Level of self-care practice among diabetic patients in Ethiopia: a systematic review and meta-analysis

Published and unpublished research articles that were conducted to assess a minimum of one variety of diabetic self-care (overall diabetic self-care, diet, physical activity, self-monitoring of glucose, foot care) in Ethiopia were included during this study. An intensive search was done from PubMed/MEDLINE, and EMBASE online databases to access articles done on the diabetes self-care practices. Moreover, Google Scholar and Science direct were accustomed to retrieve articles. Besides, reference lists of screened studies were checked. The search was administered by two authors (DBK and AA) independently. The term ‘diabetes’ was searched with all of the subsequent terms as a mix of free text and thesaurus terms in numerous variations: diabetes self-care, dietary care, foot care, physical activity, glycemic control, and Ethiopia. Moreover, the subsequent keywords were accustomed to retrieve studies from PubMed database; (((((Diabetes self-care) OR diabetes dietary care) OR diabetes foot care) OR diabetes physical activity) OR self-monitoring of blood glucose) AND Ethiopia. Studies that were relevant by title and abstract were assessed by full text to see those that provided adequate data to be included in our meta-analysis. This systematic review and meta-analysis was reported per preferred reporting items for Systematic Review and Meta-Analysis (PRISMA) guideline [36].

Diabetes self-care was defined as good self-care practice and poor self-care practice [24, 25]. In this meta-analysis, we measure diabetes self-care practice using four domains: dietary practices, engagement in regular exercise, foot-care and blood glucose monitoring [3, 6].

Patients who scored above the mean value for the response were classified as having good dietary practice and poor dietary practice otherwise.

The level of glycemic control was indicated as ‘good glycemic control’ when FBS results were between 70 and 130 mg/dL (i.e. an average of three measures at different visits).

20–30 min of aerobic exercise such as walking or swimming 3–4 days per week [37].

a total practice score of ≥50% of maximum score is categorized as good foot care practice, while a total practice less than a score of 50% categorized as poor foot care practice [38].

Seven authors (DBK, AAA, GDK, AA, AAA, and CTL) done data abstraction using pre-piloted data extraction format prepared in Microsoft™ Excel spreadsheet. The seventh author (GMK) reconciled any disagreements among the six authors. From each study, we have extracted data on the study location, region, publication year, study design, sample size, and first author name for the overall diabetes self-care and for each domains of diabetes self-care practice (Supplementary File 1, Supplementary File 2, Supplementary File 3, Supplementary File 4, and Supplementary File 5).

The quality of included studies has been assessed in accordance with the Newcastle-Ottawa Quality assessment scale [39]. Two independent reviewers (DBK & GDK) done a full-text quality assessment. Disagreement between the two reviewers was found to be very low (1.5%) and they resolved this by discussion. Then, a combined quality score ranging from 0 to 10 was assigned to each included articles.

The presence of statistical heterogeneity within the included articles was assessed using I-squared statistic. Accordingly, heterogeneity was classified as low, moderate, or high when the value of I-squared were 25, 50, and 75%, respectively [40]. Begg and Egger weighted regression method were employed to detect evidence of publication bias. P-value < 0.05 was considered as presence of significant publication bias [41].

Relevant data from every primary study was extracted using a Microsoft Excel format. Data were then exported to STATA™ Version fourteen software package for analysis. The pooled estimate was computed using metan SATA command. Results were presented using tables and forest plots with 95% confidence intervals (CI). We have used Higgins and Egger’s test to identify publication bias. As a result of high level of heterogeneity among included studies, random-effects model was used to produce Der Simonian and Laird’s pooled estimate. Additional advanced statistical analyses like univariate meta-regression and sensitivity analysis were performed to identify the potential sources of heterogeneity and to assess the influence of a single study on the overall pooled estimate respectively.

A total of 969 articles were retrieved from electronic databases and other sources. Title and abstracts were screened and duplicated or irrelevant articles were removed using EndNote × 7. Accordingly, 552 duplicate articles were removed. From the remaining 417articles, 382 articles were excluded because their titles and abstracts were not in-line with our inclusion criteria (not report the outcome of the interest, studies conducted outside of Ethiopia). Finally, a total of 35 articles were included for this systematic review and meta-analysis. The detailed selection procedures were described in Fig. 1.

We included 35 cross-sectional studies (with 11,103 participants) in the final systematic review and meta-analysis. The smallest sample size was 102 obtained from a study conducted at Ambo Hospital Oromia region of Ethiopia [35]. Whereas, the largest sample size was 419 reported from a study at Tikur Anbessa Specialized Hospital, Addis Ababa, Ethiopia [42]. The detail characteristics of included studies was presented in Table 1.

In the current meta-analysis, seven regions and one administrative town in the country were represented. Most studies took place in Oromia Region (n = 13) [2, 17, 22, 25, 28, 29, 35, 37, 48, 49, 51, 52] and Amhara Region (n = 9) [4, 16, 20, 27, 34, 38, 44, 46, 53]; followed by Tigray Region (n = 4) [23, 31, 32, 47], South Nation Nationality of People (SNNP), n = 3)) [21, 26, 33]; Addis Ababa (n = 4) [30, 42, 45, 54]; Benishangul Gumuz region [24], and Harari region [1] (n = 1 each). Most studies dealt with self-monitoring of blood glucose (SMBG) (n = 21), followed by overall diabetes self-care (n = 16), assessment of dietary practice (n = 15), and engagement in physical activity (n = 14). Only eight studies assessed diabetic foot-care. Quality assessment using the Newcastle-Ottawa Scale for cross-sectional studies showed a quality score ranging from 4 to 9 (Table 1).

In this meta-analysis, we measure diabetes self-care practice using four domains: dietary practices, engagement in regular exercise, foot-care and blood glucose monitoring. Accordingly, fourteen studies included information on dietary self-care behaviors and reported on patients’ behavior regarding dietary practice. In this meta-analysis, the pooled level estimate of good dietary practice of diabetic patients was 50% (95% CI:42, 58%) (Fig. 3). Concerning on blood glucose monitoring, twenty-one observational studies reported patients’ behavior regarding monitoring of blood glucose. As detailed in Fig. 4, the lowest percentage of diabetic patients that have good glycemic control practice were 28% (95% CI: 19, 37%). That is, 28% of diabetic patients self-monitored and kept their blood glucose levels at recommended level. In addition, fourteen observational studies measured physical activity practice of DM patients. The highest percentage of DM patients who had adequate regular exercise was reported at the University of Gondar Referral Hospital, Amhara region, Ethiopia [46]. As presented in the forest plot (Fig. 5), the pooled level of adequate physical exercise was 49% (95% CI: 38, 59%), which means 49% of diabetic patients had practiced physical activity that met the recommended guidelines. Moreover, eight observational studies reported on the diabetic patients practice on the foot-care. The pooled meta-analysis showed that 58% (95% CI: 41, 74%) of diabetic patients practiced good diabetic foot-care practice (Fig. 6).

This systematic review and meta-analysis have certain limitations. First, methods applied to measure outcome-parameters varied among included studies. Second, included studies used self-report as a measure of outcomes, and this may have introduced bias. Multiple methods may be required to detect those who report good practice but who may in fact may have poor practice.