Ergonomic risk assessments – a need for reliable
and attractive methods Mikael Forsman
[email protected] Karolinska Institutet
Stockholm, Sweden
Co-workers Reliability in ergonomic risk assessments by observation https://wearnotch.com/ Ida-Märta Rhén, Kristina Eliasson, Katarina Kjellberg, Xuelong Fan (KI) Peter Palm, Teresia Nyman (Uppsala) Per Lindberg (Gävle)
Practical objective measurements for sedentary time and body postures Liyun Yang, Carl Lind (Royal Institute of Technology)
Camilla Dahlqvist, Gert-Åke Hansson (Lund) Jens Wahlström (Umeå)
Mikael Forsman
2016-08-16
2
Mikael Forsman
2016-08-16
Introduction Work-related musculoskeletal disorders (MSDs) are still frequent. In order to decrease the disorders, risks need to be identified and quantified. Takala and co-workers, 2010, evaluated 30 observational methods, and found a need for further reliability testings. New electronic components facilitates uncomplicated measurements of biomechanical exposure.
Mikael Forsman
2016-08-16
4
Observation Reliability Project 12 experienced ergonomists were given: 10 video-recorded (2-5 minutes) work tasks supermarket work meat cutting and packing engine assembly cleaning post sorting hairdressing. data of: the work task length, pause- and restsschedules, weights of handled goods, and the employees’ ratings of discomfort, work demands and control, were given for each task. max time per work task assessment was 20 minutes. The assessments were repeated about 2 months. Mikael Forsman
2016-08-16
5
Observation Reliability Project They used 6 observation methods:
1. Occupational Repetitive Actions checklist (OCRA) 2. Quick Exposure Checklist (QEC) 3. Strain Index (SI)
4. Assessment of Repetitive Tasks (ART) 5. Hand Arm Risk-assessment Method (HARM) 6. Repetitive work model by the Swedish Work Environment Authority
Observation Reliability Project They used 6 +1 observation methods: 0. Own experience – no method
1. Occupational Repetitive Actions checklist (OCRA) 2. Quick Exposure Checklist (QEC) 3. Strain Index (SI)
4. Assessment of Repetitive Tasks (ART) 5. Hand Arm Risk-assessment Method (HARM) 6. Repetitive work model by the Swedish Work Environment Authority
Mikael Forsman
2016-08-16
7
Observation Method Example - Quick Exposure Check (QEC)
Mikael Forsman
16 augusti 2016
8
Mikael Forsman
2016-08-16
9
Katarina Kjellberg, Mikael Forsman
2016-08-16
10
Mikael Forsman
16 augusti 2016
11
Method 0 – own assessment 21 ergonomists The risk was rated into: green (no risk) yellow (investigate further) red (immediate risk) Ratings were done for 8 specific body regions neck right and left shoulders right and left arms/elbows right and left wrists/hands lower back and for the over-all risk.
Left Left elbow Left wrist shoulder Unpack groceries to shelves
Right shoulder
Neck
High risk Moderate risk Low risk
2 9 10
2 12 7
2 10 9
3 14 4
Right elbow
Right wrist Low back
2 9 10
2 8 11
4 10 7
Total risk
0 9 12
3 11 7
Put nets around roasts High risk Moderate risk Low risk
1
17
12
11
2
8
6
6
17
16
4
8
8
11
8
11
14
4
4
0
1
2
8
5
4
1
0
20 Ratings
Throw small boxes in containers High risk Moderate risk Low risk
1
2
2
4
7
7
7
0
5
9
11
13
14
13
10
12
12
15
11
8
6
3
1
4
2
9
1
Put packs of letters into boxes High risk Moderate risk Low risk
1
0
2
2
3
6
6
0
6
6
5
8
7
11
11
13
9
12
14
16
11
12
7
4
2
12
3
Debone meat High risk Moderate risk Low risk
8
5
7
6
6
14
17
2
15
9
6
10
12
8
6
4
10
5
4
10
4
3
7
1
0
9
1
0
0
0
0
0
0
0
0
0
1
3
9
7
4
9
7
1
3
20
18
12
14
17
12
14
20
18
Engine assembly High risk Moderate risk Low risk
Hair cutting High risk Moderate risk Low risk
4
6
1
0
11
6
11
2
7
16
13
6
8
10
12
7
9
12
1
2
14
13
0
3
3
10
2
0
0
0
0
5
7
6
10
5
5
4
6
7
12
11
10
11
15
16
17
15
14
4
3
5
0
1
Toilet cleaning High risk Moderate risk Low risk
Supermarket cashier work High risk Moderate risk Low risk
2
5
3
7
11
3
5
1
14
16
12
10
10
8
10
12
13
7
3
4
8
4
2
8
4
7
0
Cleaning stairs Red Yellow Green
2
1
4
7
3
6
3
1
4
16
9
10
9
15
13
14
4
14
3
11
7
5
3
2
4
16
3
OCRA over-all risk level Proportional agreement between ergonomists (39%) Linearly weighted kappa = 0.41 Video-clip Grocery work
Meat netting
Post sorting
Post handling
Meat Engine deboning assembly
Hairdressing
Toilet cleaning
Cashier work
Stair cleaning
1 2
Ergonomist
3 4
5 6
Risk level
7
5 4 3 2 1
8 9
10 11
Ida-Märta Rhén, Mikael Forsman IEA 2015
2016-08-16
14
Test-retest over-all OCRA risk level Proportional agreement within ergonomists = 46% Linearly weighted kappa = 0.51 Video-clip Grocery work
Meat netting
Post sorting
Post handling
Meat Engine deboning assembly
Hairdressing
Toilet cleaning
Cashier work
Stair cleaning
1 2
Ergonomist
3 4 5 6 7 8 9 10 = Test-retest agreement Ida-Märta Rhén, Mikael Forsman IEA 2015
2016-08-16
15
Three experts used the methods – fully according to the manuals Consensus results Method
Bedömningsmetoder Uppackning Näta kött Kasta post Fylla post SWEA AFS ART ART HARM HARM OCRA OCRA SI SI QECupper övre extremitet QEC extremity QECgeneral helkropp QEC
Bena ur
Motor
Hår
Toalett
Kassa
Trappstäd
6
Comparison with a similar study
High risk
5
Moderate risk 4 This study:
Low risk
10 9
3
8 7 6
2
High risk
5
Moderate risk 4
Low risk
1
3 2 1
0
0
SWEA
SWEA
QEC general
QEC general
ART
ART
HARM
HARM
OCRA
OCRA
QEC upper extremity
QEC upper extremity
SI
SI
QEC hand/wrist
QEC hand/wrist
Chiasson M-È, Imbeau D, Aubry K, Delisle A. Comparing the results of eight methods used to evaluate risk factors associated with musculoskeletetal disorders. International Journal of Industrial Ergonomics. 2012;42(5):478-88:
Katarina Kjellberg, Mikael Forsman
2016-08-16
17
Statistics for the inter-observer reliability The Proportional agreement, Pr(a), of the different ergonomist’s ratings (in percent) Cohen’s pairwise kappa was computed for all pairs
Pr(a) represents the actual observed proportional agreement Pr(e) represents agreement expected by chance. Linearly weighted kappa, κw, was computed and averaged in the way suggested by Davies and Fleiss (1982). Davies M, Fleiss JL. 1982. Measuring Agreement for Multinomial Data. Biometrics 38, 1047-1051.
Results
Method QEC
Reliability
Linearly Weighted Kappa
Inter-rater Intra-rater 0.49 0.60 0.44 0.58 0.67 0.76 0.87 0.88 0.55 0.68 0.41 0.61
Validity 0.44 0.44 0.79 0.91 0.47 0.36
Total
0.65
0.70
0.62
Left Hand Right Hand The Highest Risk Hand
0.39 0.48 0.47
0.49 0.51 0.55
0.41 0.41 0.40
Left Side Right Side
0.58 0.65
0.65 0.68
0.60 0.49
OCRA
Total
0.37
0.50
0.43
SWEA
Tot Risk Posture and Movements Sitting / Standing Neck Back Arm/Shoulder Leg
0.21 1.00 0.22 0.16 0.21 0.05
0.30 1.00 0.32 0.44 0.13 0.36
0.38
0.27 0.18 0.32 0.12 0.15 0.16 0.18 0.26 0.31
0.35 0.30 0.38 0.23 0.40 0.20 0.28 0.62 0.41
HARM SI
ART
Landis & Koch (1977): < 0.20 – poor 0.21 – 0.40 – fair 0.41 – 0.60 – moderate 0.61 – 0.80 – substantial 0.81 – 1.00 – almost perfect
Risk level for Back Shoulder Wrist Neck Total (with 4 levels) Total (with 3 levels)
Own assessment Neck Right Shoulder Left Shoulder Left Arm/elbow Right Arm/elbow Right Hand/wrist Left Hand/wrist Lower Back Total Risk
HARM Postures Neck-shoulder
HARM Reliability Postures Neck-shoulder Time in demanding postures: green yellow red 0–10 10–50 >50%
Inter-rater w = 0.18
κ
Intra-rater w = 0.21
κ
Landis & Koch (1977): < 0.20 – poor 0.21 – 0.40 – fair 0.41 – 0.60 – moderate 0.61 – 0.80 – substantial 0.81 – 1.00 – almost perfect
HARM Reliability Postures Arm-wrist Time in demanding postures: green yellow red 0–10 10–50 >50%
Inter-rater w = 0.14
κ
Intra-rater w = 0.25
κ
Landis & Koch (1977): < 0.20 – poor 0.21 – 0.40 – fair 0.41 – 0.60 – moderate 0.61 – 0.80 – substantial 0.81 – 1.00 – almost perfect
Observation Reliability Project 12 experienced ergonomists were given: 10 video-recorded (2-5 minutes) work tasks supermarket work meat cutting and packing engine assembly cleaning post sorting hairdressing. data of: the work task length, pause- and restsschedules, weights of handled goods, and the employees’ ratings of discomfort, work demands and control, were given for each task.
Mikael Forsman
2016-08-16
23
HARM Total risk score
HARM Total risk index (three risk levels)
Inter-rater Reliability Repetition, Movements and Postures Method
κw
low
high
QEC
0.17 - 0.44
Hand/wrist posture
Hand/wrist repetition
HARM
0.14 - 0.3
Hand/arm posture
Force duration/repetition
SI
0.04 - 0.40
Hand posture
Force repetition
ART
0.17 - 0.45
Hand/wrist posture
Hand/arm repetition
OCRA
0.03 - 0.53
Elbow posture
Repetition
SWEA
0.05-0.22
Leg posture/movements
Neck posture/movements
Left elbow
Neck
Own assessment 0.12 - 0.27 Landis & Koch (1977): < 0.20 – poor 0.21 – 0.40 – fair 0.41 – 0.60 – moderate 0.61 – 0.80 – substantial 0.81 – 1.00 – almost perfect
Reliability – other studies in comparison with Method 0 Statistics presented with Cohen’s kappa (κ), Kendall’s coefficient of concordance (KCC), Intra-class correlation (ICC) and Weighted kappa (κw) . Method
Inter-observer reliability
Method 0 own assessment
Reference
Ice-dance US championship 2015 Short program 9 judges
17 pairs Scores: 14.8 – 56.3
Inter-rater w = 0.83
κ
Madison Chock Evan Bates 2015 U.S. Champions
A DREAM!
Mikael Forsman
2016-08-16
28
A DREAM! It shall be really easy to make biomechanical risk assessments! And cheap! For high and low activity jobs!
With safe job limit values! With feedback to individual workers! Everyone can use the methods and understand the results!
Mikael Forsman
2016-08-16
29
the Smart textiles project
Long Sleeves 3-5 IMUs ECG Bioimpedance Bluetooth SmartPhone Data Collection
A DREAM!
Mikael Forsman
2016-08-16
31
the Smart textiles project The glove
Katarina Kjellberg, Mikael Forsman
2016-08-16
32
the Smart textiles project The partners
Katarina Kjellberg, Mikael Forsman
2016-08-16
33
Katarina Kjellberg, Mikael Forsman
2016-08-16
34
Mikael Forsman
Manual…
There is also a simple Excel application
alysprogram i Excel
psdel som ill höger). n så att
klinometer tt kort 2 tartas 1 sekund, Figur 5. Inklinometern blinka. som används i Excelapplikationen, fasttejpad använder för att kännas av nollvinkeln) på armen. iduret start och stopp per arbetsuppgift.
Accuracy of a posture measurement system for practitioners
Carl LIND Unit of Ergonomics, KTH, Royal Institute of Technology, Huddinge, SWEDEN
Mikael FORSMAN Institute of Environmental Medicine, Karolinska Institutet, Stockholm, SWEDEN
Results Experiment 1A
Experiment 1B
R2 = 0.9975
R2 = 0.9988
NES 2015 - Carl LIND & Mikael FORSMAN
David Dunstan IEA Key-note
Apps for measurements of sedentary times
Mikael Forsman
2016-08-16
54
Conclusion
ErgoSitMeter
Mikael Forsman
2016-08-16
55
Take Home Messages Low reliability in observational risk assessments of manual work. Also differences between different methods. Practical inexpensive inclinometer and applications in Excel and for IPhones exist that result in data of the quality comparable to that of research methods. It should be possible to combine observations with “easy measurements”. We need more research on criteria for acceptable ranges – of different parameters. Attractive! It is hard to attract ergonomists to start measure instead of doing observations.