P

Original Research

Ultrasound Measurement of Transversus AbdominisDuring Loaded, Functional Tasks in AsymptomaticIndividuals: Rater Reliability

Todd Watson, DPT, OCS, FAAOMPT, Sue McPherson, PhD, Sherry Fleeman, MD

Objective: To examine the ability of 2 clinicians to consistently measure recordedcontractions of the transversus abdominis (TrA) muscle via ultrasound imaging (USI)during lifting and reaching tasks typical of everyday or work-related activities.Design: Methodologic survey that measured inter- and intrarater reliability.Setting: University laboratory.

articipants: A subset of 54 cine-loop images that were randomly selected from 6randomly selected asymptomatic adults from a larger study that consisted of a sample ofconvenience of 20 asymptomatic adults, with a mean (SD) age of 18.8 � 2.5 years and mean(SD) body mass index of 25.5 � 3.5 kg/m2.Main Outcome Measures: Thickness changes in the TrA muscle at rest and duringcontraction while performing 5 functional, loaded tasks. Intraclass correlation coefficients(ICC) were used to estimate reliability. ICC model 2,1 was used for all reliabilityanalyses on 3 TrA muscle measures: minimum thickness, maximum thickness, andpercentage change in muscle thickness. Percentage thickness change of TrA muscle wascalculated as ([thicknessmax � thicknessmin]/thicknessmin) � 100.Results: Two clinicians both scored 9 trials of 6 randomly selected participants (54 cineloops). Rater 1 scored these images again 28 days later. All image information (participant,trial, task, and testing session) was masked. Interrater ICC (2,1) for TrA muscle measureswere moderate (0.71 rested state, 0.83 contracted state, 0.81 percentage change of musclethickness); and high for intrarater ICC (2,1) (0.97 rested state, 0.99 contracted state, and0.95% change in muscle thickness).Conclusion: This study provides data for ensuring acceptable reliability of USI measuresof TrA muscle thickness and thickness changes taken during loaded and functionalactivities. This study is the first to examine interrater and intrarater reliability of recordedcine loop images of asymptomatic adults in upright positions on 3 measures of TrA muscleactivation (minimum, maximum, and percent change in muscle thickness). Reliable USImeasures of TrA muscle thickness changes meet the ongoing need for clinicians’ knowledgeof proper and sufficiently adequate muscle activation in the clinical setting to assistprogression of lumbar stabilization exercises.

PM R 2011;3:697-705

INTRODUCTION

The abdominal drawing-in maneuver (ADIM) is a motor control exercise for the transversusabdominis (TrA) muscle. As used in the clinical setting, the ADIM is a therapeutic exercisewith instruction that typically includes verbal and tactile feedback provided by the therapist.Still, current evidence [1-3] indicates that this instruction alone may not be the mosteffective method for ADIM instruction because direct palpation of the TrA muscle (by thepatient and by the therapist) is impossible. Other issues encountered with ADIM instructioninclude the time required for client education and a lack of expertise in assessment of thecorrect technique on the part of the instructor and/or the client. To address some of theseinstructional limitations, rehabilitation researchers have begun to explore ADIM instruction

via ultrasound imaging (USI).

PM&R © 2011 by the American Academy of Physical Me1934-1482/11/$36.00

Printed in U.S.A. D

T.W. Department of Physical Therapy, 319Moore Bldg, Western Carolina University, Cul-lowhee, NC 28715. Address correspondenceto: T.W.; e-mail: twatson@wcu.eduDisclosure: nothing to disclose

S.M. Department of Physical Therapy, West-ern Carolina University, Cullowhee, NCDisclosure: nothing to disclose

S.F. Asheville Radiology Associates, Asheville,NCDisclosure: nothing to disclose

Disclosure Key can be found on the Table ofContents and at www.pmrjournal.org

Peer reviewers and all others who controlcontent have no relevant financial relation-ships to disclose.

Submitted for publication September 8, 2010;accepted March 22, 2011.

dicine and RehabilitationVol. 3, 697-705, August 2011

OI: 10.1016/j.pmrj.2011.03.015697

n

698 Watson et al ULTRASOUND OF TRANSVERSUS ABDOMINIS IN ASYMPTOMATIC INDIVIDUALS

USI has been used in the rehabilitation setting since themid 1990s. Although USI is commonly used to identifypathoanatomic conditions, such as muscle tears, hematomas,and ligamentous injuries common in sports medicine [4,5],its use in musculoskeletal rehabilitation has been in theimaging of deep muscle activation [3,6-12]. Biofeedback hasbeen recommended for patients with dysfunction in theactivation of the deeper abdominal muscles, particularlytransversus abdominis (TrA) [2,3,9,11-13]. USI shows the“corset-like action” or lateral sliding and thickening of theTrA muscle during ADIM. Patients thus are able to seechanges in their muscle cross-sectional area, and cliniciansare able to assess the muscle recruitment pattern [14].

The use of USI has been found to be a valid tool formeasuring lateral abdominal wall muscle thickness changes,particularly the TrA muscle [3,6,15-19]. Therefore, USI is

ow commonly used [20-28] to determine the degree of TrAmuscle activation and subsequent contribution to dynamicspinal stability by measurement of cross-sectional thicknessin the ADIM contracted state.

Many studies that examined ADIM training for TrA mus-cle activation have focused on simple tasks or exercises insymptomatic populations, because easy tasks often are pre-ferred to simplify instruction and/or to reduce pain duringperformances. Thus, ADIM training often is conducted inprone, supine, “hook-lying,” or 4-point kneeling positions[3,8,17,19,20,29-32]. These typically are considered “un-weighted” positions for the lumbar spine that, in the clinic,are found to reduce pain during initial exercise programonset. Accordingly, few studies have examined measurementissues regarding TrA muscle activity during functional tasksperformed in upright postures or positions [13,16,33-35]that would be of greater challenge to the lumbar stabilizingsystem.

Costa et al [36] recently conducted a systematic reviewthat examined USI reproducibility as applied to measures ofthe abdominal wall musculature in physical therapy practice.Studies examined reported any type of reliability, used anytype of USI measure, and described participants to be con-sidered for analysis. Only 21 of 315 studies located metinclusion criteria for analysis. Study designs included re-peated testing of the same images, the entire measurementprocedure, or a portion of the measurement procedure. As-pects of reliability varied with regard to participants, instruc-tions, tasks, testing, and USI measurement. Interestingly,studies reported good-to-excellent reliability for single mea-sures of thickness and poor-to-good reliability for measuresof percentage of change in TrA muscle thickness; however,no studies measured percentage of change in TrA musclethickness over time. Also of interest to our study, only 3 of 21studies in the systematic review examined activation taskswith the participants in upright postures. These 3 studiesexamined reproducibility of the whole process of measuring

via intraclass correlation coefficient (ICC) for analysis of

relative reliability and standard error of the measurement(SEM) for analysis of absolute reliability. Two of the 3 studiesexamined automatic characteristics of TrA muscle activationduring tasks, with participants not receiving training or in-structions to activate muscles [34,37,38]. Overall, Costa et al[36] noted that the studies analyzed lacked detail regardingassessors, blinding, task randomization, test order, and mea-sures. They suggested that researchers developing reliabilityprotocols for TrA muscle activation training in symptomaticadults do the following: replicate clinical contexts as much aspossible; use ICC model 2, 1 for generalizability purposes;use changes in muscle-thickness measures; and to expecttrial-to-trial variations in the performance of activation tasks.Therefore, the aim of our study was to initiate research thatexamined measurement issues concerning TrA muscle acti-vation training in healthy adult populations with the intent todevelop appropriate protocols and reliable measures for in-jury prevention programs in this area.

Moreover, the present study was designed to examinereliability of clinicians’ measurements of TrA muscle cine-loop images obtained on a subsample of 6 subjects who werepart of a larger study (N � 20). The larger study was designedto examine TrA muscle activations in asymptomatic adultsduring upright functional tasks before and after ADIM train-ing in the supine position with USI feedback. Our purposewas to examine the reproducibility of TrA muscle measure-ments via USI during functional tasks.

METHODS

Design

We present a methodologic study that examined inter- andintrarater reliability of TrA muscle measurements obtainedby 2 clinicians on 54 cine-loop images that represent 5functional tasks. The recorded images used in this reproduc-ibility study were randomly selected from a longitudinalsingle-group repeated-measures study that was designed toexamine TrA muscle activation of asymptomatic adults dur-ing upright functional tasks before and after ADIM trainingwith USI feedback with the participant in the supine posi-tion.

Participants

Recorded images were drawn from an original sample ofconvenience used to recruit healthy adults with no history oflow back pain (LBP). Twenty-four undergraduate studentsenrolled in an introductory undergraduate athletic trainingcollege course volunteered to participate in this study. Thehuman subjects institutional review board of Western Caro-lina University approved the protocol for this study. All theparticipants provided informed consent before participation,

and their rights were protected at all times.

699PM&R Vol. 3, Iss. 8, 2011

An algorithm of participant inclusion and exclusion crite-ria is presented in Figure 1. Participants were eligible forinclusion in the study if they were between 18 and 40 years ofage. Participants who reported any of the following wereexcluded: (1) previous formal training with the ADIM by aphysical therapist, (2) incidence of low back pain in the past12 months as assessed by a Physical Activity Readiness Ques-tionnaire modified for this study, (3) surgical history thatinvolved the lumbar spine or abdominal wall (eg, lumbarmicrodiscectomy, appendectomy), and (4) known spinalpathology. The participants were required to complete 3consecutive successful ADIM repetitions while in the supinehook-lying position within 10 trials during training to prog-ress in the study. The participants were not excluded from

Figure 1. Algorithm of subject inclusion criteria. EC � exclusionabdominal drawing-in maneuver.

the study for exercise and/or activity habits, including prior

instruction in yoga, Pilates, or other core training techniquesthat were not completed by a licensed physical therapist.

Two participants were excluded from progressingthrough the study because they were unable to perform 3successful isolated TrA muscle contractions during ADIMtraining while in the supine position with USI feedback. Twoparticipants were excluded because of body composition thatcaused poor-quality images that were unable to be inter-preted (some patients are USI “unfriendly”[14] because offatty infiltrate of their muscle-tissue composition, and thismakes them difficult to image satisfactorily). Thus, the sub-sample used to examine reproducibility of raters’ measure-ments consisted of 6 participants randomly selected from asample of 20 participants from a larger study. The subsample

ia; PT � physical therapy; TrA � tranversus abdominis; ADIM �

criter

consisted of 2 women and 4 men (mean [SD] age, 18.3 � 0.5

6ww

700 Watson et al ULTRASOUND OF TRANSVERSUS ABDOMINIS IN ASYMPTOMATIC INDIVIDUALS

years; height, 169.7 � 8.4 cm; weight, 68.4 � 11.4 kg, andbody mass index, 23.6 � 2.7 kg/m2). Also, all participantswere right-hand dominant.

Instruments and Procedures

Each participant was tested and trained individually by usingthe same room, equipment, experimental protocol, research-ers, and research assistants. Testing sessions were conducted

Figure 2. Task 1: “neutral standing” with neutral lumbopelvicalignment. Left: “relaxed” neutral standing. Right: neutralstanding with abdominal drawing-in maneuver (arrow).

Figure 3. Upper left, clockwise: task 2, “loaded neutral standin-inch [15-cm] reach with weighted object in position of dimineighted crate” (maintaining lumbopelvic neutral during lif

eighted crate” (moving beyond lumbopelvic neutral into flexed p

before and after ADIM training with the participant in thesupine position. Five tasks were examined for each partici-pant (performed 3 times per testing session): task 1, “un-loaded neutral standing”; task 2, “loaded neutral standing”using a 4.26-kg free weight; task 3, “standing extended reachwith object” using a 4.26-kg free weight with a 6-inch for-ward reach; task 4, “proper biomechanical lift of weightedcrate” using a 5.62-kg weighted crate; and task 5, “improperlift of weighted crate” using a 5.62-kg weighted crate. Task 1position specifically with and without ADIM is shown inFigure 2, and performances of tasks 2-5 are depicted inFigure 3.

Each participant per testing session performed eachtask randomly. Task orders were assigned a number byusing a Latin square design that consisted of 10 possibleorders. Each number was written on a separate card, andone card was drawn (without replacement) per participantper testing session. Each task was described verbally via ascript and modeled by the same research assistants beforeeach bout of trials per testing session. If a participant failedto comply with the task performance as demonstratedand/or described, then a research assistant in charge of thisduty provided visual and tactile cues until the participantcomplied. The participants were instructed to contract theTrA muscle via ADIM before each trial during the testsession that followed ADIM training and to exhale slowlyduring the functional task. To ensure that the participant’stask performances and experimenter instructions wereconsistent during data collection, test sessions were video

k 3, demonstrating “standing extended reach with object” (alordosis); task 4, demonstrating “proper biomechanical lift ofeighted crate); and task 5, demonstrating “improper lift of

g”; tasished

t of w

osition).

t

701PM&R Vol. 3, Iss. 8, 2011

recorded and an observational instrument was developedfor each task to ensure that the subjects complied with taskinstructions and/or made corrections when necessary (eg,when a subject was asked to keep his or her legs straightduring the improper lift of weighted crate task he or shedid so). Before any TrA muscle measurement, analysis oftask performances and instructions were verified live andvia digital recordings by the second author (S.L.M.). Noissues emerged with compliance concerning task perfor-mances or instructions.

ADIM training in the supine position was taught individ-ually to each participant by the first author (T.W.). Rater 1(T.W.) had 11 years of experience in ADIM training with USIin healthy and low back pain populations and currentlyconducts professional workshops on how to incorporate USIto enhance ADIM of clients in therapeutic settings. Verbaland written instructions of the technique were provided,along with diagrams of the muscles involved and the ratio-nale for this technique to prevent injury. Once the partici-pants assumed the correct supine position, rater 1 showed,via the ultrasound video display, the participants their TrAmuscle while at rest and when contracted. The participantswere told to use this visual feedback to help them to acquirethe ADIM. Success of TrA muscle contraction was providedby rater 1 in the form of verbal feedback by using USI of theTrA muscle and cues regarding ADIM. The participantsviewed the changes in muscle thickness while receiving ver-bal feedback and specific commands from rater 1. At least 3consecutive successful ADIM trials (TrA muscle contrac-tions) were required to continue the study. Up to 10 attemptswere allowed to achieve this inclusion criterion, with practicetrials, and ceased when the 3 successful contractions wereobserved. The individualized ADIM training lasted about10-15 minutes.

USI visualization of TrA muscle thickness was completedby using a portable ultrasound probe set at 7.5-MHz(InNovaSound USB; Direct Medical Systems, Pleasanton,CA). The 3.0-cm linear-array transducer (operating in Bmode) was placed on the participant’s right anterolateralabdominal wall (approximately 2 inches [5.1 cm] directlysuperior to the anterior superior iliac crest) and was used forall USI data. The transducer was manipulated to provideoptimal visualization of the fascial planes of the lateral ab-dominal muscles, identifying TrA muscle. Rater 1 collectedall the images of all the participants during the functionaltasks, a 20-second cine-loop image recording.

Measurements

A physical therapist (rater 1) measured recorded images on 2occasions, and his findings were compared with a radiolo-gist’s readings (rater 2 [S.F.]). Trials for this reproducibilitystudy were randomly selected from data collected in phase 1

of our study (before and after training), and the raters were

blinded to all participant, trial, and task information. Thesecond author, blinded to the quality of images and all otherinformation except identification number, randomly selectedidentifications and images to be measured by the raters.Measurements of resting and contracted states were includedin our analyses because it is important that clinicians accu-rately detect these states via USI during actual TrA muscleactivation training and also is important when consideringthat the adults were performing functional tasks while inupright positions. Finally, both raters had experience in USIyet were licensed professionals in different fields. Rater 1 hadextensive experience in this procedure but did not apply itfrequently in day-to-day practice, whereas rater 2 used USIfrequently in day-to-day practice but was new to this proce-dure of TrA muscle examination.

Scoring took place several weeks later to control for anyrater biases. All digital recordings were accompanied by acoding system that blinded raters to participant, trial, task,and testing session information. TrA muscle resting andcontraction states were scored by using digital recordings ofeach trial by using software calipers. Measurements of TrAmuscle thickness were made between the imaged hyper-echoic facial planes between the deep and superficialborders of the muscle. SPSS version 17.0 (SPSS, Inc, Chi-cago, IL) was used to compute the percentage of change inTrA muscle thickness of each trial ([{thicknessmax �hicknessmin}/thicknessmin] � 100).

To examine rater reliability issues, rater 1 trained rater 2 onscoring of the 20-second cine-loop image recordings. The raterhad to observe the 20-second recording, determine the mini-mum and maximum TrA muscle thickness images in the re-cording to be measured, and then use the onscreen electronicsoftware calipers. As mentioned previously, all images wereprepared by the second author and coded so that the raterswere blinded to the participant, trial, task, and testing sessioninformation. Rater 2 was not familiar with any aspects of thisstudy at the time of training or scoring.

The second author randomly selected a subsample ofparticipants’ images and tasks to be measured from the largersample; this selection process was done without viewingactual images to ensure that the quality of the image was nota source of bias. All the tasks were coded via a numberingsystem to avoid any task-selection bias. The first participantswere drawn. Next, tasks and sessions were drawn by alter-nating 1 or 2 tasks per test session per participant; all 3 trialswere measured per task drawn. The subsample of partici-pants drawn consisted of 2 women and 4 men. All the taskswere well represented because tasks 1-3 were drawn 4 timeseach and tasks 4-5 were drawn 3 times each.

Both raters measured resting and contracted states of TrAmuscle via sample images (cine loops) from trials not in-cluded in this study. Training ceased when both partiesindividually scored 10 trials of 10 participants the same.

Training lasted about 3 hours. Next, both raters scored 3

Sfocm

P

Am

hpa

(

*

I

702 Watson et al ULTRASOUND OF TRANSVERSUS ABDOMINIS IN ASYMPTOMATIC INDIVIDUALS

trials of 3 randomly selected tasks of 6 randomly selectedparticipants (54 cine loops). Scoring was done individually.Finally, rater 1 scored these images again 28 days later.

Statistical Analysis

Inter- and intrarater reliability were examined with ICCs byusing analysis of variances (ANOVA). ICC model 2 for asingle measurement (ICC 2,1) was used for all tests becausewe were interested in generalizing findings. Also, ANOVAswere used to test heterogeneity of images to ensure that therandomization process produced enough variability amongimages to support accuracy of ICC findings. Furthermore,the standard error of the measurement (SEM � pooled SD ��1 � ICC) was calculated to assess measurement precisionor actual number of millimeters by which the measurementsdiffered. A 95% confidence interval (CI) (mean � 1.96 �

EM) was constructed to determine the range of millimetersor each measurement. The “true” score was contained 95%f the time within this range. Although values below the SEMould be considered measurement error, we also calculatedinimal detectable change (MDC95 � SEM � �2 � 1.96),

which indicates that the minimal amount of change that isnot likely to be caused by chance variation in measurement.SPSS version 17.0 was used for all analyses.

Table 1. Descriptive data (mean [SD]) of characteristics, by geN � 6)

Characteristic Men (N � 9)* Wo

Age (y) 20.6 � 4.8Height (cm) 170.8 � 6.8Weight (kg) 76.0 � 18.0Body mass index (kg/m2) 25.8 � 4.5

Total study participants.†Subsample participants.

Table 2. Reliability measurements of 54 images of transversus ain thickness) via ultrasound imaging

ICC CI95

InterraterMinimum 0.710 0.083-0.887Maximum 0.830 0.098-0.946% Change 0.811 0.690-0.887

IntraraterMinimum 0.970 0.945-0.981Maximum 0.990 0.977-0.992% Change 0.950 0.908-0.968

CC � intraclass correlation coefficients (model 2, 1); CI95 � 95% confide(pooled SD � [{(n1 � 1)s1

2 � (n2 � 1)s22}/n1 � n2] ½), where n � 54 ima

of variance.*P � .05.†P � .05.‡Values are given in millimeters except percent change.§ANOVAs, examining rater differences.

�ANOVAs, examining heterogeneity of images.

RESULTS

Demographic characteristics of participants are provided inTable 1. Results for measures of inter- and intrarater reliabil-ity are presented in Table 2. Intrarater reliability coefficientswere extremely high, whereas interrater reliability coeffi-cients were less in comparison. SEM followed the same trend,because both inter- and intrarater SEM showed high preci-sion (�0.1 mm) for dynamic measurement of percentage ofchange, and response stability ranged from 0.02 to 0.30 mm(Table 2). MDC was appropriately larger than SEM but verygood for percentage of change (�0.2 mm). Overall, raterreliability was high. ANOVAs used to examine rater differ-ences indicated that rater 1 scored consistently higher thanrater 2 because significance was noted for differences be-tween the raters on measures of resting state (F1,53 � 61.855,

� .01), contracted state (F1,53 � 87.84, P � .01), andpercentage of change (F1,53 � 4.736, P � .034). Results of

NOVAs for intrarater reliability showed no significance oneasures of resting state (F1,53 � 0.529, P � .470), con-

tracted state (F1,53 � 2.896, P � .095), and percentage ofchange (F1,53 � 0.513, P � .447). ANOVAs used to examine

eterogeneity identified that there was enough variabilityresent in the data to obtain accurate reliability coefficientscross all measures. (eg, interrater reliability of percentage of

of study total participants (N � 20) and subsample participants

N � 11)* Men (N � 4)† Women (N � 2)†

� 0.5 18.0 � 0.0 19.0 � 0.0� 3.6 171.1 � 9.7 166.7 � 5.8� 9.6 68.7 � 12.7 67.6 � 12.8� 3.3 23.4 � 3.0 24.2 � 3.0

inis muscle (minimum, maximum, and percentage [%] change

‡ MDC95‡ Rater(s)§ Images�

4 0.4445 * *6 0.8248 * *8 0.1242 * *

9 0.1383 † *6 0.2012 † *3 0.0563 † *

rval; SEM � standard error of the measurement: pooled SD � �1 � ICCDC95 � minimal detectable change: SEM � �2 � 1.96; ANOVA � analysis

nder,

men (

18.3164.065.624.4

bdom

SEM§

0.1600.2970.044

0.0490.0720.020

nce integes; M

smotm

m1s

703PM&R Vol. 3, Iss. 8, 2011

change: F1,53 � 35.185, P � .01). Fifty-four trials werecored by both raters for minimum (Figure 4A) and maxi-um (Figure 4B) contractions. The 54 trials scored by rater 1

n 2 occasions are presented in Figure 5. These individualrials support absolute intrarater reliability, because measure-ents were similar on each trial on each occasion.

DISCUSSION

Our primary aim in this measurement study involved raterreliability and examined the ability of 2 clinicians of differing

Figure 5. Minimum and maximum transversus abdominis (TrA)uscle measurements of rater 1 scored on 2 occasions on trials

-3 of 3 randomly selected tasks performed by 6 randomly

Figure 4. Minimum (A) and maximum (B) transversus abdomi-nis (TrA) muscle measurements between 2 raters on 54 cine-loop images.

elected participants (54 total trials or cine loops were scored).

disciplines to consistently measure a contraction of the TrAmuscle via USI during lifting and reaching tasks typical ofeveryday- or work-related activities. Intrarater comparisonsof TrA muscle thickness (both minimum and maximum)measurements generally showed high reliability. Althoughinterrater reliability was generally lower than intrarater reli-ability, the values were acceptable. Notably, the percentage ofchange values for interrater reliability were as follows: ICC(2,1) � 0.811, SEM � 0.04 mm, MDC � 0.12 mm.

The SEM denotes the degree to which a variable can vary inthe process of measurement, because some error is present inpractically all measurements. The MDC is considered to be thechange in a measurement not due to measurement error, and,because measurement error is dependent on the level of func-tional performance, it is important for the clinician to under-stand MDC. For the clinician to be confident that a true changehas occurred during the patient’s ADIM (based on our experi-ment), MDCs of 0.14 mm and 0.20 mm must be consideredduring measurement of rest (minimum) and activation (maxi-mum), respectively. Muscle-thickness changes commonly ex-ceeded 1 mm during our task performances, therefore, observedchanges in TrA muscle activation should typically be expected tobe beyond the range of error. Applying the MDC data to apatient example can exemplify the value of considering trueclinical change. For example, a subject whose resting (mini-mum) TrA muscle thickness before a lifting task is measured at2.5 mm and during task performance with ADIM (maximum) ismeasured at 4.4 mm. Has a real change in TrA muscle thicknessoccurred? By using the 0.14-mm MDC95 for the resting mea-surement, the lower bound to be considered would be 2.64mm; likewise, the activation measurement upper bound wouldbe 4.20 mm, which results in a “true measurement” difference of1.56 mm. Therefore, according to the MDC95 that we reported,the clinician could be 95% confident that a real change occurredin TrA muscle thickness, which indicated TrA muscle activationduring ADIM.

Also, the SEM range of 0.02-0.30 mm in our study isconsistent with previous error measurements from studieswhen using both similar [3] (B-mode) and different [16,34](M-mode) scanning techniques that imaged participants insupine [3], sitting [16], and standing [34] positions.

A recent study pertinent to our research question exam-ined the ability of asymptomatic adults (N � 16) to producean increase in TrA muscle during 2 tasks and under 2instructional conditions (with ADIM and without ADIM)[39]. The tasks consisted of (1) unloaded neutral standingand (2) loaded forward reach with a 4.5-kg stick. Thus, 4conditions were examined. Intrarater reliability was obtainedin 11 participants for the uncontracted conditions only dur-ing the main data collection and again 4 days later. ICCs (3,1)without ADIM were 0.82 for unloaded neutral standing and0.95 for loaded forward reach. Post hoc paired t-tests indi-cated instructions (with and without ADIM), but not tasks,

produced significantly higher levels of muscle thickness. This

(cbtnceat

704 Watson et al ULTRASOUND OF TRANSVERSUS ABDOMINIS IN ASYMPTOMATIC INDIVIDUALS

study observed functional tasks similar to those in our ownstudy (our task 1 [unloaded neutral standing] and task 3[standing extended reach with object]) and found similarlevels of intrarater reliability.

According to Costa et al [36], measures of thickness changeswhich measure muscle activity by determining the degree ofhange in thickness between resting and contracted states) are toe considered the most important measures because they reflecthe measures used in clinical practice. However, it should beoted that, whereas measures of thickness changes may be morelinically meaningful, they do incorporate the measurementrror of both the minimum and maximum thickness measuresnd, therefore, are likely to have lower estimates of reliabilityhan those with single measures [18]. To date, to our knowl-

edge, the reliability of the percentage of change in TrA musclethickness has only been examined with the participant in thesupine position. This study is the first to examine reproducibil-ity of all 3 measures (minimum, maximum, and percentage ofchange), whereas all other studies that examined reproducibilityissues have examined muscle thickness only during upright,functional tasks.

Our findings are consistent with previous studies thatexamined TrA muscle activation in healthy, asymptomaticindividuals, which indicates that use of USI measures issufficiently reliable for clinical and research applicationsregarding lumbar stabilization exercises combined with func-tional activities [3,16,20,27,34,37,38].

Limitations

Although both clinicians were familiar with USI, the studyprocedure was new to rater 2, and additional training timemight have afforded better results. Because USI is highlyoperator dependent [22], we chose to use one clinician (rater1) to image participants. Finally, reproducibility findings aregeared toward healthy adults for injury prevention; thus, ourfindings can be generalized only to this population.

Recommendations for Future Research

Injury prevention training in this area will continue to grow,along with advances in the quality of USI equipment in termsof portability, cost, user software, and quality of images. Forexample, USI software is now providing higher-quality reso-lutions of images of TrA muscle activation in real time andeasier retrieval of past performances. These images provide awide array of various forms of feedback (or knowledge ofperformance) that may expedite motor learning and facilitateclinical outcomes. As with any new technology or trainingprotocols, measurement issues will always emerge that mustbe resolved to ensure validity of findings. To date the meth-ods used to examine lateral abdominal wall muscle activationhave been very mixed [36], so it would be helpful to establish

scoring procedures so that comparisons across studies are

possible. Future research of reproducibility of lateral abdom-inal wall activation by using USI should determine whetherthere are similar, acceptable levels of reliability when per-forming solely dynamic, challenging tasks.

We expect that injury prevention programs that use USI asfeedback and measurement will differ from rehabilitationprograms designed for symptomatic adults. For example,asymptomatic adults will be in less pain or risk of injury thantheir LBP counterparts. Thus, we anticipate that asymptom-atic adults will be able to withstand longer periods of practicewith minimal rest and to perform more challenging tasks.Also, we expect that these individuals will be highly moti-vated to learn how to activate their TrA muscle during tasks,because TrA muscle feedback via USI will be new to them.Importantly, motor learning principles and paradigms will bevital in designing and testing of injury prevention programsdesigned for healthy adults.

CONCLUSION

The findings of our study increase the ensurance of produc-ing acceptable reliability of USI measures of TrA musclethickness and thickness changes taken during loaded andfunctional activities. Reliable USI measures of TrA musclethickness changes meet the ongoing need for clinicians’knowledge of proper and sufficiently adequate muscle acti-vation in the clinical setting to assist progression of lumbarstabilization exercises.

REFERENCES1. Anderson Worth SG, Henry SM, Bunn JY. Real-time ultrasound feed-

back and abdominal hollowing exercises for people with low back pain.NZ J Physiother 2007;35:4-11.

2. Henry SM, Westervelt KC. The use of real-time ultrasound feedback inteaching abdominal hollowing exercises to healthy subjects. J OrthopSports Phys Ther 2005;35:338-345.

3. Teyhen DS, Miltenberger CE, Deiters HM, et al. The use of ultrasoundimaging of the abdominal drawing-in maneuver in subjects with lowback pain. J Orthop Sports Phys Ther 2005;35:346-355.

4. Laine HR, Peltokallio P. Ultrasonographic possibilities and findings inmost common sports injuries. Ann Chir Gynaecol 1991;80:127-133.

5. Chhem RK, Kaplan PA, Dussault RG. Ultrasonography of the musculo-skeletal system. Radiol Clin North Am 1994;32:275-289.

6. Hides JA, Richardson CA, Jull GA. Magnetic resonance imaging andultrasonography of the lumbar multifidus muscle. Comparison of twodifferent modalities. Spine 1995;20:54-58.

7. Hodges PW, Pengel LH, Herbert RD, Gandevia SC. Measurement ofmuscle contraction with ultrasound imaging. Muscle Nerve 2003;27:682-692.

8. Mannion AF, Pulkovski N, Gubler D, et al. Muscle thickness changesduring abdominal hollowing: an assessment of between-day measure-ment error in controls and patients with chronic low back pain. EurSpine J 2008;17:494-501.

9. Richardson C, Hodges P, Hides J. Therapeutic Exercise for SpinalSegmental Stabilization in Low Back Pain: Scientific Basis and ClinicalApproach. 2nd ed. Edinburgh: Churchill Livingstone; 2004.

10. Richardson CA, Snijders CJ, Hides JA, Damen L, Pas MS, Storm J. Therelation between the transversus abdominis muscles, sacroiliac joint

mechanics, and low back pain. Spine 2002;27:399-405.

705PM&R Vol. 3, Iss. 8, 2011

11. Teyhen DS, Childs JD, Flynn TW. Rehabilitative ultrasound imaging:when is a picture necessary. J Orthop Sports Phys Ther 2007;37:579-580.

12. Van K, Hides JA, Richardson CA. The use of real-time ultrasoundimaging for biofeedback of lumbar multifidus muscle contraction inhealthy subjects. J Orthop Sports Phys Ther 2006;36:920-925.

13. Teyhen DS, Rieger JL, Westrick RB, Miller AC, Molloy JM, Childs JD.Changes in deep abdominal muscle thickness during common trunk-strengthening exercises using ultrasound imaging. J Orthop SportsPhys Ther 2008;38:596-605.

14. Hides JA, Richardson CA, Jull GA. Use of real-time ultrasound imagingfor feedback in rehabilitation. Man Ther 1998;3:125-131.

15. Hides J, Wilson S, Stanton W, et al. An MRI investigation into thefunction of the transversus abdominis muscle during “drawing-in” ofthe abdominal wall. Spine 2006;31:E175-E178.

16. Kidd AW, Magee S, Richardson CA. Reliability of real-time ultrasoundfor the assessment of transversus abdominis function. J Gravit Physiol2002;9:131-132.

17. Kiesel KB, Uhl T, Underwood FB, Nitz AJ. Rehabilitative ultrasoundmeasurement of select trunk muscle activation during induced pain.Man Ther 2008;13:132-138.

18. Koppenhaver SL, Hebert JJ, Fritz JM, Parent EC, Teyhen DS, Magel JS.Reliability of rehabilitative ultrasound imaging of the transversus ab-dominis and lumbar multifidus muscles. Arch Phys Med Rehabil 2009;90:87-94.

19. McMeeken JM, Beith ID, Newham DJ, Milligan P, Critchley DJ. Therelationship between EMG and change in thickness of transversusabdominis. Clin Biomech 2004;19:337-342.

20. Hides JA, Miokovic T, Belavy DL, Stanton WR, Richardson CA. Ultra-sound imaging assessment of abdominal muscle function during draw-ing-in of the abdominal wall: an intrarater reliability study. J OrthopSports Phys Ther 2007;37:480-486.

21. Hides JA, Stanton WR, McMahon S, Sims K, Richardson CA. Effect ofstabilization training on multifidus muscle cross-sectional area amongyoung elite cricketers with low back pain. J Orthop Sports Phys Ther2008;38:101-108.

22. Hides JA, Wong I, Wilson SJ, Belavy DL, Richardson CA. Assessment ofabdominal muscle function during a simulated unilateral weight-bear-ing task using ultrasound imaging. J Orthop Sports Phys Ther 2007;37:467-471.

23. Endleman I, Critchley DJ. Transversus abdominis and obliquus inter-nus activity during pilates exercises: measurement with ultrasoundscanning. Arch Phys Med Rehabil 2008;89:2205-2212.

24. Gill NW, Teyhen DS, Lee IE. Improved contraction of the transversusabdominis immediately following spinal manipulation: a case studyusing real-time ultrasound imaging. Man Ther 2007;12:280-285.

25. O’Sullivan P, Twomey L, Allison G, Sinclair J, Miller K. Altered patternsof abdominal muscle activation in patients with chronic low back pain.Aust J Physiother 1997;43:91-98.

26. Raney NH, Teyhen DS, Childs JD. Observed changes in lateral abdom-inal muscle thickness after spinal manipulation: a case series using

CME QuestionIn this study, rater reliability for ultrasound imaging was established

a. subjects with low back pain.b. supine position.c. highly experienced examiners.d. static and dynamic functional positions.

Answer online at me.aapmr.org

rehabilitative ultrasound imaging. J Orthop Sports Phys Ther2007;37:472-479.

27. Rankin G, Stokes M, Newham DJ. Abdominal muscle size and symme-try in normal subjects. Muscle Nerve 2006;34:320-326.

28. Whittaker JL. Ultrasound imaging of the lateral abdominal wall mus-cles in individuals with lumbopelvic pain and signs of concurrenthypocapnia. Man Ther 2008;13:404-410.

29. Critchley D. Instructing pelvic floor contraction facilitates transversusabdominis thickness increase during low-abdominal hollowing. Phys-iother Res Int 2002;7:65-75.

30. John EK, Beith ID. Can activity within the external abdominal obliquebe measured using real-time ultrasound imaging? Clin Biomech 2007;22:972-979.

31. Roddey T, Brizzolara KJ, Cook KF. A comparison of two methods ofassessing transverse abdominal muscle thickness in participants usingreal-time ultrasound in a clinical setting. Orthop Phys Ther Pract2007;19:198-201.

32. Springer BA, Mielcarek BJ, Nesfield TK, Teyhen DS. Relationshipsamong lateral abdominal muscles, gender, body mass index, and handdominance. J Orthop Sports Phys Ther 2006;36:289-297.

33. Ainscough-Potts AM, Morrissey MC, Critchley D. The response of thetransverse abdominis and internal oblique muscles to different pos-tures. Man Ther 2006;11:54-60.

34. Bunce SM, Moore AP, Hough AD. M-mode ultrasound: a reliablemeasure of transversus abdominis thickness? Clin Biomech 2002;17:315-317.

35. Misuri G, Colagrande S, Gorini M, et al. In vivo ultrasound assessmentof respiratory function of abdominal muscles in normal subjects. EurRespir J 1997;10:2861-2867.

36. Costa LO, Maher CG, Latimer J, Smeets RJ. Reproducibility of rehabil-itative ultrasound imaging for the measurement of abdominal muscleactivity: a systematic review. Phys Ther 2009;89:756-769.

37. Bunce SM, Hough AD, Moore AP. Measurement of abdominal musclethickness using M-mode ultrasound imaging during functional activi-ties. Man Ther 2004;9:41-44.

38. Reeve A, Dilley A. Effects of posture on the thickness of transversusabdominis in pain-free subjects. Man Ther 2009;14:679-684.

39. McGalliard MK, Dedrick GS, Brismee JM, Cook CE, Apte GG, Sizer PSJr. Changes in transversus abdominis thickness with use of the abdom-inal drawing-in maneuver during a functional task. PM R 2010;2:187-194.

This CME activity is designated for 1.0 AMA PRA Category 1 Credit™ andcan be completed online at me.aapmr.org. Log on to www.me.aapmr.org,go to Lifelong Learning (CME) and select Journal-based CME from thedrop down menu. This activity is FREE to AAPM&R members and $25 fornon-members.

for: