General characteristics of included studies
A total of 74 studies reporting on 21,133 patients with type 2 diabetes were included in the review. The studies were conducted in 16 sub-Saharan African countries, with Ethiopia (n=24, 36.1%) being the most represented, followed by South Africa (n=11, 14.9%) and Nigeria (n=10, 13.5%). The majority of the studies (n=51, 68.9%) were conducted in the last five years (2017- 2022). Of the 74 studies selected, 55 (74.3%) were cross-sectional studies, ten (13.5%) randomized controlled studies, four (5.4%) quasi-experimental studies, three (4.1%) case-control studies, and two (2.7%) cohort studies. The general characteristics of the included studies are presented in Table 1.
Assessment of risk of bias
Of the 74 studies selected for the review, only 14 (18.9%) were assessed as being of good quality, 54 (73.0%) of moderate quality, and six (8.1%) of poor quality. Supplementary Tables 4–8 detail the assessment of study methodological quality. Of the 55 cross-sectional studies, only four (7.3%) were able to formally identify confounding factors while ten (18.2%) reported the method used to address confounding factors. In the four quasi-experimental studies, one study (25.0%), did not measure the outcomes consistently nor in a reproducible way. In two of the ten randomized controlled trials (20.0%), participants and treatment providers were not blinded to treatment allocation nor were the staff members assessing outcomes blinded to treatment allocation. Moreover, for one of these two studies, the treatment groups were not similar at baseline. For two of the three case-control studies, the confounding factors were not identified; and for one study, cases and controls were mismatched.
Assessment of glycaemic control
Glycaemic control was assessed by glycosylated haemoglobin in 43(58.1%) studies, fasting blood glucose in 25 (33.8%) studies and a combination of both methods in 6(8.1%) studies. The cut-off points for glycaemic control varied across studies.
Prevalence of glycaemic control
The estimated pooled prevalence of good glycaemic control in sub-Saharan Africa was 30% (95%CI: 28-33). The analysis showed considerable heterogeneity (I2: 93.9%, p=0.000), and glycaemic control prevalence ranged from 10 % to 60 % (Fig. 2). The subgroup analysis by region showed that most of the studies in the Central (n=5, 83.3%) and the Southern (n=5, 62.5%) regions had a prevalence of glycaemic control of <30% while most of the studies in the Eastern region had a prevalence of glycaemic control >30% (Fig.3).
Factors associated with glycaemic control
The reported sociodemographic, lifestyle, clinical, adherence, treatment factors, and reported glycaemic control optimization interventions factors are summarized in Tables 2-7.
Sociodemographic characteristics
Table 2 presents the sociodemographic factors with respect to their relationship with glycaemic control. Five studies assessed the relationship between increasing age and glycaemic control [27,31,34,58,61], two found that it was negatively associated with glycosylated haemoglobin [31,61], and one that it was associated with good glycaemic control [58]. Older age was associated with poor glycaemic control in twelve studies [22,29,32,36,39,65,68,69,73, 77,83,86]. Eight studies assessed the relationship between the female gender and glycaemic control [18,29,34,51,61,64,65,73], two studies found that the female gender was associated significantly with poor glycaemic control [18,34] versus one study that linked it to good glycaemic control [29]. Male gender in respect of glycaemic control was assessed by eleven studies [27,31,39,44,58,66,75,77,83,85,87], two studies associated it with good glycaemic control [58,75], while two studies linked it to poor glycaemic control [27,87]. Fifteen studies assessed the relationship between educational level and glycaemic control, in one study, primary, secondary, or tertiary education levels were associated with good glycaemic control [29]. The lack of formal education and low level of education were associated with poor glycaemic control in three studies [39,48,87]. In respectively two studies, low monthly income [18,87], absence of health insurance [47,58], and being a farmer [25,48] were associated with poor glycaemic control. In respectively one study, living in urban areas [49], and a high frequency of seeking traditional medicine practitioners [31] were associated with poor glycaemic control. Residing less than 100 kilometres from a health facility [25], residing in Guinea compared to residing in Cameroon [32], self-report a positive perception of family support [68], and the frequency of participating in religious activities [31] were associated with good glycaemic control in respectively one study.
Lifestyle factors
The lifestyle factors assessed were dietary adherence, the practice of exercise, smoking, and alcohol consumption (Table 3). Good dietary adherence was associated with good glycaemic control in five studies [29,36,40,61,86] while low adherence to dietary recommendations was associated with poor glycaemic control in two studies [35,67]. The practice of exercise was associated with good glycaemic control in two studies [29,36]. The inadequate practice of exercise was associated with poor glycaemic control in two studies [39,53]. In respectively one study, smoking [40], and alcohol consumption [29] were associated with poor glycaemic control.
Clinical factors
The clinical factors—history of diabetes disease and comorbidities—with respect to glycaemic control are summarized in Table 4. A family history of diabetes was significantly associated with poor glycaemic control in one study [80]. The long duration of diabetes is associated with poor glycaemic control in seven studies [18,26,27,32,58,61]. As a corollary, treatment of > 10 years was associated with poor glycaemic in one study [39]. In one study, patients who always had fluctuating/unstable blood glucose levels or had blood glucose levels not improved from diagnosis were prone to poor glycaemic control [61].
Four studies found that the presence of comorbidities was associated with poor glycaemic control [53,75,77,87]. The presence of hypertension lead to poor glycaemic control in one study [16]. Dyslipidaemia was associated with poor glycaemic control in three studies [18,53,82]. Concerning the body mass index (BMI), all the categories as being underweight [61], having a normal BMI [47], or being overweight/obese [18,34,87] have been associated significantly with poor glycaemic control. Central obesity was associated with poor glycaemic control in four studies [16,30,52,56]. In respectively one study, the presence of anaemia [17], non-alcoholic fatty liver disease [19], vitamin B12 deficiency [20], metabolic syndrome [28], cognitive impairment [32], congestive cardiac failure [46], HIV infection [46], thyroid autoimmunity [74], hypogonadism [45] had a significant association with poor glycaemic control. The presence of peripheral neuropathy [83] or high-level tooth mobility index [59] was associated with poor glycaemic control. Overall health-related quality of life was inversely associated and FBG [42]. The global disability burden was significantly associated with poor glycaemic control [70]. A unit reduction in estimated Glomerular Filtration Rate (eGFR) was also associated with HbA1c ≥ 7% [16].
Adherence to treatment plans
Adherence modalities as represented by adherence to scheduled appointments or medication adherence are presented in Table 5. Regular attendance at scheduled appointments has been associated with good glycaemic control in two studies [49,85]. Good medication adherence has been associated with good glycaemic control in two studies [40,77] while two other studies showed no association [75,78]. Low medication adherence had a significant association with poor glycaemic control in three studies [33,48,86]. Medium medication adherence was associated with poor glycaemic control in one study [48].
Treatment modalities
The findings on the treatment modalities with respect to glycaemic control are summarized in Table 6. The pill burden was associated with poor glycaemic control in one study. The combination therapy of oral hypoglycaemic agents (OHA) was associated with poor glycaemic control in two studies [48,53] while it was linked to good glycaemic control in one study [75]. Insulin plus OHA was associated with poor glycaemic control in three studies [44,48,69], while it was linked to good glycaemic control in one study [75]. The use of insulin alone was associated with poor glycaemic control in two studies [53,78]. The presence of drug-related problems was associated with poor glycaemic control as shown in one study [86]. Rwegerera et al found that being on diet and OHA was associated with suboptimal glycaemic control [73]. A South African study found that the use of statin and anti-hypertensives was associated with higher glycaemic levels [50]. Non-surgical periodontal management was associated with good glycaemic control after three months in one study [82]. Diabetes information from non-health workers [18], and high diabetes health literacy [77] were significantly associated with poor glycaemic control in respectively one study [18]. In one study, the absence of clarity in pharmacists’ advice was associated with poor glycaemic control [44].
Reported glycaemic control optimization interventions
The interventions retrieved from included studies are presented along with their effect on glycaemic control in Table 7. Only one study [43] out of five reported an educational program associated with good glycaemic control. None of the self-management programs was associated with glycaemic control as found in three studies [40,61,64]. All the exercise programs were associated with improved glycaemic control as found in four studies [37,38,52,76,84]. Adding a second OHA was associated with poor glycaemic control in one study [49]. The effectiveness of a community-based multilevel peer support intervention was associated with a significant reduction of glycosylated haemoglobin in the intervention group in one study [24].