in sleep-wake and then using mathematical modeling to predict performance. Figure 1. Participant 134 (Captain) spending 1 week on board at work.

Research Approach 19 In order to identify barriers that inhibit tug/towboat/barge personnel from adopting good sleep management practices and then develop best practices to overcome these barriers, exten- sive analyses were conducted of the new data collected as part of this research and relevant data from prior studies were re-examined. As part of these efforts, those groups of individuals who did or did not change their sleep-wake behaviors were identified, the degree determined, and why. By identifying the multivariate factors (see Table 3) that enable some crew members to obtain adequate sleep (7 to 8 hours) per 24 hours, and the factors that are associated with not obtaining sufficient sleep (<6.5 hours) per 24 hours, the research team was better able to prepare the list of best practices and assess the effectiveness of practices already in place. Table 3. Summary of factors that can impact sleep that were used in various statistical models to determine best (and worst) sleep practices in crew members. FACTOR MEASUREMENT NOTES Age * Date of birth BMI* Self-report height and weight Sleep apnea risk Berlin* STOP Bang Caffeine use* Standard shiftwork index* Sleep diary* Smoking status* Standard shiftwork index* Sleep diary* Years on maritime schedule or other shift work* History Physical health* Standard shiftwork index* Mental health* Standard shiftwork index* Job satisfaction* Standard shiftwork index* Sleepiness* Epworth Sleepiness Scale* Karolinska Sleepiness Scale* Samn Perreli Single administration Repeated at each start and end of work period Sleep duration Time in bed* Total sleep time* Modified Pittsburgh sleep quality index* For each rest interval and per 24- hour period Typical sleep period Sleep quality* Sleep diary* Sleep latency, wake after sleep onset, self-reported sleep quality Noise* Sleep diary* Survey Vibration Survey Sleep diary* Family burden* Standard shiftwork index* Stress* Sleep diary* Sleep disorders Identified risk or known diagnosis Survey Obstructive sleep apnea, restless leg syndrome, insomnia, narcolepsy Medications* Standard shiftwork index* Sleep diary* Prescription, over-the-counter, supplements Techniques used to improve sleep Survey White noise, ear plugs, relaxation techniques, etc. Gender* Survey Commute time* Standard shiftwork index* * These items were used in the Phases II-IV research on crew members on towing vessels. All sleep diary factors are available for each sleep period.

20 Enhancing Sleep Efficiency on Vessels in the Tug/Towboat/Barge Industry 2.4.1 Identify Factors That Predict the 20% Best and 20% Worst Sleepers Using daily sleep data, the average daily sleep amount for each subject was first calculated and then all subjects were ranked based on their daily sleep amount. The top 20 percentile of the subjects as long sleepers (the daily sleep amount is longer than 8.6 hours) and the bottom 20 percentile of the subjects as short sleepers (the daily sleep amount is shorter than 6.6 hours) were defined. Subjects who slept on average 6.6 to 8.6 hours a day (i.e., the remaining 60% of the subjects) were thus defined as moderate sleepers. To identify the factors that correlated with short or long sleep, various statistical approaches were used in the analysis of a number of different measures listed in Table 3. Repeated measures analysis with linear mixed-effect models were used to investigate how sleep durations during each sleep opportunity are influenced by sleep factors of interest. A univariate model was used to select sleep factors that are associated with sleep durations, and significant factors (p < 0.1) were then entered into a full model that models the linear combination of multiple sleep fac- tors. In order to identify the factors that predict long and short sleepers, the 20% longest and 20% shortest sleepers were also directly contrasted. A conventional generalized linear regression model (GLM) was used to assess the multivariate associations of the factors influencing sleep with the sleep outcomes. Similar to the mixed-effect model, each of the factors influencing sleep was entered into the GLM as a main-effect term. In the GLM, coefficients of all factors influenc- ing sleep were tested and their standardized values were ordered so as to evaluate the relative importance of the factors influencing sleep that are predictably related to sleep outcomes. The originally proposed GLM, generalized boosted model (GBM), and marginal structural model (MSM) do not handle the longitudinal (measures over several consecutive days) sleep duration measures. The longitudinal feature of the data set via linear mixed-effect models, which is better suited for reporting the factors that are related to sleep duration, was also addressed. 2.4.2 Identify Crew Members Who Changed Sleep Behaviors Since Phase IV Trial Assessment of the changes in behavior included asking crew members whether they had changed any behaviors since the previous assessment, and to provide details of the changes. Questions included (1) Following your participation in the previous sleep research study, did you change any of your usual behaviors to improve your sleep and (2) Can you provide examples of the behaviors you have changed (e.g., reduced caffeine intake close to bed, ear plugs, eye shades) as a consequence of your participation in the sleep research study? 2.4.3 Evaluate Current Best Practices In order to evaluate current best practices, both crew members and management were sepa- rately surveyed (see Section 2.3 for general approach). As part of the crew survey, participants were asked to list practices related to sleep that they currently used or have been educated about and to rate their effectiveness on a Likert Scale. Examples of the questions used included (1) Have you ever received CEMS training; and (2) If so, how was the CEMS training received (i.e., did you have a course for a day, online, handouts, etc.); (3) Did you find the CEMS training useful; and (4) If you found the CEMS training useful what aspects of the training did you find the most useful? For a complete copy of questions related to current best practices, see Questions 1 through 7 in the crew survey provided in Appendix B.

Research Approach 21 As part of the management survey the research team asked respondents to rate the effective- ness of a particular practice if their company had it in place. The research team also followed up with selected respondents by conducting interviews (see Section 3.2.1.1) to determine whether they had data that they would be willing to share that quantified the effectiveness of the practice. 2.4.4 Use the Identified Factors to Model Changes in Sleep-Wake and Then Use Mathematical Modeling to Predict Performance Following the multivariate regression modeling, key factors that impact sleep duration and sleep efficiency were determined. Once done, these factors were used to predict sleep duration and then this information was used to create predictions of fatigue using mathematical models to predict performance as modulated by circadian rhythmicity and homeostatic sleep drive (SAFTE/ FAST model). This is similar to the example provided in Figure 1. 2.5 Task 5. Prepare a List of Best Practices That Could Be Implemented by the Waterways Industry, Companies, Crews, or Individuals to Enhance Sleep Efficiency The development of the list of best practices was approached in three ways: • Determining current best practices with proven success from other industries that may either be tailored to the tug/towboat/barge industry or be directly transferable. • Determining practices/interventions already in place in the tug/towboat/barge industry that have been successful. • Proposing new best practices based on the findings of this report. As part of this task, some suggestions for the implementation of the best practices were suggested. In order to achieve the goals of Task 5, the research team utilized the findings from Tasks 2 through 4, and also conducted an extensive review of the literature related to best practices aimed at improving sleep already available from other industries, including the trucking, airline, motor coach and railroad industries (see Table 8 in Chapter 3 for a summary and Appendix A for a Bibliography). 2.6 Task 6. Prepare a Compendium of Best Practices for Enhancing Sleep Efficiency on Towboats in the U.S. Inland Waterway Industry and a Report Documenting the Results of the Research Section 3.5 presents a compendium of best practices for enhancing sleep efficiency on tow- boats in the U.S. tug/towboat/barge industry. This report documents the results of the com- pleted research.