Prevalence and associated factors of foot ulcer among diabetic patients in Ethiopia: a systematic review and meta-analysis

This systemic review and meta-analysis were conducted to assess the pooled prevalence and associated factors of diabetic foot ulcers among diabetic patients in Ethiopia. We checked the presence of systematic reviews and meta-analysis on this topic to prevent duplication. Both published and unpublished studies conducted since 2000 were searched thoroughly using electronic databases. These included Medline, Hinari, PubMed, Cochrane Library, Web of Science and Google Scholar. To find unpublished papers, some research centers, including the Addis Ababa Digital Library were used.

Pre-defined search terms were utilized to enable a comprehensive search strategy that included all the relevant studies. All fields within records and Medical Subject Headings (MeSH terms) were used to expand the search in advanced Pub Med search. The search strategy was prepared and modified for the various databases using important Boolean operators with initial keywords (“diabetes mellitus” OR “diabetic foot ulcer” OR “complication of diabetics” AND “associated factors” AND “Ethiopia”). The meta-analysis was reported using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [20]. All searched literature was downloaded to Endnote (version X7.2,) to maintain and manage citations, and facilitate the review process.

Studies that assessed the prevalence and determined potentially associated factors of diabetic foot ulcers that were written in the English language were included. The design of these studies was all observational study conducted in Ethiopia. Studies where there were difficulties in extracting necessary information and those studies published before 2000 were excluded because findings might be distorted due to changes in trend.

There were two main outcomes. The primary outcome of interest was the prevalence of diabetic foot ulcers, which was estimated as the total number of diabetic foot ulcer cases divided by the total number of diabetic patients multiplied by 100. The second outcome was identifying factors associated with a diabetic foot ulcer in diabetic patients, which were determined using the odds ratio (OR) and calculated based on binary outcomes from the included primary studies. The major factors included in this review were age (< 45 years versus 45 years), sex (male versus female), types of diabetes (type I versus type II), residence (urban versus rural), self-care practice (poor versus good), duration of diabetes (< 10 years versus), BMI (< 24.5Kg/m2 versus 24.5 Kg/m2) and callus on the feet (absent versus present).

Reference management software (endnote version X7.2) used to combine search results from databases and to remove duplicate articles. The Joanna Briggs institute meta-analysis of statistics assessment and review instrument (JBI-MAStARI) was used for critical appraisal [21]. Data were extracted by two data extractors (TT and DM) using a standardized data extraction checklist on Microsoft excel. For the first outcome (prevalence), the data extraction checklist included author name, year of publication, region (the area where studies were conducted), study design, sample size and number of participants with the outcome. For the second outcome (associated factors), data were extracted in a format of two by two tables, and then the log OR for each factor was calculated based on the findings of the original studies. Discrepancies between two independent reviewers were resolved by involving a third reviewer (BM) after discussion for possible consensus. AA has overseen the overall process of data extraction and synthesis

STATA version 14 statistical software was used to analyze the extracted data. The logarithm and standard error of the OR for each included study were generated using the “generate” command in STATA. Cochran’s Q test (reported as the p-value) and inverse variance index (I2) were used to check heterogeneity in the included studies. I2 values of 0, 25, 50, and 75% were considered as no, low, moderate, and high degrees of heterogeneity, respectively [22]. The high degree of heterogeneity was observed for the first outcome and thus a random-effects model was used to estimate the pooled prevalence. No heterogeneity was observed for six factors; hence, a fixed-effects model was computed. For the remaining three factors (sex, callus, and neuropathy) with moderate to a high degree of heterogeneity, the random-effects model was used to estimate the Der Simonian and Laird’s pooled effect. In addition, a meta-regression was conducted to identify the source of heterogeneity and there were no statistically significant results found to declare the presence of heterogeneity. A funnel plot of asymmetry was used to check the presence of publication bias. Furthermore, Egger’s statistical test was used to check the statistical significance of publication bias [23]. Subgroup analyses by region were carried out. Prevalence with 95% confidence interval (CI) and OR of the association between diabetic foot ulcer and factors in the form of forest plot was presented.

We identified 96 published and unpublished articles from different databases. Major reasons for excluded articles were duplication (14 articles) and mismatch with study objectives (64 articles). Eighteen articles were screened against eligibility criteria and only 11 studies scored 7 and above on the JBI quality appraisal criteria and included in the systematic review and meta-analysis (Fig. 1).

Table 1 describes all cross-sectional studies included in the analyses. The total sample was 2768 diabetic patients, ranging from 108 to 418 [11, 12, 15‐19, 24‐27]. Studies from 4 regions and 1 town administrative of the country were included. Oromia [12, 16, 17], Amhara [19, 25, 26], and AA [11, 15, 27] contributing three studies in each region and 1 from southern nationalities [18] and 1 from Tigray [24] region (Table 1).

The pooled prevalence of diabetic foot ulcer was 12.98% (95%CI: 7.81–18.15). High heterogeneity was observed across the included studies (I² = 96.7, p < 0.001). Both the highest (31.10% (95%CI: 26.66–35.54)) [11] and lowest (1.36%) [27] prevalence of diabetic foot ulcer was reported in Addis Ababa (Fig. 2). Sample size and year of publication were investigated to assess whether these were associated with heterogeneity using meta-regression models, although only region was statistically significant for underlying heterogeneity. Sample size (p = 0.374) and year of publication (p = 0.800) were insignificantly associated with heterogeneity.

The prevalence of diabetic foot ulcers was computed based on the regions where studies were conducted (Fig. 3). Accordingly, the prevalence ranged from 7.15% (95%CI: 3.73, 10.56) in Oromia region [12, 16, 17] to 19.31% (95%CI: 2.7. 41.37) in Addis Ababa [11, 15, 27]. The funnel plot was asymmetry, and the Egger’s test also showed statistically significant publication bias (p = 0.001) at 5% significant level (Table 2).

To identify a single study influence on the overall meta-analysis, sensitivity analysis was performed using a random-effects model and the result showed that there was no strong evidence for the effect of a single study on the overall meta-analysis result. The table showed that the estimates from a single study is closer to the combined estimate which implies the absence of a single study effect on an overall study (Table 3).

We tried to use comprehensive search strategies for this systematic review and meta-analysis by including both published and unpublished studies. A random-effects model was used to address the potential variability across studies. However, the restriction of studies written in English limited the number of studies included in a meta-analysis. In the current meta-analysis, almost all included studies were conducted at the large towns of the country, which could not be representative of all districts of the country. In addition, the small sample due to a limited number of included studies with a small sample size for all included studies was another limitation of this study. In addition, all of the studies included in this review were cross-sectional study design; as a result, the outcome variable might be affected by other confounding variables, decreases the power of the study and it decreases causal conclusion between foot ulcer and factors associated with a foot ulcer.