A Case-Control Study of Biomechanical and Psychosocial Risk Factors for Occupational Low-Back Pain

1
A Case-Control Study of Biomechanical and
Psychosocial Risk Factors for Occupational
Low-Back Pain
PREMUS 2001, Amsterdam

• Michael S. Kerr, PhD
• Institute for Work Health
• Toronto, Ontario, CANADA

Kerr et al. (2001) Am J Pub Health 911069-1075
2
Study Collaborators
Institute for Work Health Kerr MS, Frank JW,
Shannon HS, Bombardier C University of
Waterloo Norman RW, Wells RW, Neumann P
General Motors Mr. Elmer Beddome CAW Mr. John
Graham
The Ontario Universities Back Pain Study (OUBPS)
Group
Andrews D, Beaton DE, Dobbyn M, Edmonstone E,
Ferrier S, Hogg-Johnson S, Ingelman P, Mondlock
M, Peloso P, Smith J, Stanfield SA, Tarasuk V,
Woo H
3
Why a case-control study?

• Uncertainty about role of physical demands
  necessitated concentration of effort on
  developing valid workplace measures
• Cohort model required multiple measures at
  multiple times to assess valid exposure

Does a prospective cohort study with limited
(e.g. once only) exposure assessment really
provide more rigorous evidence than a
case-control study with more detailed assessments?
4
Research Question
After controlling for individual characteristics,
what are the main work-related biomechanical and
psychosocial risk factors for reported low-back
pain?
5
STUDY SETTING
GM-Oshawa autoplex
65 km east of Toronto Total hourly-paid work
force approx. 12,000 3 Divisions Car
Assembly (7,000) Truck Assembly (3,000) Fabr
ications N/A
Nursing stations handle ALL WCB reports as well
as most other health problems occurring at
work-site Truck 1 station Car 4 stations
6
The GM Study of Low-back Pain
Incidence density sampling
workers without LBP
10,000 car and truck plant workers (hourly-paid)
STUDY POPULATION

random selection (job-matched)
(n137)
CASES
accrued via workplace health stations
workers
with
LBP
CONTROLS
(n179)
baseline
conducted at home
questionnaire and

(n265)
physical exam
at work doing usual job
video, EMG, checklists, posture
physical demands
assessment

7
SUBJECT ENROLMENT
179
A total of 381 subjects enrolled in the study
Random controls
(Voluntary)
Cases
65
Job-matched controls (used for proxy data)
137
8
CASE Definition

• Full-time, hourly-paid worker with
  sprain-strain LBP (reported to nursing
  stations)
• No lost-time or WCB claim requirement
• No previous worksite LBP report (90 days)

9
Were cases and controls comparable except for LBP?
sd from male reference population
SF-36, Health-related quality of life
Physical Function
Role Physical
Bodily Pain
Social Function
Vitality
Role Emotional
Mental Health
General Health
Ref Garrat et al. (1993) BMJ 3061440-4
10
STUDY VARIABLES - 1
INDIVIDUAL Characteristics
Age, height, weight, body mass index, sex
education, marital status, preschool children,
main wage earner, non-occupational physical
activity, smoking, alcohol consumption
11
STUDY VARIABLES - 2
BIOMECHANICAL Factors
(reduced a priori to about 20 key variables, NO
EMG)

• Peak Forces
• compression shear hand force
• Cumulative Forces
• average and integrated compression
• Low-level (static) Forces
• compression
• Posture and Movement
• peak flexion time non-neutral trunk kinematics

12
STUDY VARIABLES - 3
PSYCHOPHYSICAL Factors
self-rated physical exertion measures distinct
from "objective" measurements of physical demands
analyses, but may also have additional role to
play over and above the measured demands e.g.
tolerance levels, "job stress", etc.
13
STUDY VARIABLES - 4
PSYCHOSOCIAL Factors

• Karasek-Theorell Job Content Instrument
• "psychological demand", decision latitude,
  supervisor support, coworker support, workplace
  social environment, job self-identity
• - job dissatisfaction, mastery, empowerment,
  status inconsistency ("over-education")

14
Summary of Key Risk Factors
(after adjusting for individual characteristics)
15
Study Strengths

• Directly measured physical demands data combined
  with (basic) physical exam and interview-assisted
  psychosocial data for individual subjects
• Comprehensive workplace job demands assessments
  (generalizable i.e. not specific to automobile
  manufacturing)
• Well defined study base for subject selection

16
Study Weaknesses

• Modest participation rate (approx 60)
• Relied on workplace reporting mechanism to
  identify cases
• Self-report only for psychosocial factors
• Case-control design
• But does a prospective cohort study with
  limited once only exposure assessment really
  provide more rigorous evidence than a
  case-control study with more detailed assessments?

17
How did we control for Bias?

• Directly measured physical demands data rather
  than self-report
• Job-matched controls used to examine potential
  recall bias (none observed)
• Used newly incident cases rather than prevalent
  cases
• Compared cases with non-participating
  compensation claimants (no differences)
• A priori reduction in variables eligible for
  regression modeling (multi-methods)

18
Possible Biological Mechanisms
19
Main Conclusions

• Consistency and strength indicates biomechanical
  load increases LBP risk
• Psychosocial factors also shown to be associated
  with reporting LBP
• Job dissatisfaction not a risk factor for
  reporting LBP in this setting
• Psychosocial instruments (especially Karaseks
  demands scale) may require further refinement for
  MSK studies

20
For additional information please contact
Mickey Kerr The Institute for Work Health 481
University Ave., Suite 800 Toronto, ON M5G
2E9 Phone [phone] Fax [phone] Webs
ite http//www.iwh.on.ca E-mail info_at_iwh.on.ca
The Institute for Work Health operates with the
support of the Ontario Workplace Safety
Insurance Board
Kerr et al. (2001) Am J Pub Health 911069-1075