Protocol for an Observational Study on the Effects of Early-Life Participation in Contact Sports on Later-Life Cognition in a Sample of Monozygotic and Dizygotic Swedish Twins

by   Jordan Weiss, et al.
University of Pennsylvania

A large body of work links traumatic brain injury (TBI) in adulthood to the onset of Alzheimer's disease (AD). AD is the chief cause of dementia, leading to reduced cognitive capacity and autonomy and increased mortality risk. More recently, researchers have sought to investigate whether TBI experienced in early-life may influence trajectories of cognitive dysfunction in adulthood. It has been speculated that early-life participation in contact sports---a leading cause of concussions among adolescents---may lead to poor cognitive and mental health outcomes. However, to date, the few studies to investigate this relationship have produced mixed results. We propose to extend this literature by conducting a study on the effects of early-life participation in contact sports on later-life cognitive health using the prospective Swedish Adoption/Twin Study on Aging (SATSA). The SATSA is unique in its sampling of monozygotic and dizygotic twins reared together (respectively MZT, DZT) and twins reared apart (respectively MZA, DZA). The proposed analysis is a study of 674 individuals (40 MZA, 98 DZA, 68 MZT, and 80 DZT, 102 unpaired singletons). 595 individuals in the analytic sample did not participate in contact sports and 79 did. 236 twin pairs were concordant for no participation in contact sports; 21 twin pairs were concordant for participation in contact sports; and 29 twin pairs were discordant for participation in contact sports. Our primary outcome will be a measure of global cognition assessed through the Mini-Mental State Examination (MMSE). We will also consider several secondary cognitive outcomes including verbal and spatial ability, memory, and processing speed. Our sample will be restricted to individuals with at least one MMSE score out of seven repeated assessments spaced approximately three years apart. We will adjust for age and sex in each of our models.



There are no comments yet.


page 1

page 2

page 3

page 4


Protocol for an observational study on the effects of playing football in adolescence on mental health in early adulthood

More than 1 million students play high school American football annually...

Protocol for an Observational Study of the Association of High School Football Participation on Health in Late Adulthood

American football is the most popular high school sport and is among the...

Flaws in Flawlessness: Perfectionism as a New Technology Driven Mental Disorder

Today, technologies, devices and systems play a major role in our lives....

Personalized Early Stage Alzheimer's Disease Detection: A Case Study of President Reagan's Speeches

Alzheimer`s disease (AD)-related global healthcare cost is estimated to ...

Improving Prediction of Cognitive Performance using Deep Neural Networks in Sparse Data

Cognition in midlife is an important predictor of age-related mental dec...

A Hierarchical Meta-Analysis for Settings Involving Multiple Outcomes across Multiple Cohorts

Evidence from animal models and epidemiological studies has linked prena...

Exposure to urban parks improves affect and reduces negativity on Twitter

Urbanization and the decline of access to nature have coincided with a r...
This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.

1 Background

Sports-related concussions have emerged in the past decade as a leading healthcare issue for children and adolescents in the US and abroad. The Centers for Disease Control and Prevention (CDC) define a concussion as a mild traumatic brain injury (mTBI) resulting from a direct impact or forceful motion to the head (CDC, 2018). Although concussions can occur in any sport, they are most common in contact sports such as football, hockey, and soccer (Harmon et al., 2013)

. In the US, it is estimated that children and adolescents account for 78.6% of the sports-related head trauma cases treated in emergency departments

(Gaw and Zonfrillo, 2016) although it is believed that nearly 50% of sports-related concussions go unreported (Harmon et al., 2013). In light of this underreporting and the susceptibility of this population to the effects of sports-related concussions (Halstead et al., 2010; Patel and Reddy, 2010; Yeates, 2010), there is a growing need to understand whether early-life participation in contact sports has implications for later-life cognitive function. Our goal in the proposed analysis is to expand on prior studies by investigating the link between early-life participation in contact sports and late-life cognitive impairment in a unique sample of twins reared together and twins reared apart.

2 Study Design and Materials

2.1 The Swedish Adoption/Twin Study of Aging

Detailed descriptions of the Swedish Adoption/ Twin Study of Aging (SATSA) sample have been previously published (Finkel and Pedersen, 2004; Bergeman et al., 1993). Briefly, the longitudinal SATSA sample was drawn from the population-based Swedish Twin Registry (STR) (Magnusson et al., 2013) and contains data on same-sex monozygotic and dizygotic twins that were reared apart and a control sample of twins reared together matched on birth year, country of birth, and gender with survey years in 1984, 1987, 1990, 1993, 2004, 2007, and 2010. SATSA respondents have undergone detailed in-person interviews and health examinations approximately every three years since the baseline survey in 1984. In the 1993 survey wave, responders were asked to report whether they were involved in any sport (e.g., football, ice hockey, or boxing) that may involve a hit on a head (hereafter contact sports). Examining twin pairs concordant and discordant for participation in contact sports allows us to control for biological and familial influences while comparing cognitive outcomes.The SATSA has been approved by an ethics committee at the Karolinska Institutet and the Regional Ethics Review Board in Stockholm.

2.2 Sample Selection Criteria

The SATSA surveyed 2,018 individuals (758 twin pairs; 502 unpaired singletons) in its baseline year in 1984 when participants were, on average, 60.14 years old (SD = 14.02). 1,450 responders (479 twin pairs; 492 unpaired singletons) were retained through the fourth survey wave in 1993 when responders were asked to report whether they participated in contact sports in early-life. Among these 1,450 responders, 1,240 (85.5%) reported not being involved in contact sports, 163 (11.3%) did, and data were missing for 47 (3.22%). Among the 1,403 responders with information about their participation in contact sports, 684 (48.75%) completed in-person interviews and had at least one MMSE score over the survey period. Of the 684, an additional 10 (1.5%) were dropped for reporting uncertain zygosity or being discordant on their rearing status. Of the remaining 674 responders, 79 (11.7%) reported participation in contact sports. Among the 729 responders who provided information on contact sports but had no available MMSE scores, 84 reported playing contact sports (11.6%). Fisher’s test revealed no differences in the missingness of MMSE scores by participation in contact sports (P value 0.05).

2.3 Study Outcomes

Our primary outcome is cognitive impairment suggestive of dementia or mild cognitive impairment (MCI) as assessed through the MMSE. The MMSE is commonly used to assess cognitive functioning, track changes in cognitive function over time, and screen individuals for cognitive impairment (Folstein et al., 1975)

. The SATSA implemented the MMSE in its baseline assessment in 1984 and in each wave that followed through 2010. Table 1 shows descriptive statistics for MMSE scores over the survey period. We will dichotomize MMSE scores such that scores between 0 and 27 are coded as impaired (1) and scores between 28 and 30 are coded as not impaired (0). The traditional MMSE cutoff score is 24, but higher cutoff scores have been proposed to increase diagnostic accuracy in individuals with higher levels of education, and at earlier stages of dementia severity

(Folstein et al., 1991; Rajji et al., 2009). In the absence of a conclusively defined cutoff, we will replicate our semi-parametric illness-death model for MMSE using two additional cutoffs—24 and 30—as recommended in the literature (Folstein et al., 1991; Rajji et al., 2009).

In conjunction with the MMSE, the SATSA administered an extensive cognitive battery to assess verbal and spatial ability, memory, and processing speed. Table 2 shows the frequencies of available outcomes data in our sample of eligible responders.

2.4 Primary Analysis

We will model the relationship between participation in contact sports and the onset of cognitive impairment using a semi-parametric illness-death model which accounts for (i) the interval censoring of cognitive impairment between survey waves and (ii) the competing risks with death and study drop-out (Joly et al., 2002; Touraine et al., 2017). The illness-death model provides better estimates of hazard ratios than standard Cox hazard models which ignore interval-censoring and competing risks (Leffondre et al., 2013). We will use a bootstrap clustering (Field and Welsh, 2007) approach with twin pairs as sampling units to account for the twin structure. Coefficient estimates will be obtained using the R package Smooth Hazard (Touraine et al., 2017). We will repeat this analysis in a restricted sample of 29 twin pairs (n = 58) discordant for participation in contact sports.

2.5 Secondary Analysis

We will estimate a series of age and sex adjusted growth curve models for each of seven cognitive test outcomes listed in Table 2 as a function of participation in contact sports. Growth curve models allow us to examine the effects of participation in contact sports on the level of and change in cognitive function over time by including both random (i.e., individuals nested within twin pairs) and fixed (i.e., participation in contact sports) effects. A structured modeling approach will be used to determine the functional form of the growth curve models for each cognitive outcome, independently, while accounting for participation in contact sports. The Bayesian information criterion statistic will be used to compare models. For each cognitive test, the model with the lowest BIC will be selected and re-estimated with the age and sex included. The R code used to estimate each model is provided in Table 3 using the MMSE as an example. All models will be estimated using full-information maximum likelihood. We will cluster the standard errors to account for the twin structure. In addition, we correct for multiplicity by using the Benjamini–Hochberg procedure to estimate p-values adjusted for the false discovery rate

(Cribbie, 2007).

We will include the full analytic sample (n=684). A growth curve modeling approach has been used in a previous study which used the SATSA to examine the effects of childhood social class on trajectories of cognitive aging (Ericsson et al., 2017). Because changes in cognition are age-related, we will measure time using chronological age rather than survey year.


  • Bergeman et al. (1993) Bergeman, C. S., Chipuer, H. M., Plomin, R., Pedersen, N. L., McClearn, G. E., Nesselroade, J. R., Costa, P. T., J., and McCrae, R. R. (1993). Genetic and environmental effects on openness to experience, agreeableness, and conscientiousness: an adoption/twin study. J Pers, 61(2):159–79.
  • CDC (2018) CDC (2018). Report to congress: The management of traumatic brain injury in children. National Center for Injury Prevention and Control; Division of Unintentional Injury Prevention. Atlanta, GA.
  • Cribbie (2007) Cribbie, R. A. (2007). Multiplicity control in structural equation modeling. Structural Equation Modeling: A Multidisciplinary Journal, 14(1):98–112.
  • Ericsson et al. (2017) Ericsson, M., Lundholm, C., Fors, S., Dahl Aslan, A. K., Zavala, C., Reynolds, C. A., and Pedersen, N. L. (2017). Childhood social class and cognitive aging in the swedish adoption/twin study of aging. Proc Natl Acad Sci, 114(27):7001–7006.
  • Field and Welsh (2007) Field, C. A. and Welsh, A. H. (2007). Bootstrapping clustered data. Journal of the Royal Statistical Society. Series B (Statistical Methodology), 69(3):369–390.
  • Finkel and Pedersen (2004) Finkel, D. and Pedersen, N. L. (2004). Processing speed and longitudinal trajectories of change for cognitive abilities: The swedish adoption/twin study of aging. Aging, Neuropsychology, and Cognition, 11(2-3):325–345.
  • Folstein et al. (1991) Folstein, M. F., Bassett, S. S., Anthony, J. C., Romanoski, A. J., and Nestadt, G. R. (1991). Dementia: Case ascertainment in a community survey. Journal of Gerontology, 46(4):M132–M138. 10.1093/geronj/46.4.M132.
  • Folstein et al. (1975) Folstein, M. F., Folstein, S. E., and McHugh, P. R. (1975). "mini-mental state." a practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res, 12(3):189–98.
  • Gaw and Zonfrillo (2016) Gaw, C. E. and Zonfrillo, M. R. (2016). Emergency department visits for head trauma in the united states. BMC Emerg Med, 16:5.
  • Halstead et al. (2010) Halstead, M. E., Walter, K. D., Council on Sports, M., and Fitness (2010). American academy of pediatrics. clinical report–sport-related concussion in children and adolescents. Pediatrics, 126(3):597–615.
  • Harmon et al. (2013) Harmon, K. G., Drezner, J. A., Gammons, M., Guskiewicz, K. M., Halstead, M., Herring, S. A., Kutcher, J. S., Pana, A., Putukian, M., and Roberts, W. O. (2013). American medical society for sports medicine position statement: concussion in sport. British Journal of Sports Medicine, 47(1):15.
  • Joly et al. (2002) Joly, P., Commenges, D., Helmer, C., and Letenneur, L. (2002). A penalized likelihood approach for an illness-death model with interval-censored data: application to age-specific incidence of dementia. Biostatistics, 3(3):433–43.
  • Leffondre et al. (2013) Leffondre, K., Touraine, C., Helmer, C., and Joly, P. (2013). Interval-censored time-to-event and competing risk with death: is the illness-death model more accurate than the cox model? Int J Epidemiol, 42(4):1177–86.
  • Magnusson et al. (2013) Magnusson, P. K., Almqvist, C., Rahman, I., Ganna, A., Viktorin, A., Walum, H., Halldner, L., Lundstrom, S., Ullen, F., Langstrom, N., Larsson, H., Nyman, A., Gumpert, C. H., Rastam, M., Anckarsater, H., Cnattingius, S., Johannesson, M., Ingelsson, E., Klareskog, L., de Faire, U., Pedersen, N. L., and Lichtenstein, P. (2013). The swedish twin registry: establishment of a biobank and other recent developments. Twin Res Hum Genet, 16(1):317–29.
  • Patel and Reddy (2010) Patel, D. R. and Reddy, V. (2010). Sport-related concussion in adolescents. Pediatr Clin North Am, 57(3):649–70.
  • Pedersen et al. (1991) Pedersen, N. L., McClearn, G. E., Plomin, R., Nesselroade, J. R., Berg, S., and DeFaire, U. (1991). The swedish adoption twin study of aging: an update. Acta Genet Med Gemellol (Roma), 40(1):7–20.
  • Rajji et al. (2009) Rajji, T. K., Miranda, D., Mulsant, B. H., Lotz, M., Houck, P., Zmuda, M. D., Bensasi, S., Reynolds, C. F., and Butters, M. A. (2009). The mmse is not an adequate screening cognitive instrument in studies of late-life depression. Journal of psychiatric research, 43(4):464–470.
  • Touraine et al. (2017) Touraine, C., Gerds, T. A., and Joly, P. (2017). Smoothhazard: An r package for fitting regression models to interval-censored observations of illness-death models. Journal of Statistical Software, 79(7):22.
  • Yeates (2010) Yeates, K. O. (2010). Mild traumatic brain injury and postconcussive symptoms in children and adolescents. J Int Neuropsychol Soc, 16(6):953–60.

3 Appendix

MMSE n Mean Std. Dev. Min Max.
Wave 1 451 28.1 1.6 22 30
Wave 2 505 28.4 1.3 23 30
Wave 3 520 27.9 1.9 13 30
Wave 4a 38 26.4 2.2 20 29
Wave 5 504 27.2 2.8 8 30
Wave 6 416 27.2 3.1 4 30
Wave 7 359 27.1 3.7 7 30
Table 1: Descriptive statistics for the MMSE scores across the survey period.

Note. a. Due to reasons related to funding, a telephone interview was used in place of in-person testing for Wave 4 and thus the cognitive assessment was not conducted for all individuals (Pedersen et al., 1991)

Test Wave 1 Wave 2 Wave 3 Wave 4a Wave 5 Wave 6 Wave 7
MMSE 451 505 520 38 504 416 359
65.9% 73.8% 76% 5.6% 73.7% 60.8% 52.5%
Verbal Ability 467 440 457 34 450 368 287
68.3% 64.3% 66.8% 5% 65.8% 53.8% 42%
Spatial Ability 462 438 446 32 434 345 257
67.5% 64% 65.2% 4.7% 63.5% 50.4% 37.6%
Memory 466 464 494 33 456 379 310
68.1% 67.8% 72.2% 4.8% 66.7% 55.4% 45.3%
Processing Speed 465 462 463 35 470 394 327
68% 67.5% 67.7% 5.1% 68.7% 57.6% 47.8%
Digit Span 492 501 515 35 496 409 343
71.9% 73.2% 75.3% 5.1% 72.5% 59.8% 50.1%
Delayed Recall NA NA NA NA 473 394 348
69.2% 57.6% 50.9%
Table 2: Availability of cognitive scores across the survey period (n and % of participants completing in-person testing).

Note. a. Due to reasons related to funding, a telephone interview was used in place of in-person testing for Wave 4 and thus the cognitive assessment was not conducted for all individuals (Pedersen et al., 1991)

Model Model Specification
Model 1 lmer(mmse ~ sportsYN + agec +
(agec | pair) + (agec | id), data = SATSA, REML = FALSE)
Model 2 lmer(mmse ~ sportsYN + agec + agec2 +
(agec + agec2 | pair) + (agec + agec2 | id), data = SATSA, REML = FALSE)
Model 3 lmer(mmse ~ sportsYN + agec + agec2 + sportsYN:agec + sportsYN:agec2 +
(agec + agec2 | pair) + (agec + agec2 | id), data = SATSA, REML = FALSE)
Table 3: Model specification for secondary analysis presented for the MMSE (shown in R code).

Notes. mmse = Mini-Mental State Exam; sportsYN = binary indicator for participation in contact sports; agec = mean-centered age; agec2 = mean-centered age-squared ; pair = twin-pair identifier; id = individual identifier; REML = restricted maximum likelihood.