The relation of musculoskeletal disorders to ergonomic work demands in welders and electrical workers: a prospective Canadian cohort study

Abstract

Introduction

Musculoskeletal disorders are known to result from physical demands at the workplace. Identification of risks specific to particular trades may encourage work modification to prevent new onset conditions.

Methods

In a Canadian cohort study of male and female welders and electrical workers, we collected self-reports of low-back pain, shoulder pain, and symptoms suggestive of vibration white finger (VWF) at each 6-monthly contact for up to 5 yr. Physician records of back and shoulder/joint disorders and Raynaud’s syndrome were extracted from the Alberta administrative health database (AHDB). At each contact, participants completed questions on ergonomic work factors. We computed cumulative hours exposed for 6 factors: whole-body vibration, hand-arm vibration, manipulating heavy objects, working at or above shoulder height, working while crouching or kneeling, and work in cold environments. The relation of current and log cumulative exposures to health outcomes was identified by proportional hazards regression, adjusting for sex, age, body mass index, smoking, anxiety, and depression.

Results

Of 1,885 workers recruited January 2011 to September 2017, 872 welders and 812 electrical workers recorded symptoms and workplace demands on at least one occasion, with 693 welders and 567 electrical workers matched to the AHDB. In final models, whole-body vibration was related overall to each self-reported health outcome with backpain risk most in evidence for male welders (HR = 1.10 log increase/h of exposure: 95% CI, 1.05 to 1.15, P < 0.001). Working in a crouching position and in cold temperatures also contributed to back pain in welders. Cumulative hours working at or above shoulder height increased welders’ risk of shoulder pain (HR = 1.07 log increase/h of exposure: 95% CI, 1.03 to 1.11, P = 0.001). Working at or above shoulder height was related to both back and shoulder pain in the electrical trades, where cumulative exposure to hand-arm vibration was an additional factor for shoulder pain (HR = 1.06 log increase/h of exposure: 95% CI, 1.01 to 1.10, P = 0.007). Manipulating heavy loads was a further risk factor for back and shoulder pain for women in electrical work. There were only 3 incident cases of Raynaud’s syndrome in physician records: symptoms suggestive of VWF related strongly to work in cold environments but not to hand-arm vibration. Physician records of back pain did not reflect specific workplace demands, but physician records of shoulder/joint conditions were greater, overall, in those with longer exposure to whole-body vibration and to current hand-arm vibration in electrical workers.

Conclusion

Vibration, a well-recognized hazard, was a risk factor for all outcomes but with whole-body vibration more evidently a risk for welders and hand-arm vibration for electrical workers. The final models of risk factors differed importantly between the two trades, suggesting specific targets for intervention.

What’s Important About This Paper?

Information is needed about musculoskeletal risks specific to clearly defined work groups such that interventions can be well-targeted. This paper develops risk profiles for welders and electrical workers that differ importantly between the trades. Risks differed little for women and men within trades, confirming that interventions indicated, here particularly for vibration, should be determined by features of the workplace rather than the worker.

Introduction

At least 30% of workplace injuries are musculoskeletal disorders (MSDs; WorkSafe Australia 2019; US Bureau of Labor Statistics 2020; Health and Safety Executive UK 2022) with high costs to the employer (Bevan 2015; Walker-Bone et al. 2022), compensation systems (WorkSafe Australia 2019), and the health of the worker (March et al. 2014). There is a vast literature on the prevalence and causes of MSDs and on interventions, with limited success, to reduce work-related MSDs and to return injured workers to employment. The design of interventions to prevent injury requires a clear understanding of risk factors that impact a specific workforce. The epidemiological evidence for a causal relation of MSDs to work demands was extensively and critically reviewed by the US National Institute for Occupational Health (NIOSH 1997) which concluded that there was strong evidence for a relation of low-back pain to workplace lifting and whole-body vibration and for vascular symptoms of vibration white finger (VWF) with high levels of hand-arm vibration. Evidence for ergonomic demands and shoulder disorders was judged to be less conclusive, but recent well-designed studies (Dalbøge et al. 2020; Sun et al. 2024) have strengthened the evidence base for this outcome. Where causality is supported, MSDs are preventable. Demonstration of increased risk with clearly identified exposures, such as VWF in forestry workers (Bovenzi et al. 1995) or low-back pain in baggage handlers (Brauer et al. 2020), may be more effective than generic guidelines in focusing attention on specific interventions.

The most convincing evidence for causality comes from studies in which there is objective measurement of both exposure and health outcome but assessment of cumulative exposures over time or over disparate workplaces may require other approaches. Job exposure matrices (JEMs) have been developed and used to examine MSD in community-based studies (Dalbøge et al. 2016; Yung et al. 2020; Yanik et al. 2023), but JEMs account poorly for individual variation in exposures and are a blunt tool for detecting differences between trades. While there are limitations to using self-reported exposures to assess musculoskeletal demands in epidemiological studies (Hagberg 1992; Burdorf 1992; Palmer et al. 2000; Stock et al. 2005; Chiasson et al. 2015), they may provide greater credibility for intervention than exposures inferred from job title.

There is a strong correlation between mental distress and MSDs (NIOSH 1997; Nahit et al. 2001; Vargas-Prada and Coggon 2015). Workers who are unhappy may be less tolerant of workplace physical demands. Insofar as distress is caused by psychosocial factors at work, this may, in itself, point to areas for intervention to reduce MSDs. Studies that do not consider mental health as a confounder may be difficult to interpret, particularly with a cross-sectional study design. Self-report exposure data collected prospectively prior to the outcome should provide stronger evidence.

The cohort used in this report was set up to investigate the effects of work as a welder on the outcome of pregnancy (Cherry et al. 2022; Galarneau et al. 2022). Electrical workers were recruited as a comparison group. With the successful establishment of this female cohort, a parallel male cohort was recruited to investigate new onset occupational ill-health, including MSDs, in male and female workers, with the objective of identifying modifiable workplace exposures. This paper reports on models for MSDs in welding and electrical trades and considers risk factors specific to each.

Materials and methods

Four linked cohorts were set up, including women and men who had started an apprenticeship in a welding trade (welding, boiler making, and steam fitter/pipefitter) or an electrical trade after 2005. To ensure sufficient numbers, women were recruited from every Canadian province or territory while, with a more plentiful pool, men in the same trades were recruited only in Alberta (Cherry et al. 2018). Potential participants were approached through the Apprenticeship Board (or equivalent) of the province or territory, with recruitment of women starting in 2011 and of men in 2014. The study of the female cohorts became known as WHAT-ME (Women’s Health in Apprenticeship Trades—Metalworking and Electrical) and the men’s cohorts as WHAT-MEN (Workers’ Health in Alberta Trades—Men). For women, to maximize the number of pregnancies recorded, follow-up continued for up to 5 y. For men, where the focus was on new onset ill-health, follow-up was for 3 y. Questionnaires could be completed in English or French, by telephone or online. No financial incentive was offered to join.

Recruitment information was collected as the tradesperson returned the consent form sent by the Apprenticeship Board, consenting to be contacted for periodic follow-up and, in Alberta, for linkage to the administrative health database (AHDB). At baseline, we collected demographic, substance use, reproductive and health data together with an occupational history since leaving secondary school and the date of starting in the trade. At each periodic (6 mo) follow-up, data were collected on current health conditions, smoking and jobs since the last contact. Detailed trade-specific questionnaires were completed about tasks and activities on the last day at work. All those who were working, in trade or not, gave information on ergonomic demands.

For those consenting in Alberta, information on physician contacts was extracted from the AHDB for the period 1 April 2002 to 31 March 2018: in Canada, health care is free at the point of service and essentially all residents have a personal health number (PHN). For a physician to receive a fee for service, they must record both the diagnoses and the PHN. The diagnoses may be made available for research, if the patient consents. Physician data were not collected for those in other provinces.

Database construction

From the recruitment questionnaire and follow-up contacts, a month by month account of each participant’s working life since leaving secondary school was constructed, using the dates of starting and ending jobs. Ergonomic demands in a particular job were assigned for each month the participant was in that job, using the latest responses for the most recent day in that job. Symptoms (see below) were only collected for the day the questionnaire was completed but were assigned to each month until new symptoms were recorded. Diagnoses from the AHDB could be assigned to the specific month. Only diagnoses recorded for the period between leaving school and exiting the survey were entered into the database.

Exposure

Six exposure variables were examined as potentially related to low-back pain, shoulder pain, or symptoms of VWF (Supplementary Materials 1). These were the recorded hours spent

• With whole-body vibration (Q15.7: Work with tools or equipment causing whole-body vibration).

• With hand-arm vibration (Q15.6: Work with tools or equipment causing hand/arm vibration).

• Manipulating heavy objects (Q15.1: Lift, push, pull or otherwise manipulate heavy objects or goods).

• Working at or above shoulder height (Q14.1: Work with elbows at or above shoulder level).

• Crouching or kneeling down (Q14.2).

• Working in unpleasantly cold temperatures (Q15.3).

The hours for each activity were calculated from the most recent day at work and cumulated since the first report of exposures. Cumulative ergonomic indices included all periods of employment, whether or not in trade.

Current exposures were considered as binary contrasts, using the cut points from previous analyses (Cherry et al. 2022). Cumulative exposures were retained as total hours. To reduce skewness, natural logarithms were taken of cumulative measures and used throughout the analysis.

Outcomes

Self-report

At recruitment and at each follow-up questionnaire, the participant was asked, for each of 8 health conditions, “Do you have any of the following health problems now?” (Supplementary Material 2 question 3). The first 3 conditions, asthma/wheeze, rhinitis, and dermatitis, have been addressed in previous analyses (Cherry and Galarneau 2021; Galarneau et al. 2024a). Low-back pain and shoulder pain are addressed here. Symptoms of VWF, anxiety, and depression were added in March 2013, before the start of the male cohorts but were missing in the early questionnaires for women. The data on anxiety and depression have previously been examined as an outcome (Galarneau et al. 2024b) but are used here as potential confounders.

The questions used as outcomes in this report were:

• Shoulder pain sufficient that it interferes with your normal activities.

• Low-back pain sufficient that it interferes with your normal activities.

• Episodes when your fingers turn white, with numbness or tingling, in the cold?

Responses to these are labeled shoulder pain, back pain, and blanching, respectively.

At baseline (only), the participant was asked (if yes to the symptom now): How old were you when this started?

Physician reports

For those in Alberta linked to the AHDB, all physician diagnoses specified in the study protocol were assembled for the period from April 2002 to the date the participant entered the trade and, from the date of entering the trade to March 2018. Four groups of diagnoses were extracted: details of the International Classification of Diseases codes used to define them are given in Supplementary Materials 3.

• Low-back pain.

• Shoulder conditions.

• Raynaud’s syndrome.

• Joint unspecified conditions.

The composite code “Shoulder or joint unspecified conditions” reflected a physician report of either a shoulder or a joint unspecified condition.

A physician could record multiple codes for a single consultation. Mention of one of these codes was taken as an indication of the condition. It did not need to be identified as the primary reason for the visit. All diagnoses recorded by a physician were included, whether the physician was seen during a hospital admission, an ambulatory care visit or was recorded during a family physician/walk-in clinic consultation.

Potential confounders

Age, sex (as recorded by the Apprenticeship Board), body mass index (BMI), and current cigarette smoking were considered as potential confounders for each outcome. Responses to the mental health questions were also investigated as potential confounders. The questions were: Do you have:

• Days when you feel sad, empty, or depressed most of the time? (a positive response labeled Depression).

• Days when you feel worried or anxious most of the time? (a positive response labeled Anxiety).

If a respondent replied “yes” to these questions, they were asked whether this (anxiety or depression) was made worse by work.

Mental ill-health recorded by a physician was also examined, using previously published grouping of codes for anxiety disorders, stress and adjustment reactions, and depressive disorders (Galarneau et al. 2024c) with the addition of codes for personality disorders (Supplementary Materials 3). A physician’s report of any of these codes was considered a report of mental ill-health.

Statistical methods

Cox regression allowing for multiple reports within the same participant was used to investigate the relation between ergonomic work factors and both self-reported and physician recorded conditions. Each health report was related to each current and cumulative exposure in analyses allowing for confounders, for those in paid employment, whether or not that was in trade and then, for each trade separately, limited to those working in their trade. For analysis of self-reported symptoms, the dataset was limited to exposures reported on the same day or, for cumulative exposures, up to that day. For physician-reported conditions, exposure was that reported for the job, extrapolated from the most recently completed task questionnaire. The final step was to develop multivariable proportional hazard models. Ergonomic factors individually related to a health outcome (adjusting for confounders) with P < 0.10 were considered for the multivariable model and retained if they contributed with P < 0.05. Variable selection was conducted with forward and backward testing of Wald statistics. The final model for each outcome, overall and within trade, was then stratified by sex to examine whether the factors identified carried equivalent risks for men and women entering the trade.

The statistical analysis was carried out in Stata (StataCorp. 2023. Stata Statistical Software: Release 18. College Station, TX: StataCorp LLC.).

Results

Overall 1,885 participants were recruited, 1,001 welders (447 female and 554 male) and 884 workers in electrical trades (438 female and 446 male). At recruitment, 80% of those in the welding trades reported being welders, 11% steam/gas fitters, 6% pipefitters, and 3% boilermakers. The apprenticeship types for electrical workers differed between provinces: 60% were industrial electricians with smaller numbers in commercial or residential work. All but 2 reported the presence (or absence) of symptoms at least once. Among those with symptom reports, 1,684 reported workplace ergonomic demands at least once (Fig. 1). Among welders 79% (693/872) of those with symptom and ergonomic reports lived in Alberta and were matched to the AHDB. The proportion of electrical workers matched was somewhat lower (70%, 567/812) reflecting the greater proportion of female electrical workers recruited outside Alberta.

Fig. 1.

Flow diagram for inclusion in analysis.

Characteristics of the cohort are shown by trade and sex in Table 1. At recruitment, welders were more likely to be current smokers (27.9%) than those in the electrical trades (13.2%). Both welders and those in electrical work had, on average, been in their trade for 5.9 yr when they joined the study. Women in welding were somewhat younger and had less time in the trade (3.7 yr) than the other groups.

Table 1.

Characteristics of the cohort by trade

	Welders						Electrical Trade						All
	Female		Male		Overall		Female		Male		Overall
	N	%	N	%	N	%	N	%	N	%	N	%	N	%
Current smoker at recruitment	130	29.1	149	26.9	279	27.9	77	17.6	40	9.0	117	13.2	396	21.0
	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD	Mean	SD
Age at recruitment	29.5	8.6	34.9	9.4	32.5	9.4	32.3	9.2	34.8	9.7	33.5	9.5	33.0	9.5
Body mass index	25.2	4.9	27.4	4.5	26.4	4.8	25.5	4.5	26.9	4.4	26.2	4.5	26.3	4.7
Years in trade at recruitment	3.7	4.4	7.7	6.7	5.9	6.1	5.0	5.2	6.7	5.1	5.9	5.2	5.9	5.7
Years in trade at end of study	5.5	4.9	9.4	6.8	7.7	6.3	7.4	5.6	8.8	5.4	8.1	5.5	7.9	6.0
Number of follow-ups completed	5.1	3.5	3.8	2.2	4.4	2.9	6.2	3.5	4.5	1.9	5.4	2.9	4.8	2.9
Total in cohort	447		554		1,001		438		446		884		1,885

Differences in ergonomic demands for those starting welding and electrical apprenticeships were evident both in the first job for which there was ergonomic information and in cumulative exposures at the last contact (Tables 2 and 3). Those starting in welding trades were somewhat more likely to report whole-body vibration and manipulating heavy objects than those in the electrical trades. Welders were much more likely to report hand-arm vibration. Those in the electrical trades had more exposure to working above shoulder height and, cumulatively, to working in a crouching or kneeling position and to working in unpleasantly cold conditions.

Table 2.

Ergonomic demands by trade: first month with ergonomic data

	Welder		Electrical		Both		P
	n	%	n	%	n	%
Whole-body vibration
None	793	90.9	759	93.5	1,552	92.2	0.057
Some	79	9.1	53	6.5	132	7.8
Hand-arm vibration
≤1 h	451	51.7	658	81.0	1,109	65.9	<0.001
>1 h	421	48.3	154	19.0	575	34.1
Manipulate heavy objects
<1.5 h	594	68.1	610	75.1	1,204	71.5	0.002
≥1.5 h	278	31.8	202	24.9	480	28.5
Working above shoulder height
≤2 h	585	67.1	449	55.3	1,034	61.4	<0.001
≥2 h	287	32.9	363	44.7	650	38.6
Hours crouching or kneeling
<2 h	513	58.8	454	55.9	967	57.4	0.237
≥2 h	359	41.2	358	44.1	717	42.6
Working in unpleasantly cold temperatures
None	728	83.5	654	80.5	1,382	82.1	0.127
Some	144	16.5	158	19.5	302	17.9
N	872	100.0	812	100.0	1,684	100.0

Table 3.

Ergonomic demands by trade: mean log hours of cumulative exposures at last month with ergonomic data

	Welder		Electrical		Both		P
	Mean log hours	SD	Mean log hours	SD	Mean log hours	SD
Whole-body vibration
	−3.29	2.79	−3.54	2.51	−3.41	2.67	0.056
Hand-arm vibration
	2.26	3.23	0.37	3.85	1.35	3.66	<0.001
Manipulate heavy objects
	2.35	3.01	2.05	3.20	2.21	3.11	0.044
Working above shoulder height
	2.16	3.13	3.41	2.57	2.77	2.94	<0.001
Hours crouching or kneeling
	3.15	2.34	3.61	2.48	3.37	2.42	<0.001
Working in unpleasantly cold temperatures
	−1.70	3.81	−1.06	4.03	−1.39	3.93	<0.001
N	872		812		1,684

Overall 38% had at some time reported low-back pain that started after they joined their trade, 34% shoulder pain and 33% symptoms suggestive of VWF (blanching). The proportions ever receiving a physician record of a health condition starting since entering the trade were lower: 21% for back conditions and 9% for conditions coded as shoulder or unspecified joint (Table 4). Only 3 participants received an incident diagnosis of Raynaud’s syndrome. Few differences were seen, by sex or trade, in reports of health conditions. Male, but not female, welders were more likely than electrical workers to have a physician report of shoulder or unspecified joint conditions.

Table 4.

Any report of condition since joining trade: incident only

	Self-Report
	Low-Back Pain			`			Blanching
Women	N	n	%	N	n	%	N	n	%
Welder	392	145	37.0	413	131	31.7	372	134	36.0
Electrical	387	152	39.3	405	145	35.8	394	124	31.5
Both	779	297	38.1	818	276	33.7	766	258	33.7
P	0.555			0.237			0.194
Men
Welder	499	206	41.3	533	196	36.8	502	170	33.9
Electrical	401	143	35.7	416	133	32.0	401	129	32.2
Both	900	349	38.8	949	329	34.7	903	299	33.1
P	0.099			0.131			0.619
All
Welder	891	351	39.4	946	327	34.6	874	304	34.8
Electrical	788	295	37.4	821	278	33.9	795	253	31.8
Both	1,679	646	38.5	1,767	605	34.2	1,669	557	33.4
P	0.422			0.763			0.212
	Physician Report
	Low Back Condition			Shoulder Condition			Shoulder or joint unspecified condition
Women	N	n	%	N	n	%	N	n	%
Welder	251	53	21.1	276	10	3.6	276	23	8.3
Electrical	174	36	20.7	191	10	5.2	189	17	9.0
Both	425	89	20.9	467	20	4.3	465	40	8.6
P	1.00			0.487			0.867
Men
Welder	487	106	21.8	500	29	5.8	499	53	10.6
Electrical	394	83	21.1	415	13	3.1	413	26	6.3
Both	881	189	21.5	915	42	4.6	912	79	8.7
P	0.869			0.058			0.024
All
Welder	738	159	21.5	776	39	5.0	775	76	9.8
Electrical	568	119	21.0	606	23	3.8	602	43	7.1
Both	1,306	278	21.3	1,382	62	4.5	1,377	119	8.6
P	0.838			0.297			0.083

Self-reports of health conditions were strongly related to a corresponding physician report (Table S1) but only 31% of those who ever reported low-back pain had a physician report of the condition. Of those with self-reported shoulder pain, 16% had a record from a physician consultation of shoulder or unspecified joint conditions. While all three participants with an incident record of Renaud’s syndrome complained of symptoms of white finger, there were many who reported suggestive symptoms but with no corresponding physician diagnosis, and it is unclear how physician consultation for these symptoms were recorded (if indeed there were consultations).

The relation of potential confounders to health reports is shown in Table S2. A history of smoking since joining the trade was related to ever-reports of each health outcome. Those who ever reported feeling anxiety or depression were also more likely to have a report of each outcome other than physician reports of a shoulder condition. In those with at least one physician consultation, a physician report of a low back condition since entering the trade was related to having some physician record of mental ill-health. Similarly, those with a physician record of a shoulder or unspecified joint condition were more likely to have been recorded as having mental ill-health.

Table 5 shows the multivariable relation of confounders to health conditions. Unlike Table S2, which considered “ever” for both confounders and outcome, Table 5 considers whether the outcome and confounder occur in the same month. It also includes age and BMI (see Table 1) and differentiates anxiety and depression made worse by work. For each of the self-reported outcomes, current smoking, anxiety, and depression were related to a greater likelihood of reporting the symptom, with anxiety made worse by work associated with a higher risk of back pain and, more marginally, shoulder pain. Age showed only a slight increase in reports of shoulder pain. A higher BMI was associated with lower reports of blanching. Men were slightly more likely to report back pain and women to report blanching. Confounders were rather different for physician reports. Age and BMI were related to physician records of low back conditions and age to shoulder or unspecified joint conditions. There was no difference between men and women in reports of any of the conditions and no relation was seen between back and shoulder conditions and anxiety and depression, apart from an increased risk of joint conditions in those with depression made worse by work. To simplify the comparison of effects of different job demands, all analyses were adjusted for all potential confounders shown in Table 5. Having adjusted for these confounders, there was little difference in symptom reporting in welders and electricians, with somewhat greater self-reporting of back and shoulder pain in male welders (Table S3).

Table 5.

Relation of confounders to complaints: multivariate Cox regression in those working and without the condition before starting in trade

		Self-Report
		Low-Back Pain			Shoulder Pain			Blanching
		HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
Work not in trade		1.19	0.94–1.50	0.139	0.93	0.75–1.15	0.503	0.96	0.77–1.21	0.731
Age		1.00	0.98–1.01	0.466	1.01	1.00–1.03	0.058	1.01	1.00–1.02	0.107
Female		0.82	0.67–1.01	0.058	1.07	0.88–1.31	0.497	1.27	1.03–1.56	0.023
BMI		1.01	1.00–1.03	0.123	1.00	0.98–1.02	0.931	0.94	0.92–0.96	<0.001
Current smoker		1.46	1.19–1.79	<0.001	1.21	0.96–1.52	0.101	1.41	1.13–1.76	0.003
Current anxiety		1.31	1.01–1.69	0.039	1.62	1.25–2.10	<0.001	1.92	1.51–2.45	<0.001
Anxiety made		1.59	1.25–2.04	<0.001	1.29	1.00–1.66	0.051	1.16	0.95–1.48	0.242
Worse with work
Current depression
Depression made		1.77	1.39–2.25	<0.001	1.39	1.08–1.80	0.011	1.38	1.06–1.79	0.016
Worse with work		0.98	0.75–1.26	0.852	1.22	0.93–1.61	0.156	1.17	0.88–1.58	0.283
N	Observations	7,040			7,296			7,165
	Workers	1,542			1,615			1,576
	Conditions	1,099			1,046			1,074
		Physician Reports
		Low Back Condition			Shoulder Condition			Shoulder or Unspecified Joint Condition
		HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
Work not in trade		0.98	0.46–2.07	0.954	1.15	0.51–2.62	0.736	1.01	0.55–1.88	0.971
Age		1.03	1.01–1.05	0.005	1.05	1.00–1.09	0.038	1.08	1.01–1.15	0.020
Female		1.02	0.60–1.74	0.943	1.58	0.74–3.36	0.234	1.09	0.57–2.10	0.795
BMI		1.04	1.02–1.07	0.002	1.01	0.96–1.06	0.716	1.04	0.99–1.09	0.128
Current smoker		1.03	0.66–1.62	0.896	1.66	0.88–3.12	0.119	2.15	0.91–5.07	0.082
Current anxiety		1.61	0.77–3.35	0.207	1.65	0.69–3.92	0.261	1.70	0.87–3.32	0.121
Anxiety made		1.22	0.51–2.91	0.660	1.26	0.28–5.54	0.763	0.93	0.38–2.31	0.881
Worse by work
Current depression		1.05	0.49–2.25	0.904	0.50	0.18–1.37	0.178	0.48	0.23–1.00	0.050
Depression made		1.25	0.52–3.01	0.624	1.64	0.30–8.98	0.568	2.73	0.99–7.47	0.051
Worse by work
N	Observations	113,935			118,053			117,703
	Workers	1,233			1,301			1,297
	Health Condition	478			55			125

The relation of each of the job demands individually to the outcome measures is shown in Tables S4–S8. The number of included events ranged from 988 for self-reported back pain (Table S4) to 15 for physician-reported shoulder pain (with only 3 of these for electrical workers): the alternative joint outcome (shoulder or unspecified joint condition) with 46 reports was examined here.

Self-reported back pain was related to each of the current job demands in participants overall (Table S4). Among welders in trade, hand-arm vibration, crouching or kneeling, and working in unpleasantly cold temperatures, increased the HR and in electrical workers all demands except whole-body vibration. Overall, and for welders in trade, cumulative hours working with whole-body vibration and in cold conditions increased the likelihood of reporting low-back pain. Cumulative hours manipulating heavy objects also increased the HR overall and in electrical workers. None of the ergonomic demands related significantly to an increased risk of physician-reported low-back pain (Table S5), but the size and direction of the HRs suggest whole-body vibration and hours working in a crouching or kneeling position as potential risk factors.

Self-reported shoulder pain was also related to each of the work demands except working in unpleasantly cold temperatures in the cohort as a whole (Table S6): hand-arm vibration and working above shoulder height were risk factors in all subgroups. Cumulative exposure to whole-body vibration, hand-arm vibration, manipulating heavy objects, and working above shoulder height were related to reports of shoulder pain overall, with hours working above shoulder height more strongly related to shoulder pain in welders and whole-body and hand-arm vibration more strongly in electrical workers. The analysis of job demands in relation to physician reports of shoulder or unspecified joint pain (Table S7) showed an increased risk with hand-arm vibration overall and in electrical workers and with cumulative whole-body vibration. Working above shoulder height or in a crouching position were also suggested by the size of the HR as a possible risk for a physician report of joint conditions in electrical workers.

Factors associated with a self-report of symptoms consistent with early VWF (blanching) are shown in Table S8. Each subgroup showed a strong relation of such symptoms to working in unpleasantly cold conditions, both currently and cumulatively, but there was little relationship to hand-arm vibration either currently or cumulatively. Hours working above shoulder height were related to this symptom overall, more evidently for welders than electrical workers, while hours crouching or kneeling was more evident for electrical workers.

The final multivariable models for self-reported back pain, shoulder pain, and blanching are given in Table 6. The model for low-back pain for the whole sample identified reports of currently manipulating heavy objects, working above shoulder height and in cold environments, together with cumulative exposure to whole-body vibration. For welders in trade, cumulative exposure to whole-body vibration contributed to reports of low-back pain, together with working while crouching or kneeling and in a cold environment. For electrical workers, only current work that involved manipulating heavy objects and working above shoulder height determined reports of low-back pain. Shoulder pain was related to current manipulation of heavy objects and working above shoulder height and to cumulative exposure to whole-body vibration in the cohort as a whole, with whole-body vibration replaced by hand-arm vibration for electrical workers. In welders, shoulder pain was related only to cumulative work above shoulder height. Self-report of symptoms suggesting VWF (blanching) were related to current exposures to whole-body vibration and cold work environments and to cumulative exposure to crouching or kneeling in the group as a whole, to cumulative exposure to cold environments in welders and to current exposure to crouching or kneeling and cold environments in electrical workers. Supplementary analyses (data not shown) showed no evidence of an interaction between cold exposure and hand-arm vibration increasing the risk of such symptoms.

Table 6.

Relation of ergonomic factors to health outcomes in the whole cohort and by trade. Final model from multivariable Cox regression, adjusting for confounders^a

		Self-Report									Physician Report
Whole Cohort		Back pain			Shoulder pain			Blanching			Shoulder or joint
Current exposure		HR	95% CI	P	HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
Whole-body vibration: any		—	—	—	—	—	—	1.34	1.01–1.77	0.042	—	—	—
Manipulating heavy objects: ≥1.5 h		1.24	1.04–1.48	0.015	1.22	1.02–1.47	0.030	—	—	—	—	—	—
Working above shoulder height: ≥2 h		1.31	1.11–1.55	0.002	1.39	1.18–1.65	<0.001	—	—	—	—	—	—
Unpleasantly cold temperatures: any		1.25	1.03–1.53	0.027	—	—	—	1.39	1.16–1.66	<0.001	—	—	—
Cumulative exposure (log hours)
Whole-body vibration		1.04	1.00–1.08	0.029	1.05	1.02–1.07	0.001	—	—	—	1.12	1.01–1.26	0.036
Crouching or kneeling		—	—	—	—	—	—	1.03	1.00–1.07	0.045	—	—	—
Unpleasantly cold temperatures		—	—	—	—	—	—	—	—	—	—	—	—
N:	Observations	6,340			6,561			6,467			50,330
	Participants	1,425			1,488			1,460			1,196
	Health reports	988			949			975			46
Welders		Back pain			Shoulder pain			Blanching			Shoulder or joint
Current exposure		HR	95% CI	P	HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
Crouching or kneeling: ≥2 h		1.33	1.06–1.68	0.015	—	—	—	—	—	—	—	—	—
Unpleasantly cold temperatures: any		1.42	1.01–2.00	0.045	—	—	—	—	—	—	—	—	—
Cumulative exposure (log hours)
Whole-body vibration		1.06	1.01–1.11	0.018	—	—	—	—	—	—	—	—	—
Working above shoulder height		—	—	—	1.07	1.03–1.11	0.001	—	—	—	—	—	—
Unpleasantly cold temperatures		—	—	—	—	—	—	1.05	1.01–1.09	0.012
N:	Observations	2,362			2,477			2,359			20,266
	Participants	654			683			660			591
	Health reports	420			395			372			28
Electrical workers		Back pain			Shoulder pain			Blanching			Shoulder or joint
Current exposure		HR	95% CI	P	HR	95% CI	P	HR	95% CI	P	HR	95% CI	P
Hand-arm vibration: >1 h		—	—	—	—	—	—	—	—	—	4.50	1.18–17.08	0.027
Manipulating heavy objects: ≥1.5 h		1.39	1.06–1.81	0.017	1.37	1.03–1.83	0.030	—	—	—	—	—	—
Working above shoulder height: ≥2 h		1.67	1.31–2.12	<0.001	1.79	1.34–2.17	<0.001	—	—	—	—	—	—
Crouching or kneeling: ≥2 h		—	—	—	—	—	—	1.47	1.18–1.83	0.001	—	—	—
Unpleasantly cold temperatures: any		—	—	—	—	—	—	1.55	1.21–1.99	0.001	—	—	—
Cumulative exposure (log hours)
Hand-arm vibration		—	—	—	1.06	1.01–1.10	0.007	—	—	—	—	—	—
N:	Observations	2,932			2,990			3,022			23,944
	Participants	629			652			653			500
	Health reports	392			387			430			7

^aAdjusted for in trade, age, sex, body mass index, current smoker, current anxiety, and current depression.

As physicians’ reports of back pain showed no significant positive relation to ergonomic factors (Table S5), no further modelling was carried out. For joint conditions, cumulative whole-body vibration and current hand-arm vibration were identified as potential risk factors for a physician report overall and for electrical workers (Table S7): in multivariable analysis, only whole-body vibration was related to physician report in the cohort as a whole (Table 6) and only current hand-arm vibration in electrical workers.

Supplementary analyses (Tables S9–S14) considered whether the final models applied equally to men and women within trades. Women were somewhat less likely to self-report backpain than men, overall and in welding (Tables 4 and 5). The final models for back pain showed cumulative exposure to whole-body vibration strongly related to back pain for men overall (Table S9) and in welding (Table S11). Risk of reporting back pain for women was unrelated to whole-body vibration (Table S10 and S12) with working in cold environments as the strongest risk factor. For female electricians (Table S14), but not for men (Table S13), currently manipulating heavy objects contributed to back pain. For self-reported shoulder pain and blanching, where there was no difference between men and women in reporting of symptoms, the final models in Table 6 were closely approximated when stratified by sex (Tables S9–S14).

There was no difference between men and women in the likelihood of receiving a physician record of back pain or joint conditions. For physician reports of joint conditions, the higher risks with whole-body vibration overall and hand-arm vibration in electrical workers (Table 6) were reflected in elevated HRs for both men (Tables S9 and S11) and women (Tables S10 and S12) with somewhat greater effect seen in men.

Discussion

This analysis aimed to develop models of ergonomic risk factors for MSDs in welders and electrical workers using both cumulative and current exposures, adjusting for key confounders. In welders, cumulative whole-body vibration was related to self-reported back pain, with cumulative hours working at or above shoulder height a risk factor for shoulder pain and cumulative exposure to cold environments for symptoms suggestive of VWF. In electrical workers, current exposures were more predictive than cumulative ones, although cumulative exposure to hand-arm vibration was related to self-reported shoulder pain. Risk patterns were largely similar for men and women in these trades, although cumulative whole-body vibration, more often reported by men, was predictive of back pain only for male welders concordant with the higher reporting of back pain in male than female welders. The relationship of cumulative exposures to MSDs would suggest that these are chronic conditions, caused by repeated job demands.

These results are consistent with earlier reports of ergonomic demands, with whole-body vibration being strongly linked to back pain in the 1997 NIOSH report. The relation of working above shoulder height and hand-arm vibration to shoulder pain in the present study is also consistent with earlier reports of occupational shoulder exposures and shoulder conditions (Roquelaure et al. 2009; Dalbøge et al. 2020; Thorsen et al. 2024; Sun et al. 2024). Reports in welders (Burdorf et al. 1998; Weyh et al. 2020; Lourenço and Luís 2021; Elvis C et al. 2022) confirm the high rate of MSDs but without identifying physical demands that might be causally related. A study of metal workers, that included welders, found cumulative exposure to hand-arm vibration to be related to VWF, not seen in the present study (Sauni et al. 2009). In electrical workers, the high prevalence of MSDs has been documented (Hunting et al. 1994; Inaba and Mirbod 2010) but without detailed consideration of job demands.

A strength of the present study is that it followed welders and electrical workers prospectively, allowing exclusion from analysis of participants developing the symptom before entering the trade and computation of cumulative exposures prospectively, before the condition became apparent. The concurrent collection of data on mental distress as well as other confounders was also a strength, allowing us to adjust statistically for the tendency of those who are distressed to report pain. The linkage to the AHDB was a further strength, allowing the examination of conditions recorded by a physician rather than by self-report and the confirmation, through medical records, that those with low-back pain or a joint condition were also more likely to have a mental ill-health condition. The inclusion of a large proportion of women allowed us to consider differences between men and women in the relation between job demands and health outcome.

There are also limitations. We considered 6 ergonomic factors for each of 3 MSDs, and this number of comparisons might well give rise to spurious relationships. Furthermore, we cannot comment on disorders in sites (neck, elbow, hand, wrist, knee, or hip) other than the 3 discussed here. Moreover, we considered each site in isolation: 1 in 7 of those reporting back pain also reported shoulder pain and blanching in the same period, and it may be that, as found in the CUPID study (Coggon et al. 2013), the influence of physical work demands would be attenuated in those with widespread pain. There was no objective measurement of exposure or independent observations to verify the self-report information: with participants working for hundreds of employers across Canada, such validation was not feasible. Furthermore, even for conditions identified through the AHDB, we do not have a clear understanding of the underlying pathology, with the coding for joint conditions frequently not specifying the joint, resulting in disparate codes to identify “shoulder or joint conditions unspecified.” It is of interest that cumulative whole-body vibration (and current hand-arm vibration) increased the likelihood of a condition so poorly defined in physician records. We were not able to carry out objective testing for VWF and it is not clear whether the lack of any relation of symptoms to vibration reflects cumulative hand-arm vibration of too low intensity or duration to cause VWF, or whether the question on “episodes when your fingers turn white, with numbness or tingling, in the cold” was too unspecific, receiving positive answers from many who had experienced bitterly cold Canadian winters, but without reported vibration exposure. Working in unpleasantly cold conditions was the prime predictor of a positive response. Burström et al. (2010) found that those working in colder climates had an increased risk of VWF.

Only a minority of participants reporting musculoskeletal pain consulted a physician about this, and we know nothing about why they did not do so. It might be that the respondent expected the pain to resolve without intervention or that long hours of work made accessing a physician more difficult. It is unlikely that a family physician would initiate extensive intervention or treatment for low-back or shoulder pain, unless there were severe limitation in the capacity for daily living. Nevertheless, some tasks may have been exceptionally difficult or painful to carry out if suffering from an MSD, whether or not caused by work. Such demands may have led to a physician consultation, perhaps for a medically sanctioned work reassignment or time off work. Furthermore, we know nothing about the psychosocial demands of the workplace, although we do have a record of mental distress made worse by work. Earlier analysis showed that those harassed by fellow workers or supervisors were more likely to report distress made worse by work (Galarneau et al. 2024b). Similar work-related stresses, or poor workplace support, might explain the relation seen here between anxiety made worse by work and reports of back pain. Even given the relatively small proportion of those with low-back pain who consulted a physician, the lack of any clear relation to work demands is surprising and suggests that factors other than intensity or persistence of pain may drive the decision to consult a physician.

In summary, using largely self-report data, we have developed models of factors resulting in MSDs that are consistent with the literature, largely congruent for men and women, and differ for those in welding and electrical work. This has not been done previously for either trade. Although much of the pain reported had not yet reached a physician, it must be remembered that this is a relatively young cohort, with decades of work ahead. Even at this early stage, pain was seen to be related to cumulative exposures. With aging, and continued employment, cumulative demands may lead to debilitating arthroses, limitations of movement and early retirement (Palmer and Goodson 2015). There is no easy panacea. The American Conference of Governmental Industrial Hygienists (ACGIH) publishes Threshold Limit Values for each of the ergonomic factors, other than crouching or kneeling, in our final models (ACGIH 2024). Reduction of vibration may require engineering solutions with replacement or modification of tools (Jetzer et al. 2003), following limit values from the ACGIH or from local jurisdictions including the European Union Directive 2002/44/ European Union (2019) and Health and Safety Executive UK (2019) . While use of exoskeletons may be feasible to reduce strain from working overhead, changes to position of work are likely to need innovative approaches, coupled with close knowledge of the trade (Tahmasebi et al. 2020; Stone et al. 2011). The analyses reported here have highlighted key ergonomic factors that pose risks for welders and electrical workers and need to be addressed to maintain a heathy and productive workforce in these trades.

Supplementary material

Supplementary material is available at Annals of Work Exposures and Health online.

Funding

This work was funded by WorkSafe BC, the Government of Alberta (OHS Futures program), and the Canadian Institutes for Health Research (FRN 130235).

Conflict of interest

The authors declare no conflict of interest relating to the material presented in this Article. Its contents, including any opinions and/or conclusions expressed, are solely those of the authors.

Ethics and informed consent

The project was reviewed by the Health Ethics Review Board of the University of Alberta (Pro00017851). All participants gave written informed consent.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.