Study Selection
A total of 1,114 articles were searched from the included electronic databases (Web of Science, SCOPUS, PubMed, Google Scholar, CINAHL, African Index Medicus, African Journals Online databases and Science direct) from 10 September 2020 to 12 October 2020. The search was carried out by the authors (DAM, YMD and AA) from the included electronic databases independently. Then, 285 duplicate articles were removed using the ENDNOTE software version X5 (Thomson Reuters, USA). A total of 706 articles were excluded after the initial screening based on title and abstract. Thirty-five articles were excluded after full-text articles were evaluated for eligibility, of which 20 articles were included in the systematic review and meta-analysis (Figure 1).
Characteristics of Included Articles
In this study, a total of 9,410 participants were included in 20 articles conducted in Ethiopia and published from 2017 to 2020 (15, 19, 20, 28-44); nine (45.0%) articles (15, 19, 29, 31, 36, 38, 40, 43, 44) were conducted in Oromia, three (15%) in Tigray (20, 30, 41), three (15%) in SNNP (28, 33, 34), three (15%) articles in Addis Ababa (32, 37, 42) and two (10%) articles in Amhara region state (19, 35). The included studies were cross-sectional studies with a sample size ranging from 264 (41) to 771 (37) study participants.
Among the included articles, 10 (50%) (15, 19, 20, 29, 32, 33, 36, 37, 39) articles reported the prevalence of low back pain alone, 9(45%) (28, 30, 31, 34, 38, 40-42, 44) articles reported both low back pain and upper back pains and 1(5%) (35) article reported the prevalence of upper back pain alone.
Furthermore, 8 (40%) articles published in 2020 (28, 30, 31, 33, 40-43), followed by studies published in 2019 (15, 19, 20, 35, 38, 40) that represented 6(30%) of the included articles. Based on JBI Critical Assessment tool (46), all included articles had a low risk of bias. Occupational-related prevalence of low and upper back pain in the previous year ranged from 25.5% (38) to 74.8% (20) and 10.4% (28) to 60.4% (34), respectively.
Sex of the study participants specified in 18 (90%) articles (15, 19, 20, 28-31, 33-37, 39-44); 4906 (57.2%) males and 3672 (42.8%) females. Nordic Musculoskeletal Questionnaire was used for data collection in 18 (90%) studies (15, 19, 20, 28-37, 39-43) (Table 1).
Prevalence of Occupational-Related Upper and Low Back Pain
Meta-analysis was performed using Comprehensive Meta-Analysis (CMA) Version 3 statistical package (software) to determine pooled prevalence of occupational related low and upper back pain in Ethiopia.
Prevalence of Occupational-Related Upper Back Pain
The pooled prevalence of occupational-related upper back pain in the previous year was 27.1% (95% CI: 18.4, 37.9%) (Figure 2). After a subgroup analysis was performed based on occupation, the pooled prevalence of upper back pain in the previous year was 34.7% (95% CI: 33.1, 36.2%). The lowest prevalence [10.4% (95% CI 7.6, 14.1%)] was reported among vehicle repair workers, while the highest prevalence [60.4% (95% CI 55.7, 65.0%)] was reported among pedestrian back-loading women (Supplementary file I; Figure 1).
After the subgroup analysis was performed based on the publication year, the pooled prevalence of occupational-related upper back pain in the previous year was 43.8 % [95% CI 39.9, 47.7%)]. The lowest prevalence [15.3% (95% CI: 11.7, 19.8%)] was reported in the study published in 2018, while the highest prevalence [60.4% (95% CI: 55.7, 65.0%)] observed among the study published in 2017 (Supplementary file I; Figure 2).
Furthermore, based on the study region, the pooled prevalence of upper back pain was 36.2 % (95% CI: 33.6, 39.0%). The lowest prevalence [22.1% (95% CI: 9.2, 44.5%)] was reported among the studies conducted in the Oromia regional state, while the highest prevalence [38.8% (95% CI: 34.2, 43.6%)] was reported by the study conducted in Amhara regional state (Supplementary file I; Figure 3).
Prevalence of Occupational-Related Low Back Pain
The pooled prevalence of occupational-related low back pain in the previous year was 54.2% (95% CI: 48.2, 60.0) (Figure 3). Based on the subgroup analysis by occupation, the pooled prevalence of low back pain in the previous year was 52.8% (95% CI: 51.3, 54.3%). The lowest prevalence [25.5% (95% CI: 21.5, 29.9%)] was reported among construction workers, while the highest prevalence [67.3% (95% CI: 62.7, 71.6%) was reported among pedestrian back-loading women (Supplementary File I; Figure 4).
After subgroup analysis was performed based on the publication year, the pooled prevalence of low back pain was 61.8 % (95% CI: 58.9, 64.6%). The lowest pooled prevalence [46.9% (95% CI: 39.9, 54.0%)] was reported among studies published in 2020, while the highest prevalence [65.7%, (95% CI: 62.5, 68.9%)] was reported among the studies published in 2017 (Supplementary file I; Figure 5).
Furthermore, after the subgroup analysis was performed by study region, the pooled prevalence of low back pain was 55.2 % (95% CI: 51.4, 59.0%). The lowest pooled prevalence [50.7% (95% CI: 25.0, 76.0%)] was reported among the studies conducted in Tigray regional state, while the highest prevalence [56.3% (95% CI: 37.1, 73.9%)] was reported among the studies conducted in Southern Nations, Nationalities, and Peoples (Supplementary File I; Figure 6).
The result of sensitivity analysis indicated that there is no significant difference between the pooled prevalence, before and after sensitivity analysis. There is no significant difference between the overall pooled prevalence of upper and low back pain before and after sensitivity analyses (Table 2).
Subgroup Analysis based on the Outcome
After subgroup analysis was performed based on the outcomes, the pooled prevalence of occupational-related upper and low back pain in the previous year was 54.2% [95% CI: 48.2 to 60.0%] and 31.0% [95% CI: 21.6 to 42.3%], respectively (Figure 4).
Publication bias
In this meta-analysis, publication biases were visualized using funnel plots. Publication bias was examined using Begg and Egger tests and was determined as p= 0.8065 and p= 0.8367, respectively for low back pain, and p= 0.9287 and p= 0.9994, respectively for upper back pain. These results indicated that the probability of publication bias was not statistically significant (Figure 5 and Figure 6).