Health-Promoting Behaviors and Their Associations With Factors Related to Well-Being Among Former National Football League Players: An NFL-LONG Study

Participants

Former NFL players who played at least 1 NFL season were eligible to participate in this study.² Individual NFL teams, former players representing individual team alumni associations, and former players in the NFL Legends community were contacted and offered participation via hard-copy mailing through an NFL Legends partner to the entire NFL Legends community (n = 15 025) and via electronic mail to an overlapping sample through the same NFL Legends partner (n = 11 645). Hard-copy questionnaires were distributed in the fall of 2019 and electronic communications were sent in 3 waves, including spring 2019, summer 2019, and spring 2020. All questionnaire respondents were screened by a member of the study team to determine their eligibility to participate.

Data Collection

The General Health Survey was modified from prior versions with input from epidemiologists, athletic trainers, neuropsychologists, physicians, and other sports medicine professionals.^7,11 The present survey was updated from previous versions with more extensive and detailed queries of health history (eg, medical diagnoses), current health status (comorbidities, symptoms, and treatment), self-reported functioning (PROMIS measures across multiple domains), experiences related to the transition from playing football, and past and current concerns about their health. Five former NFL players critically reviewed the survey instrument and provided specific feedback to the study team to improve the questionnaire’s flow and clarity. The study population historically consisted of predominantly non-Hispanic White/Caucasian- and Black/African-American-identifying men from across a wide range of socioeconomic statuses. Our study recruitment methods using paper and electronic formats were intended to reach and be accessible to all former NFL players. Participants selected and completed either the electronic or hard-copy questionnaire. A member of the study team was available via phone and email to address any questions or concerns regarding participation or completion of the survey, regardless of the mode of self-administration.

The present study instrument included demographic information, participation in selected health-related behaviors, and current self-reported functioning and well-being measures. Demographic information included age, race/ethnicity, body mass and height (which were used to calculate body mass index), and current relationship status (married or cohabitating, not married or cohabitating, or other relationship status). Behavioral information captured included frequencies of participation in moderate-to-vigorous physical activities, resistance training, other wellness activities, sedentary/sitting activities, screen-related activities, and alcohol intake. Self-reported functioning and well-being measures included validated inventories of items related to meaning and purpose, self-efficacy, ability to participate in social roles and activities, physical functioning, pain interference, and emotional support. Specific details for these items are described below.

Current Health-Promoting Behaviors: Exercise, Diet, and Sleep

Exercise and other wellness activity frequencies were reported as the number of days per week over the past 4 weeks that participants engaged in each type of activity. Moderate-to-vigorous aerobic exercise was defined as “exercise that increases breathing rate and heart rate (eg, jogging, cycling),” resistance training was defined as “resistance training or weight-lifting,” and other wellness related activities were defined as “other wellness activities (eg, yoga, meditation).” No minimum session durations were required for any participant response.

The Rapid Eating Assessment for Participants-Shortened Version (REAPS) was used to assess diet quality.²⁴ The REAPS 16-item survey was validated in comparison to an established food frequency questionnaire (Pearson r correlation values of up to 0.516)²⁴ and a longer measure of dietary quality (r = 0.227),²⁵ and moderate internal consistency (Cronbach’s α = 0.65)²⁶ has been recently reported. This instrument captures information about food consumption frequency by type (eg, whole grains, fruits, and processed meats) in an average week as well as meal preparation and willingness to make dietary changes to improve health via 5-point Likert scale from 1 (very willing) to 5 (not at all willing). Diet quality was calculated as the sum of each of the 13 scorable REAPS items, ranging from 13 (worst) to 39 (best).

Sleep duration was calculated as the average number of self-reported hours slept per night weighted by weekday and weekend nights when considering the past 4 weeks.² Sleep disturbance was self-reported using the PROMIS-29 Profile version 2.0 items related to sleep disturbance.¹⁹ These items were each scored on 5-point Likert scales from very poor to very good in response to the item “My sleep quality was …” and from not at all to very much in response to “My sleep was refreshing,” “I had a problem with my sleep,” and “I had difficulty falling asleep.” This 4-item sleep disturbance inventory has strong internal consistency (Cronbach’s α of up to 0.88) across large groups of individuals with heterogenous demographic and health histories.¹⁹ Individual item scores were summed and standardized as T-scores based on national normative data (mean ± standard deviation [SD] = 50 ± 10), with a higher T-score indicating more sleep disturbance (a less adaptive/worse outcome).

Self-Reported Well-Being: Meaning and Purpose, Self-Efficacy, and Social Participation

Factors related to well-being were measured via the PROMIS Adult Meaning and Purpose SF-4a,²⁰ PROMIS Adult Self-Efficacy SF-4a,²¹ and the Ability to Participate in Social Roles and Activities 4-item scale from the PROMIS-29 Profile version 2.0.¹⁹ Each scale consisted of 4 separate items rated on a 5-point Likert scale regarding the previous 7 days. Meaning and Purpose responses ranged from not at all to very much in response to items such as “My life has meaning.” Self-Efficacy responses ranged from I am not at all confident to I am very confident in response to items such as “I can handle whatever comes my way.” Social participation responses ranged from never to always in response to items such as “I have trouble doing all of the family activities I want to do.” The summed items within each measure were standardized as T-scores based on national normative data (mean ± SD = 50 ± 10). Higher T-scores on each measure indicated more favorable outcomes, and scores of ≥60 may be interpreted as 1 SD better outcome scores than the general population, whereas the opposite is true for scores of ≤40.

Analysis Covariates

Covariates were selected a priori based on previous literature.^{2,4,16,27–31} Demographic covariates included age, race/ethnicity (dichotomous; identified as White/non-Hispanic versus non-White), and current marital status (dichotomous; married/cohabitating with a partner versus not cohabitating). Behavioral covariates included current frequency of alcohol intake (ordinal; never, monthly or less, 2 to 4 times a month, 2 to 3 times a week, or 4 or more times a week) and number of days/week spent sitting (ordinal; described as at a desk, reading, or talking to friends), and screen time (ordinal; defined as looking at a screen to watch a show [television, computer, tablet, phone]). Hours of weekly sitting and screen time were not calculable with the present data. Self-reported functioning covariates were Physical Function and Pain Interference domains from the PROMIS-29 Profile version 2.0 and the PROMIS Emotional Support SF-4a (Supplemental Tables 5–7, available online at http://dx.doi.org/10.4085/1062-6050-0537.23.S1).^19,30

Self-reported lifetime concussion history was also considered as a covariate based on previous work with this same sample reporting statistically significant associations between concussion history and other self-reported measures of cognitive function, depression, anxiety, and emotional-behavioral dyscontrol.² A definition of concussion was provided to participants who were then asked to report the number of concussions they suspected they sustained on an ordinal level (0, 1, 2, 3 …10+) in each level of football participation (before high school, during high school, during college, and during professional play) as well as from other sources (eg, military or motor vehicle accident). The sum of concussions reported across all categories was binned into 5 categories (0, 1 to 2, 3 to 5, 6 to 9, and 10+) to represent lifetime concussion history.

Data Processing and Statistical Analyses

To describe former NFL player’s engagement in health-promoting behaviors, summary statistics were computed for the frequency of participating in exercise, diet quality, sleep duration, and sleep disturbance. Additionally, proportions of participants engaging in moderate-to-vigorous exercise ≥5 days/week as well as those engaging in resistance training ≥2 days/week were computed, with reference to physical activity guidelines from the Department of Health and Human Services in the United States, and proportions of participants sleeping between 7 and 9 hours/night were computed, with reference to recommendations provided by the American Academy of Sleep Medicine and Sleep Research Society.^32,33 Descriptive statistics were computed for the full study sample and for the following individual age subgroups: <30 years (n = 150), 30 to 39 years (n = 360), 40 to 49 years (n = 285), 50 to 59 years (n = 321), 60 to 69 years (n = 360), and 70+ years (n = 308). Multiple imputation methods were used to address missingness, similar to previous work (see Supplemental Material).² Summary scores, but not individual items, were imputed for the REAPS and PROMIS measures.

To measure the strength of the associations between health-promoting behaviors and well-being-related factors, we compared the effects of health-promoting behaviors on player self-reported Meaning and Purpose, Self-Efficacy, and Ability to Participate in Social Roles and Activities. Separate multivariable linear regression models were fit for each outcome, first in the full sample and separately within each individual age group, for a total of 21 models. We included frequencies participating in moderate-to-vigorous activities, resistance exercise, and other wellness activities, REAPS summary score, sleep duration, and sleep disturbance as the primary predictors of interest alongside the demographic, behavioral, and self-reported functioning and well-being covariates described above. Age was not included as a separate covariate in models within each individual age group. When fitting the regression models from the multiple imputation datasets, standardized z score values with a mean of 0 and SD of 1 were used for each predictor and outcome. Variance inflation factors were calculated to probe multicollinearity using a recommended cutoff point of 10.³⁴ Adjusted R² and standardized β coefficients (β) were calculated for each model, with a priori α set to .05. All analyses were performed with SPSS version 28.0.

Current Health-Promoting Behaviors: Exercise, Diet, and Sleep

Overall, former players reported participating in approximately 3 days/week of moderate-to-vigorous exercise, 2 days/week of resistance training, and 1 day/week of other wellness activities (Table 2). Approximately 24% of all players reported participating in moderate-to-vigorous exercise ≥5 days/week, and 66% reported resistance training ≥2 days/week. Age-related trends for meeting these cutoffs were similar to the age-related trends observed in median weekly frequency of each exercise type (Table 2). The highest proportion of participants engaged in moderate-to-vigorous exercise ≥5 days/week was observed in former players aged 70+ years (28.3%), whereas the smallest proportion was observed in former players aged 40 to 49 years (20.9%). Conversely, the youngest age group of former players, those aged <30 years, reported the highest proportion of meeting strength training recommendations (79.5%), whereas former players aged 70+ years reported the smallest frequency (58.5%).

Table 2. Health-Promoting Behaviors Across the Lifetime^a

View Fullscreen

REAPS summary scores for diet quality were similar across age groups, with average scores near 28 for each individual age group (Table 2). Nearly half of the former players ate ≥8 ounces of meat per day, whereas about 20% had ≤2 servings of fruit and 2 servings of whole grain products per day (Table 3). Among REAPS items not included in the summary score, 82% indicated that they eat at home rather than eating out, and 80% reported a high degree of willingness to make changes to eating habits to be healthier (Table 3).

Table 3. REAPS Individual Scored Item Responses^a

View Fullscreen

Average nightly sleep duration for this sample was just under 8 hours/night (Table 2). Former players between 30 and 49 years of age reported the lowest average number of hours/night of sleep (7.3), and those aged 40 to 49 years also reported the lowest frequency of sleeping between 7 and 9 hours/night (60.1%). The longest average nightly sleep duration was observed in the oldest former player group, those aged 70+ years, at 8.4 hours/night; however, the highest proportion of participants sleeping within the recommended range was observed in the participants <30 years old (72.9%; Table 2). Sleep disturbance T-scores approximated normative values (mean ± SD = 50 ± 10) in the full sample and for each separate age group (Table 2).

Associations of Health-Promoting Behaviors With Factors Related to Well-Being

Meaning and Purpose, Self-Efficacy, and Ability to Participate in Social Roles and Activities in our sample approximated normative values for the general population across all age groups (Table 1 and Figure).

View Figure

• Download as Powerpoint
• Download Figure

Meaning and Purpose

In the full sample, more frequent moderate-to-vigorous exercise and less disturbed sleep were associated with greater Meaning and Purpose (Table 4). Among separate age groups, moderate-to-vigorous exercise was only associated with Meaning and Purpose among participants 50 to 59 years of age, and sleep disturbance was associated with all age ranges except <30 years of age. Longer sleep durations were also associated with lower Meaning and Purpose only among those 70+ years of age (Table 4).

Table 4. Associations Between Health-Promoting Behaviors and Well-Being in Former National Football League Players

View Fullscreen

Exercise and Other Wellness Activities

Overall, only 23.8% of all participants participated in ≥5 days/week of moderate-to-vigorous activity, whereas most former NFL players reported participating in moderate-to-vigorous exercise fewer than half of the days in each week. Study participants did not report duration of participation in each exercise session, so we cannot directly classify whether these former players were meeting recommendations for American adults regarding moderate-to-vigorous exercise: 150 to 300 minutes per week of moderate or 75 to 150 minutes of vigorous activity or equivalent combination.³² Although this is a limitation of this study, our data suggest that many former NFL players may not have been meeting those recommendations, and this is something that should be evaluated more directly in clinical settings and future research to determine aerobic exercise engagement more specifically. Conversely, most players reported >2 days/week of resistance training, which is in line with current recommendations.³² Taken together, these seminal data on exercise-related behaviors in former NFL players show that most players do not abandon exercise after their professional career has ended, and approximately one-quarter may still be highly engaged in aerobic and resistance training activities as evidenced by a median of 5 combined sessions per week. As a result, former players may be open to clinical exercise prescriptions and recommendations to improve their health and well-being.

Clinical and lifestyle interventions that help former NFL players engage in aerobic exercise activities may be important given findings from this study. More frequent moderate-to-vigorous exercise was associated with more adaptive (ie, better) outcomes among all of the well-being-related measures in our study, particularly among those between 40 and 59 years of age (Table 4). These findings suggest that moderate-to-vigorous exercise has a particularly meaningful relationship with well-being among former professional football players in these midlife years, despite negligible differences in weekly moderate-to-vigorous exercise participation between age groups. It is possible that former players in this age group have the most to gain from exercise in this period of life, when they are also reporting worse functioning and well-being overall than their younger and older counterparts. For example, in recent work with an overlapping sample, former players between 40 and 59 years of age reported less adaptive (ie, worse) functioning than both younger and older players on measures of anxiety, depressive symptoms, and emotional-behavioral dyscontrol, and moderate-to-vigorous exercise participation was also associated with better outcomes across each of these measures.^2,17 In combination with previous work, our study supports the need to identify exercise intervention targets, such as engaging in more frequent moderate-to-vigorous aerobic exercise, as a strategy for improving well-being alongside the prevention of numerous noncommunicable disorders that commonly occur in mid-to-later life, including cardiovascular disease and dementia.^2–4 Clinicians treating patients in this population should also consider the roles that injury, surgery, and other comorbidity history play in affecting both engagement with exercise and feelings of well-being over time.

Diet Quality

Overall diet quality scores were relatively homogenous across age groups. Most participants reported that they usually/often or sometimes did not meet REAPS item benchmarks (Table 3). The lowest scoring item overall was related to frequent consumption of ≥8 ounces of meat per day, consistent with a prior study of former football players using a longer REAPS instrument.³ We did not measure dietary preferences in this study, so we cannot speak to whether participants were following vegetarian, vegan, or other dietary patterns that intentionally limited consumption of meat. Still, responses for individual REAPS items in our study denote room for improvement in multiple domains of dietary intake. More than 80% of participants indicated that they would be willing to make changes in their eating habits to be healthier, and, therefore, dietary intervention could be a beneficial clinical target alongside exercise and sleep. There is no well-established threshold or screening criteria for REAPS summary scores to interpret “healthy” eating versus not healthy eating; however, the average summary score for former NFL players in our study (28.0 ± 4.3) was below average scores reported in generally healthy young adults (33.6 ± 3.1) and similar to scores reported in a sample of middle-aged individuals seeking weight-loss surgery (28.3 ± 4.0).^25,26 REAPS summary scores can be used as a starting point for identifying who may need specific improvements to dietary habits, as may be informed by addressing specific items within the REAPS questionnaire. Interventions centered on individual self-efficacy may promote healthy eating behaviors,³⁵ and this framework may benefit former NFL players such as those in our study who are willing to change their eating to improve their health.

In a study of former football players by Hinton et al, better diet quality (higher REAP score) was positively associated with better outcome scores on crude, single-item measures of overall physical and mental health as well as fewer reported cognitive difficulties.³ However, that study used a longer version of the instrument (31 items) than was used in our study (16 items), so we cannot directly compare summary scores between studies. In the present study, diet quality was significantly associated with better self-efficacy only among participants <40 years of age. It is possible that higher perceived self-efficacy in this sample influenced younger former players’ choices to eat healthier. Alternatively, the ability to make healthy dietary intake choices, including financial freedom and access to a variety of foods, may support these players’ feelings of self-efficacy.^35,36 Regardless, more information is needed as to why this pattern was observed in only the youngest age groups. Where dietary intake patterns such as the Mediterranean diet have been associated with reduced risks for numerous disease processes, there is little known about the effects of specific nutrients on subjective well-being.^37–40 It is reasonable to consider that because dietary intake is related to physical health status, better subjective well-being and related factors could be downstream effects of overall good physical health. Yet, the opposite may also be true that better subjective well-being facilitates engagement in health-promoting activities such as healthy eating. In this light, dietary intake is a valuable component of health and well-being. Dietary intake should be considered a readily modifiable intervention target in former football players.

Sleep Duration and Quality

Average sleep durations for all age groups were in accordance with current recommendations for healthy adults to obtain 7 to 9 hours of sleep/night.³³ Similarly, sleep disturbance scores approximated US norms across all ages.¹⁹ Compared with the other age groups, former players aged 40 to 59 years reported the lowest proportions for meeting recommended sleep durations, and this coincided with the highest relative sleep disturbance scores. However, these differences were relatively small, particularly for sleep disturbance scores. Possible reasons for these age-dependent patterns could be familial (eg, number of young children) or vocational (eg, early building of a new career after professional football) or related to myriad other factors that were not measured in this study. Former players 70+ years of age were the only ones to report average sleep of more than 8 hours per night, and longer sleep durations were associated with worse outcomes in this group alone. In the general population, sleep duration generally declines with older age, where average nightly sleep durations appear to plateau around 7 hours/night after age 40.⁴¹ We are unable to determine what specific factors are associated with the discrepancy between participants in our study and expected changes in the general population. One possible explanation could be the relative prevalence of mild cognitive impairment and dementia in former football players, which, although not included in the present study, also increase in prevalence with older age.^{10–12,42–44} Mild cognitive impairment and dementia-related conditions are simultaneously related to longer sleep durations and worse health-related quality of life, potentially resulting from reduced independence in daily life, which are each related to measures of well-being.^45,46 More work is needed to understand the prevalence and effects of mild cognitive impairment and dementia-related conditions in former NFL players, particularly toward measuring their associations with other facets of health and well-being and identifying points for intervention. Monitoring clinical history and comorbidity in former football players may be essential for understanding and intervening on health-related behaviors like sleep in clinical practice. Our study findings support the premise that sleep duration and other facets of sleep hygiene are potentially actionable clinical targets to improve well-being in former football players.

Further Considerations

Comorbidities, injury history, and social-behavioral factors might be associated with both participation in health-promoting behaviors and self-reported factors related to functioning and well-being. However, health promotion and morbidity prevention strategies in clinical practice also need to account for the application of this knowledge in real-world settings, including factors that impede or support their implementation. Therefore, evaluation of barriers and facilitators for meeting established health-behavior guidelines and recommendations by former NFL players could help inform the implementation of interventions to improve health and well-being in this population, potentially including behavioral change approaches.^35,47

Limitations

Despite our large sample size, the survey response was relatively low (a minimum response frequency of 1784/15 025 [11.9%]), and as a result, our findings may lack generalizability across all former NFL players.² Given the self-reported nature of the questionnaire, participants may have under-estimated/overestimated their engagement in health-promoting behaviors. Further, we can only make inferences on behaviors and factors related to well-being within the relatively narrow time frame antecedent to their completion of the study questionnaire. However, the health-promoting behaviors and well-being outcomes were reflective of the month preceding study participation and may therefore be a fair representation of the relationship in that frame. Importantly, some participants submitted their questionnaire to us in electronic or paper format during the first month after the coronavirus disease 2019 national emergency declaration (March 13, 2020), but we are unable to determine how many of those forms were actually completed within this timeframe. The decision to extract data for analyses was informed, in part, by drastic slowing of survey responses and the beginning of the pandemic. That said, there is not likely to be a meaningful proportion of participants who completed the questionnaire in this time frame compared with those who did not. Ultimately, it cannot be determined how this phase of the pandemic might have affected participants’ engagement in exercise, diet, and sleep behaviors. Longitudinal studies are needed to understand the evolving nature of health, well-being, and behaviors in former NFL players over time. Future studies should use objective measures where possible (eg, actigraphy) and also account for medical diagnoses, disabilities, socioeconomic status, and other barriers and facilitators to meeting recommendations for health-related behaviors, such as personal (eg, chronic injury), interpersonal (eg, social support), and environmental (eg, presence of sidewalks in the community) factors.