2.1 Study population

Cases were recruited primarily through the Motor Neurone Disease Association of New Zealand (MNDANZ) over a period of 3 years (2013–2016). This was supplemented with searches for all hospital outpatients with a primary or secondary diagnosis of MND (ICD 10 code – G122) from the National Minimum Dataset (2013–2015). The inclusion criterion for cases was a diagnosis by a neurologist, with all forms of MND included. A total of 2400 controls (two per case) were randomly selected from the New Zealand Electoral Roll (2008), frequency matched by age (based on the age distribution of the Unite Kingdom MND incidence distribution, as New Zealand MND incidence age distribution was not available)³¹ and sex. Controls with any neurodegenerative disease were excluded.

All participants gave written informed consent. Ethical approval was obtained by the Multi-region Ethics Committee in New Zealand (ref: MEC/12/01/005).

2.2 Data collection

Depending on the preference of the participants, a face-to-face (59% in cases vs 16% in controls) or a telephone interview (23% vs 66%), or a postal questionnaire (18% vs 18%) was used to collect information on demographic and personal characteristics, family history, physical trauma/injury, emotional trauma (life events) physical activities (sports), smoking, alcohol and a lifetime occupational history as described previously.³² All controls completed the questionnaire themselves, while nine cases used a proxy (three required proxy assistance with a face-to-face interview, and six used proxy assistance for reading and writing only).

2.3 Exposure assessment

2.4 Statistical analyses

Statistical analyses were conducted using SAS version 9.4. Differences in general characteristics between cases and controls were tested using chi-squared tests, and unconditional logistic regression was used to estimate odds ratios (ORs) and 95% confidence intervals (CIs), for ever-exposed to a particular event (sports, head injury, spine injury, emotional traumas), compared to never being exposed to that particular event.

Analyses were adjusted for age (5-year categories); sex; ethnicity (European; Māori, the indigenous population of New Zealand; other); highest achieved education level (primary and secondary school, technical or trade school diploma, undergraduate university degree, postgraduate university degree); smoking status (never, ex, current smoker (smoker at the time of diagnosis for cases)); alcohol consumption before diagnosis (≤once a month,1–2 times/week, 3–5 times/week, daily); and socioeconomic status (SES) using the New Zealand Deprivation Index (NZDep2006, quintiles). All analyses were repeated separately for males and females. Additional analyses were conducted mutually adjusting for all other exposures (sports, head injury, spine injury, and emotional trauma). We checked for multicollinearity by comparing the standard errors for the main effect estimates between the full model, and a minimally adjusted model,³³ this showed no evidence of collinearity. Analyses were also repeated while controlling for the interview method.

We also assessed associations with the lifetime number of years played in a particular sport (for each sport separately). Duration was categorized based on quartiles in the controls, specific to each sport. A test for trend was performed by assigning scores to the categories of the categorical duration variables, which were subsequently fitted as continuous variables.

As it cannot be excluded that accidents and injuries may be an early manifestation of MND, even before diagnosis,²³ analyses were repeated excluding all injuries that occurred within 3 years before the index date (diagnosis date for cases, interview date for controls).

3.1 Population characteristics

A total of 321 cases and 605 controls were included in the analyses. Of the cases, 63% were male versus 55% in controls. The majority of cases (68%) were aged >60 years. While the 70+ age group was overrepresented in the controls, there was little difference between cases and controls in ethnicity, socioeconomic status (SES), education, tobacco smoking and alcohol consumption (Table 1).

3.2 Head injury with concussion and spine injury

Among cases, 33% reported having ever had a head injury with a concussion, compared to 22% among controls (OR 1.49, 95% CI 1.09, 2.05; Table 2). This difference remained after excluding all brain injuries that occurred within 3 years prior to diagnosis (OR 1.51, 95% CI: 1.09, 2.09). Reporting only one head injury with a concussion was not associated with an increased risk (OR 1.16, 95% CI 0.80, 1.69), while statistically significant increased risks were observed for reporting two (OR 4.01, 95% CI 1.82, 8.86; based on 22 cases and 10 controls), and 3 or more (OR 2.34, 95% CI 1.00, 5.45; based on 14 cases and 11 controls) head injuries. Elevated risks were observed for self-reported mild, moderate and severe head injuries, but this reached statistical significance only for moderate injury (OR 1.77, 95% CI 1.14, 2.77).

When considering the age at which head injury occurred (childhood only, adulthood only, or both childhood and adulthood), the highest risk was observed for having had head injuries in both childhood and adulthood (OR 1.87, 95% CI 1.09, 3.21). While more males reported head injuries with concussion than females (28% of male and 16% of female controls), the observed pattern of an increased risk associated with multiple head injuries was similar for males and females. Head injuries in two or more different circumstances (work-related, sport-related, leisure-related, traffic accident, other accidents and multiple circumstances) were associated with an almost 2.5-fold increased risk (OR 2.46, 95% CI 1.11, 5.44; Table 3). When the number of head injuries in each circumstance was considered, elevated risks were observed for having had more than one sport-related head injury (OR 3.05, 95% CI 1.06, 8.77) and more than one leisure-related head injury (OR 6.89, 95% CI 1.18, 40.3). Having had one work-related head injury was also associated with an increased risk (OR 2.48, 95% CI 1.01, 6.09; Table 3).

Having ever had a spine injury (9% for cases and controls), repeated spine injury, the severity of the injury, and the age at the injury, were not associated with the risk of MND (Table S1).

3.3 Sports

None of the participants reported having been professional athletes. The majority of study participants participated in leisure sports throughout childhood and adulthood (73% of male and 64% of female controls), which was associated with an almost doubling of the MND risk (OR 1.81, 95% CI 1.01, 3.25, Table 4).

Analysis by type of sports did not reveal statistically significant associations for any specific sport, although elevated ORs were observed for several common sports including football (soccer) (OR 1.40, 95% CI 0.95, 2.06), running (OR 1.48, 95% CI 0.92, 2.38), golf (OR 1.38, 95% CI 0.88, 2.16) and rugby (OR 1.26, 95% CI 0.86, 1.83). Analysis by duration for these sports revealed a positive association with longer duration of playing football (soccer) (p for trend = .04), with playing football (soccer) for >12 years associated with a more than twofold MND risk (OR 2.35, 95% CI 1.19, 4.65; Table S2). For rugby, running and golf, a trend with duration was not observed. For rugby and golf, only the shortest duration category was associated with a statistically significant increased risk (Table S2).

3.4 Emotional trauma

The majority of cases (76%) and controls (72%) reported a traumatic life event within at least one of the 14 categories (Table 5), but this was not associated with MND (OR 1.17, 95% CI 0.84, 1.63). Multiple traumatic life events in different categories were associated with elevated risks, with the highest risk observed for reporting traumatic life events in ≥4 categories (OR 1.88, 95% CI 1.17, 3.03); elevated ORs were observed in both males (OR 1.76, 95% CI 0.92, 2.34) and females (OR 2.30, 95% CI 1.06, 5.00).

Results for the individual emotional trauma categories, mutually adjusted for life events in the other emotional trauma categories, showed that having been beaten by a family member or a carer during childhood was associated with an increased risk for MND (OR 1.82, 95% CI 1.14, 2.90), with similar ORs for males (OR 2.10, 95% CI 1.16, 3.83) and females (OR 2.06, 95% CI 0.88, 4.82). For females who witnessed another person being killed or seriously injured, an increased risk was also observed (OR 2.68, 95% CI 1.06, 6.76; Table 5). For females, having been bullied by a close relation was also associated with an increased risk, but this did not reach statistical significance (OR1.81, 95% CI 0.87, 3.38; Table 5).

More detailed analyses on the association between self-reported childhood physical abuse and MND are presented in Table 6. Increased risks were observed independent of the age at which the physical abuse first occurred (before the age of 7 or after). Reporting multiple episodes of physical abuse was associated with an increased risk, being statistically significant for having experienced physical abuse as a child 2–4 times (OR 5.72, 95% CI 1.41, 23.24) and more than 10 times (OR 2.04, 95% CI 1.13, 3.66). The pattern of elevated ORs for the higher frequency categories was observed for both males and females but reached statistical significance only in males.

Analyses by duration of childhood physical abuse suggested that longer duration was associated with a higher risk, that is physical abuse lasting 5–8 years was associated with a twofold risk (OR 2.26, 95% CI 1.14, 4.49) and physical abuse lasting >8 years with a threefold risk (OR 3.01, 95% CI 1.18, 7.70).

Of those reporting childhood physical abuse, half reported that this resulted in physical injury. While similar ORs were observed for physical abuse with (OR 1.73) and without (OR 1.89) physical injury, statistical significance was reached only for physical abuse without physical injury (OR 1.89; 95% CI 1.06, 3.38).

When considering the type of force used (fist, belt, stick, smacking on the hands, combined fist/belt/stick, other non-specific types), the use of fist, belt and stick were all associated with a more than twofold MND risk, reaching statistical significance for the use of a stick (OR 2.57, 95% CI 1.10, 5.97), with a particularly high risk observed for females (OR 6.35, 95% CI 1.31, 30.82; Table 6).

4.1 Head injury

Having had multiple head injuries with concussion was associated with a twofold risk of MND. These findings are consistent with several meta-analyses^24-27 reporting increased MND risk for head injury, and particularly for repeated head injuries.²⁵ It has been argued that this may, at least in part, be due to reverse causation²³ as accidents and injuries may be an early manifestation of MND. Our findings did not change after excluding head injuries that occurred 3 years before the index date, suggesting that reverse causation is unlikely to explain our findings.

A recent meta-analysis²⁴ showed that severe head injury was associated with a higher risk of MND. We did not observe a clear association with injury severity, although misclassification due to the self-reported nature of the injury severity may have contributed to this. However, several reports have suggested that repetitive mild traumatic brain injuries alone could be sufficient to trigger the physiological changes required to increase the risk of MND.^36-38 Head trauma is known to disrupt and deteriorate the blood–brain barrier,³⁹ which may play a role in MND pathogenesis.^{40, 41} In ALS animal models, traumatic brain injury has been found to trigger TAR DNA-binding protein 43 (TDP-43) associated pathology, which is a neuropathological hallmark lesion in the brain of patients with ALS.⁴² TDP-43 pathology has also been shown in the brains of athletes that experienced repetitive head injury.³⁷

In this study population, head injury with a concussion most commonly occurred as part of leisure sports (9% of male and 3% of female controls reported at least one sport-related head injury) and having more than one sport-related head injury was associated with a threefold MND risk. This identifies leisure sports as an important setting of preventable head injury that may be contributing to MND risk.

4.2 Sports

In this study participating in sports throughout childhood and adulthood was associated with an increased risk for MND, which is consistent with several previous studies.^{11, 12, 34, 43} The association between physical activities and the risk of MND is biologically plausible as vigorous exercise may induce oxidative stress and glutamate excitotoxicity, two well-established mechanisms for MND.⁴⁴

MND risk differed by the type of sport played. For many individual sports, we did not observe an increase in MND risk, including cricket, basketball/netball, tennis, swimming, hockey, volleyball, badminton, lawn bowl, yachting, rowing, diving and skiing. Taking part in endurance sports (when considered collectively) was associated with an increased MND risk, but when considering specific endurance sports separately, none of the associations reached statistical significance. A number of common sports in New Zealand were associated with elevated ORs that did not reach statistical significance, including rugby, football (soccer), running and golf.

Only for playing football (soccer), we observed a statistically significant positive association with duration, with a twofold MND risk for those who played football (soccer) >12 years. This is of interest, as several studies^{5, 45, 46} have reported an increased MND risk for professional football (soccer) players. This study adds to this body of evidence and suggests that non-professional football (soccer) may also increase the risk of MND.

The association observed for football (soccer) may be mediated through strenuous physical activities and/or head injury.^{5, 26} As our analyses were adjusted for head injury with a concussion, diagnosed concussions are unlikely to explain the observed association. Therefore, with football (soccer) being the only sport in which players purposefully use the head to deflect the ball, and it emerging as the only specific sport associated with an increased risk in our study, a role for sub-concussive impacts to the head in football (soccer) players cannot be excluded,^{5, 47} and merits further study.

4.3 Emotional trauma

In the current study, elevated risks were associated with having experienced more than two different types of traumatic life events, with an almost twofold risk associated with having experienced four or more types of traumatic life events. An Australian ALS case–control study,³⁰ which used a checklist for potentially stressful life events as part of an anonymous online questionnaire, found no difference in life events inventory scores between cases and controls, but results for specific traumatic life events were not presented. We are not aware of other studies that evaluated the association between traumatic life events and MND. Our findings show some parallels with a Japanese study,¹³ which found that self-reported stress and type A behaviour pattern (related to emotional distress and to more keenly perceived life stress) were associated with an increased ALS risk. Emotional trauma has also been proposed as a risk factor for other neurodegenerative diseases, including Alzheimer disease⁴⁸ and Parkinson disease,⁴⁹ and mechanistic studies have demonstrated that psychological stress is associated with increased oxidative stress,^{28, 50} which plays an important role in the pathology of neurodegenerative diseases including MND.⁵¹

It cannot be fully excluded that self-reports of traumatic events may systematically differ between cases and controls, either because of differences in recall, or because of possible differences in personality type and the perception of life stress between cases and controls, as suggested by the Japanese study.¹³ Our relative risk estimates for specific categories of traumatic events, which identified childhood physical abuse as a risk factor for MND, are, however, less likely to be affected by this, as these were adjusted for the report of all other categories of traumatic events.

It is possible that physical trauma, including head injury, may play a role in the observed associations for emotionally traumatic events and MND, as several of the emotional trauma categories relate to physical violence that may result in physical trauma. In particular, we observed a twofold MND risk for having experienced physical abuse as a child. This association was consistently observed for both males and females, and risks increased for higher frequency, as well as longer duration of child abuse. While this association was adjusted for self-reported head injuries with concussion, physical child abuse may result in undiagnosed and unreported physical trauma that could not be taken into account in this study. A role for emotional trauma cannot be excluded either, as similar ORs were observed for physical child abuse with and without physical injury. While the underlying contributing factors could not be identified, this new finding warrants further research.