Study screening and selection processes
A total of 3861 studies have been identified through the searching strategy from different databases, of which four of them were from University repository and unpublished articles. One hundred fifty studies remained after removing the duplications. Finally, nineteen studies left eligible and included in the study (Figure 1).
Characteristics of included primary studies
A total of nineteen studies yielding a total study population of 6595 was eligible and included for both qualitative and quantitative analyses. Amhara, Oromia, and Hareri regions contributed 26.3%(24-28), 21% (6, 11, 29, 30), and 157% (2, 7, 31) to the total of included studies, respectively. Three regions: Southern Nations Nationalities and Peoples (SNNPs) (10, 32), Tigray (33, 34), and Addis Abeba (8, 35) each contributed two articles to the review. While the remaining study was from Benishangul Gumuz region(36). Significant variation was observed with the sample size of the included studies; the highest sample size was 660 from a study conducted in Addis Abeba (8) and the least was 191 from a study done in SNNPs(32). Referring to the response rate, the lowest and the highest response rates were 95% and 100% consecutively. Furthermore, the prevalence of diabetics self-care practice falls between 28.4% and 76.8% from a study conducted in Amhara (27) and SNNPs(10) regions, respectively. All the studies incorporated were simple and analytical cross-section by design (Table 1).
Table1. General characteristics of included studies that report the prevalence of diabetic’s self-care practice, 2019.
Authors Name
|
study year
|
Region
|
Sample size
|
Response rate
|
Prevalence
|
JBI Score
|
Addisu Y et.al(10)
|
2013
|
SNNPE
|
310
|
100
|
76.8
|
75%
|
Dedefo et al(29)
|
2016
|
Oromia
|
252
|
100
|
60.7
|
75%
|
Feyissa et.al(35)
|
2014
|
AddisAbeba
|
328
|
98.8
|
51.5
|
62%
|
Fikadu A et.al(24)
|
2017
|
Amhara
|
344
|
100
|
50
|
75%
|
Getie et.al(7)
|
2018
|
DireDawa
|
513
|
98.6
|
55.9
|
100
|
Gurmu Y et al(11)
|
2017
|
Oromia
|
257
|
100
|
54.5
|
75%
|
Haftom et.al(33)
|
2016
|
Tigray
|
338
|
100
|
74.5
|
75%
|
Kassahun T et.al(6)
|
2014
|
Oromia
|
309
|
95
|
50.9
|
62%
|
Ketema et.al(31)
|
2011
|
Harer
|
222
|
100
|
39.2
|
50%
|
Melat et.al(8)
|
2011
|
AddisAbeba
|
660
|
97.8
|
60.3
|
62%
|
Mende et.al(32)
|
2015
|
SNNPE
|
194
|
100
|
87.1
|
75%
|
Segni et.al(36)
|
2018
|
B.Gumuz
|
383
|
96
|
54.3
|
62%
|
Sofonyas A et.al(25)
|
2018
|
Amhara
|
385
|
95
|
63.1
|
62%
|
Solomon A et al(26)
|
2012
|
Amhara
|
410
|
97.2
|
36.8
|
62%
|
Tadele et.al(30)
|
2013
|
Oromia
|
260
|
98
|
55
|
62%
|
Teklewoini et.al(34)
|
2017
|
Tigray
|
284
|
100
|
37.3
|
62%
|
Teshager etal(27)
|
2017
|
Amhara
|
416
|
99.5
|
28.4
|
75%
|
Behailu H et.al(2)
|
2017
|
Harer
|
327
|
97.8
|
38.1
|
75%
|
Andualem et.al
|
2019
|
Amhara
|
403
|
100
|
51.86
|
75%
|
In the random-effects model, diabetic self-care practice was 54.04% (47.07-61%, I2 =97.3%, P<0.001) (Figure 2.).
Sensitivity analysis
There was no influential study that caused variation between studies, according to the sensitivity analysis.
The subgroup analysis by the regions where studies were conducted suggested that notable difference among regions has been observed; the lowest (45.33%) and highest (81.93%) prevalence was recorded in Amhara and SNNPs regions, respectively. For the sake of analysis, one study from Benishangul Gumuz region was included into Amhara region based on physical proximity as well as the proportion of study population in the area (Figure3).
The absence of influential study has revealed through sensitivity analysis, and the presence of publication bias has ruled out by the funnel plot (Figure 4.) and Egger’s regression test (P value=0.568).
Determinants of diabetics’ self-care practice
Five published articles were included to explore and determine the association between diabetic’s knowledge and self-care practice. Of the included studies, studies from Diredawa (7), Oromia (11), Addis Ababa (8), and Amhara (25) regions have shown a significant association; patients who had good knowledge about diabetics were good at applying self-care practice. However, a study conducted in the Tigray region did not show an association (34).
The meta-analysis result revealed that having diabetic knowledge was about 2.7 times more likely to perform self-care than their counterparts 2.69(1.62, 4.46), I2 =99%, P<0.001) (Figure 4).
No influential study according to the sensitivity analysis. Publication bias was not seen as to the egger’s test demonstrated (p= 0.892).
The effect of social support on diabetic self-care practice has been explored among six studies (7, 8, 11, 25, 29, 36). Of these, five of the studies revealed that good social support promotes self-care practice. The odds of performing self-care among diabetic victims having good social support were increased two folds 2.25((1.49-3.39); I2=99%, p=0.00) as compared to those who had no good social support (Figure 6).
There was no influential study that resulted in variation among studies and publication bias was not seen (p = 0.752).
Further to the aforementioned factors, the relationship between having glucometer at home and diabetic’s self-care practice has been explored, and six studies (6, 7, 25, 29, 34, 36) were involved in to the analysis; all the included studies were significantly associated with self-care practice. in other words, DM patients who have glucometer at home were three times more likely to practice diabetics self-care compared to patients who didn’t have access to glucometer at home 3.04(1.64, 5.65; I2=97.4, P<=0.001)(Figure 7). The sensitivity analysis depicted no study that accounted for the variation. The funnel plot illustrates symmetric distribution of studies and Egger’s regression supports the finding (P = 0.631).
Four primary studies (29, 34-36) were screened and included to examine the relationship between co-morbidities and diabetics self-care practice. A significant association was observed among five studies (7, 25, 29, 34, 36), However, a study showed a protective association (6). As a result, the pooled effect size of the presence of co-morbidity showed an absence of significant association with diabetics self-care practice 1.24(0.8, 1.91); I2=97.8, p<=0.001) (Figure 8). No influential study and publication bias (p = 0.904) has been detected.
Furthermore, the relationship between diabetic’s self-care practice and residence of DM patients was reviewed by five articles. All studies (25, 28, 29, 33, 36) demonstrate that being an urban residence promoted diabetic self-care practice. The effect size of the residence revealed that living in an urban area has a threefold chance of practicing diabetics self-care than DM patients residing in the rural area 3.26(2.24, 4.74), I2=96.3%, P<0.001) (Figure9). The analysis result indicated that there is no small study that affects the pooled estimate and the absence of publication bias (P = 0.296).