Balancing Efficiencies & Tradeoffs: Evaluating EMG Exposure Assessment for Low Back Injury Risk Factors in Heavy Industry Catherine Trask 2008.

Balancing Efficiencies & Tradeoffs:Balancing Efficiencies & Tradeoffs:

Evaluating EMG Exposure Evaluating EMG Exposure Assessment for Low Back Injury Risk Assessment for Low Back Injury Risk

Factors in Heavy IndustryFactors in Heavy Industry

Catherine Trask 2008Catherine Trask 2008

Back injury is a prevalent and expensive problem, particularly in heavy industry

‘‘Solving’ Back InjurySolving’ Back Injury

How should exposure be measured?

For what duration? Who should be

measured? How many times should

they be measured?

Thesis ObjectivesThesis Objectives

Thesis ChaptersThesis Chapters

Chapters 4 and 5

Chapter 6 Chapter 7 Chapter 7

How should exposure be measured?

For what duration? Who should be

measured? How many times should

they be measured?

Thesis ChaptersThesis Chapters

Chapters 1

Chapter 2 and 3 Chapters 4 and 5

Chapter 6 Chapter 7 Chapter 7

Introduction to exposure assessment

Introduction to methods How should exposure be

measured? For what duration? Who should be

measured? How many times should

they be measured?

Introduction to Exposure Introduction to Exposure AssessmentAssessment

Available Exposure Available Exposure Assessment MethodsAssessment Methods

Direct Measure

using electronic devices

Observation

by trained experts

Self-report

by the workers

Continuum of MethodsContinuum of Methods

High-resolution – lots of detail

Objective

Expensive

Few people

Short time

Wider scope – ‘big picture’

Subjective

Inexpensive

More people

Longer time

Direct Measure Observation Self-report

Data CollectionData Collection

Worker RecruitmentWorker Recruitment

Contacted workers in heavy industry with accepted back injury claims

Contacted employer to gain access to the worksite

Recruited co-workers at each worksite

126 individuals Repeated measures 223 measurement days

The Measurement DayThe Measurement Day

Direct Measure

by electronic devices

Observation

by trained experts

Self-report

by the workers

Measured all methods concurrently Full shift

Back Back InjuryInjuryManual Materials

Handling

Risk Factors for Back Risk Factors for Back Injury:Injury:

Self-ReportSelf-Report

Working

Postures

Asked for the amount of time in each activity Used pictographs for most questions

Self-report

Back Back InjuryInjuryManual Materials

Handling

Risk Factors for Back Risk Factors for Back Injury:Injury:

ObservationObservation

Working

Postures

‘Snapshots’ of 15 variables at 1 minute intervals Full-shift, excluding breaks

Observation

Back Back InjuryInjury

Manual Materials Handling

Risk Factors for Back Injury:Risk Factors for Back Injury:Direct MeasurementDirect Measurement

Working

Postures

Inclinometer

Whole body

vibration

Seat pad accelerometer

Mean90th %

CumulativeRCM

EMGBack

muscle activity

Chapter 4: Measuring low back injury risk factors in challenging work environments:

an evaluation of cost and feasibility

A version of this chapter has been published. Trask, C., Teschke, K., Village, J., Chow, A version of this chapter has been published. Trask, C., Teschke, K., Village, J., Chow, Y., Johnson, P., Luong, N., and Koehoorn, M. (2007). Evaluating methods to measure Y., Johnson, P., Luong, N., and Koehoorn, M. (2007). Evaluating methods to measure low back injury risk factors in challenging work environments. American Journal of low back injury risk factors in challenging work environments. American Journal of Industrial Medicine 50(9):687-96.Industrial Medicine 50(9):687-96.

Cost and FeasibilityCost and Feasibility

Success rate = successful measurement/ attempted measurement

Cost ($CDN) per successful measurement

Measurement Success Measurement Success RatesRates

41.9%

61.9%

89.2%97.8% 99.6%

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

Vibration EMG Inclinometer Interview Observation

measurement method

su

ce

ss

ra

te (

%)

Measurement Costs Measurement Costs (per successful (per successful measurement)measurement)

$544.76

$320.91

$192.05

$21.70

$213.52

$0.00

$100.00

$200.00

$300.00

$400.00

$500.00

$600.00

Vibration EMG Inclinometer Interview Observation

measurement method

cost

per

mea

sure

($C

DN

)

ConclusionsConclusions

Inverse relationship between cost and feasibility

Industrial environments are demanding on mechanical equipment Cold, dusty, wet, explosive, Rough handling/vibration

Consider costs and feasibility when planning field work!

A version of this chapter has been submitted for publication. Trask, C., Teschke, K., A version of this chapter has been submitted for publication. Trask, C., Teschke, K., Morrison, J., Village, J., Johnson, P., Koehoorn, M. (2008) Predicting Exposure for Morrison, J., Village, J., Johnson, P., Koehoorn, M. (2008) Predicting Exposure for Mean, 90th Percentile, and Cumulative EMG Activity in Heavy Industry. Submitted Mean, 90th Percentile, and Cumulative EMG Activity in Heavy Industry. Submitted February 2008 to: Applied Ergonomics.February 2008 to: Applied Ergonomics.

Chapter 5: Predicting exposure for Chapter 5: Predicting exposure for mean, 90th percentile, and cumulative mean, 90th percentile, and cumulative

EMG activity in heavy industryEMG activity in heavy industry

Modeling determinants of Modeling determinants of exposure exposure

%RC = %RC = ββ1(observed variable 1) + 1(observed variable 1) + ββ2(observed 2(observed variable 2) + variable 2) + ββ3(observed variable 3)…3(observed variable 3)…

0

20

40

60

80

100

120

140

0 10 20 30 40 50 60 70

Low Back EMG

Observation or self report

Observation-based ModelObservation-based Model

Variable Mean EMG (in %RC)

β (slope) p

Intercept (average for all subjects) 19.8

Standing (% time) 0.115 <0.001*

Trunk position >60o (% time) 0.612 .0018*

4.5-10kg load in hands (% time) 0.910 <0.001*

10-20 kg load in hands (% time) 0.325 .0641

Self-report ModelSelf-report Model

Variable Mean EMG (in %RC)

Β (slope) p

Intercept (average of subjects) 33.4

Sitting (% time) -0.181 0.0023*

Industry

Construction industry 14.8 0.0054*

Forestry industry 13.3 0.0109*

Wood product industry 4.44 0.369

Warehousing industry 8.75 0.1024

Transportation industry 0 Reference

Model PerformanceModel Performance

variance explained,

47%

not explained,

53%

Self-report based modelSelf-report based model Observation based model Observation based model

variance explained,

36%not

explained, 64%

ConclusionConclusion

Is this enough to conduct injury research?Is this enough to conduct injury research? Chemical exposure studies often predict 30-60% Chemical exposure studies often predict 30-60% Many studies using self-report and observation have Many studies using self-report and observation have

found a relationship with back injury in the pastfound a relationship with back injury in the past Epidemiology often uses categorical exposure Epidemiology often uses categorical exposure

variables, not continuous variablesvariables, not continuous variables One can predict some of the variability in EMG by One can predict some of the variability in EMG by

asking a few questions or observing a few asking a few questions or observing a few exposuresexposures

Tradeoff is in measuring more individuals, more Tradeoff is in measuring more individuals, more timestimes

A version of this chapter has been accepted for publication. Trask, C., Koehoorn, M., Village, J., Johnson, P., Teschke, K. (2008) How long is long enough? Evaluating sampling durations for low-back EMG assessment. Journal of Occupational and Environmental Hygiene. Submission number: JOEH-07-0094.R1.

Chapter 6: How long is long enough? Chapter 6: How long is long enough? Selecting efficient sampling durations Selecting efficient sampling durations

for low-back EMG assessmentfor low-back EMG assessment

Sampling Duration RationaleSampling Duration Rationale

Direct measurements were made for a whole shift

Do you really need to measure a whole shift?

How much information is lost if you measure a portion of the shift?

Selecting sampling durationsSelecting sampling durations

Compared 7 different sampling durations of the same work shift:

Whole shift (5.5 to 7.5 hours)4 hours2 hours1 hour10 minute2 minute2 shifts

Re-sampled post hocRandomized start time

Sampling durationsSampling durations

Whole shift

4 hour

1 hour

2 hour

Red = left back musclesGreen = right back muscles

Absolute errorAbsolute error between sampling durations between sampling durations

1.64789991.0148922 1.4294942

2.43776282.94

1.4

3.03

5.55.33

1.8

4.7

9.3

7.14

2.4

6.2

14.8

10.3

4.2

9.1

20.9

13.2

4.7

11.4

25.2

0

5

10

15

20

25

30

mean 10 %ile 50 %ile 90%ile

ab

so

lute

err

or

in %

RC

full shift - 2 shifts full shift - 4 hours full shift - 2 hours

full shift - 1 hours full shift - 10min full shift - 2min

ConclusionConclusion

8% error for 4-hour and 14% error for 2-hour durations: reasonable estimates

1 hour or less produces very large errors

Balance cost with data precision and sample size Shorter duration but more

workers measured

A version of this chapter has been submitted for publication. Trask, C., Teschke, K., Morrison, J., Koehoorn, M. (2007) Optimizing Sampling Strategies: Components of Low-Back EMG Variability in Five Heavy Industries. Submitted February 2008 to: Occupational and Environmental Medicine. Submission number: OEM/2008/039826

Chapter 7: Optimizing sampling Chapter 7: Optimizing sampling strategies: components of low-back strategies: components of low-back

EMG variability in five heavy EMG variability in five heavy industriesindustries

How many individuals? How many repeats? (How) should we group

measurements?

Components of Variability Components of Variability RationaleRationale

Grouping schemes make for less attenuation of an exposure-response relationship

Attenuation can be estimated based on the exposure data, even when the response is not measured

Sample Exposure-Sample Exposure-ResponseResponse

Relationship Relationship Back injury outcome = intercept + Back injury outcome = intercept + ββ1(exposure variable 1)1(exposure variable 1)

0

20

40

60

80

100

120

140

0 10 20 30 40 50 60 70

Response

Exposure

Grouping SchemesGrouping Schemes

No grouping No grouping Job titleJob title CompanyCompany IndustryIndustry Post hoc ranking of Post hoc ranking of

industry/job title groupsindustry/job title groups

Percentage of true E-R Percentage of true E-R by grouping schemeby grouping scheme

0.876 0.895 0.906 0.9220.993

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

Grouping byCompany

No Grouping Grouping byJob

Grouping byIndustry

Post hocGrouping

att

en

ua

tio

n f

ac

tor

Workers per group (k) Workers per group (k) required for 95% of true required for 95% of true

E-RE-R

Grouping strategyNumber

of groups

25% repeats

50% repeats

100% repeats

Grouping by Industry 5 58.6 44.3 26.4

Grouping by Company 31 143.5 104.7 58.5

Grouping by Job 24 13 10.3 6.9

Post hoc Grouping 5 3.3 2.7 1.9

ConclusionConclusion

The post hoc grouping scheme was the most efficient grouping scheme Lowest estimated attenuation Lowest number of measurements required

Measurement and recruitment challenges mean one should aim for a larger number of measurements

Attenuation isn't everything when selecting a sampling strategy – want to choose sample size to be robust

SummarySummary

There are always tradeoffs in exposure assessmentThere are always tradeoffs in exposure assessment

Lots of decisions to make!Lots of decisions to make! How you ‘tip the scales’ toward more samples How you ‘tip the scales’ toward more samples

or more precision depends on the purpose of or more precision depends on the purpose of the study and the characteristics of the the study and the characteristics of the populationpopulation

Contribution is in the ways of framing these Contribution is in the ways of framing these questions and starting to quantify the answersquestions and starting to quantify the answers

AcknowledgementsAcknowledgements

Participating Workers and WorksitesParticipating Workers and Worksites

WorkSafe BCWorkSafe BC

Michael Smith Foundation for Health ResearchMichael Smith Foundation for Health Research

CIHR Bridge Fellowship ProgramCIHR Bridge Fellowship Program

Mieke Koehoorn Mieke Koehoorn

Kay TeschkeKay Teschke

Jim MorrisonJim Morrison

Kevin HongKevin Hong

Nancy LuongNancy Luong

Melissa KnottMelissa Knott

James CooperJames Cooper

Judy VillageJudy Village

Pete JohnsonPete Johnson

Jim PlogerJim Ploger

Yat ChowYat Chow

QuestionsQuestions