A 12-month prospective cohort study on symptoms of mental health disorders among Dutch former elite athletes

ABSTRACT

Objectives

The primary objectives of the study were to (i) establish the 12-month incidence and comorbidity of symptoms of mental health disorders (distress, anxiety/depression, sleep disturbance, alcohol misuse, disordered eating) among Dutch former elite athletes and (ii) explore the potential relationship with the stressors involuntary retirement, recent life events, career dissatisfaction. The secondary objective was to investigate whether the incidence of symptoms of mental health disorders and their potential relationship with stressors are related to duration since retirement.

Methods

An observational prospective cohort study with a 12-month follow-up was conducted among Dutch former elite athletes. Symptoms of mental health disorders and potential stressors were assessed using validated questionnaires.

Results

A total of 282 participants were included at baseline and 193 completed follow-up. The incidence ranged from 7% for alcohol misuse to 28% for anxiety/depression. Comorbidity of two or three symptoms of mental health disorders was reported in 7% and 4%, respectively. Adverse life events were significantly related to disordered eating (RR = 1.30, 95%CI = 1.05–1.61), while career dissatisfaction also showed a significant relationship with disordered eating (RR = 3.98, 95%CI = 1.32–11.99) and sleep disturbance (RR = 3.23, 95%CI = 1.10–9.51). The stressor involuntary retirement did not have a significant relationship with any symptoms of mental health disorders. The effect of duration since retirement seemed to be most present in the first 15 years since athletic career retirement in the plotted graphs.

Conclusions

The 12-month incidence of symptoms of mental health disorders ranged from 7% to 28% and comorbidity of symptoms of mental health disorders was present in 12%. Adverse life events and career dissatisfaction only increased the risk of certain symptoms of mental health disorders. Duration since retirement might affect the incidence of symptoms of mental health disorders and its relationship with stressors in the first 15 years.

KEYWORDS:

• Sport
• athlete
• mental health
• incidence

1. Introduction

Worldwide, around 20% of the general population experiences one or more symptoms of mental health disorders in a period of 12 months [1]. Symptoms of mental health disorders describe a mental and emotional state of self-reported adverse or abnormal thoughts, feelings, and/or behavior that might lead to functional impairments, but are generally not as severe as clinically diagnosed mental health disorders [2]. Symptoms of mental health disorders include symptoms of distress, anxiety/depression, sleep disturbance, alcohol misuse, and disordered eating. Physical activity and exercise can contribute to the prevention of symptoms of mental health disorders and can counterbalance the effects of symptoms of mental health disorders on quality of life [3]. However, at elite and competitive levels, sport might lead to symptoms of mental health disorders. In a systematic review among international current elite athletes, symptoms of mental health disorders ranged from 19% for alcohol misuse to 34% for anxiety/depression [4].

During their sports career, elite athletes encounter many organizational stressors, which can be divided into the categories: leadership and personnel issues, cultural and team issues, logistical and environmental issues, and performance and personal issues [5]. These stressors can cause strain, displayed by emotional, attitudinal, and behavioral responses, and can induce symptoms of mental health disorders [6]. However, after athletic career retirement, elite athletes face new challenges. Former elite athletes can experience loss of daily structure, athletic identity, support, and purpose [7]. Furthermore, they are confronted with an altered social status, vocational responsibilities, and life events [8]. In addition to these potential generic stressors, former elite athletes might also have to deal with more sport-specific stressors such as involuntary retirement, bodily changes, and career dissatisfaction [8]. These potential stressors predispose former elite athletes to the development of symptoms of mental health disorders. Among retired professional football (soccer) players, the 12-month incidence of symptoms of mental health disorders ranged from 11% for distress to 29% for anxiety/depression [9]. Retired professional football players who experienced one or more life events reported a higher incidence of distress and disordered sleeping [9]. In retired professional ice-hockey players, the 6-month incidence of symptoms of mental health disorders ranged from 8% for distress, anxiety/depression, and disordered eating to 25% for sleep disturbance [10]. Career dissatisfaction was associated with an increase in distress and sleep disturbance in these retired professional ice-hockey players [10]. Likewise, former professional rugby players reported a higher incidence of symptoms of distress after involuntary retirement compared to voluntary retirement [11].

Although symptoms of mental health disorders in former elite athletes have been studied increasingly during the past decade, current knowledge is primarily based on cross-sectional studies. So far, three prospective studies about symptoms of mental health disorders and potential stressors were conducted among male athletes, retired from team sports [9,10,12]. Yet, the representativeness of these findings was limited by low response rates and/or small sample sizes. A prospective study identifying the incidence of symptoms of mental health disorders and their relationship with potential stressors in a large group of former elite athletes from various sports would enhance the current knowledge.

Therefore, the primary objectives of the study were to (i) establish the 12-month incidence and comorbidity of symptoms of mental health disorders (distress, anxiety/depression, sleep disturbance, alcohol misuse, disordered eating) among former elite athletes with a wide range of duration since retirement and (ii) explore the potential relationship with the stressors involuntary retirement, recent life events, and career dissatisfaction. We hypothesized that former elite athletes experience a higher incidence and comorbidity of symptoms of mental health disorders compared to the general population and that former elite athletes exposed to a stressor (or combination of stressors) have a higher risk of developing symptoms of mental health disorders than former elite athletes not exposed to one or more stressors. The secondary objective of the study was to investigate whether the incidence of symptoms of mental health disorders and their potential relationship with stressors are related to duration since retirement. We hypothesized that former elite athletes with a shorter duration since retirement report a higher incidence of symptoms of mental health disorders and a stronger relationship with potential stressors compared to former elite athletes with a longer duration since retirement.

2. Methods

2.1. Design

Reported in compliance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement, the present study was an observational prospective cohort study with three measurements over a follow-up period of 12 months by means of questionnaires [13]. The study design was similar to a previous study that was conducted among current Dutch elite athletes [7]. Ethical approval was provided by the Medical Ethics Review Committee of the Academic Medical Center (W15_051#15.0062; Amsterdam, the Netherlands). The present research was conducted in accordance with the Declaration of Helsinki [14]. Baseline questionnaires were distributed online between October and December 2015.

2.2. Study setting, participants and sample size

Participants were former elite athletes from the Netherlands Olympic Committee*Netherlands Sports Confederation (NOC*NSF). NOC*NSF is the Dutch association that provides financial, material, performance, or social advice and support to both current and former elite athletes. Former elite athletes from NOC*NSF have committed significant time to elite sports and have competed at international or highest national sport level (eligible for Olympic games and national team competitions). Inclusion criteria for the participants were: (i) being on the NOC*NSF list of official former elite athletes and (ii) being fluent in Dutch. With regard to the first aim of the study (12-month incidence), sample size calculation indicated that at least 130 former athletes were required (power of 80%; confidence interval (CI) of 95%; precision of 5%) under the assumption that one out of five former athletes might suffer from a mental health symptom [15].

2.3. Dependent variables: symptoms of mental health disorders

Symptoms of distress, anxiety/depression, sleep disturbance, alcohol misuse, and disordered eating were measured at baseline (presence/absence in previous 4 weeks) and twice during the follow-up (presence/absence in previous 6 months). All dependent variables were measured using validated questionnaires and dichotomized according to the optimal cutoff point of the relevant questionnaires. Distress was measured using the Distress Screener (three items scored on a 3-point scale) which is based on the four-dimensional symptom questionnaire (4DSQ) (e.g. ‘Did you suffer from worry?’) [16,17]. The 4DSQ, i.e. Distress Screener, has been validated in several languages including Dutch and/or used for different recall periods (internal consistency: 0.6–0.7; test–retest coefficients: ≥0.9; criterion-related validity: sensitivity 0.8, specificity 0.7, Area Under ROC Curve ≥0.8) [16,17]. A total score ranging from 0 to 6 was obtained by summing up the answers on the three items, a total score of 4 or more indicating the presence of distress [16,17]. The 12-item General Health Questionnaire (GHQ-12) was used to assess psychological symptoms related to anxiety/depression (e.g. ‘Have you felt under strain?’) [18]. The GHQ-12 has been validated in several languages including Dutch and/or used for different recall periods (internal consistency: 0.7–0.9; criterion-related validity: sensitivity 0.8, specificity 0.8, Area Under ROC Curve ≥0.8) [18]. Based on the traditional scoring system, a total score ranging from 0 to 12 was calculated by summing up the answers on the 12 items, with a score of 3 or more indicating signs of anxiety/depression (Area Under Curve = 0.9) [18]. Based on the Patient-Reported Outcomes Measurement Information Service, PROMIS (short form), sleep disturbance was assessed through four single questions (e.g. ‘Did you have some problem to sleep?’) scored on a 5-point scale (from ‘not at all’ to ‘very much’) [19]. The PROMIS has been validated in several languages including Dutch and/or used for different recall periods (internal consistency: >0.9; construct validity: product-moment correlations ≥0.9) (for detailed information, see www.nihpromis.org). A total score ranging from 1 to 20 is obtained by summing up the answers to the four questions, a score of 13 or more indicating the presence of sleep disturbance [19]. Alcohol misuse was detected using the 3-item Alcohol Use Disorders Identification Test, AUDIT-C (e.g. ‘How many standard drinks containing alcohol do you have on a typical day?’) [20]. The AUDIT-C has been validated in several languages including Dutch and/or used for different recall periods (test–retest coefficients: 0.6–0.9; criterion-related validity: sensitivity 0.8, specificity 0.8, Area Under ROC Curve ≥0.7) [20]. A total score ranging from 0 to 12 was obtained by summing up the answers on the three items, a score of 5 or more indicating the presence of alcohol misuse [20,21]. The Eating disorder Screen for Primary care (5-items scored as ‘yes’ or ‘no’; ‘0ʹ for a normal response, ‘1ʹ for an abnormal response) was used as a screening instrument to detect disordered eating (e.g. ‘Were you satisfied with your eating patterns?’, a ‘no’ would be considered as an abnormal response) [22]. The Eating disorder Screen for Primary care has been validated in several languages including Dutch and/or used for different recall periods (criterion-related validity: sensitivity 1.0, specificity 0.7) [22]. A total score ranging from 0 to 5 is obtained by summing up the answers on the five items, a score of 2 or more indicating the presence of disordered eating [22].

2.4. Independent variables: stressors

Involuntary retirement, number of recent life events, and career dissatisfaction were measured at baseline (and during follow-up). Involuntary retirement (i.e. forced to retire), was explored through a single question (‘Did you retire from elite sport voluntarily’; yes or no). If retired involuntarily, respondents were subsequently asked to mention the principal reason (single question). Based on the validated Social Athletic Readjustment Rating Scale, the occurrence of life events (e.g. ‘Death of spouse,’ ‘Change in financial state’) in the previous 6 months was explored by 14 single questions (yes or no) [23]. The score was calculated by summing up the life events that occurred in the previous 6 months and the sum score was also subsequently transformed into a dichotomous variable: not exposed to a recent life event or exposed to one or more recent life events. Elite sports career dissatisfaction was explored through the validated Greenhaus scale (e.g. ‘I am satisfied with the success I have achieved in my career’) (five items on a 5-point scale) [24]. A total score (5–25) was obtained by summing up the answers to the five items and this sum score was also subsequently transformed as a dichotomous variable: 5–12 as being dissatisfied with elite sports career and 13–25 as being satisfied [24].

2.5. Procedures

Based on the independent and dependent variables included in the study, baseline and two follow-up electronic anonymous questionnaires available in Dutch were created (FluidSurveys^TM). At baseline, the following descriptive variables were inquired: age, gender, height, weight, type of sport, duration of elite sports career, duration since retirement, educational level, employment status, and the number of working or studying hours. In order to guarantee the strict confidentiality of the responses, no personally identifiable information was included in the questionnaires and codes were used. Each questionnaire took about 15 to 20 minutes to complete. Information about the purpose and procedures of the study was sent per e-mail to potential participants by the medical staff of NOC*NSF. Participants interested in the study gave their informed electronic consent and were given access to the baseline online questionnaire which they were asked to complete within 2 weeks. After completion of the baseline questionnaire, participants could give informed electronic consent for the follow-up questionnaire. At 6 and 12 months of follow-up, participants were invited by e-mail to complete the second and third questionnaires within 2 weeks. Reminders at baseline and follow-up were sent after two and 4 weeks. Once completed, the electronic questionnaires were saved automatically on a secured electronic server (Amsterdam UMC-location AMC) that only the principal investigator could access. Participants participated voluntarily in the study and did not receive any reward for their participation.

2.6. Statistical methods

All data analyses were performed using the statistical software IBM SPSS Statistics 25.0 for Windows. Descriptive data analyses (mean, standard deviation, frequency, median, interquartile range) were performed with the descriptive and independent variables measured at baseline. To explore whether the loss to follow-up was selective, we compared baseline characteristics (age, gender, height, weight, duration of elite sports career, duration since retirement, and sport type) of non-responders and responders at follow-up by means of independent T-tests for normal distribution and Mann–Whitney U-test for no normal distribution of continuous variables, while using the Pearson Chi-square test for dichotomous variables [15].

The 12-month incidence and comorbidity of symptoms of mental health disorders were calculated over the follow-up period of 12 months, using the adjusted Wald method (incidence: a sample size of 150 persons or less) for 95% confidence intervals (95% CI) [15]. Incidence (expressed as a percentage) was calculated as the proportion of the number of participants with a newly given symptom of mental health disorders during the 12-month follow-up relative to the total number of participants without any symptoms of mental health disorders at baseline [15]. Comorbidity of two, three, four, or five simultaneous symptoms of mental health disorders was defined as the simultaneous presence of two, three, four, or five symptoms of mental health disorders (respectively) among the participants. Comorbidity was expressed as a percentage and calculated as the proportion of the number of participants with newly developed two, three, four, or five simultaneous symptoms of mental health disorders during the 12-month follow-up relative to the total number of participants without any symptoms of mental health disorders at baseline.

3In order to explore the strength of the relationship between potential stressors and the onset of symptoms of mental health disorders (dichotomous dependent variable) during the subsequent 12-month follow-up, relative risks (RR) and related 95% CI were calculated through a logistic regression analysis [15]. The logistic regression analysis was adjusted for duration since retirement since we recruited participants with a wide range of duration since retirement. If the 95% CI did not include the null value, the relationship between a symptom of mental health disorders and stressor was regarded as significant.

The effect of duration since retirement was descriptively explored by plotting a graph of the incidence of symptoms of mental health disorders and its potential relationship with stressors in Python using the library Matplotlib.pyplot.

2.7. Bias

Potential confounding variables were examined by calculating whether the loss to follow-up was selective and by searching the literature for known confounders. Recall bias was addressed by splitting the 12-month follow-up period into two periods of 6 months. Bias due to self-report was reduced by using short, clear questions to decrease misunderstanding and by making the survey anonymous to decrease social-desirable answers [25]. At baseline, participants that failed to report duration since retirement and symptoms of mental health disorders were excluded from further analysis. At follow-up, missing data were not included in the analysis.

3. Results

3.1. Participants

The study contacted 297 participants, from which 282 were included at baseline. However, 12 participants were excluded from the baseline analysis due to missing data. A total of 193 former elite athletes completed both follow-up questionnaires. The flowchart of the recruitment and follow-up of participants is presented in Figure 1. The mean age of the participants at baseline was 50 years (SD = 15 years) and the median duration since retirement was 17 years (interquartile range (IQR) = 6–31). Among these former elite athletes, 19% retired involuntarily and 21% were dissatisfied with their elite sports career (from which 31% retired involuntarily). At baseline, 18% of the participants reported experiencing distress, 30% anxiety/depression, 22% sleep disturbance, 23% alcohol misuse, and 28% disordered eating in the previous 4 weeks. All baseline characteristics of the participants are presented in Table 1. The reasons for nonparticipation and loss to follow-up could not be assessed. However, we found that the age of non-responders at follow-up was significantly higher (p = 0.018) than the age of responders, with a mean difference of approximately 5 years. Since we assumed that age and duration since retirement were related, we performed a post hoc analysis and tested for linearity. The deviation from linearity (sig. 0.838) in the ANOVA table was larger than 0.05, so we concluded that there was a linear relationship between the two variables. Therefore, data analysis was not adjusted for age.

Figure 1. Flowchart of the recruitment and follow-up of Dutch former elite athletes.

Table 1. Baseline characteristics of Dutch former elite athletes

	Total (N = 270)
Age (in years; mean ± SD)	50 ± 15
Male/female (%)	49/51
Height (in cm; mean ± SD)	179 ± 9
Weight (in kg; mean ± SD)	78 ± 15
Duration of elite sport career (in years; mean ± SD)	11 ± 5
Duration retirement (in years; median (IQR))	17 (6–31)
Team/Individual sport (%)	60/40
Educational level (%) No schooling completed Nursery/Elementary school High school Vocational/technical school College, university or equivalent	2 0 5 14 79
Currently employed (%)	76
Life events (%) None 1 or 2 3 or more	43 48 9
Career dissatisfaction (%)	21
Involuntary retirement (%)	19
Baseline prevalence (%) Distress Anxiety/depression Sleep disturbance Alcohol misuse Disordered eating	18 29 22 23 27
Comorbidity (%) 2 symptoms 3 symptoms 4 symptoms 5 symptoms	16 11 6 0

SR since retirement, N number of participants, SD standard deviation, % percentage, cm centimeters, kg kilograms, IQR interquartile range

3.2. Incidence and comorbidity of symptoms of mental health disorders

The 12-month incidence of symptoms of mental health disorders among Dutch former elite athletes ranged from 7% for alcohol misuse to 13% for distress, 15% for sleep disturbance, 20% for disordered eating, and 28% for anxiety/depression. Among this group, 7% reported two and 4% reported three simultaneous symptoms of mental health disorders. All incidence rates and associated 95% confidence intervals are represented in Table 2.

Table 2. Incidence and comorbidity of symptoms of mental health disorders among Dutch former elite athletes at 12-month follow-up

	Incidence (95% CI) for total group
Distress	13.4 (8.9–19.7)
Anxiety/depression	27.5 (20.7–35.6)
Sleep disturbance	15.0 (10.0–21.7)
Alcohol misuse	6.8 (3.6–12.3)
Disordered eating	19.6 (13.8–27.0)
Comorbidity 2 symptoms 3 symptoms	7.2 (2.8–16.2) 4.3 (1.0–12.5)

1SR since retirement, CI confidence interval

3.3. Potential relationship between stressors and symptoms of mental health disorders

Adverse life events were significantly related to disordered eating (RR = 1.30, 95%CI = 1.05–1.61), but showed no statistically meaningful relationship with distress, anxiety/depression, sleep disturbance, and alcohol misuse. Career dissatisfaction also showed a significant relationship with disordered eating (RR = 3.98, 95%CI = 1.32–11.99), and was also significantly related to sleep disturbance (RR = 3.23, 95%CI = 1.10–9.51). However, there was no significant relationship between career dissatisfaction and distress, anxiety/depression, and alcohol misuse. The stressor involuntary retirement did not have a significant relationship with any of the measured symptoms of mental health disorders. All relationships between stressors and symptoms of mental health disorders are presented in Table 3.

Table 3. Potential relationship (relative risk and 95% CI) between symptoms of mental health disorders and stressors in Dutch former elite athletes adjusted for duration since retirement

	Distress	Anxiety/Depression	Sleep disturbance	Alcohol misuse	Disordered eating
Adverse life events	1.24 (0.97–1.58)	1.41 (0.44–4.49)	1.20 (0.96–1.50)	1.00 (0.71–1.40)	1.30 (1.05–1.61)
Career dissatisfaction	2.80 (0.82–9.57)	1.31 (0.35–4.97)	3.23 (1.10–9.51)	1.80 (0.42–7.71)	3.98 (1.32–11.99)
Involuntary retirement	1.41 (0.44–4.49)	1.06 (0.38–2.98)	0.98 (0.29–3.23)	n/a	2.06 (0.75–5.69)

SR since retirement CI confidence interval

3.4. Duration since retirement

The incidence of symptoms of mental health disorders is plotted against duration since retirement in Figure 2, showing that the onset of symptoms of distress, anxiety/depression, and sleep disturbance were mostly reported in the first 15 years since retirement. Symptoms of alcohol misuse and disordered eating were distributed more evenly across duration since retirement.

Figure 2. Incidence of symptoms of mental health disorders among Dutch former elite athletes at 12-months follow-up plotted against duration since retirement.

4. Discussion

The 12-month incidence of symptoms of mental health disorders among former elite athletes ranged from 7% for alcohol misuse to 28% for anxiety/depression. Comorbidity of two and three simultaneous symptoms of mental health disorders was reported in around 7% and 4% of the participants, respectively. Former elite athletes exposed to career dissatisfaction had a three to four-time higher risk of developing symptoms of sleep disturbance and disordered eating compared to former elite athletes satisfied with their career. Former elite athletes exposed to one or more adverse life events showed to have a small increased risk (RR = 1.30) of developing symptoms of disordered eating. Participants with 15 years or less since their athletic retirement seemed to report symptoms of mental health disorders more often than participants with a longer duration since retirement.

4.1. Comparison with the general population

In the Dutch general population, the incidence of symptoms of mental health disorders is 6% for depression, 6% for anxiety, 4% for alcohol misuse [26]. This might indicate that Dutch former elite athletes experience a higher incidence of symptoms of mental health disorders than the general population. However, valid comparisons are difficult to make because symptoms of mental health disorders among the general population were measured using the Composite International Diagnostic Interview [26]. Although scientific knowledge about the incidence of comorbid symptoms of mental health disorders is lacking, research has suggested that symptoms of mental health disorders can cause the onset of new symptoms of mental health disorders [27]. In the network theory of mental disorders, direct causal interactions between symptoms are argued to be the result of various biological, psychological, and societal mechanisms [27]. Research among the general population also aids the understanding of the effect of the duration of athletic career retirement and age on symptoms of mental health disorders. Most symptoms of mental health disorders are elevated among young people (18–24 years) [28]. Since our participants did not fall under this age group, the effect of career retirement was not strengthened by the effect of a young age. However, the potential stressor adverse life events is presumed to have the strongest effect on the development of symptoms of mental health disorders among people between 35 and 45 years old [29]. This could have increased the strength of the relationship between adverse life events and symptoms of mental health disorders of participants with a relatively short (≤15 years) duration since retirement.

A study among the British general population indicates that gender could also play an important role in the development of symptoms of mental health disorders, as they found that women reported a higher percentage of anxiety/depression than males (21% vs 16%) [30]. In a post hoc analysis of our baseline data, females (compared to males) reported a higher prevalence of distress (23% vs 13%), anxiety/depression (36% vs 22%), sleep disturbance (26% vs 17%), and disordered eating (31% vs 24%), while males reported a higher prevalence of alcohol misuse than females (34% vs 12%).

4.2. Comparison with other former elite athletes

Retired male professional footballers (N = 212; mean age: 35 years; mean duration of retirement: 4 years) reported a 12-month incidence of 11% for distress, 29% for anxiety/depression, 28% for sleep disturbance, and 15% for alcohol misuse [9]. Retired male professional ice-hockey players (N = 123; mean age: 35 years; mean duration of retirement: 7 years) reported a 6-month incidence of 8% for distress, 8% for anxiety/depression, 25% for sleep disturbance, 17% for alcohol misuse, and 8% for disordered eating [10]. While both studies used the same design and measurement scales as our study, differences in demographics as well as sport-cultural differences could have impacted the comparability of these studies. For example, a study among male Finnish athletes showed that around 18% of former team athletes reported alcohol misuse, whereas individual athletes reported around 6% of alcohol misuse [31]. This also holds for our population, since post-hoc analysis showed that 26% of team sport athletes versus 19% of individual sport athletes reported alcohol misuse at baseline. Therefore, sport-culture should be taken into account when assessing symptoms of mental health disorders.

Former elite athletes exposed to stressors (e.g. adverse life events, career dissatisfaction, and involuntary retirement) were only at increased risk for certain symptoms of mental health disorders. The wide confidence intervals indicate that there was a large variability in the groups. Contrary to our findings, adverse life events were associated with the onset of distress, anxiety/depression, sleep disturbance, and disordered eating among retired professional football players [9,32]. However, the onset of alcohol misuse was not related to adverse life events, as was seen in our study. The lack of relationship between adverse life events and alcohol misuse is consistent with studies among the general population that suggest that only specific groups, including males (not females) with an avoidant form of emotional coping and people with two or more adverse childhood events, are at increased risk for alcohol misuse when exposed to adverse life events [33–35].

Former elite athletes exposed to career dissatisfaction were approximately 3–4 times more likely to experience sleep disturbance and disordered eating. In retired professional ice-hockey players, career dissatisfaction was related to distress and sleep disturbance, but not to disordered eating [10]. Knowledge about the relationship between disordered eating and potential stressors is scarce, while former elite athletes frequently face disordered eating: around 78% of former Brazilian professional football players are overweight and around 73% of former elite female gymnasts reported disorder eating behaviors [36,37].

Our results suggest that involuntary retirement is not related to the measured symptoms of mental health disorders among Dutch former elite athletes. In contrast to these findings, a study among former professional rugby players reported that involuntary retirement was significantly related to distress, and associated with anxiety/depression, alcohol misuse, and sleep disturbance [11]. Among this population of rugby players, 41% retired involuntarily, while 19% of our participants retired involuntarily. This could have contributed to the difference in outcomes.

From the plotted figure, we assumed that symptoms of mental health disorders were more frequently reported among former elite athletes with 15 years or less of duration since retirement. However, we could not determine the exact years in which the effect of retirement was the strongest. Previous research about the effect of duration since retirement has not been unambiguous. Among retired professional skiers, the relationship between involuntary retirement and symptoms of mental health disorders decreased over time [38]. Though, a study among retired professional footballers did not find a relationship between symptoms of mental health disorders and the number of years since retirement [9].

4.3. Strength and Limitations

One of the strengths of this study is the high response rate, which could indicate that this study addressed an important topic for many former elite athletes. Another strength is the longitudinal design, which allowed us to follow participants over time. This study also included females and individual sport athletes, populations that are often not included in research about symptoms of mental health disorders among former elite athletes [4].

One of the limitations of our study was the sample size, which was only calculated for the first primary aim. Consequently, the potential relationship between symptoms of mental health disorders and stressors could not always be calculated or contained a high variability. Furthermore, the effect of career retirement, assumed to be most protruding in the initial years, could only be investigated descriptively [39]. The limited sample size could also be explained by the low prevalence of career dissatisfaction (21%) and involuntary retirement (19%), and the high prevalence of symptoms of mental health disorders at baseline, as these participants could not be included in the calculations at follow-up. The design of the questionnaires could also have contributed to the limited sample size, since important questions such as duration since retirement and symptoms of mental health disorders could be skipped. Yet, requiring participants to answer every question could be counterproductive, as forced answering in online surveys can increase dropout rates and decrease the quality of answers [40].

Another limitation was the absence of a control group matched from the general population. Therefore, we could not control for variables that could influence our results and we could not determine the extent of the problem. A control group of the general population would have provided us with a better context to situate our study and would have enabled us to determine the anticipated changes. Furthermore, selective bias was present due to a selective drop-out of older participants. While online questionnaires have been suggested to be a feasible method of surveying older adults, in 2015, around 29% of the Dutch population of 65 years and older has never used the internet and might thus be more likely to drop-out [41,42]. Additionally, the symptoms of mental health disorders were self-reported and not clinically diagnosed by a medical professional. While this could give subjective results and could overestimate the extent of the problem, symptoms of mental health disorders could indicate the need for treatment, and adequate treatment could prevent the development of symptoms of mental health disorders into mental disorders [43]. Furthermore, our study used a recall period of 6 months to retrieve information about symptoms of mental health disorders and stressors. Since the results are based on self-report, recall bias could have affected the accuracy of the data. Over time, the incidence of symptoms of mental health disorders and stressors could be decreased by denial and repression provoked by negative emotions/experiences, while on the other hand, the incidence could be increased due to a stronger impact of negative emotions/experiences on people compared to positive emotions/experiences [44,45]. Therefore, the results of our study represent estimates based on memory, which are still useful for retrospective data collection.

4.4. Practical implications

Because many former elite athletes experience symptoms of mental health disorders, they might benefit from receiving assistance in their retirement process. Before athletic career retirement, education programs, mental and life skill training, and employment counseling could prepare elite athletes for life after retirement [46,47]. After athletic career retirement, elite athletes could benefit from an exit health examination or after career counseling [48,49]. The exit health examination gives an overview of the past medical history, examines the athlete for current medical issues, and provides the athlete recommendations [48]. Similarly, after career consultation focuses on detraining, remission of injuries, employment, and education, preventing mental and cognitive health problems, and promotion of healthy lifestyles [49]. This long-term care of elite athletes should be the responsibility of the sports industry, which includes clubs, players’ unions, and governing bodies [11,49].

We recommend that further studies create a control group to make the comparison with the general population more valid. Furthermore, we recommend generating monthly questionnaires to reduce recall bias and to investigate the cause and effect relationship. We also recommend recruiting a larger number of former elite athletes to make the relationship between stressors and symptoms of mental health disorders clearer, to investigate the effect of athletic career retirement in the initial years, and to explore how gender and sport type influence symptoms of mental health disorders. Lastly, we recommend that studies further investigate the need and feasibility of support before and after retirement.

5. Conclusions

Our study showed that the 12-month incidence of symptoms of mental health disorders among former elite athletes ranged from 7% for alcohol misuse to 28% for anxiety/depression and that the onset of two or three simultaneous symptoms of mental health disorders was reported in 7% and 4% of the participants, respectively. We found that the stressors: career dissatisfaction and involuntary retirement could put former elite athletes at increased risk of the onset of sleep disturbance and/or disordered eating. Furthermore, Dutch former elite athletes seemed to report symptoms of mental health disorders more frequently in the first 15 years since athletic career retirement.

Declaration of interest

No potential conflict of interest was reported by the authors.

Acknowledgments

The authors would like to thank the Athletes Committee and medical staff of the Netherlands Olympic Committee*Netherlands Sports Confederation (NOC*NSF) for their support in the study (especially former elite athlete Margriet de Schutter and Olympic champions Stefan Groothuis and Mieke Cabout). We are grateful to all former elite athletes who participated in the study.