Andrew Hua Tiffany Bullard Jason Cohen

Daniel Palac Edward McAuley

Arthur Kramer Sean Mullen

A Novel Cognitive

Training Intervention

Reduces Back Pain in

Middle-aged Adults

Prevalence of Back Pain

23.2% global 1-month prevalence

(Hoy et al., 2012)

11-12% are disabled (Balagué et al.,

2012)

Total cost exceeds $100 billion (Katz,

2006)

Back pain results in a 5.3 hr/week

loss of productive time (Stewart et

al., 2003)

Treating Back Pain

Analgesics (Kuritski and Samraj, 2012)

Acetaminophen

Nonsteroidal anti-inflammatory drugs

Opioids

Massage/chiropractic therapy (Airaksinen et al., 2006)

Surgery/intradiscal injection (Bron et

al., 2009; Khot et al., 2004; Urrútia et al.,

2004)

Understanding

Pain Treatments

Pain travels along the spinal

cord to the brain

Insula (Starr et al., 2009)

Dorsolateral prefrontal cortex

(DLPFC) (Lorenz, Minoshima,

and Casey, 2003)

ACC (Guan et al., 2015)

Patients with chronic low back

pain exhibit significantly less

activation in the DLPFC and

dACC (Mao et al., 2014)

Figure from Bushnell, M. C., Čeko, M., & Low, L. A. (2013).

Cognitive and emotional control of pain and its disruption in

chronic pain. Nature Reviews Neuroscience, 14(7), 502–511.

http://doi.org/10.1038/nrn3516

Coping with Pain

Active coping (Lynn Snow-Turek,

Norris, and Tan, 1996)

Consume cognitive resources

Goal setting

Planning and executing activities

Passive coping

Avoiding challenge

Relying on doctors

Praying

Rationale

Chronic pain has a

complex relationship

with cognitive,

emotional, and

behavioral components

(Nes, Roach, and

Segerstrom, 2009)

Self-regulatory demands to

exert control over emotions,

thoughts, relationships, and

behaviors

Self-regulatory fatigue

Executive functions

PainTrait

capacity

CORTEXCognitive Regulation Training

and Exercise

Primary Aim: To test the effects of

cognitive training on middle-

aged adults exercise adherence

Secondary data analysis: Can

cognitive training improve self-

regulation of pain?

2 year study funded by the National Heart, Lung, &

Blood Institute (#RHL113410A1)

CORTEX – Study Design

Month 0 Screening and Randomization

Cognitive Testing

Fitness Assessment

Psychometric Surveys

Month 1 4 Wk Cognitive Training

Cognitive Testing

Fitness Assessment

Psychometric Surveys

Month 2 Psychometric Surveys

Month 5 4 Month Ex. Program

Cognitive Testing

Fitness Assessment

Psychometric Surveys

CORTEX – sample

characteristics

Recruited from Central

Illinois: Urbana-

Champaign

N = 133 (79% female)

Mean age = 53.88 years

78.2% Caucasian (15.1%

African-American, 3.4%

Asian, 3.3% other)

72.1% with college

degree

6.7% non-native English

speaker

Initial Contacts:n =453

Randomized: n=133Gaming: n=68Video: n=65

Completed Randomized:

n=117Gaming: n=56Video: n=61

Excluded: n=320

Not meeting inclusion criteria (n=61)

• Too active

• GDS>5

• TICS<21

• Age

• Participating in other exercise

study

• Brain training games

• No internet

Declined to participate (n=55)• Program length/time

• Not interested

Other reasons (n=204)• Unreachable

• Passed and unresponsive

• Inactive phone number

• Partially screened

• Distance

• No MD clearance

• Injury

• No primary care

• Hospitalization

• Moved

CORTEX – The InterventionIn

terv

en

tio

n •1 hour exergaming

•Xbox Kinect

•WiiFit

•Playstation Move

•1 hour computer-based training

•Dual-task

•Stroop, attention network, and task-switch

•Exercise-related self-priming task

Co

ntr

ol •Attention control

•Health and educational videos

N = 68 N = 65

CORTEX – exergaming

A theoretically-driven and evidence-based intervention

emphasizing stationary and active dual task paradigms (see

systematic review by Ogawa, You, and Leveille, 2016)

20 min

Xbox Kinect

20 min

WiiFit

20 min

Playstation Move

CORTEX – Computer-based

Training

30 min

↓

dlPFC

Dual Task

5 min

↓

ACC

Task Switch

5 min

↓

dlPFC

Attention Network

5 min

↓

Insula

Stroop

CORTEX – Measuring Pain

7 item questionnaire

Headache

Neck

Shoulders

Back

Arms/hands

Legs/feet

Joint

5 point Likert scale

1 – Never

2 – Rarely

3 – Sometimes

4 – Often

5 – Very Often

Pain Experience Questionnaire (PEQ; Fuchs, Goehner,

and Seelig, 2011)

Plan of Analysis

CFA Overall

CFA 4-item

Group Invariance

Longitudinal Invariance

Configural

Metric

Scalar

Strict

Configural

Metric

Scalar

Strict

LGM 4-item

LGM Back Only

GMM 2 Class

Confirmatory Factory Analysis

CFA of the complete 7-item PEQ revealed a poor fit of the model

CFA of the reduced 4-item PEQ

Back, arms/hands, legs/foot, and joint pain (arthritic pain)

Χ2 = 4.31, p = 0.116, df = 2

CFI = 0.983

RMSEA = 0.096

SRMR = 0.033

Longitudinal Invariance

Group

Invarianceχ2 p RMSEA CFI SRMR Δχ2 ΔCFI

Configural 25.555 0.043 0.075 0.967 0.047 - -

Metric 30.908 0.041 0.071 0.962 0.068 5.453 0.005

Scalar 34.763 0.055 0.064 0.963 0.066 3.855 0.001

Strict 38.051 0.077 0.057 0.965 0.064 3.288 0.002

Group Invariance

Group

Invarianceχ2 p RMSEA CFI SRMR Δχ2 ΔCFI

Configural 5.764 0.218 0.084 0.983 0.038 - -

Metric 7.630 0.470 0.000 1.000 0.061 1.866 0.017

Scalar 9.347 0.673 0.000 1.000 0.068 1.717 0.000

Strict 12.304 0.723 0.000 1.000 0.074 2.957 0.000

Latent Growth Model

Covariates:

Group

Baseline Fitbit Activity

Injury/Illness

Dropouts

BMI

Age

Gender

Education

Reduced 4 item pain measure

No change by group

Single item back pain

Group had a

significant decrease in

back pain (β = -0.111,

p = 0.002) across the 4-

month exercise

program.

LGM Results

2.3

2.5

2.7

2.9

3.1

Month 1 Month 2 Month 5

Mean Back Pain at Months 1, 2, and 5

CT Video

Growth Mixture Modeling

Pre-

exercise

drop

Injury

Or

Illness

Education

level

N-Back

RT

Global

Task-

Switching RT

Digit-

Symbol

Accuracy

Class 1 0.062 0.079 0.751 1168.708 125.763 76.037

Class 2 0.011 0.014 0.232 1069.014 174.412 62.381

p 0.007 0.002 0.078 0.025 0.105 0.001

Class 1 – 90% of the sample – decrease in 4-item pain

Class 2 – 10% of the sample – increase in 4-item pain

Study Summary

Cognitive training and exergaming

appeared to reduce back pain with no

identified mediators

A decrease in back pain was associated

with better cognitive function

CT provides a cheap, effective alternative

to existing treatments for CLBP

Limitations

Secondary data analyses increase likelihood

of Type I error

Subjective measurement of pain fails to

differentiate types of pain (e.g., acute vs.

chronic pain; nonspecific pain) and may not

be invariant across subgroups.

Heterogeneous, un-impaired sample

Questions?ahua5@Illinois.edu

Next Steps

How large of an effect can CT have on

reducing pain?

Need a measure of change in

function/performance for the

exergaming tasks

MRI scans during pain stimulus to

measure activity in the dlPFC, ACC,

and insula