An Electromyographic Analysis ofCommercial and Common AbdominalExercises: Implications for Rehabilitationand TrainingRafael F. Escamilla, PT, PhD, CSCS1

Michael S.C. McTaggart, MS2

Ethan J. Fricklas, MSE3

Ryan DeWitt, MPT4

Peter Kelleher, MPT4

Marcus K.Taylor, PhD5

Alan Hreljac, PhD6

Claude T. Moorman, III, MD7

Study Design: A repeated-measures, counterbalanced design.Objectives: To test the effectiveness of 7 commercial abdominal machines (Ab Slide, Ab Twister,Ab Rocker, Ab Roller, Ab Doer, Torso Track, SAM) and 2 common abdominal exercises (crunch,bent-knee sit-up) on activating abdominal and extraneous (nonabdominal) musculature.Background: Numerous abdominal machine exercises are believed to be effective in activatingabdominal musculature and minimizing low back stress, but there are minimal data to substantiatethese claims. Many of these exercises also activate nonabdominal musculature, which may or maynot be beneficial.Methods and Measures: A convenience sample of 14 subjects performed 5 repetitions for eachexercise. Electromyographic (EMG) data were recorded for upper and lower rectus abdominis,external and internal oblique, pectoralis major, triceps brachii, latissimus dorsi, lumbarparaspinals, and rectus femoris, and then normalized by maximum muscle contractions.Results: Upper and lower rectus abdominis EMG activities were greatest for the Ab Slide, TorsoTrack, crunch, and Ab Roller, while external and internal oblique EMG activities were greatest forthe Ab Slide, Torso Track, crunch, and bent-knee sit-up. Pectoralis major, triceps brachii, andlatissimus dorsi EMG activities were greatest for the Ab Slide and Torso Track. Lumbar paraspinalEMG activities were greatest for the Ab Doer, while rectus femoris EMG activities were greatest forthe bent-knee sit-up, SAM, Ab Twister, Ab Rocker, and Ab Doer.

1 Associate Professor of Physical Therapy, California State University Sacramento, Sacramento, CA.2 Graduate student (at the time of study), Duke University Medical Center, Durham, NC.3 Student (at the time of study), Duke University Medical Center, Durham, NC.4 Student (at the time of study), California State University Sacramento, Sacramento, CA.5 Lieutenant, Medical Service Corps, US Navy, Naval Health Research Center, San Diego, CA.6 Associate Professor of Kinesiology and Health Science, California State University Sacromento,Sacromento, CA.7 Associate Professor of Orthopaedic Surgery, Duke University Medical Center, Durham, NC.The protocol used in this study was approved by the Institutional Review Board at Duke UniversityMedical Center, Durham, NC. The authors of this manuscript affirm we have no financial affiliation(including research funding) or involvement with any commercial organization that has a direct financialinterest in any matter included in this manuscript.Address correspondence to Rafael Escamilla, Associate Professor of Physical Therapy, California StateUniversity Sacramento, Department of Physical Therapy, 6000 J Street, Sacramento, CA 95819-6020.E-mail: rescamil@csus.edu

Conclusions: The Ab Slide and Torso Trackwere the most effective exercises in activatingabdominal and upper extremity muscles whileminimizing low back and rectus femoris (hipflexion) activity. The Ab Doer, Ab Twister, AbRocker, SAM, and bent-knee sit-up may beproblematic for individuals with low back pa-thologies due to relatively high rectus femorisactivity. J Orthop Sports Phys Ther 2006;36:45-57.

Key words: EMG, low back pain,lumbar spine, rectus abdominis,sit-up

Strong abdominals are im-portant for stabilizingthe trunk and helpingunload stress in the lum-bar spine.3,13 Abdominal

muscles (rectus abdominis, exter-nal oblique, internal oblique, andtransverse abdominal) are com-monly strengthened by activelyflexing the trunk with a concentricmuscle action or by resisting trunkextension (due to an externalforce such as gravity) with an iso-metric or eccentric muscle action.

Journal of Orthopaedic & Sports Physical Therapy 45

RE

SE

AR

CH

RE

PO

RT

There are numerous exercises used for abdominalstrengthening. Many of these exercises also activateextraneous (nonabdominal) muscles, such as the hipflexors, lumbar paraspinals, or upper extremity mus-culature, which may or may not be beneficial. Forexample, high activation levels from the hip flexorsand lumbar paraspinals tend to generate a forcecouple that attempts to anteriorly rotate the pelvisand increase lumbar lordosis. When coupled withweak abdominal musculature, activation of these ex-traneous muscles may increase the risk of low backpathologies.

Understanding the muscle activation generated bydifferent abdominal exercises is useful to therapistsand other health care or fitness specialists whodevelop specific abdominal exercises for their pa-tients or clients to facilitate their rehabilitation ortraining needs and objectives. For example, abdomi-nal exercises that actively flex the trunk may beproblematic for individuals with lumbar disk patholo-gies due to increased intradiscal pressure18 and lum-bar spine compression,3 and for individuals withosteoporosis due to the risk of vertebral compressionfractures.21 However, these same individuals may beasymptomatic during abdominal exercises that main-tain a relatively neutral spine and pelvis. In contrast,individuals with facet joint syndrome, spondylolis-thesis, and vertebral or intervertebral foramenstenosis may not tolerate exercises such as the AbSlide and Torso Track due to the extended spineposition.

There are numerous commercially available ab-dominal machines that are believed to be effective inactivating abdominal musculature and minimizinglow back stress, but there are little or no scientificresearch data to substantiate these beliefs. Whilethere are numerous studies that examined muscleactivity during more traditional abdominal exercises,such as the sit-up or crunch exercises,3,14,16,26,27 thereis a scarcity of data related to the use of abdominalmachines. A limited number of studies comparedselect abdominal muscle activity while performingexercises using the Torso Track, Ab Doer, Ab Shaper,Ab Flex, and Ab Roller,6,7,12,24,26 but there are nostudies that we are aware of that have quantifiedabdominal muscle activity while using the Ab Twister,Ab Rocker, Super Abdominal Machine (SAM), andAb Slide. Moreover, when using abdominal machines,the extent of recruitment of extraneous musculature,such as low back or upper and lower extremitymusculature, is currently unknown because therehave been no studies that have reported extraneousmuscle activity while performing these exercises. It isalso unknown how abdominal machines compare totraditional abdominal exercises, such as the sit-up andcrunch, in recruiting the abdominal musculature.

Abdominal machines use various techniques totarget different muscles. For example, some abdomi-

FIGURE 1. Ab Rocker.

FIGURE 2. Ab Roller.

nal machines allow only uniplanar motions, such astrunk flexion, while others use multiplanar motions,such as trunk flexion and rotation. It is commonlybelieved that performing simultaneous trunk flexionand rotation recruits the external and internal ob-lique musculature to a greater extent compared totrunk flexion only. However, there is currently noscientific evidence to support this assertion.

46 J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006

The purpose of this study was to test the effective-ness of 7 popular commercial abdominal machinesand 2 common abdominal strengthening exercises onactivating abdominal and extraneous musculature. Itwas hypothesized that significant differences would befound in the normalized electromyographic (EMG)data of both abdominal and extraneous muscle activ-ity among exercises.

METHODS

Subjects

To optimize the EMG signal, this study was limitedto a convenience sample of 14 healthy, young subjects(7 male and 7 female) who had normal or belownormal amounts of body fat for their age group.Baseline skinfold calipers (model 68900; CountryTechnology, Inc, Gays Mill, WI) and appropriateregression equations were used to assess percent bodyfat, and standards set by the American College ofSports Medicine were used to determine normal orbelow normal amounts of body fat.4 Mean (±SD) age,mass, height, and percent body fat were 24.1 ± 5.4years, 58.7 ± 4.9 kg, 166.8 ± 5.9 cm, and 22.7% ±1.9%, respectively, for females, and 24.0 ± 7.1 years,78.6 ± 13.9 kg, 179.8 ± 4.1 cm, and 9.7% ± 4.1%,respectively, for males. All subjects provided writteninformed consent in accordance with the InstitutionalReview Board at Duke University Medical Center.

FIGURE 3. Ab Doer.

FIGURE 4. Ab Twister.

FIGURE 5. Torso Track.

FIGURE 6. Ab Slide.

Individuals were excluded from the study if they hada history of abdominal or back pain, or were unableto perform all exercises pain free and with properform and technique for 12 consecutive repetitions.

Exercise Descriptions

The 7 abdominal machine exercises were the AbRocker (Figure 1), Ab Roller (Figure 2), Ab Doer(Figure 3), Ab Twister (Figure 4), Torso Track(Figure 5), Ab Slide (Figure 6), and SAM (Figure 7).The 2 common abdominal exercises tested were the

J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006 47

RE

SE

AR

CH

RE

PO

RT

bent-knee sit-up (Figure 8) and crunch (Figure 9).No subject had prior experience in performing the 7commercial abdominal exercises, but they had mod-erate experience in performing the crunch andbent-knee sit-up.

The Ab Rocker, Ab Twister, Ab Doer, and SAMexercises started and ended in a seated position witha neutral spine and pelvis. The 2 common move-ments advertised for the Ab Rocker and Ab Twisterwere the crunch (involving sagittal plane trunk flex-ion) and the oblique crunch (moving obliquelyacross the body by simultaneously flexing and rotat-ing the trunk), and both were tested with rotationoccurring to the left. Three common movementsadvertised for the Ab Doer were the body bob(frontal plane side-to-side motion), body boogie(moving in a circular motion), and good morning(involving sagittal plane trunk flexion), and all 3 ofthese variations were tested. The movement for theSAM involved sagittal plane trunk flexion, similar tohow the Ab Rocker (crunch), Ab Twister (crunch),and Ab Doer (good morning) were performed.

The Ab Roller, crunch, and bent-knee sit-up startedand ended in a supine position. The crunch and AbRoller both had 2 variations: a normal crunch involv-ing sagittal plane trunk flexion and an obliquecrunch involving moving obliquely across the body by

FIGURE 7. SAM.

FIGURE 8. Bent-knee sit-up.

FIGURE 9. Crunch.

simultaneously flexing and rotating the trunk to theleft. The primary differences between the 2 exerciseswere that during the Ab Roller the head was sup-ported by a head pad and the arms were supportedby a supporting bar (Figure 2), while during thecrunch the thumbs were positioned in the ears andthe hands were relaxed against the head (this handposition was standardized for comfort for both thecrunch and bent-knee sit-up) (Figure 9). Both varia-tions for the crunch and Ab Roller involved acurling-up motion (trunk flexion or trunk flexionwith left rotation) until both scapulae were off theground. During the bent-knee sit-up the thumbs werepositioned in the ears with the hands relaxed againstthe head, the feet were supported and held down,the knees were flexed approximately 90°, and fromthis supine position the subject simultaneously flexedthe trunk and hips until the elbows were even withthe knees (Figure 8).

The Ab Slide and Torso Track started and ended inthe quadruped position (on hands and knees withhips and shoulders flexed approximately 90°), with aneutral spine and pelvis. From this position thesubject straightened out the body by rolling forwardin a straight line (Torso Track and Ab Slide straight)or a curved line to the left (Ab Slide curved), whilemaintaining a neutral spine and pelvis (Figures 5 and6).

48 J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006

Procedures

All subjects became familiar with all abdominalexercises during a pretest session that took placeapproximately 1 week prior to the testing session.During the pretest session, each subject receivedinstructions explaining how to correctly performeach of the abdominal exercises (each abdominalmachine came with written or video instructions forits use). All exercises were performed with a 3-secondcadence (1 second from start of exercise to endrange, 1-second isometric hold at end range,1-second return to starting position) and a 1-secondrest between repetitions. The subjects practiced allexercises under the supervision of trained researchpersonnel. All of the commercial exercises except theAb Roller and Ab Slide had adjustable elastic bandsto make the exercise easier or harder. To betternormalize the intensity of each exercise, resistancewas adjusted according to each subject’s preferenceand the manufacturer’s recommendation (eg, using aresistance that was not too hard but hard enough toallow the execution of at least 15 repetitions)—similar to how each subject would adjust the resis-tance and use the equipment if they purchased it forhome use. The selected resistance during the pretestsession was also used for that subject during thetesting session. For each exercise, each subject used aresistance that enabled the subject to correctly per-form at least 15 consecutive repetitions using the3-second cadence described above. It was not possibleto normalize all exercises with exactly the samerelative intensity because the Ab Roller, Ab Slide,crunch, and bent-knee sit-up used the body only asan external resistance, while the remaining exercisesused resistance bands in addition to body as externalresistance. In addition, even when the maximumresistance possible was used for the Ab Doer, AbTwister, and Ab Rocker, all subjects indicated thatthey were capable of performing these exercises withmore resistance. A metronome (set at 1 beat persecond) was used to help ensure proper cadenceboth during the pretest and testing sessions. Once asubject was able to correctly perform each exercisewith the proper cadence, a testing session was sched-uled.

Neuroline (Medicotest Marketing, Inc, Ballwin,MO) disposable surface electrodes (type 720-00-S)were used to collect EMG data. These oval-shapedelectrodes (22 mm wide and 30 mm long) wereplaced in a bipolar electrode configuration along thelongitudinal axis of each muscle, with a center-to-center distance of approximately 3 cm between elec-trodes. Prior to positioning the electrodes over eachmuscle, the skin was prepared by shaving, abrading,and cleaning with isopropyl alcohol wipes to reduceskin impedance values, which typically were less than10 k�. Electrode pairs were then placed on the

subject’s right side (except for the internal oblique,which was positioned on the subject’s left side) forthe following muscles in accordance with procedurespreviously described5,8,17,20: (a) upper rectusabdominis, positioned vertically and centered on themuscle belly (not on tendinous intersection) near themidpoint between umbilicus and xiphoid process and3 cm lateral from midline; (b) lower rectusabdominis, positioned 8° from vertical ininferomedial direction and centered on the musclebelly near the midpoint between umbilicus and pubicsymphysis and 3 cm lateral from midline; (c) externaloblique, positioned obliquely approximately 45° (par-allel to a line connecting the most inferior point ofthe costal margin of the ribs and the contralateralpubic tubercle) above anterior superior iliac spine(ASIS) near the level of the umbilicus; (d) internaloblique, positioned horizontally 2 cm inferomedial tothe ASIS, within a triangle confined by the inguinalligament, lateral border of the rectus sheath, and aline connecting the ASISs (it has been demonstratedthat in this region only the aponeurosis of theexternal oblique, and not the external obliquemuscle, covers the internal oblique)20; (e) sternalpectoralis major, positioned horizontally 2 cm medialto the axillary fold; (f) triceps brachii long head,positioned vertically over the long head muscle bellynear midline of the arm approximately halfway be-tween the acromion and olecranon; (g) latissimusdorsi, positioned obliquely (approximately 25° fromhorizontal in an inferomedial direction) 4 cm belowinferior angle of the scapula; (h) rectus femoris,positioned vertically near midline of thigh approxi-mately halfway between ASIS and proximal patella;and (i) lumbar paraspinals, positioned vertically 3 cmlateral to spine and near level of iliac crest betweenL3 and L4 vertebrae. A ground (reference) electrodewas positioned over the skin of the right acromionprocess. Electrode cables were connected to theelectrodes and taped to skin appropriately to mini-mize pull on the electrodes and movement of thecables.

Once the electrodes were positioned, the subjectwarmed up and practiced the exercises as needed,then data collection commenced. EMG data werecollected using a Noraxon 16-channel telemyo EMGunit (Noraxon USA, Inc, Scottsdale, AZ), and theamplifier bandwidth frequency was 10 to 500 Hz. Theinput impedance of the amplifier was 20 000 k� andthe common-mode rejection ratio was 130 dB. EMGdata were sampled at 1000 Hz, recorded by atransmitter and amplifier, and broadcast to a receiverinterfaced to a computer. The recorded signals wereprocessed through an analog-to-digital (A/D) con-verter by a 16-bit A/D board.

EMG data were collected during 5 repetitions foreach exercise, with all exercises performed in arandomized order. Each repetition was performed in

J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006 49

RE

SE

AR

CH

RE

PO

RT

a slow and controlled manner using the 3-secondcadence previously described and 1-second rest be-tween repetitions. With a relatively low number ofrepetitions performed, all subjects acknowledged thatfatigue was minimized. Each testing session tookapproximately 45 minutes to complete.

Randomly interspersed within the exercise testingsession, EMG data from each muscle tested werecollected during two 5-second maximum voluntaryisometric contractions (MVICs). After conducting pi-lot work, we adopted the following protocols forMVIC testing, which were based on the positions thatelicited the greatest MVIC for each respective muscle(all MVICs were collected on a plinth with subject ina prone, supine, or short-sitting position): (a) upperand lower rectus abdominis, body supine with hipsand knees flexed 90°, feet supported, and trunkmaximally flexed (ie, curl-up position), with resis-tance at the shoulders in the trunk extension direc-tion; (b) external and internal oblique, body supinewith hips and knees flexed 90°, feet supported, andtrunk maximally flexed and rotated to the left, withresistance at the shoulders in the trunk extensionand right rotation directions; (c) sternal pectoralismajor, body supine with right shoulder flexed 90°and internally rotated, the right forearm supinated,and the right elbow slightly flexed, with resistance atthe right distal arm and forearm in the horizontalabduction direction; (d) triceps long head, bodyprone with right shoulder abducted 90° and rightelbow flexed 45°, with resistance at the right distalforearm in the elbow flexion direction; (e) latissimusdorsi, body prone with right shoulder abducted 0°and extended maximally, with resistance at the rightdistal arm in the direction of shoulder flexion; (f)lumbar paraspinals, body prone with trunk fullyextended and hands clasped behind head, with resis-tance at the shoulders in the direction of trunkflexion; and (g) rectus femoris, body in short-sittingposition with hips and knees flexed 90°, with resis-tance at the distal leg in the knee flexion direction.The MVICs were collected to normalize the EMGdata collected during the abdominal exercises. Eachsubject was given similar verbal encouragement foreach MVIC to help ensure a maximum effortthroughout the 5-second duration, and the subjectwas asked after each MVIC if he/she felt it was amaximum effort. If not, the MVIC was repeated. Anapproximately 1-minute rest was given between eachMVIC and an approximately 2-minute rest was givenbetween each exercise trial.

Data ProcessingRaw EMG signals were full-wave rectified, smoothed

with a 10-millisecond moving average window, andlinear enveloped, then averaged over the entireduration of each exercise repetition. For each repeti-tion the EMG data were normalized for each muscle

and expressed as a percentage of a subject’s highestcorresponding MVIC trial, which was determined bycalculating throughout the 5-second MVIC the high-est average EMG signal over a 1-second time interval.Normalized EMG data were then averaged over the 5repetition trials performed for each exercise andused in statistical analyses.

Data Analysis

A 1-factor repeated-measures analysis of variancewas employed to assess differences in normalizedEMG muscle activity among the different exercisevariations (P� .01). Post hoc analyses were performedusing the Bonferroni test to evaluate the significanceof between-exercise pairwise comparisons (P� .01).

RESULTS

Normalized EMG data for each muscle and exer-cise are shown in Table 1. Among all exercises tested,upper rectus abdominis EMG activities were greatestfor the Ab Slide (straight and curved), Torso Track,crunch (normal and oblique), and Ab Roller (crunchand oblique) exercises, and lowest for the Ab Twister(crunch and oblique), Ab Rocker (crunch and ob-lique), and Ab Doer (good morning, body boogie,and body bob) exercises. Lower rectus abdominisEMG activities were greatest for the Ab Slide (straightand curved) and Torso Track exercises, and lowestfor the Ab Twister (crunch and oblique), Ab Rocker(crunch and oblique), and Ab Doer (good morning,body boogie, and body bob) exercises. Graphicalrepresentations of upper and lower rectus abdominisactivity ranked from highest to lowest among allexercises are shown in Figures 10 and 11.

The external oblique EMG activity for the crunch(normal), Ab Roller (crunch), and Ab Doer (goodmorning) exercises were significantly lower comparedto the Ab Slide (straight and curved) and bent-kneesit-up exercises. Internal oblique EMG activities weregreatest for the Ab Slide (straight and curved), TorsoTrack, bent-knee sit-up, and crunch (normal andoblique) exercises, and lowest for the Ab Roller(oblique), Ab Twister (crunch and oblique), AbRocker (crunch and oblique), and Ab Doer (goodmorning) exercises. Graphical representation of ex-ternal and internal oblique activity ranked fromhighest to lowest among all exercises are shown inFigures 12 and 13.

Sternal pectoralis major EMG activities were great-est for the Ab Slide (straight and curved), TorsoTrack, SAM, and Ab Twister (crunch and oblique)exercises, and lowest for the Ab Rocker (crunch andoblique), Ab Doer (good morning, body boogie, andbody bob), Ab Roller (crunch and oblique), andcrunch (normal and oblique) exercises. Tricepsbrachii long head EMG activities were significantly

50 J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006

TABL

E1.

Aver

age

EMG

(±SD

)for

each

mus

cle

and

exer

cise

expr

esse

das

a%

ofm

axim

umiso

met

ricvo

lunt

ary

cont

ract

ion.

Upp

erRe

ctus

Abdo

min

is*Lo

wer

Rect

usAb

dom

inis*

Exte

rnal

Obl

ique

*In

tern

alO

bliq

ue*

Ster

nal

Pect

oral

isM

ajor

*Tr

icep

sLo

ngH

ead*

Latis

simus

Dor

si*Lu

mba

rPa

rasp

inal

s*Re

ctus

Fem

oris*

AbSl

ide

(stra

ight

)67

±26

72±

1940

±16

53±

1523

±7

30±

1210

±4

3±

2k5

±3fh

j

AbSl

ide

(cur

ved)

61±

2466

±19

42±

1751

±15

20±

926

±12

10±

32

±2k

9±

7f

Tors

oTr

ack

67±

2572

±17

32±

1858

±14

20±

826

±11

10±

52

±2k

6±

5fhj

Crun

ch(n

orm

al)

51±

950

±8ab

16±

11af

41±

94

±3ab

gi1

±1ab

5±

1g2

±1k

3±

2fhj

Crun

ch(o

bliq

ue)

50±

1539

±14

ab32

±22

40±

116

±5ag

i2

±2ab

8±

55

±3k

3±

2fhj

Bent

knee

sit-u

p38

±12

ab44

±13

ab41

±16

49±

218

±6ag

2±

2ab6

±3g

4±

2k36

±16

SAM

42±

17ab

50±

20ab

31±

2136

±13

b26

±15

10±

6ab12

±6

4±

2k20

±15

AbRo

ller

(cru

nch)

46±

1742

±12

ab13

±8af

38±

9b7

±5ag

i3

±2ab

5±

2g3

±2k

1±

1fhj

AbRo

ller

(obl

ique

)49

±12

36±

16ab

20±

925

±11

abf

5±

3abgi

3±

2ab6

±2g

3±

2k2

±2fh

j

AbTw

ister

(cru

nch)

19±

8abcd

e19

±10

abcd

efg

21±

1222

±9ab

f13

±11

7±

3ab5

±2g

4±

3k27

±19

AbTw

ister

(obl

ique

)20

±7ab

cde

22±

11ab

cfg

33±

1828

±11

abf

22±

155

±4ab

6±

2g5

±6k

24±

14Ab

Rock

er(c

runc

h)15

±8ab

cdef

g13

±5ab

cdef

g22

±11

24±

8abf

7±

7agi

6±

4ab6

±3g

4±

3k30

±21

AbRo

cker

(obl

ique

)14

±10

abcd

efg

14±

8abcd

efg

31±

1823

±8ab

f6

±6ag

i7

±4ab

5±

2g3

±1k

21±

16Ab

Doe

r(g

ood

mor

ning

)14

±7ab

cdef

g14

±5ab

cdef

g16

±11

af22

±13

abf

4±

4abgi

2±

1ab2

±2ab

g15

±7

12±

11f

AbD

oer

(bod

ybo

ogie

)12

±4ab

cdef

g11

±6ab

cdef

g24

±10

31±

13b

3±

2abgi

1±

1ab2

±1ab

g13

±8

24±

19Ab

Doe

r(b

ody

bob)

7±

5abcd

efg

7±

4abcd

efg

30±

1937

±18

b2

±2ab

gi2

±1ab

1±

1abg

8±

316

±14

f

*Si

gnifi

cant

diffe

renc

e(P

�.0

01)i

nEM

Gac

tivity

amon

gab

dom

inal

exer

cise

sba

sed

ona

1-w

ayre

peat

ed-m

easu

res

anal

ysis

ofva

rianc

e.Ke

yto

pairw

iseco

mpa

rison

s(P

�.0

1):

a.Si

gnifi

cant

lyle

ssEM

Gac

tivity

com

pare

dto

the

AbSl

ide

(stra

ight

and

curv

ed).

b.Si

gnifi

cant

lyle

ssEM

Gac

tivity

com

pare

dto

the

Tors

oTr

ack.

c.Si

gnifi

cant

lyle

ssEM

Gac

tivity

com

pare

dto

the

crun

ch(n

orm

al).

d.Si

gnifi

cant

lyle

ssEM

Gac

tivity

com

pare

dto

the

crun

ch(o

bliq

ue).

e.Si

gnifi

cant

lyle

ssEM

Gac

tivity

com

pare

dto

the

AbRo

ller(

crun

chan

dob

lique

).f.

Sign

ifica

ntly

less

EMG

activ

ityco

mpa

red

toth

ebe

nt-k

nee

sit-u

p.g.

Sign

ifica

ntly

less

EMG

activ

ityco

mpa

red

toth

eSA

M.

h.Si

gnifi

cant

lyle

ssEM

Gac

tivity

com

pare

dto

the

AbTw

ister

(cru

nch)

.i.

Sign

ifica

ntly

less

EMG

activ

ityco

mpa

red

toth

eAb

Twist

er(o

bliq

ue).

j.Si

gnifi

cant

lyle

ssEM

Gac

tivity

com

pare

dto

the

AbRo

cker

(cru

nch)

.k.

Sign

ifica

ntly

less

EMG

activ

ityco

mpa

red

toth

eAb

Doe

r(go

odm

orni

ngan

dbo

dybo

ogie

).

J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006 51

RE

SE

AR

CH

RE

PO

RT

FIGURE 10. Upper rectus abdominis normalized mean (SD) EMG activity among exercises.

FIGURE 11. Lower rectus abdominis normalized mean (SD) EMG activity among exercises.

greater for the Ab Slide (straight and curved) andTorso Track exercises compared to all other exercises.Latissimus dorsi EMG activities were greatest for theAb Slide (straight and curved), Torso Track, SAM,crunch (oblique), and Ab Twister (oblique) exercises,and lowest for the Ab Doer (good morning, bodyboogie, and body bob) and Ab Roller (crunch)exercises.

Lumbar paraspinal EMG activities were significantlygreater for the Ab Doer (good morning, body

boogie, and body bob) exercises compared to allother exercises. Rectus femoris EMG activities weregreatest for the bent-knee sit-up, SAM, Ab Twister(crunch and oblique), Ab Rocker (crunch and ob-lique), and Ab Doer (body boogie) exercises, andlowest for the Ab Roller (crunch and oblique), AbSlide (straight), Torso Track, and crunch (normaland oblique) exercises. The relative effectiveness ofexercises in muscle recruitment of the trunk, upperextremity, and hip musculature is shown in Table 2.

0

10

20

30

40

50

60

70

80

90

100

Ab Slid

e (s

traigh

t)

Torso

Tra

ck

Ab Slid

e (c

urve

d)

Crunc

h (n

orm

al)

Crunc

h (o

bliqu

e)

Ab Roll

er (o

bliqu

e)

Ab Roll

er (c

runc

h)SAM

Bent-k

nee

sit-u

p

Ab Twist

er (o

bliqu

e)

Ab Twist

er (c

runc

h)

Ab Roc

ker (

crun

ch)

Ab Roc

ker (

obliq

ue)

Ab Doe

r (go

od m

ornin

g)

Ab Doe

r (bo

dy b

oogie

)

Ab Doe

r (bo

dy b

ob)

Nor

mal

ized

EM

G (

%M

VIC

)

0

10

20

30

40

50

60

70

80

90

100

Ab Slide

(stra

ight)

Torso

Tra

ck

Ab Slid

e (c

urve

d)

Crunc

h (n

orm

al)SAM

Bent-k

nee

sit-u

p

Ab Roll

er (c

runc

h)

Crunc

h (o

bliqu

e)

Ab Roll

er (o

bliqu

e)

Ab Twist

er (o

bliqu

e)

Ab Twist

er (c

runc

h)

Ab Roc

ker (

obliq

ue)

Ab Doe

r (go

od m

ornin

g)

Ab Roc

ker (

crun

ch)

Ab Doe

r (bo

dy b

oogie

)

Ab Doe

r (bo

dy b

ob)

Nor

mal

ized

EM

G (

%M

VIC

)

52 J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006

FIGURE 12. External oblique normalized mean (SD) EMG activity among exercises.

FIGURE 13. Internal oblique normalized mean (SD) EMG activity among exercises.

DISCUSSION

Biomechanical Differences Between Flexion andExtension Exercises

The Ab Slide and Torso Track were the mosteffective exercises in activating abdominal muscula-ture, including the upper and lower rectus abdominisand the external and internal oblique. While per-forming these exercises, the abdominal muscles con-tract in a different manner compared to performingtraditional trunk flexion exercises. During the ‘‘roll-

out’’ portion in performing the Ab Slide and TorsoTrack, the abdominal musculature contracts eccentri-cally or isometrically to resist the attempt of gravity toextend the trunk and rotate the pelvis. During thereturn motion, the abdominal musculature contractsconcentrically or isometrically. If the pelvis and spineare stabilized and maintained in a neutral positionthroughout the roll-out and return movements, thenthe abdominal musculature primarily would contractisometrically. While performing these exercises, arelatively neutral pelvis and spine were maintainedthroughout the movement. In contrast, all other

0

10

20

30

40

50

60

70

Ab Slid

e (c

urve

d)

Bent-k

nee

sit-u

p

Ab Slid

e (s

traigh

t)

Ab Twist

er (o

bliqu

e)

Torso

Tra

ck

Crunc

h (o

bliqu

e)SAM

Ab Roc

ker (

obliq

ue)

Ab Doe

r (bo

dy b

ob)

Ab Doe

r (bo

dy b

oogie

)

Ab Roc

ker (

crun

ch)

Ab Twist

er (c

runc

h)

Ab Roll

er (o

bliqu

e)

Crunc

h (n

orm

al)

Ab Doe

r (go

od m

ornin

g)

Ab Roll

er (c

runc

h)

Nor

mal

ized

EM

G (

%M

VIC

)

0

10

20

30

40

50

60

70

80

Ab Slid

e (c

urve

d)

Ab Slid

e (s

traigh

t)

Torso

Tra

ck

Ab Roll

er (o

bliqu

e)

Crunc

h (n

orm

al)

Crunc

h (o

bliqu

e)SAM

Ab Doe

r (bo

dy b

ob)

Ab Roll

er (c

runc

h)

Ab Doe

r (bo

dy b

oogie

)

Ab Twist

er (c

runc

h)

Bent-k

nee

sit-u

p

Ab Roc

ker (

crun

ch)

Ab Roc

ker (

obliq

ue)

Ab Twist

er (o

bliqu

e)

Ab Doe

r (go

od m

ornin

g)

Nor

mal

ized

EM

G (

%M

VIC

)

J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006 53

RE

SE

AR

CH

RE

PO

RT

exercises activated abdominal musculature by activelyflexing the trunk by concentric contractions duringthe initial portion of the motion, an isometric con-traction during the middle portion, and an eccentriccontraction during the final portion of the motion.

Understanding biomechanical differences betweenexercises is important because trunk flexion may becontraindicated in certain populations, such as thosewith lumbar disk pathologies or osteoporosis. Main-taining a neutral pelvis and spine (such as perform-ing the Ab Slide or Torso Track exercise), ratherthan forceful flexion of the lumbar spine (such asduring the bent-knee sit-up), may be desirable forthese individuals. In contrast, an individual with facetjoint pain, spondylolisthesis, and vertebral orintervertebral foramen stenosis may not benefit fromexercises that maintain the spine and pelvis in aneutral or extended position, such as when using theAb Slide and Torso Track. These exercises may in factcontribute to the nerve compression. However, trunkflexion exercises, such as the crunch, bent-kneesit-up, SAM, Ab Roller, Ab Twister, Ab Rocker, and AbDoer may be beneficial.

When the lumbar spine is forcefully flexed, whichmay occur when performing many of the commercialabdominal machines used in the current study, the

anterior fibers of the intervertebral disk are com-pressed, while the posterior fibers are in tension. Inaddition, in extreme lumbar flexion intradiscal pres-sure may increase several times above the normalintradiscal pressure from a resting supine posi-tion.18,19 While these stresses on the disk may not beproblematic for the normal healthy disk, they may bedetrimental to the degenerative disk or pathologicspine.

There have only been a few studies that havecompared abdominal machine exercises to the tradi-tional crunch or bent-knee sit-up exercises.6,7,12,24,26

Most of these studies compared the crunch to the AbRoller, and like the results of the current study, therewere generally no significant differences in abdomi-nal muscle activity between these 2 exercises. Thebiggest difference between these exercises is that theAb Roller provides head support, which may make itmore comfortable to perform compared to thecrunch. The only known study to investigate abdomi-nal muscle activity between the crunch and the TorsoTrack and Ab Doer (good morning) was by Sternlichtand Rugg,24 and these authors found similar resultsas the current study: that abdominal muscle activitywas significantly greater in the crunch and TorsoTrack compared to the Ab Doer (good morning).

TABLE 2. Relative muscle recruitment of the trunk, upper extremity, and hip musculature. Note: the Ab Slide (straight and curved) andTorso Track were the exercises that produced the greatest activation of the abdominal, oblique, and upper extremity musculature, whileonly minimally recruiting the hip flexors.

Abdominal andOblique Muscles

Upper ExtremityMuscles Low Back Muscles Hip Flexor Muscles

Greatest recruitment • Ab Slide (straight andcurved)

• Torso Track

• Ab Slide (straight andcurved)

• Torso Track• SAM

• Ab Doer (good morn-ing and body boogie)

• Bent knee sit-up• Ab Rocker (crunch)• Ab Twister (crunch)• Ab Doer (body boogie)• Ab Twister (oblique)

Intermediate recruitment • Crunch (normal andoblique)

• Bent-knee sit-up• SAM• Ab Roller (crunch and

oblique)

• Ab Twister (crunchand oblique)

• Ab Rocker (crunchand oblique)

• Crunch (oblique)• Bent-knee sit-up• Ab Roller (crunch and

oblique)

• Ab Doer (body bob) • Ab Rocker (oblique)• SAM• Ab Doer (body bob)• Ab Doer (good morn-

ing)

Least recruitment • Ab Twister (crunchand oblique)

• Ab Rocker (crunchand oblique)

• Ab Doer (good morn-ing, body boogie, andbody bob)

• Crunch (normal)• Ab Doer (good morn-

ing, body boogie, andbody bob)

• Ab Twister (crunchand oblique)

• Ab Rocker (crunchand oblique)

• Ab Roller (crunch andoblique)

• Bent-knee sit-up• Crunch (normal and

oblique)• SAM• Ab Slide (straight and

curved)• Torso Track

• Ab Slide (straight andcurved)

• Torso Track• Crunch (normal and

oblique)• Ab roller (crunch and

oblique)

54 J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006

Biomechanical Differences Between the Crunch andBent-Knee Sit-up

It should be noted that not all abdominal exercisesinvolve the same degree of flexion of the lumbarspine. Halpern and Bleck14 have demonstrated thatlumbar spinal flexion was only 3° during the crunchbut approximately 30° during the bent-knee sit-up. Inaddition, the bent-knee sit-up has been shown togenerate greater intradiscal pressure18,19 and lumbarcompression3 compared to exercises similar to thecrunch, largely due to increased lumbar flexion andhip flexor activity. This implies the crunch may be asafer exercise to perform than the bent-knee sit-upfor individuals who need to minimize lumbar spinalflexion or compressive forces due to lumbar pathol-ogy.

Although the crunch and bent-knee sit-up wereboth effective in recruiting abdominal musculature,there were some differences. Several studies, includ-ing the current study, have shown that externaloblique activity, and to a lesser extent internal ob-lique activity, are significantly greater in the bent-knee sit-up compared to the crunch.1-3,16 However,upper and lower rectus abdominis activities havebeen shown to be greater in the crunch compared tothe bent-knee sit-up.6,14 In addition, like the currentstudy, hip flexor activity has been shown to be greaterin the bent-knee sit-up compared to the crunch.3,16

The Role of Abdominal Muscles in Trunk Stability

The role of the abdominal muscles, especially thetransverse abdominal and internal oblique, in en-hancing spinal and pelvic stabilization and increasingintra-abdominal pressure (IAP) has been well studied,but still remains controversial.9,10,15,22,25 IAP has beenshown to unload the spine by generating a trunkextensor moment and tensile loading to the spine.11

By making the trunk a more solid cylinder by the IAPmechanism, there is a reduction in spinal axialcompression and shear loads. The attachments of thetransverse abdominal and internal oblique into thethoracolumbar fascia may enhance spinal and pelvicstabilization, because when these muscles contractthey tense the thoracolumbar fascia. The transverseabdominal, which is the deepest of the 4 abdominalmuscles, has been shown to exhibit a similar (within15%) muscle activation pattern and amplitude as theinternal oblique muscle during many of the sametrunk flexion movements (eg, bent-knee sit-up,crunch) used in the current study.16,17 The highestEMG activities from the internal oblique were for theAb Slide, Torso Track, crunch, and bent-knee sit-upexercises, which implies that these exercises may offermore effective stabilization to the spine and pelviscompared to the other exercises.

Technique Variations

Despite slightly greater external oblique EMG ac-tivities in oblique and curved techniques, meanabdominal and oblique EMG activities generally werenot significantly different between technique varia-tions for exercises such as the Ab Slide, crunch, AbRoller, Ab Twister, and Ab Rocker (eg, normalcrunch versus oblique crunch, straight Ab Slideversus curved Ab Slide). Because simultaneous trunkflexion and rotation have been shown to increase therisk of torsional injury to the annulus fibrosis of theintervertebral disk, as well as generate relatively highlumbar compressive forces,3 and because abdominaland oblique EMG activities were generally not differ-ent between uniplanar and multiplanar trunk move-ments, the additional risks involved when performingmultiplanar trunk flexion and rotation motions arenot warranted for individuals who have lumbar diskpathologies.

Exercise Intensity

The Ab Slide and Ab Roller were the only 2commercial exercises in which resistance could notbe adjusted. This may account for more moderateamounts of muscle activity in the Ab Roller (becauseresistance could not be added to make it harder) andhigher amounts of muscle activity in the Ab Slide(because there was no way to make it easier).However, the Torso Track, which is performed in thesame manner as the Ab Slide and did have resistancebands that could be adjusted to make it easier orharder, had nearly identical muscle activity comparedto the Ab Slide. The subjects used in the currentstudy were all relatively young, active individuals whoall used the Torso Track in a more difficult resistancesetting. This more difficult resistance may be appro-priate for younger more active individuals, but older,less active, or weaker individuals may not be able tocorrectly perform the Ab Slide due to its difficultylevel. In addition, all subjects set the resistance forthe Ab Doer, Ab Rocker, and Ab Twister to themaximum number of resistance bands that could fiton each device. Even with maximal resistance, these 3commercial abdominal devices recorded the lowestamount of abdominal activity. In contrast, the AbSlide, Ab Roller, and Torso Track generated signifi-cantly greater abdominal and oblique muscle activitycompared to the Ab Doer, Ab Rocker, and Ab Twister.

Extraneous (Nonabdominal) Muscle Activity

There are no studies that we are aware of that havereported extraneous muscle activity for abdominalmachine exercises. Of the exercises tested, the AbSlide and Torso Track produced the greatest activa-tion of the upper extremity musculature, includingthe sternal pectoralis major, triceps brachii, and

J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006 55

RE

SE

AR

CH

RE

PO

RT

latissimus dorsi. Both the sternal pectoralis major(lower fibers) and latissimus dorsi contract eccentri-cally during the initial roll-out phase to control therate of shoulder flexion due to gravity, and concentri-cally in the return phase as the shoulders extend.Because the elbows typically remain slightly flexedand at a fixed elbow angle throughout the move-ment, the triceps brachii primarily contract isometri-cally throughout the movement. Although the hipflexors would appear to also contract eccentricallyduring the initial roll-out phase to control the rate ofhip extension and concentrically during the returnphase to aid in hip flexion, we unexpectedly foundlow rectus femoris activity for both the Ab Slide andTorso Track exercises. It appears that upper extremitymuscles may have a greater role compared to the hipflexors in controlling and causing exercise move-ments during these 2 exercises. Although the activityof the psoas muscle was not measured in the currentstudy due to being a deep muscle, it has beendemonstrated that during exercises performed in asimilar position and manner as the Ab Slide andTorso Track that psoas EMG magnitudes are low andthat psoas EMG magnitudes are typically within ap-proximately 10% of rectus femoris EMG magni-tudes.16,17 From these data it can be hypothesizedthat psoas activity, like rectus femoris activity, isrelatively low during the Ab Slide and Torso Track.However, because the Torso Track and Ab Slide areunique exercises in which psoas activity has not yetbeen quantified, additional studies are needed to testthis hypothesis.

Because the Ab Slide and Torso Track exercisesproduced the greatest activation of both abdominaland upper extremity musculature, these exercises maybe beneficial for individuals with limited workouttime and whose goal is to perform exercises that notonly provide an abdominal workout but also anupper body workout. The greater relative intensityand number of muscles used during the Ab Slide andTorso Track exercises implies that these exercises mayalso achieve a greater energy expenditure comparedto the other exercises. Moreover, tension in thelatissimus dorsi in addition to the internal oblique(and presumably the transverse abdominal), which alltense the thoracolumbar fascia, may enhance trunkstabilization while performing these exercises. Itshould also be emphasized that cocontraction of thelumbar paraspinal muscles, along with abdominal andlatissimus dorsi musculature, may enhance trunkstability and spine stiffness. Although excessive activityfrom the lumbar paraspinals can cause high compres-sive and shear (especially at the L5-S1) forces on thelumbar spine,3,16,23 the relatively low lumbarparaspinal activity in all the exercises tested is prob-ably not high enough to by itself cause deleteriouseffects to the lumbar spine.

Performing exercises that generate high activityfrom the hip flexors and lumbar paraspinals may notbe advantageous for those with weak abdominalmuscles or lumbar instability, because the forcesgenerated when these muscles act to anteriorly rotatethe pelvis may increase the lordotic curve of thelumbar spine. Individuals with weak abdominalmuscles or lumbar instability may want to avoid thebent-knee sit-up, SAM, Ab Twister, Ab Rocker, and AbDoer exercises due to the relatively high rectusfemoris activity. In exercises performed similarly tothe exercises in the current study, psoas and iliacusactivities have been shown to be similar in magnitudeand recruitment pattern as rectus femoris activity.1,2,17

The psoas muscle, by its attachments into the lumbarspine, acts to hyperextend the spine as it flexes thehip during the bent-knee sit-up and similar types ofhip flexion exercises, which may be detrimental toindividuals with lumbar instability. It has also beendemonstrated that the psoas muscle can generatecompression of the lumbar spine and anterior shearforce at L5-S1,16,23 which may be problematic forindividuals with lumbar disk pathologies. In addition,the role of gravity in generating L5-S1 shear forces insome exercises, such as the Torso Track and Ab Slide,should also be considered when examining injury riskto the low back. Unfortunately, it is unknown howmuch L5-S1 shear force is generated while perform-ing the Torso Track and Ab Slide, and whether ornot these forces are high enough to be problematicin some patients with low back pathologies.

Effects of Electrode Placement on EMG CrosstalkThe electrode positions used in the current study

have been shown to minimize EMG crosstalk fromother muscles.5,8,20 This is especially true for theinternal oblique, which was the only muscle testedthat is not a superficial muscle. The internal obliquenormally lies deep to the external oblique, andtherefore is susceptible to considerable EMG crosstalkfrom this muscle. However, it has been shown thatthe internal oblique is only covered by theaponeurosis of the external oblique, and not theexternal oblique muscle, within the triangle confinedby the inguinal ligament, lateral border of the rectussheath, and a line connecting the ASISs.20 Therefore,surface electrodes are appropriate to use for theinternal oblique when electrode placement is withinthis area, especially when clinical questions are beingdiscussed and if a small percentage of EMG crosstalkis acceptable. In fact, it has been shown that whenperforming trunk flexion exercises similar to those inthe current study, mean internal and external ob-lique EMG data from surface electrodes (similarlylocated as in the current study) were only approxi-mately 10% different compared to mean internal andexternal oblique EMG data from intramuscular elec-trodes.17 These authors demonstrated that appropri-

56 J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006

ately placed surface electrodes accurately reflect(within 10%) the muscle activity within the internalor external oblique muscles.

CONCLUSIONS

The exercises in the current study activated ab-dominal muscles by actively flexing the trunk(crunch, bent-knee sit-up, SAM, Ab Roller, AbTwister, Ab Rocker, Ab Doer) or by resisting trunkextension (Ab Slide and Torso Track). The Ab Slideand Torso Track exercises produced the highestactivation of the abdominal and upper extremitymuscles while minimizing low back and hip flexionactivity. Both the bent-knee sit-up and crunch exer-cises demonstrated similar amounts of abdominalactivation, while the Ab Twister, Ab Rocker, SAM, AbDoer, and bent-knee sit-up exercises exhibited thegreatest rectus femoris activity. The Ab Doer (goodmorning and body boogie) exhibited the greatestamount of lumbar paraspinal activity.

ACKNOWLEDGEMENTSWe would like to acknowledge Mike Andrawes,

Tracy Lowry, and Mark Adams for all their help indata collection and analyses in this project.

REFERENCES1. Andersson EA, Ma Z, Thorstensson A. Relative EMG

levels in training exercises for abdominal and hip flexormuscles. Scand J Rehabil Med. 1998;30:175-183.

2. Andersson EA, Nilsson J, Ma Z, Thorstensson A. Ab-dominal and hip flexor muscle activation during varioustraining exercises. Eur J Appl Physiol Occup Physiol.1997;75:115-123.

3. Axler CT, McGill SM. Low back loads over a variety ofabdominal exercises: searching for the safest abdominalchallenge. Med Sci Sports Exerc. 1997;29:804-811.

4. Balady GJ, Franklin BA, Whaley MH, Howley ET.ACSM’s Guidelines for Exercise Testing and Prescrip-tion. 6th ed. Philadelphia, PA: Lippincott Williams &Wilkins; 2000.

5. Basmajian J, Blumenstein R. Electrode Placement inEMG Biofeedback. Baltimore, MD: Williams & Wilkins;1980.

6. Beim GM, Giraldo JL, Pincivero DM, Borror MJ, Fu FH.Abdominal strengthening exercises: a comparative EMGstudy. J Sport Rehab. 1997;6:11-20.

7. Clark KM, Holt LE, Sinyard J. Electromyographic com-parison of the upper and lower rectus abdominis duringabdominal exercises. J Strength Cond Res. 2003;17:475-483.

8. Cram JR, Kasman GS. Introduction to SurfaceElectromyography. Gaithersburg, MD: Aspen Publishers,Inc; 1998.

9. Cresswell AG, Grundstrom H, Thorstensson A. Observa-tions on intra-abdominal pressure and patterns of ab-dominal intra-muscular activity in man. Acta PhysiolScand. 1992;144:409-418.

10. Cresswell AG, Blake PL, Thorstensson A. The effect ofan abdominal muscle training program on intra-abdominal pressure. Scand J Rehabil Med. 1994;26:79-86.

11. Daggfeldt K, Thorstensson A. The role of intra-abdominal pressure in spinal unloading. J Biomech.1997;30:1149-1155.

12. Demont RG, Lephart SM, Giraldo JL, Giannantonio FP,Yuktanandana P, Fu FH. Comparison of two abdominaltraining devices with an abdominal crunch usingstrength and EMG measurements. J Sports Med PhysFitness. 1999;39:253-258.

13. Gardner-Morse MG, Stokes IA. The effects of abdominalmuscle coactivation on lumbar spine stability. Spine.1998;23:86-91; discussion 91-82.

14. Halpern AA, Bleck EE. Sit-up exercises: anelectromyographic study. Clin Orthop Relat Res.1979;172-178.

15. Hodges PW, Richardson CA. Contraction of the ab-dominal muscles associated with movement of thelower limb. Phys Ther. 1997;77:132-142; discussion142-134.

16. Juker D, McGill S, Kropf P, Steffen T. Quantitativeintramuscular myoelectric activity of lumbar portions ofpsoas and the abdominal wall during a wide variety oftasks. Med Sci Sports Exerc. 1998;30:301-310.

17. McGill S, Juker D, Kropf P. Appropriately placed surfaceEMG electrodes reflect deep muscle activity (psoas,quadratus lumborum, abdominal wall) in the lumbarspine. J Biomech. 1996;29:1503-1507.

18. Nachemson A. Lumbar intradiscal pressure. In: JaysonMIV, ed. The lumbar Spine and Back Pain. Edinburg,Scotland: Churchill Livingstone; 1987:191-203.

19. Nachemson AL. The lumbar spine: an orthopaedicchallenge. Spine. 1976;1:59-71.

20. Ng JK, Kippers V, Richardson CA. Muscle fibre orienta-tion of abdominal muscles and suggested surface EMGelectrode positions. Electromyogr Clin Neurophysiol.1998;38:51-58.

21. Ralston SH, Urquhart GD, Brzeski M, Sturrock RD.Prevalence of vertebral compression fractures due toosteoporosis in ankylosing spondylitis. BMJ.1990;300:563-565.

22. Richardson CA, Snijders CJ, Hides JA, Damen L, PasMS, Storm J. The relation between the transversusabdominis muscles, sacroiliac joint mechanics, and lowback pain. Spine. 2002;27:399-405.

23. Santaguida PL, McGill SM. The psoas major muscle: athree-dimensional geometric study. J Biomech.1995;28:339-345.

24. Sternlicht E, Rugg S. Electromyographic analysis ofabdominal muscle activity using portable abdominalexercise devices and a traditional crunch. J StrengthCond Res. 2003;17:463-468.

25. Thomson KD. Estimation of loads and stresses inabdominal muscles during slow lifts. Proc Inst MechEng [H]. 1997;211:271-274.

26. Warden SJ, Wajswelner H, Bennell KL. Comparison ofAbshaper and conventionally performed abdominal ex-ercises using surface electromyography. Med Sci SportsExerc. 1999;31:1656-1664.

27. Willett GM, Hyde JE, Uhrlaub MB, Wendel CL, KarstGM. Relative activity of abdominal muscles duringcommonly prescribed strengthening exercises. J StrengthCond Res. 2001;15:480-485.

J Orthop Sports Phys Ther • Volume 36 • Number 2 • February 2006 57

RE

SE

AR

CH

RE

PO

RT