Work related injury and its disparity across selected occupations in Ethiopia: systematic review and meta-analysis

Background Objective

and 160 million victims of work-related illnesses are occurred annually and over 6000 deaths every single day (3).
Despite magnitude and extent disparities, occupational injury and illness is the issue of both developed and developing nations. For instance, in Canada around 951 workers died due to work related causes in 2017(4).
Work related injury has been continued as a major public health problem and leads to devastating economic and health impact. According to World Health Organization (WHO) about 11% of the burden of disease is from work place related illnesses (5).
In Ethiopia there is high report of injuries from health facilities (6,7). However, health facilities reports did not show how many cases of injuries were work places related and could not well address what factors were associated with those injuries and illness (7,8). There are also few fragmented empirical studies on predictors or determinants of occupational injury and illness (9)(10)(11).
As to the knowledge of the authors of this study in Ethiopia there is no a national representative empirical study on the magnitude and determinants of occupational injury among municipal solid management workers, textile and construction industry workers.
Therefore this study aims to determine magnitude of work related injury and its disparity across selected occupations in Ethiopia. The results of this study may be helpful for decision and policy makers to have a nationwide insight and to take targeted corrective measures to promote health and safety and prevent non-fatal and fatal injuries and illnesses.

Identi cation and study selection
Through international data bases (MEDLINE/PubMed, Science Direct, Google scholar and Cochrane Library) and national digital library repositories (Addis Ababa University's digital library repository) were searched for potentially relevant published and unpublished research articles; between the periods 12 October, 2019 to 15 December, 2019. The key terms used for the searching articles were "Magnitude", "Epidemiology" , "Prevalence" , "occupational injury", "Work related injury", " work places injury", "Illness", "diseases", "Municipal Solid waste management", "Textile industry", "Garment", "Construction industry", "Workers" and "Ethiopia". Boolean operators "AND" and "OR" were used as connector. The Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guideline was used to undertake this study (12).

Eligibility criteria
The authors of this meta-analysis reviewed abstracts from initial search using well de ned inclusion and exclusion criteria.

Inclusion criteria
Both peer reviewed published and unpublished observational articles written in English language and conducted in Ethiopia which reports magnitude of work related injury among workers of municipal solid waste management, textile and construction industry were included.

Exclusion criteria
Studies which were not fully accessed, di cult for data extraction and didn't report the outcome of interest were excluded.

Data extraction
Using a Microsoft Excel two authors (WG and MA) independently extracted all the necessary data .The data extraction format for the rst outcome of interest (magnitude of work related injury) consists; name of the rst author, region publication year, study design, sample size, response rate, cases, total, type of occupation, data collection period). For the second outcome of interest (predictors), the data extraction format was prepared for each most frequently reported predictors by the studies included in this metanalysis. These predictors includes educational status (below high school vs high school and above high Married), working hours per day (8 and below 8 hrs. vs more than 8 hours) and job related stress (Yes vs No) .Variables in this meta-analysis were considered as a predictor because at least two or more studies reported them as a predictor.

Outcome measurements and Quality assessment
This study has two main outcomes. The rst outcome of interest was to determine the magnitude of work related injury. The second outcome of the study was to identify the predictors of work related injury. The magnitude was calculated by dividing the number of respondents having at least 1 occupational injury within 12 months prior to the data collection period of included studies to the total number of participants who have been included in the study (number of respondents) multiplied by 100. Regarding predictor variables, the authors of this study calculated the odds ratio from the primary studies using the two by two tables. To assess the methodological qualities of the included articles, a modi ed version of the Newcastle-Ottawa Scale adapted for cross-sectional studies was used (13). By taking the mean score of the two researchers, disagreements of their assessment results were resolved. Finally, articles assessed with a score of ≥6 out of 10 were considered as achieving high quality.

Data processing and analysis
Microsoft Excel for data extraction and STATA Version 14.0 software were used for analysis. The characteristics of included original articles were described using a table and forest plot. Statistical tests like heterogeneity χ2 test, I 2 test and the p-values were done to check the presence of heterogeneity among the reported prevalence of included studies (14).Random effects meta-analysis model was used to estimate overall magnitude of work related injury. Univariate meta-regression model was conducted by taking the publication year and number of respondents. Potential publication bias was also assessed objectively by using Egger's correlation and Begg's regression intercept tests at 95% signi cant level respectively (15,16). In addition, subgroup analysis was done based on region of studies and type of occupation.

Search results
In the rst step of our search, 695 articles were retrieved regarding magnitude and predictors of occupational injury among workers of municipal solid waste management, construction, and textile workers. Of these initial records, 294 articles were excluded due to duplication. From the remaining 401 articles, 307 articles were excluded after review of their titles and abstracts being assessed as nonrelevant to this review. Therefore, 34 full text articles were accessed, and assessed for eligibility based on the pre-set criteria, which resulted in further exclusion of 17 articles primarily due to the study population and outcome of interest. Finally, 17 studies were eligible and included in the meta-analysis ( g: 1) In this study to estimate the pooled magnitude of work related injury a total of 8025 workers from MSWM, construction and textile industry were involved. Concerning geographical distribution of the studies, the 17 studies were obtained from the four regions of the country: six studies from Addis Ababa (10,19,(21)(22)(23)30), three from Oromia (24,27,29), seven (7) from Amhara (11,17,18,20,25,26,28), and one from Southern Nations, Nationalities and Peoples' Region (SNNPR) (29). The highest prevalence of occupational injury and illness (84.7%) was reported in Addis Ababa(10) whereas the lowest prevalence (31.4%) was reported from a study done in Arba Minch (SNNPR) (9). In addition, the original studies included in the meta-analysis had a response rate ranging from 83-100% and almost all the studies had good response rate having a response rate of above 80%. Regarding the publication condition of the studies, two of the 17 studies were unpublished (29,30), and 15 of the studies were published in reputable journals. Finally, the quality score of the studies ranges from 6-9 out of 10 points.

Meta-analysis
As shown in the forest plot, the result of the 17 included studies revealed that the pooled magnitude of occupational injury was 43.59% (95% CI: 34.48, 52.70%) (Fig: 2). There was high heterogeneity across the studies which was shown by I 2 statistic (I 2 = 98.7, p value = 0.000). As a result, a random effect model was employed to estimate pooled magnitude of work related injury. In addition, a univariate metaregression model to identify the possible sources of heterogeneity, by considering factors like data collection year, number of studies per region and sample size, but none of these variables was found to be statistically signi cant. Eggers' test revealed the presence of statistically signi cant publication bias (p = 0.019). Therefore, trim and ll analysis was also done to adjust the nal pooled estimate.

Discussion
In this study the overall pooled magnitude of work related injury among workers of selected occupations was 39.35% (95% CI: 27.40, 51.30). This prevalence is higher than the reports of previous meta-analysis done in Ethiopia which was particularly on needle stick injury among health professionals 28.8% (95%CI: 23.0, 34.5)(32) and studies done in Malaysia which was reported as 5.4% (33). However it was lower than the which was reported at 57%(34) and the results of this meta-analysis (current) was comparable with reports of a meta-analysis done on Prevalence of needle stick and sharps injuries in healthcare workers of Iranian hospitals which was 42.5% (95% CI 37-48) (35).
The subgroup analysis of this study revealed that the magnitude of work related injury was signi cantly varied across regions. The highest prevalence of occupational injury was observed in Addis Ababa with a prevalence of 45.82% (95% CI: 21.39, 70.25) followed by Amhara region, 40.89% (95% CI: 32.44, 49.34) and lowest was from SNNPR region 31.40% (95% CI: 27.03, 35.77). The possible justi cations for this variation could be due to socio-cultural differences across the regions. The other possible reason for the higher prevalence of occupational injury in Addis Ababa could be in the fact that Addis Ababa (the capital) is the center of both politics and economy as a result there is high expansion of textile, construction and solid waste management activities. In spite of the emphasis given to expansion of these sectors, occupational and safety practice among workers and concerned bodies emphasis give to occupational safety may be minimal. With regard to type of occupation, the magnitude of work related injury was slightly higher in those studies done among MSWM workers 45.68% (95%CI: 34.00, 57.36).This difference may be due to MSWM sector is more risky than the rest in that the occupation is linked with wastes and it worsen by poor handling of wastes at home level.
The predictors of work related injury was identi ed based on the reports of included empirical primary studies. Hence in this study, drinking alcohol, and smoking cigarette and khat chewing habit and having work related stress were found to be predictors of work related injury. In addition, in this study chance of occupational injury was signi cantly lower among those workers who were using PPE at work places compared to their counterparts. This nding is consistent with studies conducted in sub-Saharan Africa which was reporting not using PPE (OR 1.91, 95% CI 1.10-3.32) exposed nurses to injury (34). This could be explained by the fact that using protective devices highly reduces chance of injury occurrence.
The magnitude of occupational injury was highly associated with workers level of education. Those workers who complete high school education had lower risk of injury at workplaces. This nding is in line with different studies carried out so far. The possible reason for the association may be, as anyone becomes more educated he/she is become aware of occupational health and safety issues. Furthermore, in this meta-analysis it was observed that those workers who had drinking alcohol, smoking cigarette and chewing khat habit and having work related stress had higher work related injury as compared to their counterparts. This nding is supported by previous studies conducted so far (32,34,35). The possible explanation could be because of a worker's substance use habit and work related stress reduces his/her concentration at work and use of PPE habit during work.

Limitations Of The Study
Majority of the studies included in this review were cross-sectional in nature as a result; the outcome variable might be affected by other confounding variables. Therefore, this factor could affect the estimated report. In addition this meta-analysis was considering only English articles or reports. Lastly, this meta-analysis represented only studies reported from four larger regions of Ethiopia. Even though these regions covers majority of the country geographically, population and economic activity, still the rest regions may be under-represented.

Conclusions And Recommendations
The study found that magnitude of work related injury among workers of textile, construction and MSWM was quite common. Drinking alcohol, smoking cigarette chewing khat practice, having work related stress, level education and utilization of PPE were found to be predictors of work related injury. Hence, it is strongly recommend that concerned stakeholders should launch health education programs about the risk of substance use on occupational injury and apply strict occupational safety practices regulations at such rapidly growing sectors. Availability data and materials:

Abbreviations
Data will be available from the corresponding author upon reasonable request.
Competing interests: The authors declare that they have no competing interests.

Funding:
No funding was obtained for this study.
Authors' contribution: The rst author WG involved in Conception of research protocol and developed the protocol, in addition he involved in study design, selection of study, literature review, data extraction, data analysis, interpretation and drafting the manuscript. MA involved in selection of study, data analysis, reviewing and editing the manuscript, data extraction and quality assessment. DS, MAA, and MM were involved in data analysis, reviewing and editing the manuscript. All authors have read and approved nal draft of the manuscript.  Forest plot of the pooled prevalence of occupational injury and illness among workers of MSWM, construction and textile industry in Ethiopia before trim and ll analysis, 2019.