Study design
This was a cross-sectional study, conducted among workers in Twangiza Gold Mine, a subsidiary of Banro Corporation in the DR Congo in 2018. In 2011, Twangiza became the first commercial gold mine built in the DRC in over 50 years. It is an open pit gold mine which started commercial gold production in late 2012. It is located 35 kilometres west of the Burundi border and 45 kilometres to the southeast of Bukavu in South Kivu.
Study population and sample size.
The study was conducted amongst workers of Twangiza gold mine for four weeks from 1st June to 30th June 2018. Eligibility for participants is shown in Table 1.
Table 1: Eligibility criteria for recruiting participants in the Musculoskeletal disorders study in Twangiza Gold Mine
Inclusion criteria
|
Exclusion criteria
|
- Minimum of one year experience at the mining site.
- No history of trauma or injuries and psychological problems (information obtained from annual medical reports at the mine’s medical office)
- Signed (or thumb-printed) informed consent document
|
- Child miners below 18 years of age.
- Previous trauma or ongoing psychological problem (information obtained from annual medical reports at the mine’s medical office)
- Refusal to consent to participate.
|
The gold mine employed 691 workers at the time. Workers were stratified into six departments (strata) with each performing different tasks. The different strata were defined by grouping jobs together based on similar job demands. These departments include the mining department (133 members) which does extraction and blasting of gold ore requiring high efforts, engineering (82 members) and maintenance departments (137 members) which are involved in construction and maintenance of equipment respectively thus demanding high impact efforts. Mineral resources (MRM) department (86 members) explores the ore reserves including drilling requiring high efforts. Metallurgy department (116 members) refines and processes gold ore into pure gold with moderate effort requirements. The “others” (137 members) are involved in the transportation of goods/personnel to and from the mining sites, planning, educating and maintaining health and safety for the company with minimal effort requirements.
The Kish Leslie formula (1965) was used to determine the sample size (16). We assumed an alpha of 0.05, power (1-beta) of 0.80, a sampling error of 5%, and prevalence (P) of WRMSDs of 42.6% was considered from a similar study in Malawi (2). The sample size was 376 but since the proportion of sample to population was larger than 5%, we then used a finite population correction formula by Daniel (17) with an additional 10% to account for refusal to participate/non-response giving an adjusted sample size of 273 (18, 19)
To make the sample size, we set out to randomly select 46 participants from each of the 6 departments (strata). Random sampling was applied instead of proportionate sampling since the departments had similar number of workers. We used a table of random numbers to select the participants per department. Complete data were obtained from 196 gold mining workers hence a response rate of 71.8%. The reasons for non-response include incomplete data, difficulties to schedule appointments since some workers had left for their regular day offs/leave while others had changed their minds not to participate.
Data collection and measurements
A self-administered questionnaire was developed from existing surveys of musculoskeletal disorders and risk factors. Various questionnaires guided the design of the questionnaire for this study including the standardized Nordic, and the modified versions of the Washington state risk factor checklist and the upper limb Core QX checklist used by Kunda et al, (2, 20). The Nordic questionnaire is a widely accepted, easy to administer, and cost efficient tool for collecting data on self-reported musculoskeletal discomfort and sickness absence and it has been shown to have high validity for capturing MSDs in various settings (21) and for different body regions(22, 23).
The questionnaire had four sections with the first section providing data on background variables such as age, sex, educational level. The second part consisted of questions on the MSD injury/complaints profile.
The modified version of the Nordic questionnaire measured the subjective ache/pain/numbness/injury on the different body parts. Work related MSD was defined as developing an ache/pain/numbness/injury after starting work at mines or either aggravated by working conditions while on duty (work related) during the last 12 months. A “yes” response to complaints from duty was used to ascertain the prevalence of WRMSDs. The affected body parts (Neck, shoulder, elbows, wrists/hands, upper back, lower back, hips/thighs, knee, ankles/feet) were also recorded under this section.
Section three consisted of the risk factors such as the work-environment characteristics and work-practices such as machines used, postures adopted at work and total work duration per day. The workers reported on health hazards at their job types to provide estimates of safety hazards about risk factors. The workers indicated the length of exposure to an activity which determined the exposure level as being lower risk /cautionary (occasionally/less than 2 hours per day or less than 10 times per day) or higher risk / hazardous ( ≥2hours/day or ≥10 times/day).
The fourth section recorded the psychological and psychosocial risk factors which were measured using a modified version of the upper limb Core QX checklist (2). Five questions were asked on job demand and they had responses with the format 1=strongly disagree; 2=disagree; 3=agree; and 4=strongly agree), These responses were collapsed into two categories (agree or disagree) during analysis. Agreement that some aspects of the job are demanding was considered “high job demand” or otherwise low job demand. Job security was considered present if the participant felt he/she is indispensable and less likely to lose their job, otherwise considered job insecure.
We also asked four questions on job control (variety amount, pace and duration of tasks) with responses “very little”, “little”, moderate”, “much” and “very much”. These were each collapsed into “little” and “much”. A participant was considered to have job control if they indicated much control on any of the four aspects, otherwise considered to have low job control.
Regarding work manship, four questions were asked about receiving support from supervisors with options “Very much” , much or (easy)”, “little” or “Not at all”. These were collapsed to two categories; “little” or “much”. Workmanship was then categorised as “good” if workers received much support from supervisors on any of the four aspects or bad if otherwise.
Mental state was evaluated using five questions on anxiety and depression with options rarely or none (Coded 1), sometimes (coded 2), Often (coded3), Most or all of the time (coded4). These were collapsed to two categorises (rarely/ none/sometimes and Often/most or all of the time). Mental state was considered “normal” if the participants indicated none/rarely or sometimes on any of the questions and otherwise considered not normal. The full English questionnaire for the study is provided as supplementary file 1.
Four individuals were trained as research assistants for three days on study aims, procedures, ethics, MSDs, associated factors of MSDs and preventive measures. They distributed and collected the questionnaires and assisted the workers with difficulties encountered during the filling of questionnaires. The research assistants were introduced to the participants and an appointment was made with the mine sectional supervisors and all the participants who were available on the day and time of questionnaire distribution. The questionnaires were distributed by the researcher or research assistants and by the sectional supervisors who were on duty. Those on the night shift had the questionnaires distributed to them by the sectional heads operating at night who had received a briefing from the daytime sectional heads.
The questionnaires were completed over a four weeks period with the researcher and research assistants collecting the completed questionnaires daily and also reminding those who had not yet completed to do so if possible. The original questionnaire was designed in English and later translated into French since most of the workers spoke French and the minority English.
Statistical analysis
All generated data were entered into a Microsoft Excel database, cleaned and exported to Stata version 14.0 (StataCorp, Texas). Continuous data were expressed as mean and standard deviation. Categorical data variables such as sex, age groups, the prevalence of WRMSDs, prevalence of WRMSDs by age categories, working (shift) hours, body parts affected, department, exposure to the ergonomic and psychosocial factors were expressed as frequencies and proportions. Prevalence ratios (PRs) were computed using a multivariable modified Poisson regression with the logarithm as the link function, with robust error variances to measure the association between the WRMSDs and independent variables. Simple models consisting of the outcome and one independent variable were run to obtain the crude PRs. Variables that had p values under liberal threshold of 0.1 in bivariate models were included in the multivariable model (24). Backward stepwise elimination method was applied until only variables with p value ≤0.05 and those significantly improved the fit of the model were retained. The goodness of fit test showed an insignificant p-value of 0.7553 suggesting that the model fitted the data reasonably well. The adjusted PRs and their 95% confidence intervals are presented.